1 2017-11-15 Martin Liska <mliska@suse.cz>
4 * config/sh/sh-mem.cc: Use proper probability for
7 2017-11-14 Jeff Law <law@redhat.com>
9 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
10 the red zone for stack_clash_protection_final_dynamic_probe targets
11 when the total dynamic stack size is zero bytes.
13 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
16 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
18 * dumpfile.h (TDF_COMPARE_DEBUG): New.
19 * final.c (rest_of_clean_state): Set it for the
21 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
22 class when TDF_COMPARE_DEBUG is set.
24 * dwarf2out.c (gen_producer_string): Discard
27 2017-11-15 Joseph Myers <joseph@codesourcery.com>
30 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
31 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
32 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
33 (__TGMATH_REAL_2_3): Remove macros.
34 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
35 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
37 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
39 (remquo): Define using __TGMATH_REAL_3.
41 2017-11-14 Jeff Law <law@redhat.com>
43 * vr-values.c: New file with contents extracted from tree-vrp.c.
44 * Makefile.in (OBJS): Add vr-values.o
45 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
46 (set_value_range, set_and_canonicalize_value_range): Likewise.
47 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
48 (value_range_constant_singleton, symbolic_range_p): Likewise.
49 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
50 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
51 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
52 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
53 (find_case_label_range, find_case_label_index): Likewise.
54 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
55 (range_int_cst_singleton_p, value_inside_range): Likewise.
56 (get_single_symbol): Likewise.
57 (switch_update): Move structure definition here.
58 (to_remove_edges, to_update_switch_stmts): Provide externs.
59 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
60 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
61 (vrp_val_is_min, set_value_range): Likewise.
62 (set_and_canonicalize_value_range, copy_value_range): Likewise.
63 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
64 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
65 (range_is_nonnull, range_int_cst_p): Likewwise.
66 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
67 (get_single_symbol, operand_less_p): Likewise
68 (compare_values_warnv, compare_values): Likewise.
69 (value_inside_range, value_range_constant_singleton): Likewise.
70 (zero_nonzero_bitgs_from_vr): Likewise.
71 (extract_range_from_binary_expr_1): Likewise.
72 (overflow_comparison_p): Likewise.
73 (to_remove_edges, to_update_switch_stmts): Likewise.
74 (find_case_label-index, find_case_label_range): Likewise.
75 (switch_update, set_value_range_to_nonnegative): Remove.
76 (set_value_range_to_truthvalue): Likewise.
77 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
78 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
79 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
80 (find_case_label_ranges, test_for_singularity): Likewise.
81 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
82 (x_vr_values): Move to its remaining use site.
84 2017-11-10 Jeff Law <law@redhat.com>
86 * vr-values.h (VR_INITIALIZER): Move #define here.
87 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
88 * Makefile.in (OBJS): Add tree-evrp.o
89 * tree-vrp.h (assert_info): Move structure definition here.
90 (set_value_range_to_varying): Prototype.
91 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
92 (infer_value_range, register_edge_assert_for): Likewise.
93 (stmt_interesting_for_vrp): Likewise.
94 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
95 (set_value_range_to_varying): No longer static.
96 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
97 (infer_value_range, register_edge_assert_for): Likewise.
99 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
101 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
102 generate the XXBRD instruction.
104 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
105 (rs6000_builtin_type_compatible): Treat _Float128 and long double
106 as being compatible if -mabi=ieeelongdouble.
107 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
108 to setup float128 built-ins with hardware support.
109 (BU_FLOAT128_HW_2): Likewise.
110 (BU_FLOAT128_HW_3): Likewise.
111 (BU_FLOAT128_HW_VSX_1): Likewise.
112 (BU_FLOAT128_HW_VSX_2): Likewise.
113 (scalar_extract_expq): Change float128 built-in functions to
114 accommodate having both KFmode and TFmode functions. Use the
115 KFmode variant as the default.
116 (scalar_extract_sigq): Likewise.
117 (scalar_test_neg_qp): Likewise.
118 (scalar_insert_exp_q): Likewise.
119 (scalar_insert_exp_qp): Likewise.
120 (scalar_test_data_class_qp): Likewise.
121 (sqrtf128_round_to_odd): Delete processing the round to odd
122 built-in functions as special built-in functions, and define them
123 as float128 built-ins. Use the KFmode variant as the default.
124 (truncf128_round_to_odd): Likewise.
125 (addf128_round_to_odd): Likewise.
126 (subf128_round_to_odd): Likewise.
127 (mulf128_round_to_odd): Likewise.
128 (divf128_round_to_odd): Likewise.
129 (fmaf128_round_to_odd): Likewise.
130 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
131 support for KFmode and TFmode xststdcqp calls.
132 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
133 point, switch the built-in handlers for the get/set float128
134 exponent, get float128 mantissa, float128 test built-ins, and the
135 float128 round to odd built-in functions. Eliminate creating the
136 float128 round to odd built-in functions as special built-ins.
137 (rs6000_init_builtins): Eliminate special creation of the float128
138 round to odd built-in functions.
139 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
140 function insns to support both TFmode and KFmode variants.
141 (xsxsigqp_<mode>): Likewise.
142 (xsiexpqpf_<mode>): Likewise.
143 (xsiexpqp_<mode>): Likewise.
144 (xststdcqp_<mode>): Likewise.
145 (xststdcnegqp_<mode>): Likewise.
146 (xststdcqp_<mode>): Likewise.
148 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
150 * tree-ssa-threadupdate.c (compute_path_counts): Remove
151 unused path_in_freq_ptr parameter.
152 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
154 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
156 * ipa-inline.c (edge_badness): Dump sreal frequency.
157 (compute_inlined_call_time): Match natural implementaiton ...
158 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
159 forgotten division by CGRAPH_FREQ_BASE.
161 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
163 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
164 Solaris 11. Update comment.
165 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
167 * config/sol2.h (STARTFILE_SPEC): Likewise.
168 * configure: Regenerate.
170 2017-11-14 Carl Love <cel@us.ibm.com>
172 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
173 le_ and be_ prefixes to swap* variables. Remove
174 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
177 2017-11-14 Jason Merrill <jason@redhat.com>
179 Support GTY((cache)) on hash_map.
180 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
181 (ggc_cache_remove): Override it instead of ggc_mx.
182 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
183 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
184 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
185 (simple_cache_map_traits): Override maybe_mx.
186 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
187 (hash_map): Friend gt_cleare_cache.
188 (gt_cleare_cache): New.
189 * tree.h (tree_cache_traits): New hash_map traits class.
190 (tree_cache_map): New typedef.
192 2017-11-14 Richard Biener <rguenther@suse.de>
194 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
195 paramter and handling.
196 (cleanup_control_flow_bb): Likewise.
197 (cleanup_control_flow_pre): New helper performing a DFS walk
198 to call cleanup_control_flow_bb in PRE order.
199 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
200 via cleanup_control_flow_pre.
202 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
204 * config/aarch64/aarch64-simd.md
205 (aarch64_simd_bsl<mode>_internal): Remove DImode.
206 (*aarch64_simd_bsl<mode>_alt): Likewise.
207 (aarch64_simd_bsldi_internal): New.
208 (aarch64_simd_bsldi_alt): Likewise.
210 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
212 * tracer.c (better_p): Do not compare frequencies.
213 * reg-stack.c (better_edge): Likewise.
214 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
217 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
219 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
220 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
221 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
222 * lto-streamer-in.c (input_function): Use update_max_bb_count.
223 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
224 * predict.c (maybe_hot_frequency_p): Inline to ...
225 (maybe_hot_count_p): ... here; rewrite to counts.
226 (counts_to_freqs): Rename to ...
227 (update_max_bb_count): ... this one.
228 (expensive_function_p): Use counts.
229 (estimate_bb_frequencies): Update.
230 (rebuild_frequencies): Update.
231 * predict.h (counts_to_freqs): Rename to ...
232 (update_max_bb_count): ... this one.
233 * profile.c (compute_branch_probabilities): Add debug info
234 * tree-inline.c (expand_call_inline): Update debug info.
235 (optimize_inline_calls): Use update_max_bb_count..
236 (tree_function_versioning): Use update_max_bb_count..
237 * value-prof.c (gimple_value_profile_transformations):
238 Do not use update_max_bb_count.
240 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
242 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
243 always use frequencies.
245 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
247 * bb-reorder.c: Remove frequencies from comments.
248 (better_edge_p): Use profile counts.
249 (find_traces): Dump profile counts.
250 (rotate_loop): Use profile counts.
251 (find_traces_1_round): Likewise.
252 (connect_better_edge_p): Use counts instead of probabilities for
254 (copy_bb_p): Drop early check for non-0 frequency.
255 (sanitize_hot_paths): Update comments.
257 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
259 * ipa-split.c (struct split_point): Add count.
260 (consider_split): Do not compute incoming frequency; compute incoming
261 count and store it to split_point.
262 (split_function): Set count of the call to split part correctly.
264 2017-11-13 Carl Love <cel@us.ibm.com>
266 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
268 2017-11-13 Tom Tromey <tom@tromey.com>
270 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
272 2017-11-13 Carl Love <cel@us.ibm.com>
274 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
275 Add support for builtins:
276 unsigned int vec_first_{,miss}_match_{,or_eos}index,
277 vector {un,}signed {char,int,short},
278 vector {un,}signed {char,int,short}) arguments.
279 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
280 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
281 Add BU_P9V_AV_2 expansions for the builtins.
282 * config/rs6000/altivec.h (vec_first_match_index,
283 vec_first_mismatch_index, vec_first_match_or_eos_index,
284 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
285 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
286 new extern declaration.
287 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
288 * config/rs6000/vsx.md (first_match_index_<mode>,
289 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
290 first_mismatch_or_eos_index_<mode>): Add define expand.
291 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
292 * doc/extend.texi: Update the built-in documenation file for the new
295 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
297 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
298 into the min/max operations for _Float<N> and _Float<N>X types.
300 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
303 * dwarf2out.c (do_eh_frame): New static variable.
304 (dwarf2out_begin_prologue): Set it.
305 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
307 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
309 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
311 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
313 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
316 * debug.h (dwarf2out_do_eh_frame): Declare.
317 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
318 (dwarf2out_do_frame): Use it.
319 (dwarf2out_do_cfi_asm): Likewise.
320 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
321 (dwarf2out_assembly_start): Likewise.
322 (dwarf2out_begin_prologue): Fix comment.
323 * toplev.c (compile_file): Always call dwarf2out_frame_finish
324 if the target needs either debug or unwind DWARF2 info.
325 * lto-opts.c (lto_write_options): Do not save -fexceptions,
326 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
327 -ftrapping-math, -ftrapv and -fwrapv.
329 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
331 * cgraph.c (cgraph_edge::sreal_frequency): New function.
332 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
333 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
334 (estimate_edge_size_and_time): Likewise.
335 (ipa_merge_fn_summary_after_inlining): Likewise.
336 * ipa-inline.c (cgraph_freq_base_rec): Remove.
337 (compute_uninlined_call_time): Use sreal_frequency.
338 (compute_inlined_call_time): Likewise.
339 (ipa_inline): Do not initialize cgraph_freq_base_rec.
340 * profile-count.c: Include sreal.h.
341 (profile_count::to_sreal_scale): New.
342 * profile-count.h: Forward declare sreal.
343 (profile_count::to_sreal_scale): Declare.
345 2017-11-13 Nathan Sidwell <nathan@acm.org>
347 * diagnostic.c (maybe_line_and_column): New.
348 (diagnostic_get_location_text): Use it.
349 (diagnostic_report_current_module): Likewise.
350 (test_diagnostic_get_location_text): Add tests.
352 2017-11-13 Luis Machado <luis.machado@linaro.org>
354 * doc/md.texi (Specifying processor pipeline description): Fix
355 incorrect latency for the div instruction example.
357 2017-11-13 Jakub Jelinek <jakub@redhat.com>
359 PR tree-optimization/78821
360 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
361 that bit_not_p is the same.
362 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
363 (split_group): Count precisely bit_not_p bits in each statement.
364 (invert_op): New function.
365 (imm_store_chain_info::output_merged_store): Use invert_op to
366 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
367 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
369 2017-11-13 Martin Liska <mliska@suse.cz>
371 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
372 (struct source_info): Likewise.
373 (add_branch_counts): Likewise.
374 (add_line_counts): Likewise.
375 (function_summary): Likewise.
376 (output_intermediate_line): Likewise.
377 (generate_results): Likewise.
379 2017-11-13 Martin Liska <mliska@suse.cz>
381 * gcov.c (struct block_info): Remove typedef for block_t.
382 (struct line_info): Likewise.
383 (line_info::has_block): Likewise.
384 (EXIT_BLOCK): Likewise.
387 (get_cycles_count): Likewise.
388 (process_file): Likewise.
389 (read_graph_file): Likewise.
390 (solve_flow_graph): Likewise.
391 (find_exception_blocks): Likewise.
392 (add_line_counts): Likewise.
393 (accumulate_line_info): Likewise.
394 (output_line_details): Likewise.
396 2017-11-13 Martin Liska <mliska@suse.cz>
398 * gcov.c (struct arc_info): Remove typedef for arc_t.
399 (struct line_info): Likewise.
400 (add_branch_counts): Likewise.
401 (output_branch_count): Likewise.
402 (function_info::~function_info): Likewise.
404 (output_intermediate_line): Likewise.
405 (read_graph_file): Likewise.
406 (solve_flow_graph): Likewise.
407 (find_exception_blocks): Likewise.
408 (add_line_counts): Likewise.
409 (accumulate_line_info): Likewise.
410 (output_line_details): Likewise.
411 (output_function_details): Likewise.
413 2017-11-13 Martin Liska <mliska@suse.cz>
415 * gcov.c (struct function_info): Remove typedef for function_t.
416 (struct source_info): Likewise.
417 (source_info::get_functions_at_location): Likewise.
418 (solve_flow_graph): Likewise.
419 (find_exception_blocks): Likewise.
420 (add_line_counts): Likewise.
421 (output_intermediate_file): Likewise.
422 (process_file): Likewise.
423 (generate_results): Likewise.
424 (release_structures): Likewise.
425 (read_graph_file): Likewise.
426 (read_count_file): Likewise.
427 (accumulate_line_counts): Likewise.
428 (output_lines): Likewise.
430 2017-11-13 Martin Liska <mliska@suse.cz>
432 * gcov.c (function_info::function_info): Remove num_counts
433 and add vector<gcov_type>.
434 (function_info::~function_info): Use the vector.
435 (process_file): Likewise.
436 (read_graph_file): Likewise.
437 (read_count_file): Likewise.
438 (solve_flow_graph): Likewise.
440 2017-11-13 Martin Liska <mliska@suse.cz>
442 * gcov.c (function_info::is_artificial): New function.
443 (process_file): Erase all artificial early.
444 (generate_results): Skip as all artificial are already
447 2017-11-13 Martin Liska <mliska@suse.cz>
449 * gcov.c (read_graph_file): Store to global vector of functions.
450 (read_count_file): Iterate the vector.
451 (process_file): Likewise.
452 (generate_results): Likewise.
453 (release_structures): Likewise.
455 2017-11-13 Jakub Jelinek <jakub@redhat.com>
457 PR tree-optimization/82954
458 * gimple-ssa-store-merging.c
459 (imm_store_chain_info::coalesce_immediate_stores): If
460 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
462 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
464 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
465 Upddate call to ENDIAN_LANE_N.
466 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
467 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
468 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
469 and use aarch64_endian_lane_rtx.
470 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
472 2017-11-12 Tom de Vries <tom@codesourcery.com>
474 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
476 2017-11-12 Tom de Vries <tom@codesourcery.com>
478 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
480 2017-11-12 Tom de Vries <tom@codesourcery.com>
482 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
484 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
485 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
486 * defaults.h (ASM_OUTPUT_LABELREF): Same.
488 2017-11-11 Martin Sebor <msebor@redhat.com>
491 * doc/extend.texi (attribute nonstring): Remove spurious argument.
494 * prefic.c (translate_name): Replace strncpy with memcpy to
495 avoid -Wstringop-truncation.
497 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
499 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
501 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
503 * predict.c (maybe_hot_frequency_p): Do not use cfun.
505 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
507 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
510 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
512 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
513 enable generating XXBRH if the value is in a vector register.
514 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
515 value is in a vector register.
516 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
517 register to register bswap64's instead of doing the GPR sequence
518 used on previous machines.
519 (bswapdi2_xxbrd): New insn.
520 (bswapdi2_reg): Disallow on ISA 3.0.
521 (register to register bswap64 splitter): Do not split the insn on
522 ISA 3.0 systems that use XXBRD.
524 2017-11-10 Martin Sebor <msebor@redhat.com>
527 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
529 (find_subframework_file): Same.
531 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
533 * auto-profile.c (afdo_indirect_call): Drop frequency.
534 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
535 (cgraph_node::create_edge): Drop frequency argument.
536 (cgraph_node::create_indirect_edge): Drop frequency argument.
537 (cgraph_edge::make_speculative): Drop frequency arguments.
538 (cgraph_edge::resolve_speculation): Do not update frequencies
539 (cgraph_edge::dump_edge_flags): Do not dump frequency.
540 (cgraph_node::dump): Check consistency in IPA mode.
541 (cgraph_edge::maybe_hot_p): Use IPA counter.
542 (cgraph_edge::verify_count_and_frequency): Rename to ...
543 (cgraph_edge::verify_count): ... this one; drop frequency checking.
544 (cgraph_node::verify_node): Update.
545 * cgraph.h (struct cgraph_edge): Drop frequency.
546 (cgraph_edge::frequency): New function.
547 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
549 (cgraph_edge::rebuild_edges): Likewise.
550 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
551 (duplicate_thunk_for_node): Do not pass frequency.
552 (cgraph_node::create_clone): Scale only counts.
553 (cgraph_node::create_virtual_clone): Do not pass frequency.
554 (cgraph_node::create_edge_including_clones): Do not pass frequency.
555 (cgraph_node::create_version_clone): Do not pass frequency.
556 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
557 (cgraph_node::expand_thunk): Do not pass frequency.
558 (cgraph_node::create_wrapper): Do not pass frequency.
559 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
561 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
562 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
563 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
564 (gather_caller_stats): Use frequency function.
565 (ipcp_cloning_candidate_p): Use frequency function.
566 (ipcp_propagate_stage): Use frequency function.
567 (get_info_about_necessary_edges): Use frequency function.
568 (update_profiling_info): Update only IPA profile.
569 (update_specialized_profile): Use frequency functoin.
570 (perhaps_add_new_callers): Update only IPA profile.
571 * ipa-devirt.c (ipa_devirt): Use IPA profile.
572 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
573 (dump_ipa_call_summary): Use frequency function.
574 (estimate_edge_size_and_time): Use frequency function.
575 (ipa_merge_fn_summary_after_inlining): Use frequency function.
576 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
577 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
578 (update_noncloned_counts): ... ths one; scale counts only.
579 (clone_inlined_nodes): Do not scale frequency.
580 (inline_call): Do not pass frequency.
581 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
582 (compute_inlined_call_time): Use IPA profile.
583 (want_inline_small_function_p): Use IPA profile.
584 (want_inline_self_recursive_call_p): Use IPA profile.
585 (edge_badness): Use IPA profile.
586 (lookup_recursive_calls): Use IPA profile.
587 (recursive_inlining): Do not pass frequency.
588 (resolve_noninline_speculation): Do not update frequency.
589 (inline_small_functions): Collect max of IPA profile.
590 (dump_overall_stats): Dump IPA porfile.
591 (dump_inline_stats): Dump IPA porfile.
592 (ipa_inline): Collect IPA stats.
593 * ipa-inline.h (clone_inlined_nodes): Update prototype.
594 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
595 (ipa_propagate_frequency): Use frequency function.
596 (ipa_profile): Cleanup.
597 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
598 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
599 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
600 (input_node): Do not stream frequency.
601 (input_edge): Do not stream frequency.
602 (merge_profile_summaries): Scale only IPA profiles.
603 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
604 * predict.c (drop_profile): Do not recompute frequency.
605 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
606 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
607 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
608 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
609 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
610 * tree-inline.c (copy_bb): Do not scale frequency.
611 (expand_call_inline): Do not scale frequency.
612 (tree_function_versioning): Do not scale frequency.
613 * ubsan.c (ubsan_create_edge): Do not pass frequency.
615 2017-11-10 Julia Koval <julia.koval@intel.com>
617 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
618 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
619 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
620 _mm256_maskz_gf2p8affine_epi64_epi8,
621 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
622 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
623 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
624 __builtin_ia32_vgf2p8affineqb_v32qi,
625 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
626 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
628 2017-11-10 Tamar Christina <tamar.christina@arm.com>
632 (arm_option_override): Refactor.
633 (arm_option_reconfigure_globals): New.
634 (arm_options_perform_arch_sanity_checks): New.
635 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
636 (arm_options_perform_arch_sanity_checks): Likewise
638 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
640 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
642 * rs6000/rs6000.md ("type" attr): Add qmul.
643 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
644 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
645 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
647 2017-11-10 Martin Sebor <msebor@redhat.com>
650 * builtins.c (compute_objsize): Handle arrays that
651 compute_builtin_object_size likes to fail for. Make extern.
652 * builtins.h (compute_objsize): Declare.
653 (check_strncpy_sizes): New function.
654 (expand_builtin_strncpy): Call check_strncpy_sizes.
655 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
656 -Wstringop-truncation.
657 (gimple_fold_builtin_strncat): Same.
658 * gimple.c (gimple_build_call_from_tree): Set call location.
659 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
660 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
661 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
662 (handle_builtin_strlen): Use strlen_to_stridx.
663 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
665 Use strlen_to_stridx.
666 (pass_strlen::execute): Release strlen_to_stridx.
667 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
668 (-Wstringop-truncation): Document new option.
670 2017-11-10 Martin Liska <mliska@suse.cz>
672 PR gcov-profile/82702
673 * gcov.c (main): Handle intermediate files in a different
675 (get_gcov_intermediate_filename): New function.
676 (output_gcov_file): Remove support of intermediate files.
677 (generate_results): Allocate intermediate file.
678 (release_structures): Clean-up properly fn_end.
679 (output_intermediate_file): Start iterating with line 1.
681 2017-11-10 Jakub Jelinek <jakub@redhat.com>
683 PR tree-optimization/82929
684 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
685 ops_swapped_p non-static data member.
686 (store_immediate_info::store_immediate_info): Clear it.
687 (imm_store_chain_info::coalesce_immediate_stores): If swapping
688 ops set ops_swapped_p.
689 (count_multiple_uses): Handle ops_swapped_p.
691 2017-11-10 Martin Liska <mliska@suse.cz>
693 * coverage.c (coverage_init): Stream information about
694 support of has_unexecuted_blocks.
695 * doc/gcov.texi: Document that.
696 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
697 * gcov.c (read_graph_file): Likewise.
698 (output_line_beginning): Fix a small issue with
701 2017-11-10 Bin Cheng <bin.cheng@arm.com>
703 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
704 reference of trivial component.
706 2017-11-10 Jakub Jelinek <jakub@redhat.com>
709 * gimple-ssa-store-merging.c
710 (pass_store_merging::terminate_all_aliasing_chains): For
711 gimple_store_p stmts also call refs_output_dependent_p.
713 PR rtl-optimization/82913
714 * compare-elim.c (try_merge_compare): Punt if def_insn is not
717 2017-11-09 Jeff Law <law@redhat.com>
719 * vr-values.h: New file with vr_values class.
720 * tree-vrp.c: Include vr-values.h
721 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
722 data objects into the vr_values class.
723 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
724 (get_value_range): Make it a member function within vr_values class.
725 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
726 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
727 (op_with_constant_singleton_value_range): Likewise.
728 (extract_range_for_var_from_comparison_expr): Likewise.
729 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
730 (extract_range_from_binary_expr): Likewise.
731 (extract_range_from_unary_expr): Likewise.
732 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
733 (check_for_binary_op_overflow, extract_range_basic): Likewise.
734 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
735 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
736 (compare_name_with_value, compare_names): Likewise.
737 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
738 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
739 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
740 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
741 (extract_range_from_phi_node): Likewise.
742 (simplify_truth_ops_using_ranges): Likewise.
743 (simplify_div_or_mod_using_ranges): Likewise.
744 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
745 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
746 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
747 (simplify_float_conversion_using_ranges): Likewise.
748 (simplify_internal_call_using_ranges): Likewise.
749 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
750 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
751 poitner via x_vr_values for calls into gimple folder.
752 (vrp_initialize_lattice): Make this the vr_values ctor.
753 (vrp_free_lattice): Make this the vr_values dtor.
754 (set_vr_value): New function.
755 (class vrp_prop): Add vr_values data member. Add various member
756 functions as well as member functions that delegate to vr_values.
757 (check_array_ref): Make a member function within vrp_prop class.
758 (search_for_addr_array, vrp_initialize): Likewise.
759 (vrp_finalize): Likewise. Revamp to avoid direct access to
760 vr_value, values_propagated, etc.
761 (check_array_bounds): Extract vrp_prop class instance pointer from
762 walk info structure. Use it to call member functions.
763 (check_all_array_refs): Make a member function within vrp_prop class.
764 Smuggle class instance pointer via walk info structure.
765 (x_vr_values): New local static.
766 (vrp_valueize): Use x_vr_values to get class instance.
767 (vr_valueize_1): Likewise.
768 (class vrp_folder): Add vr_values data member. Add various member
769 functions as well as member functions that delegate to vr_values.
770 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
771 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
772 class instance from vr_values. Use it to call member functions.
773 (vrp_dom_walker): Add vr_values data member.
774 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
775 instance via x_vr_values.
776 (identify_jump_threads): Accept vr_values as argument. Store
777 it into the walker structure.
778 (evrp_dom_walker): Add vr_values class data member. Add various
780 (evrp_dom_walker::try_find_new_range): Use vr_values data
781 member to access the memory allocator.
782 (evrp_dom_walker::before_dom_children): Store vr_values class
783 instance into the vrp_folder class.
784 (evrp_dom_walker::push_value_range): Rework to avoid direct
785 access to num_vr_values and vr_value.
786 (evrp_dom_walker::pop_value_range): Likewise.
787 (execute_early_vrp): Remove call to vrp_initialize_lattice.
788 Use vr_values to get to dump_all_value_ranges member function.
789 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
790 and simplify_cond_using_ranges_2 via vrp_prop class instance.
791 Pass vr_values class instance down to identify_jump_threads.
792 Remove call to vrp_free_lattice.
793 (debug_all_value_ranges): Remove.
795 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
796 (vrp_folder): Likewise.
798 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
799 Get it from the existing bitmap instead.
800 (vrp_intersect_ranges_1): Likewise.
802 2017-11-09 Jakub Jelinek <jakub@redhat.com>
804 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
806 (store_immediate_info::store_immediate_info): Add bitnotp argument,
808 (imm_store_chain_info::coalesce_immediate_stores): Break group
809 if bit_not_p is different.
810 (count_multiple_uses, split_group,
811 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
812 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
813 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
814 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
815 to store_immediate_info ctor.
817 2017-11-09 Jim Wilson <jimw@sifive.com>
819 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
820 (scan_prog_file): Likewise.
822 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
824 * bb-reorder.c (max_entry_frequency): Remove.
825 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
826 connect_traces, push_to_next_round_p): Remove prototypes.
827 (find_traces_1_round): Use counts only.
828 (push_to_next_round_p): Likewise.
829 (find_traces): Likewise.
830 (rotate_loop): Likewise.
831 (find_traces_1_round): Likewise.
832 (connect_traces): Likewise.
833 (edge_order): Likewise.
835 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
837 * config/arm/arm.c (output_return_instruction): Add comments to
838 indicate requirement for cmse_nonsecure_entry return to account
839 for the size of clearing instruction output here.
840 (thumb_exit): Likewise.
841 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
842 return in hardfloat mode.
844 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
846 * config/rs6000/rs6000.c (machine_function): Add a bool,
847 "toc_is_wrapped_separately".
848 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
849 if it wasn't explicitly set or unset, we are optimizing for speed, and
850 doing separate shrink-wrapping.
851 (rs6000_get_separate_components): Enable the TOC component if
852 saving the TOC register in the prologue.
853 (rs6000_components_for_bb): Handle the TOC component.
854 (rs6000_emit_prologue_components): Store the TOC register where needed.
855 (rs6000_set_handled_components): Mark TOC as handled, if handled.
856 (rs6000_emit_prologue): Don't save the TOC if that is already done.
858 2017-11-09 Martin Jambor <mjambor@suse.cz>
860 * ipa-param-manipulation.c: New file.
861 * ipa-param-manipulation.h: Likewise.
862 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
863 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
864 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
865 (ipa_parm_adjustment): Likewise.
866 (ipa_parm_adjustment_vec): Likewise.
867 (ipa_get_vector_of_formal_parms): Moved declaration to
868 ipa-param-manipulation.h.
869 (ipa_get_vector_of_formal_parm_types): Likewise.
870 (ipa_modify_formal_parameters): Likewise.
871 (ipa_modify_call_arguments): Likewise.
872 (ipa_combine_adjustments): Likewise.
873 (ipa_dump_param_adjustments): Likewise.
874 (ipa_modify_expr): Likewise.
875 (ipa_get_adjustment_candidate): Likewise.
876 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
877 ipa-param-manipulation.c.
878 (ipa_get_vector_of_formal_parm_types): Likewise.
879 (ipa_modify_formal_parameters): Likewise.
880 (ipa_modify_call_arguments): Likewise.
881 (ipa_modify_expr): Likewise.
882 (get_ssa_base_param): Likewise.
883 (ipa_get_adjustment_candidate): Likewise.
884 (index_in_adjustments_multiple_times_p): Likewise.
885 (ipa_combine_adjustments): Likewise.
886 (ipa_dump_param_adjustments): Likewise.
887 * tree-sra.c: Also include ipa-param-manipulation.h
888 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
891 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
892 Alan Hayward <alan.hayward@arm.com>
893 David Sherwood <david.sherwood@arm.com>
895 * doc/sourcebuild.texi (vect_masked_store): Document.
897 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
898 Alan Hayward <alan.hayward@arm.com>
899 David Sherwood <david.sherwood@arm.com>
901 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
903 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
904 Alan Hayward <alan.hayward@arm.com>
905 David Sherwood <david.sherwood@arm.com>
907 * doc/sourcebuild.texi (vect_variable_length): Document.
909 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
915 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
916 Alan Hayward <alan.hayward@arm.com>
917 David Sherwood <david.sherwood@arm.com>
919 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
921 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
922 Alan Hayward <alan.hayward@arm.com>
923 David Sherwood <david.sherwood@arm.com>
925 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
926 previously undocumented selectors.
927 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
929 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
931 * doc/rtl.texi (const_vector): Say that elements can be
933 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
934 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
935 (gen_vec_duplicate): Use it instead of CONSTANT_P.
936 * optabs.c (expand_vector_broadcast): Likewise.
938 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
939 Alan Hayward <alan.hayward@arm.com>
940 David Sherwood <david.sherwood@arm.com>
942 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
943 scaled index even if the unscaled address was invalid.
944 Don't increase the complexity of using a scale in that case.
946 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
947 Alan Hayward <alan.hayward@arm.com>
948 David Sherwood <david.sherwood@arm.com>
950 * doc/rtl.texi: Rewrite the subreg rules so that they partition
951 the inner register into REGMODE_NATURAL_SIZE bytes rather than
952 UNITS_PER_WORD bytes.
953 * emit-rtl.c (validate_subreg): Divide subregs into blocks
954 based on REGMODE_NATURAL_SIZE of the inner mode.
955 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
956 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
957 * expmed.c (lowpart_bit_field_p): Divide the value up into
958 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
959 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
960 whether something is likely to occupy more than one register.
962 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
965 * ipa-inline-transform.c (update_noncloned_frequencies): Use
966 profile_count::adjust_for_ipa_scaling.
967 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
968 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
970 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
972 2017-11-09 Jakub Jelinek <jakub@redhat.com>
974 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
975 (split_group): Add total_orig and total_new arguments, estimate the
976 number of statements related to the store group without store merging
977 and with store merging.
978 (imm_store_chain_info::output_merged_store): Adjust split_group
979 callers, punt if estimated number of statements with store merging
980 is not smaller than estimated number of statements without it.
982 (handled_load): Remove has_single_use checks.
983 (pass_store_merging::process_store): Likewise.
985 2017-11-09 Richard Biener <rguenther@suse.de>
987 PR tree-optimization/82902
988 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
990 2017-11-09 Martin Liska <mliska@suse.cz>
993 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
996 2017-11-09 Martin Liska <mliska@suse.cz>
998 PR tree-optimization/82669
999 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1001 2017-11-09 Martin Liska <mliska@suse.cz>
1003 PR gcov-profile/48463
1004 * coverage.c (coverage_begin_function): Output also end locus
1005 of a function and information whether the function is
1007 * gcov-dump.c (tag_function): Parse and print the information.
1008 * gcov.c (INCLUDE_MAP): Add include.
1009 (INCLUDE_SET): Likewise.
1010 (struct line_info): Move earlier in the source file because
1011 of vector<line_info> in function_info structure.
1012 (line_info::line_info): Likewise.
1013 (line_info::has_block): Likewise.
1014 (struct source_info): Add new member index.
1015 (source_info::get_functions_at_location): New function.
1016 (function_info::group_line_p): New function.
1017 (output_intermediate_line): New function.
1018 (output_intermediate_file): Use the mentioned function.
1019 (struct function_start): New.
1020 (struct function_start_pair_hash): Likewise.
1021 (process_file): Add code that identifies group functions.
1022 Assign lines either to global or function scope.
1023 (generate_results): Skip artificial functions.
1024 (find_source): Assign index for each source file.
1025 (read_graph_file): Read new flag artificial and end_line.
1026 (add_line_counts): Assign it either to global of function scope.
1027 (accumulate_line_counts): Isolate core of the function to
1028 accumulate_line_info and call it for both function and global
1030 (accumulate_line_info): New function.
1031 (output_line_beginning): Fix GNU coding style.
1032 (print_source_line): New function.
1033 (output_line_details): Likewise.
1034 (output_function_details): Likewise.
1035 (output_lines): Iterate both source (global) scope and function
1037 (struct function_line_start_cmp): New class.
1038 * doc/gcov.texi: Reflect changes in documentation.
1040 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1043 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1044 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1045 and similarly for not instead of neg.
1047 2017-11-08 Andi Kleen <ak@linux.intel.com>
1049 * config/i386/i386.opt: Add -mforce-indirect-call.
1050 * config/i386/predicates.md: Check for flag_force_indirect_call.
1051 * doc/invoke.texi: Document -mforce-indirect-call
1053 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1055 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1057 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1058 config/riscv/riscv.c (predict.h): New include.
1059 (riscv_slow_unaligned_access_p): No longer static.
1060 (riscv_block_move_straight): Add require.
1061 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1063 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1066 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1067 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1068 nonimmediate_operand predicate for operand 1 instead of
1071 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1073 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1075 * config/aarch64/constraints.md (Uml): New constraint.
1076 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1079 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1081 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1082 of two vec_duplicates into a vec_concat.
1084 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1086 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1087 Simplify vec_merge of vec_duplicate and vec_concat.
1088 * config/aarch64/constraints.md (Utq): New constraint.
1089 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1092 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1094 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1095 Simplify vec_merge of vec_duplicate and const_vector.
1096 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1098 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1099 mode iterator. Update predicate on operand 1 to
1100 handle non-const_vec constants. Delete constraints.
1101 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1103 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1105 PR tree-optimization/78821
1106 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1108 (store_operand_info::store_operand_info): Initialize it to false.
1109 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1110 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1111 store in the group, and if chain_info is non-NULL, to ignore altogether
1113 (compatible_load_p): Fail if bit_not_p does not match.
1114 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1115 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1116 (pass_store_merging::process_store): Adjust
1117 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1118 call terminate_all_aliasing_chains newly when adding a store into
1119 a chain with non-NULL chain_info.
1121 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1123 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1125 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1127 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1129 (aarch64_layout_frame): Initialise emit_frame_chain.
1130 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1131 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1133 2017-11-08 Martin Liska <mliska@suse.cz>
1135 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1136 of gimple_call_internal_p.
1138 2017-11-07 Tom de Vries <tom@codesourcery.com>
1140 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1142 2017-11-07 Tom de Vries <tom@codesourcery.com>
1144 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1147 2017-11-08 Martin Liska <mliska@suse.cz>
1150 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1152 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1154 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1156 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1159 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1160 (*cmp<mode>_ccz_1): New insn with $k alternative.
1163 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1164 RTX_COMM_COMPARE as commutative as well.
1165 (ix86_binary_operator_ok): Formatting fix.
1166 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1167 *<code><mode>3<mask_name><round_saeonly_name>,
1168 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1169 *<s>mul<mode>3_highpart<mask_name>,
1170 *vec_widen_umult_even_v16si<mask_name>,
1171 *vec_widen_umult_even_v8si<mask_name>,
1172 *vec_widen_umult_even_v4si<mask_name>,
1173 *vec_widen_smult_even_v16si<mask_name>,
1174 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1175 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1176 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1177 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1178 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1179 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1180 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1181 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1182 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1183 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1184 ix86_binary_operator_ok. Formatting fixes.
1185 (*<plusminus_insn><mode>3<mask_name><round_name>,
1186 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1189 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1191 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1192 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1193 eq and ne if TARGET_ISEL.
1194 (cmp): New code_iterator.
1195 (UNS, UNSU_, UNSIK): New code_attrs.
1196 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1197 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1199 ("eq<mode>3"): ... this.
1200 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1202 ("ne<mode>3"): ... this.
1204 2017-11-07 Julia Koval <julia.koval@intel.com>
1207 * common/config/i386/i386-common.c
1208 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1209 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1210 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1211 * config/i386/i386.opt: Ditto.
1212 * config/i386/i386.c (ix86_target_string): Ditto.
1213 (ix86_option_override_internal): Ditto.
1214 (ix86_init_mpx_builtins): Move MPX to args2.
1215 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1216 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1217 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1218 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1219 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1220 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1221 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1223 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1226 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1227 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1228 (zero-extendsidi peephole2): Remove peephole.
1230 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1233 * stor-layout.c: Include attribs.h.
1234 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1235 explicit lookup of "aligned" attribute.
1237 2017-11-07 Andrew Waterman <andrew@sifive.com>
1239 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1240 (riscv_expand_block_move): Likewise.
1241 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1243 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1244 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1245 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1246 (riscv_adjust_block_mem): Likewise.
1247 (riscv_block_move_loop): Likewise.
1248 (riscv_expand_block_move): Likewise.
1249 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1251 2017-11-07 Michael Clark <michaeljclark@mac.com>
1253 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1254 (MUSL_DYNAMIC_LINKER): Likewise.
1256 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1258 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1259 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1262 2017-11-07 Richard Biener <rguenther@suse.de>
1264 * match.pd: Fix build.
1266 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1267 Jackson Woodruff <jackson.woodruff@arm.com>
1269 PR tree-optimization/71026
1270 * match.pd: Canonicalize negate in division.
1272 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1275 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1277 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1279 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1280 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1282 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1284 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1285 non-scalar integral types.
1286 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1287 (-(A-B), -(~A)): New transformations.
1289 2017-11-07 Tom de Vries <tom@codesourcery.com>
1291 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1292 semicolon after "do {} while (0)".
1293 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1294 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1295 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1296 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1297 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1299 2017-11-07 Tom de Vries <tom@codesourcery.com>
1301 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1302 after "do {} while (0)".
1303 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1304 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1305 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1306 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1307 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1309 2017-11-07 Tom de Vries <tom@codesourcery.com>
1312 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1314 (arm_rtx_costs_internal): Add missing semicolon after
1315 HANDLE_NARROW_SHIFT_ARITH call.
1317 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1320 disable isel if it was not set explicitly.
1322 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1324 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1326 (add_type_attribute) likewise.
1328 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1330 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1331 of struct ix86_frame.
1332 (ix86_initial_elimination_offset): Likewise.
1333 (ix86_expand_split_stack_prologue): Likewise.
1335 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1337 * tree-vrp.h (enum value_range_type): Update stale comment.
1339 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1340 Alan Hayward <alan.hayward@arm.com>
1341 David Sherwood <david.sherwood@arm.com>
1343 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1344 (aarch64_expand_vec_perm_const): Take the number of units too.
1345 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1346 (aarch64_expand_vec_perm_const): Likewise.
1347 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1348 (vec_perm<mode>): Update accordingly.
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_simd_vect_par_cnst_half):
1355 Take the number of units too.
1356 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1357 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1358 but check for a vector mode before rather than after the call.
1359 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1360 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1361 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1362 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1363 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1364 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1365 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1366 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1367 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1368 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1369 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1370 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1371 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1372 (aarch64_sqdmull2_n<mode>): Update accordingly.
1374 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1379 the number of units too.
1380 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1381 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1382 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1383 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1384 (vec_store_lanesxi<mode>): Update accordingly.
1386 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1391 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1392 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1393 of units rather than the mode.
1394 * config/aarch64/iterators.md (nunits): New mode attribute.
1395 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1396 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1397 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1398 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1399 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1400 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1401 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1402 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1403 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1404 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1405 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1406 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1407 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1408 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1409 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1410 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1411 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1412 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1413 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1414 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1415 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1416 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1417 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1418 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1419 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1420 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1421 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1422 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1423 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1424 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1425 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1426 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1427 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1428 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1429 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1430 (aarch64_simd_vec_setv2di): Likewise.
1432 2017-11-06 Carl Love <cel@us.ibm.com>
1434 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1436 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1437 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1439 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1441 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1442 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1443 Add power 8 macro expansions.
1444 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1445 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1446 power 8 instructions. (VSX_XXBR): Add iterator.
1448 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1450 * config/arm/arm.md (predicable_short_it): Change default to "no",
1451 improve documentation, remove uses that are identical to the default.
1452 (enabled_for_depr_it): Rename to enabled_for_short_it.
1453 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1454 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1455 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1456 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1457 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1459 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1462 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1463 float128 helper macros, which are no longer used after deleting
1464 the old 'q' built-in functions, and moving the round to odd
1465 built-in functions to being special built-in functions.
1466 (BU_FLOAT128_2): Likewise.
1467 (BU_FLOAT128_1_HW): Likewise.
1468 (BU_FLOAT128_2_HW): Likewise.
1469 (BU_FLOAT128_3_HW): Likewise.
1470 (FABSQ): Delete old 'q' built-in functions.
1471 (COPYSIGNQ): Likewise.
1472 (SQRTF128_ODD): Move round to odd built-in functions to be
1473 special built-in functions, so that we can handle
1474 -mabi=ieeelongdouble.
1475 (TRUNCF128_ODD): Likewise.
1476 (ADDF128_ODD): Likewise.
1477 (SUBF128_ODD): Likewise.
1478 (MULF128_ODD): Likewise.
1479 (DIVF128_ODD): Likewise.
1480 (FMAF128_ODD): Likewise.
1481 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1482 built-in names to 'f128'.
1483 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1484 old 'q' built-in functions, as the machine independent code for
1485 'f128' built-in functions handles this.
1486 (rs6000_expand_builtin): Add expansion for float128 round to odd
1487 functions, keying off on -mabi=ieeelongdouble of whether to use
1488 the KFmode or TFmode variant.
1489 (rs6000_init_builtins): Initialize the _Float128 round to odd
1491 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1492 _Float128 'q' built-in functions are now mapped into the new
1493 'f128' built-in functions.
1495 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1497 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1498 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1500 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1503 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1504 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1505 * toplev.c: Include "ipa-fnsummary.h".
1506 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1508 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1510 PR tree-optimization/82838
1511 * gimple-ssa-store-merging.c
1512 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1513 on a separate gimple_seq which is then appended to seq.
1515 2017-11-06 Jeff Law <law@redhat.com>
1518 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1519 (get_probe_interval): New functions.
1520 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1521 (ix86_adjust_stack_and_probe): Likewise.
1522 (output_adjust_stack_and_probe): Likewise.
1523 (ix86_emit_probe_stack_range): Likewise.
1524 (ix86_expand_prologue): Likewise.
1526 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1528 PR tree-optimization/82816
1529 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1530 if the modes of the two types are the same.
1531 (convert_plusminus_to_widen): Likewise.
1533 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1535 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1537 (usadv16qi): New define_expand.
1538 (usadv8hi): New define_expand.
1540 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1543 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1545 (inline_transform): Likewise.
1546 * predict.c (counts_to_freqs): Remove useless conditional.
1547 * profile-count.h (profile_count::apply_scale): Move sanity check.
1548 * tree-inline.c (copy_bb): Always scale.
1549 (copy_cfg_body): Likewise.
1551 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1554 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1556 (*cmp_ite0): Add enabled_for_depr_it attribute.
1557 (*cmp_ite1): Likewise.
1559 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1561 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1564 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1566 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1567 return the value by pointer.
1568 (extract_range_from_multiplicative_op_1): Update accordingly.
1569 Return as soon as an operation fails.
1571 2017-11-05 Tom de Vries <tom@codesourcery.com>
1574 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1575 (DEF_SANITIZER_BUILTIN): ... here.
1576 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1577 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1579 2017-11-05 Tom de Vries <tom@codesourcery.com>
1582 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1584 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1585 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1586 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1588 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1589 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1590 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1592 2017-11-05 Tom de Vries <tom@codesourcery.com>
1595 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1598 2017-11-04 Michael Clark <michaeljclark@mac.com>
1600 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1601 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1603 (*addsi3_extended): Likewise.
1604 (*addsi3_extended2): Likewise.
1605 (<optab>si3): Likewise.
1606 (<optab>di3): Likewise.
1607 (<optab><mode>3): Likewise.
1608 (<*optabe>si3_internal): Likewise.
1609 (zero_extendqi<SUPERQI:mode>2): Likewise.
1610 (*add<mode>hi3): Likewise.
1611 (*xor<mode>hi3): Likewise.
1612 (<optab>di3): Likewise.
1613 (*<optab>si3_extend): Likewise.
1614 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1615 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1616 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1618 2017-11-04 Andrew Waterman <andrew@sifive.com>
1620 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1621 TARGET_STRICT_ALIGN based upon -mtune argument.
1623 2017-11-04 Andrew Waterman <andrew@sifive.com>
1625 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1627 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1629 config/i386/i386.c (choose_basereg): Use optional scratch
1630 register and add assertion.
1631 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1632 needed, and don't allocate stack.
1633 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1634 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1635 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1637 2017-11-03 Jeff Law <law@redhat.com>
1639 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1640 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1641 to probe at the start of a noreturn function.
1643 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1645 PR tree-optimization/78821
1646 * gimple-ssa-store-merging.c: Update the file comment.
1647 (MAX_STORE_ALIAS_CHECKS): Define.
1648 (struct store_operand_info): New type.
1649 (store_operand_info::store_operand_info): New constructor.
1650 (struct store_immediate_info): Add rhs_code and ops data members.
1651 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1652 and op1r arguments to the ctor, initialize corresponding data members.
1653 (struct merged_store_group): Add load_align_base and load_align
1655 (merged_store_group::merged_store_group): Initialize them.
1656 (merged_store_group::do_merge): Update them.
1657 (merged_store_group::apply_stores): Pick the constant for
1658 encode_tree_to_bitpos from one of the two operands, or skip
1659 encode_tree_to_bitpos if neither operand is a constant.
1660 (class pass_store_merging): Add process_store method decl. Remove
1661 bool argument from terminate_all_aliasing_chains method decl.
1662 (pass_store_merging::terminate_all_aliasing_chains): Remove
1663 var_offset_p argument and corresponding handling.
1664 (stmts_may_clobber_ref_p): New function.
1665 (compatible_load_p): New function.
1666 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1667 if there is overlap and rhs_code is not INTEGER_CST. For
1668 non-overlapping stores terminate group if rhs is not mergeable.
1669 (get_alias_type_for_stmts): Change first argument from
1670 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1671 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1672 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1674 (get_location_for_stmts): Change first argument from
1675 auto_vec<gimple *> & to vec<gimple *> &.
1676 (struct split_store): Remove orig_stmts data member, add orig_stores.
1677 (split_store::split_store): Create orig_stores rather than orig_stmts.
1678 (find_constituent_stmts): Renamed to ...
1679 (find_constituent_stores): ... this. Change second argument from
1680 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1681 to info structures rather than the statements.
1682 (split_group): Rename ALLOW_UNALIGNED argument to
1683 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1684 it. Adjust find_constituent_stores caller.
1685 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1686 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1687 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1688 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1689 (mem_valid_for_store_merging): New function.
1690 (handled_load): New function.
1691 (pass_store_merging::process_store): New method.
1692 (pass_store_merging::execute): Use process_store method. Adjust
1693 terminate_all_aliasing_chains caller.
1695 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1697 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1698 Return true for more constants, symbols and label references.
1699 (aarch64_valid_floating_const): Remove unused function.
1701 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1704 * config/aarch64/aarch64.c (aarch64_layout_frame):
1705 Undo forcing of LR at bottom of frame.
1707 2017-11-03 Jeff Law <law@redhat.com>
1710 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1711 for int_registers_saved.
1713 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1714 extracted from tree-ssa-dom.c.
1715 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1716 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1717 (record_equivalences_from_incoming_edge): Add additional argument
1718 to single_pred_edge_ignoring_loop_edges call.
1719 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1720 (uncprop_dom_walker::before_dom_children): Add additional argument
1721 to single_pred_edge_ignoring_loop_edges call.
1722 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1723 single_pred_edge_ignoring_loop_edges rather than open coding.
1724 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1726 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1728 * match.pd (-(-A)): Rewrite.
1730 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1732 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1733 (rs6000_emit_int_cmove): New declaration.
1734 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1735 (rs6000_emit_sISEL): Delete.
1736 (rs6000_emit_int_cmove): Make non-static.
1737 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1738 instead of rs6000_emit_sISEL.
1740 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1742 * asan.c (create_cond_insert_point): Maintain profile.
1743 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1745 * basic-block.h (struct basic_block_def): Remove frequency.
1746 (EDGE_FREQUENCY): Use to_frequency
1747 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1749 (find_traces): Update to use to_frequency.
1750 (find_traces_1_round): Likewise; use only IPA counts.
1751 (bb_to_key): Likewise.
1752 (connect_traces): Use IPA counts only.
1753 (copy_bb_p): Update to use to_frequency.
1754 (fix_up_crossing_landing_pad): Likewise.
1755 (sanitize_hot_paths): Likewise.
1756 * bt-load.c (basic_block_freq): Likewise.
1757 * cfg.c (init_flow): Set count_max to uninitialized.
1758 (check_bb_profile): Remove frequencies; check counts.
1759 (dump_bb_info): Do not dump frequencies.
1760 (update_bb_profile_for_threading): Update counts only.
1761 (scale_bbs_frequencies_int): Likewise.
1762 (MAX_SAFE_MULTIPLIER): Remove.
1763 (scale_bbs_frequencies_gcov_type): Update counts only.
1764 (scale_bbs_frequencies_profile_count): Update counts only.
1765 (scale_bbs_frequencies): Update counts only.
1766 * cfg.h (struct control_flow_graph): Add count-max.
1767 (update_bb_profile_for_threading): Update prototype.
1768 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1769 (find_many_sub_basic_blocks): Likewise.
1770 * cfgcleanup.c (try_forward_edges): Likewise.
1771 (try_crossjump_to_edge): Likewise.
1772 * cfgexpand.c (expand_gimple_cond): Likewise.
1773 (expand_gimple_tailcall): Likewise.
1774 (construct_init_block): Likewise.
1775 (construct_exit_block): Likewise.
1776 * cfghooks.c (verify_flow_info): Check consistency of counts.
1777 (dump_bb_for_graph): Do not dump frequencies.
1778 (split_block_1): Do not update frequencies.
1779 (split_edge): Do not update frequencies.
1780 (make_forwarder_block): Do not update frequencies.
1781 (duplicate_block): Do not update frequencies.
1782 (account_profile_record): Do not update frequencies.
1783 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1784 for global heuristics.
1785 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1786 (expected_loop_iterations_unbounded): Use counts only.
1787 * cfgloopmanip.c (scale_loop_profile): Simplify.
1788 (create_empty_loop_on_edge): Simplify
1790 (duplicate_loop_to_header_edge): Simplify
1791 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1792 (update_br_prob_note): Take care of removing note when profile
1794 (relink_block_chain): Do not dump frequency.
1795 (rtl_account_profile_record): Use to_frequency.
1796 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1797 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1798 (cgraph_update_edges_for_call_stmt_node): Likewise.
1799 (cgraph_edge::verify_count_and_frequency): Update.
1800 (cgraph_node::verify_node): Temporarily disable frequency verification.
1801 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1802 to_cgraph_frequency.
1803 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1804 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1806 (cgraph_node::expand_thunk): Update profile.
1807 * except.c (dw2_build_landing_pads): Do not update frequency.
1808 * final.c (compute_alignments): Use to_frequency.
1809 (dump_basic_block_info): Do not dump frequency.
1810 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1811 (dump_gimple_bb_header): Do not dump frequency.
1812 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1814 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1815 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1816 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1817 (init_before_recovery): Do not update frequency.
1818 (sched_create_recovery_edges): Do not update frequency.
1819 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1820 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1821 (ipa_cp_c_finalize): Set max_count to uninitialized.
1822 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1823 (param_change_prob): Use counts.
1824 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1826 * ipa-split.c (consider_split): Use to_frequency.
1827 (split_function): Use to_frequency.
1828 * ira-build.c (loop_compare_func): Likewise.
1829 (mark_loops_for_removal): Likewise.
1830 (mark_all_loops_for_removal): Likewise.
1831 * loop-doloop.c (doloop_modify): Do not update frequency.
1832 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1834 * lto-streamer-in.c (input_function): Update count_max.
1835 * omp-expand.c (expand_omp_taskreg): Update count_max.
1836 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1837 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1838 (maybe_hot_count_p): Use ipa counts only.
1839 (maybe_hot_bb_p): Simplify.
1840 (maybe_hot_edge_p): Simplify.
1841 (probably_never_executed): Do not take frequency argument.
1842 (probably_never_executed_bb_p): Do not pass frequency.
1843 (probably_never_executed_edge_p): Likewise.
1844 (combine_predictions_for_bb): Check that profile is nonzero.
1845 (propagate_freq): Do not set frequency.
1846 (drop_profile): Simplify.
1847 (counts_to_freqs): Simplify.
1848 (expensive_function_p): Use to_frequency.
1849 (propagate_unlikely_bbs_forward): Simplify.
1850 (determine_unlikely_bbs): Simplify.
1851 (estimate_bb_frequencies): Add hack to silence graphite issues.
1852 (compute_function_frequency): Use ipa counts.
1853 (pass_profile::execute): Update.
1854 (rebuild_frequencies): Use counts only.
1855 (force_edge_cold): Use counts only.
1856 * profile-count.c (profile_count::dump): Dump new count types.
1857 (profile_count::differs_from_p): Check compatiblity.
1858 (profile_count::to_frequency): New function.
1859 (profile_count::to_cgraph_frequency): New function.
1860 * profile-count.h (struct function): Declare.
1861 (enum profile_quality): Add profile_guessed_local and
1862 profile_guessed_global0.
1863 (class profile_proability): Decrease number of bits to 29;
1864 update from_reg_br_prob_note and to_reg_br_prob_note.
1865 (class profile_count: Update comment; decrease number of bits
1866 to 61. Check compatibility.
1867 (profile_count::compatible_p): New private member function.
1868 (profile_count::ipa_p): New member function.
1869 (profile_count::operator<): Handle global zero correctly.
1870 (profile_count::operator>): Handle global zero correctly.
1871 (profile_count::operator<=): Handle global zero correctly.
1872 (profile_count::operator>=): Handle global zero correctly.
1873 (profile_count::nonzero_p): New member function.
1874 (profile_count::force_nonzero): New member function.
1875 (profile_count::max): New member function.
1876 (profile_count::apply_scale): Handle IPA scalling.
1877 (profile_count::guessed_local): New member function.
1878 (profile_count::global0): New member function.
1879 (profile_count::ipa): New member function.
1880 (profile_count::to_frequency): Declare.
1881 (profile_count::to_cgraph_frequency): Declare.
1882 * profile.c (OVERLAP_BASE): Delete.
1883 (compute_frequency_overlap): Delete.
1884 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1885 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1886 * sched-ebb.c (rank): Use counts only.
1887 * shrink-wrap.c (handle_simple_exit): Use counts only.
1888 (try_shrink_wrapping): Use counts only.
1889 (place_prologue_for_one_component): Use counts only.
1890 * tracer.c (find_best_predecessor): Use to_frequency.
1891 (find_trace): Use to_frequency.
1892 (tail_duplicate): Use to_frequency.
1893 * trans-mem.c (expand_transaction): Do not update frequency.
1894 * tree-call-cdce.c: Do not update frequency.
1895 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1896 (gimple_merge_blocks): Likewise.
1897 (gimple_split_edge): Likewise.
1898 (gimple_duplicate_sese_region): Likewise.
1899 (gimple_duplicate_sese_tail): Likewise.
1900 (move_sese_region_to_fn): Likewise.
1901 (gimple_account_profile_record): Likewise.
1902 (insert_cond_bb): Likewise.
1903 * tree-complex.c (expand_complex_div_wide): Likewise.
1904 * tree-eh.c (lower_resx): Update profile.
1905 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1907 (initialize_cfun): Do not initialize frequencies
1908 (freqs_to_counts): Delete.
1909 (copy_cfg_body): Ignore count parameter.
1910 (copy_body): Update.
1911 (expand_call_inline): Update count_max.
1912 (optimize_inline_calls): Update count_max.
1913 (tree_function_versioning): Update count_max.
1914 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1915 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1917 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1918 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1919 (try_peel_loop): Likewise.
1920 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1922 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1923 (tree_transform_and_unroll_loop): Do not use frequencies
1924 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1925 Use reliable prediction only.
1926 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1927 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1928 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1929 probability scaling.
1930 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1931 not update frequency
1932 (any_remaining_duplicated_blocks): Likewise.
1933 (update_profile): Likewise.
1934 (estimated_freqs_path): Delete.
1935 (freqs_to_counts_path): Delete.
1936 (clear_counts_path): Delete.
1937 (ssa_fix_duplicate_block_edges): Likewise.
1938 (duplicate_thread_path): Likewise.
1939 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1940 * tree-tailcall.c (decrease_profile): Do not update frequency.
1941 (eliminate_tail_call): Likewise.
1942 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1943 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1944 (optimize_mask_stores): Likewise.
1945 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1946 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1947 (ubsan_expand_ptr_ifn): Update profile.
1948 * value-prof.c (gimple_ic): Simplify.
1949 * value-prof.h (gimple_ic): Update prototype.
1950 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1951 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1953 (want_inline_self_recursive_call_p): Likewise.
1954 (resolve_noninline_speculation): Only cummulate defined counts.
1955 (inline_small_functions): Use nonzero_p.
1956 (ipa_inline): Do not access freed node.
1958 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1960 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1961 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1963 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1965 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1966 non-legitimate address.
1968 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1970 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1971 (*lt0_<mode>di, *lt0_<mode>si): New.
1973 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1975 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1976 TARGET_PAIRED_FLOAT.
1978 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1979 Jim Wilson <jim.wilson@linaro.org>
1981 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1982 * config/aarch64/aarch64-tune.md: Regenerated.
1983 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1984 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1986 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1988 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1989 (arc_expand_prologue): Restore blink for millicode.
1990 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1992 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1995 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1996 gen_vec_duplicate after forcing the scalar into a register.
1998 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2000 * combine (try_combine): Print the insns input to try_combine to the
2003 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2006 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2007 Remove second argument from aarch64_process_target_attr call.
2008 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2010 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2012 (aarch64_handle_attr_arch): Remove second argument.
2013 (aarch64_handle_attr_cpu): Ditto.
2014 (aarch64_handle_attr_tune): Ditto.
2015 (aarch64_handle_attr_isa_flags): Ditto.
2016 (aarch64_process_one_target_attr): Ditto.
2017 (aarch64_process_target_attr): Ditto.
2018 (aarch64_option_valid_attribute_p): Remove second argument.
2019 on aarch64_process_target_attr call.
2021 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2023 * diagnostic.c: Include "selftest-diagnostic.h".
2024 (selftest::assert_location_text): New function.
2025 (selftest::test_diagnostic_get_location_text): New function.
2026 (selftest::diagnostic_c_tests): Call it.
2028 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2030 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2031 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2032 (class selftest::test_diagnostic_context): Move to...
2033 * selftest-diagnostic.c: New file.
2034 * selftest-diagnostic.h: New file.
2036 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2038 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2039 offset range for FT32B.
2040 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2041 * config/ft32/ft32.md: Add TARGET_NOPM.
2042 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2043 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2045 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2047 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2049 2017-11-02 Jeff Law <law@redhat.com>
2051 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2052 virtual keyword on FINAL OVERRIDE members.
2054 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2055 virtuals together. Add virtual destructor.
2056 (substitute_and_fold_engine): Similarly.
2058 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2060 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2062 2017-11-02 Richard Biener <rguenther@suse.de>
2064 PR tree-optimization/82795
2065 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2067 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2069 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2071 (gcc_GAS_CHECK_FEATURE): Remove.
2072 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2073 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2074 * configure: Regenerate.
2076 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2078 * config/arc/arc.c (hwloop_optimize): Account for empty
2081 2017-11-02 Richard Biener <rguenther@suse.de>
2084 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2085 Truncate ARRAY_REF index and element size.
2087 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2089 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2091 2017-11-01 Jeff Law <law@redhat.com>
2093 * tree-ssa-ccp.c (ccp_folder): New class derived from
2094 substitute_and_fold_engine.
2095 (ccp_folder::get_value): New member function.
2096 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2097 (ccp_fold_stmt): Remove prototype.
2098 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2099 * tree-ssa-copy.c (copy_folder): New class derived from
2100 substitute_and_fold_engine.
2101 (copy_folder::get_value): Renamed from get_value.
2102 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2103 * tree-vrp.c (vrp_folder): New class derived from
2104 substitute_and_fold_engine.
2105 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2106 (vrp_folder::get_value): New member function.
2107 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2108 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2109 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2110 provide a class interface to folder/substitute routines.
2111 (ssa_prop_fold_stmt_fn): Remove typedef.
2112 (ssa_prop_get_value_fn): Likewise.
2113 (subsitute_and_fold): Remove prototype.
2114 (replace_uses_in): Likewise.
2115 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2116 Renamed from replace_uses_in. Call the virtual member function
2117 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2118 (substitute_and_fold_dom_walker): Remove initialization of
2119 data member entries for calbacks. Add substitute_and_fold_engine
2120 member and initialize it.
2121 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2122 member functions for get_value, replace_phi_args_in c
2123 replace_uses_in, and fold_stmt calls.
2124 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2125 substitute_and_fold. Remove assert. Update ctor call.
2127 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2128 (ssa_prop_visit_phi_fn): Likewise.
2129 (class ssa_propagation_engine): New class to provide an interface
2131 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2133 (ssa_prop_visit_phi): Likewise.
2134 (ssa_propagation_engine::simulate_stmt): Moved into class.
2135 Call visit_phi/visit_stmt from the class rather than via
2136 file scoped static variables.
2137 (ssa_propagation_engine::simulate_block): Moved into class.
2138 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2139 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2140 set file scoped statics for the visit_stmt/visit_phi callbacks.
2141 * tree-complex.c (complex_propagate): New class derived from
2142 ssa_propagation_engine.
2143 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2144 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2145 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2147 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2148 ssa_propagation_engine.
2149 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2150 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2151 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2152 * tree-ssa-copy.c (copy_prop): New class derived from
2153 ssa_propagation_engine.
2154 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2155 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2156 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2157 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2158 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2159 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2160 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2162 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2164 PR rtl-optimization/82778
2165 PR rtl-optimization/82597
2166 * compare-elim.c (struct comparison): Add in_a_setter field.
2167 (find_comparison_dom_walker::before_dom_children): Remove killed
2168 bitmap and df_simulate_find_defs call, instead walk the defs.
2169 Compute last_setter and initialize in_a_setter. Merge definitions
2170 with first initialization for a few variables.
2171 (try_validate_parallel): Use insn_invalid_p instead of
2172 recog_memoized. Return insn rather than just the pattern.
2173 (try_merge_compare): Fix up comment. Don't uselessly test if
2174 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2176 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2179 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2180 Alan Hayward <alan.hayward@arm.com>
2181 David Sherwood <david.sherwood@arm.com>
2183 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2184 aarch64_hard_regno_nregs to get the number of registers
2187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2188 Alan Hayward <alan.hayward@arm.com>
2189 David Sherwood <david.sherwood@arm.com>
2191 * config/aarch64/constraints.md (Upl): Rename to...
2193 * config/aarch64/aarch64.md
2194 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2197 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2198 Alan Hayward <alan.hayward@arm.com>
2199 David Sherwood <david.sherwood@arm.com>
2201 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2202 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2205 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2206 Alan Hayward <alan.hayward@arm.com>
2207 David Sherwood <david.sherwood@arm.com>
2209 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2210 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2211 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2213 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2214 of a single element.
2215 * config/aarch64/iterators.md: Add a comment above the permute
2216 unspecs to say that they are generated directly by
2217 aarch64_expand_vec_perm_const.
2218 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2220 2017-11-01 Nathan Sidwell <nathan@acm.org>
2222 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2224 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2226 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2227 and medany code models, and describe what they do.
2229 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2231 Revert accidental duplicate:
2233 * combine.c (can_change_dest_mode): Reject changes in
2234 REGMODE_NATURAL_SIZE.
2236 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2238 PR rtl-optimization/64682
2239 PR rtl-optimization/69567
2240 PR rtl-optimization/69737
2241 PR rtl-optimization/82683
2242 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2243 register mentioned in the note, drop the note, unless it came from I3,
2244 in which case it should go to I3 again.
2246 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2248 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2249 and return false if not.
2250 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2252 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2254 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2255 range is known to be empty.
2257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2258 Alan Hayward <alan.hayward@arm.com>
2259 David Sherwood <david.sherwood@arm.com>
2261 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2262 and CONST_VECTOR_NUNITS instead of computing the number of units from
2263 the byte sizes of the vector and element.
2264 (simplify_binary_operation_1): Likewise.
2265 (simplify_const_binary_operation): Likewise.
2266 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2273 (int_mem_offset): ...this new function.
2274 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2275 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2276 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
2284 (compute_costs, find_decomposable_subregs, decompose_register)
2285 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2286 (resolve_clobber, dump_choices): Use it.
2288 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2289 Alan Hayward <alan.hayward@arm.com>
2290 David Sherwood <david.sherwood@arm.com>
2292 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2294 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2295 Alan Hayward <alan.hayward@arm.com>
2296 David Sherwood <david.sherwood@arm.com>
2298 * alias.c (find_base_value, find_base_term): Only process integer
2299 truncations. Check the precision rather than the size.
2301 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2302 Alan Hayward <alan.hayward@arm.com>
2303 David Sherwood <david.sherwood@arm.com>
2305 * machmode.h (is_narrower_int_mode): New function
2306 * optabs.c (expand_float, expand_fix): Use it.
2307 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2309 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * rtl.h (narrower_subreg_mode): New function.
2314 * ira-color.c (update_costs_from_allocno): Use it.
2316 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2317 Alan Hayward <alan.hayward@arm.com>
2318 David Sherwood <david.sherwood@arm.com>
2320 * optabs-query.h (convert_optab_p): New function, split out from...
2321 (convert_optab_handler): ...here.
2322 (widening_optab_handler): Delete.
2323 (find_widening_optab_handler): Remove permit_non_widening parameter.
2324 (find_widening_optab_handler_and_mode): Likewise. Provide an
2325 override that operates on mode class wrappers.
2326 * optabs-query.c (widening_optab_handler): Delete.
2327 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2328 parameter. Assert that the two modes are the same class and that
2329 the "from" mode is narrower than the "to" mode. Use
2330 convert_optab_handler instead of widening_optab_handler.
2331 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2332 instead of widening_optab_handler.
2333 * expr.c (expand_expr_real_2): Update calls to
2334 find_widening_optab_handler.
2335 * optabs.c (expand_widen_pattern_expr): Likewise.
2336 (expand_binop_directly): Take the insn_code as a parameter.
2337 (expand_binop): Only call find_widening_optab_handler for
2338 conversion optabs; use optab_handler otherwise. Update calls
2339 to find_widening_optab_handler and expand_binop_directly.
2340 Use convert_optab_handler instead of widening_optab_handler.
2341 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2342 find_widening_optab_handler and use scalar_mode rather than
2344 (convert_plusminus_to_widen): Likewise.
2346 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2347 Alan Hayward <alan.hayward@arm.com>
2348 David Sherwood <david.sherwood@arm.com>
2350 * machmode.h (fixed_size_mode): New class.
2351 * rtl.h (get_pool_mode): Return fixed_size_mode.
2352 * gengtype.c (main): Add fixed_size_mode.
2353 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2354 (get_raw_arg_mode): Likewise.
2355 * doc/tm.texi: Regenerate.
2356 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2357 * targhooks.c (default_get_reg_raw_mode): Likewise.
2358 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2359 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2360 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2361 (msp430_get_raw_result_mode): Likewise.
2362 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2363 * dbxout.c (dbxout_parms): Require fixed-size modes.
2364 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2365 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2366 * omp-low.c (lower_oacc_reductions): Likewise.
2367 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2368 (simplify_subreg): Update accordingly.
2369 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2370 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2371 that aren't fixed-size.
2372 (get_pool_mode): Return a fixed_size_mode.
2373 (output_constant_pool_2): Take a fixed_size_mode.
2375 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2376 Alan Hayward <alan.hayward@arm.com>
2377 David Sherwood <david.sherwood@arm.com>
2379 * doc/rtl.texi (vec_series): Document.
2380 (const): Say that the operand can be a vec_series.
2381 * rtl.def (VEC_SERIES): New rtx code.
2382 * rtl.h (const_vec_series_p_1): Declare.
2383 (const_vec_series_p): New function.
2384 * emit-rtl.h (gen_const_vec_series): Declare.
2385 (gen_vec_series): Likewise.
2386 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2387 (gen_vec_series): Likewise.
2388 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2389 * simplify-rtx.c (simplify_unary_operation): Handle negations
2391 (simplify_binary_operation_series): New function.
2392 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2393 (test_vector_ops_series): New function.
2394 (test_vector_ops): Call it.
2395 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2396 gen_const_vec_series.
2397 (altivec_lvsr): Likewise.
2398 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2400 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2401 Alan Hayward <alan.hayward@arm.com>
2402 David Sherwood <david.sherwood@arm.com>
2404 * doc/rtl.texi (const): Update description of address constants.
2405 Say that vector constants are allowed too.
2406 * common.md (E, F): Use CONSTANT_P instead of checking for
2408 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2409 checking for CONST_VECTOR.
2410 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2412 * expr.c (expand_expr_real_2): Check for vector modes instead
2413 of checking for CONST_VECTOR.
2414 * rtl.h (const_vec_p): New function.
2415 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2416 (unwrap_const_vec_duplicate): Handle them here too.
2418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2419 David Malcolm <dmalcolm@redhat.com>
2420 Alan Hayward <alan.hayward@arm.com>
2421 David Sherwood <david.sherwood@arm.com>
2423 * rtl.h (vec_duplicate_p): New function.
2424 * selftest-rtl.c (assert_rtx_eq_at): New function.
2425 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2426 (assert_rtx_eq_at): Declare.
2427 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2428 * selftest-run-tests.c (selftest::run_tests): Call it.
2429 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2430 (simplify_unary_operation_1): Recursively handle vector duplicates.
2431 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2433 (simplify_subreg): Handle subregs of vector duplicates.
2434 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2435 (selftest::simplify_rtx_c_tests): New functions.
2437 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2438 Alan Hayward <alan.hayward@arm.com>
2439 David Sherwood <david.sherwood@arm.com>
2441 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2442 (gen_vec_duplicate): Likewise.
2443 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2445 (gen_const_vector): ...here.
2446 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2447 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2448 whose elements are all equal.
2449 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2450 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2451 (simplify_relational_operation): Likewise.
2452 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2454 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2455 (aarch64_expand_vector_init): Likewise.
2456 * config/arm/arm.c (neon_vdup_constant): Likewise.
2457 (neon_expand_vector_init): Likewise.
2458 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2459 (arm_block_set_unaligned_vect): Likewise.
2460 (arm_block_set_aligned_vect): Likewise.
2461 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2462 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2463 (expand_vec_perm_even_odd_pack): Likewise.
2464 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2465 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2466 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2467 gen_const_vec_duplicate.
2468 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2469 * config/mips/mips.c (mips_gen_const_int_vector): Use
2470 gen_const_vec_duplicate.
2471 (mips_expand_vector_init): Use CONST0_RTX.
2472 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2473 (define_split): Use gen_const_vec_duplicate.
2474 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2475 (define_split): Use gen_const_vec_duplicate.
2476 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2477 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2478 * config/spu/spu.c (spu_const): Likewise.
2480 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2481 Alan Hayward <alan.hayward@arm.com>
2482 David Sherwood <david.sherwood@arm.com>
2484 * combine.c (can_change_dest_mode): Reject changes in
2485 REGMODE_NATURAL_SIZE.
2487 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2489 * configure.ac (--enable-libssp): New.
2490 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2491 trying to determine target-specific default. Adjust indentation.
2492 * configure: Regenerated.
2493 * doc/install.texi (Configuration): Expand --disable-libssp
2496 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2498 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2501 2017-10-31 Martin Jambor <mjambor@suse.cz>
2504 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2506 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2508 * auto-profile.c (autofdo_source_profile::read): Use
2509 UNKNOWN_LOCATION rather than 0.
2510 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2511 (warning_at): ...this overload.
2512 (warning_at_rich_loc_n): Rename to...
2513 (warning_n): ...this overload.
2514 (error_at_rich_loc): Rename to...
2515 (error_at): ...this overload.
2516 (pedwarn_at_rich_loc): Rename to...
2517 (pedwarn): ...this overload.
2518 (permerror_at_rich_loc): Rename to...
2519 (permerror): ...this overload.
2520 (inform_at_rich_loc): Rename to...
2521 (inform): ...this overload.
2522 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2523 (diagnostic_n_impl_richloc): Rename to...
2524 (diagnostic_n_impl): ...this rich_location *-based decl.
2525 (inform_at_rich_loc): Rename to...
2526 (inform): ...this, and add an assertion.
2527 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2528 (warning_at_rich_loc): Rename to...
2529 (warning_at): ...this, and add an assertion.
2530 (warning_at_rich_loc_n): Rename to...
2531 (warning_n): ...this, and add an assertion.
2532 (warning_n): Update location_t-based implementation for removal of
2533 location_t-based diagnostic_n_impl.
2534 (pedwarn_at_rich_loc): Rename to...
2535 (pedwarn): ...this, and add an assertion.
2536 (permerror_at_rich_loc): Rename to...
2537 (permerror): ...this, and add an assertion.
2538 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2539 (error_at_rich_loc): Rename to...
2540 (error_at): ...this, and add an assertion.
2541 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2542 (driver::do_spec_on_infiles): Likewise.
2543 * substring-locations.c (format_warning_va): Update for renaming
2544 of inform_at_rich_loc.
2546 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2548 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2549 _Float<N>X built-in functions so that the variant without the
2550 "__builtin_" prefix is only enabled for the GNU C and Objective C
2551 languages when they are in non-strict ANSI/ISO mode.
2552 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2553 * target.def (floatn_builtin_p): Add a target hook to control
2554 whether _Float<N> and _Float<N>X built-in functions without the
2555 "__builtin_" prefix are enabled, and return true for C and
2556 Objective C in the default hook. Include langhooks.h in
2558 * targhooks.h (default_floatn_builtin_p): Likewise.
2559 * targhooks.c (default_floatn_builtin_p): Likewise.
2560 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2561 floatn_builtin_p target hook.
2562 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2564 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2565 Eric Botcazou <ebotcazou@adacore.com>
2567 PR rtl-optimization/81803
2568 * lra-constraints.c (curr_insn_transform): Also reload the whole
2569 register for a strict subreg no wider than a word if this is for
2570 a WORD_REGISTER_OPERATIONS target.
2572 2017-10-31 Jason Merrill <jason@redhat.com>
2574 * gdbinit.in: Skip over inlines from timevar.h.
2576 2017-10-31 Martin Liska <mliska@suse.cz>
2578 * doc/gcov.texi: Document new option.
2579 * gcov.c (print_usage): Likewise print it.
2580 (process_args): Support the argument.
2581 (format_count): New function.
2582 (format_gcov): Use the function.
2584 2017-10-31 Martin Liska <mliska@suse.cz>
2586 * gcov.c (struct name_map): do not use typedef.
2587 Define operator== and operator<.
2588 (name_search): Remove.
2589 (name_sort): Remove.
2590 (main): Do not allocate names.
2591 (process_file): Add vertical space.
2592 (generate_results): Use std::find.
2593 (release_structures): Do not release memory.
2594 (find_source): Use std::find.
2596 2017-10-31 Martin Liska <mliska@suse.cz>
2598 * gcov.c (struct line_info): Remove it's typedef.
2599 (line_info::line_info): Add proper ctor.
2600 (line_info::has_block): Do not use a typedef.
2601 (struct source_info): Do not use typedef.
2602 (circuit): Likewise.
2603 (get_cycles_count): Likewise.
2604 (output_intermediate_file): Iterate via vector iterator.
2605 (add_line_counts): Use std::vector methods.
2606 (accumulate_line_counts): Likewise.
2607 (output_lines): Likewise.
2609 2017-10-31 Martin Liska <mliska@suse.cz>
2611 * gcov.c (struct source_info): Remove typedef.
2612 (source_info::source_info): Add proper ctor.
2613 (accumulate_line_counts): Use struct, not it's typedef.
2614 (output_gcov_file): Likewise.
2615 (output_lines): Likewise.
2616 (main): Do not allocate an array.
2617 (output_intermediate_file): Use size of vector container.
2618 (process_file): Resize the vector.
2619 (generate_results): Do not preallocate, use newly added vector
2621 (release_structures): Do not release sources.
2622 (find_source): Use vector methods.
2623 (add_line_counts): Do not use typedef.
2625 2017-10-31 Martin Liska <mliska@suse.cz>
2627 * doc/gcov.texi: Document that.
2628 * gcov.c (add_line_counts): Mark lines with a non-executed
2630 (output_line_beginning): Handle such lines.
2631 (output_lines): Pass new argument.
2632 (output_intermediate_file): Print it in intermediate format.
2634 2017-10-31 Martin Liska <mliska@suse.cz>
2636 * color-macros.h: New file.
2637 * diagnostic-color.c: Factor out color related to macros to
2639 * doc/gcov.texi: Document -k option.
2640 * gcov.c (INCLUDE_STRING): Include string.h.
2641 (print_usage): Add -k option.
2642 (process_args): Parse it.
2643 (pad_count_string): New function.
2644 (output_line_beginning): Likewise.
2645 (DEFAULT_LINE_START): New macro.
2646 (output_lines): Support color output.
2648 2017-10-31 Martin Liska <mliska@suse.cz>
2650 PR gcov-profile/82633
2651 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2652 their interaction with GCOV infrastructure.
2653 * configure.ac: Add -fkeep-{inline,static}-functions to
2655 * configure: Regenerate.
2657 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2660 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2662 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2665 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2666 into a register if it does not fit into an immediate offset field.
2668 2017-10-31 Olivier Hainque <hainque@adacore.com>
2670 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2672 2017-10-31 Julia Koval <julia.koval@intel.com>
2674 * config.gcc: Add gfniintrin.h.
2675 * config/i386/gfniintrin.h: New.
2676 * config/i386/i386-builtin-types.def
2677 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2678 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2679 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2680 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2681 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2682 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2683 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2684 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2685 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2686 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2687 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2688 * config/i386/immintrin.h: Include gfniintrin.h.
2689 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2691 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2693 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2695 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2697 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2698 (arm_ashldi3_1bit): Remove pattern.
2699 (ashrdi3): Remove shift by 1 expansion.
2700 (arm_ashrdi3_1bit): Remove pattern.
2701 (lshrdi3): Remove shift by 1 expansion.
2702 (arm_lshrdi3_1bit): Remove pattern.
2703 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2704 cost of ashldi3 by 1.
2705 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2706 (<shift>di3_neon): Likewise.
2708 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2710 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2711 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2712 and (*aarch64_simd_mov<VQ:mode>).
2713 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2714 pattern alternative.
2715 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2716 attributes to match pattern alternative.
2718 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2720 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2721 * config/rs6000/emmintrin.h: New file.
2722 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2724 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2726 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2727 * (movdi_vfp_cortexa8): Remove pattern.
2729 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2731 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2733 (Specific, alpha*-dec-osf5.1): Remove.
2734 (Specific, mips-sgi-irix5): Remove.
2735 (Specific, mips-sgi-irix6): Remove.
2737 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2740 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2741 arguments to clear_bit_region_be.
2743 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2745 * gimplify.c: Include memmodel.h.
2747 2017-10-30 Martin Jambor <mjambor@suse.cz>
2749 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2750 condition whether loop should be executed at all.
2752 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2755 gimple folding of vec_madd() intrinsics.
2756 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2757 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2758 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2760 2017-10-30 Richard Biener <rguenther@suse.de>
2762 PR tree-optimization/82762
2764 2017-10-23 Richard Biener <rguenther@suse.de>
2766 PR tree-optimization/82129
2768 2017-08-01 Richard Biener <rguenther@suse.de>
2770 PR tree-optimization/81181
2771 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2772 (compute_antic): ... end of iteration here.
2774 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2776 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2777 -std=iso9899:2017 and -std=gnu17.
2778 * doc/standards.texi (C Language): Document C17 support.
2779 * doc/cpp.texi (Overview): Mention -std=c17.
2780 (Standard Predefined Macros): Document C11 and C17 values of
2781 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2782 * doc/extend.texi (Inline): Do not list individual options for
2783 standards newer than C99.
2784 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2786 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2789 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2791 * asan.c (asan_finish_file): Align asan globals array by shadow
2794 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2797 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2798 (struct store_immediate_info): Add bitregion_start and bitregion_end
2800 (store_immediate_info::store_immediate_info): Add brs and bre
2801 arguments and initialize bitregion_{start,end} from those.
2802 (struct merged_store_group): Add bitregion_start, bitregion_end,
2803 align_base and mask fields. Drop unnecessary struct keyword from
2804 struct store_immediate_info. Add do_merge method.
2805 (clear_bit_region_be): Use memset instead of loop storing zeros.
2806 (merged_store_group::do_merge): New method.
2807 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2808 stores as long as the surrounding bitregions have no gaps.
2809 (merged_store_group::merge_overlapping): Use do_merge.
2810 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2811 is byte aligned, rather than requiring that start and width are
2812 byte aligned. Drop unnecessary struct keyword from
2813 struct store_immediate_info. Allocate and populate also mask array.
2814 Make start of the arrays relative to bitregion_start rather than
2815 start and size them according to bitregion_{end,start} difference.
2816 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2817 struct store_immediate_info.
2818 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2819 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2820 struct keyword from struct store_immediate_info.
2821 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2822 between stores as long as the surrounding bitregions have no gaps.
2824 (struct split_store): Add orig non-static data member.
2825 (split_store::split_store): Initialize orig to false.
2826 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2827 if there is exactly a single original stmt. Change stmts argument
2828 to pointer from reference, if NULL, don't push anything to it. Add
2829 first argument, use it to optimize skipping over orig stmts that
2830 are known to be before bitpos already. Simplify.
2831 (split_group): Return unsigned int count how many stores are or
2832 would be needed rather than a bool. Add allow_unaligned argument.
2833 Change split_stores argument from reference to pointer, if NULL,
2834 only do a dry run computing how many stores would be produced.
2835 Rewritten algorithm to use both alignment and misalign if
2836 !allow_unaligned and handle bitfield stores with gaps.
2837 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2838 from bitregion_start instead of start. Compute allow_unaligned
2839 here, if true, do 2 split_group dry runs to compute which one
2840 produces fewer stores and prefer aligned if equal. Punt if
2841 new count is bigger or equal than original before emitting any
2842 statements, rather than during that. Remove no longer needed
2843 new_ssa_names tracking. Replace num_stmts with
2844 split_stores.length (). Use 32-bit stack allocated entries
2845 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2846 if possible. Handle bitfields with gaps.
2847 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2848 Compute bitregion_{start,end} for the stores and construct
2849 store_immediate_info with that. Formatting fixes.
2851 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2854 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2855 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2857 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2859 * gimplify.c: Include tm_p.h.
2861 * common.opt (gcoff): Re-add as ignored option.
2862 (gcoff1, gcoff2, gcoff3): Likewise.
2864 * Makefile.in (OBJS): Delete sdbout.o.
2865 (GTFILES): Delete $(srcdir)/sdbout.c.
2866 * debug.h: Delete sdb_debug_hooks.
2867 * final.c: Delete sdbout.h include.
2868 (final_scan_insn): Delete SDB_DEBUG check.
2869 (rest_of_clean_state): Likewise.
2870 * output.h: Delete sdb_begin_function_line.
2873 * toplev.c: Delete sdbout.h include.
2874 (process_options): Delete SDB_DEBUG check.
2875 * tree-core.h (tree_type_common): Delete pointer field of
2877 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2878 TYPE_SYMTAB_POINTER.
2879 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2880 (TYPE_SYMTAB_IS_POINTER): Delete.
2881 (TYPE_SYMTAB_IS_DIE): Renumber.
2882 * xcoffout.c: Refer to former sdbout.c file.
2883 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2885 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2886 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2887 (Debugging Options): Delete -gcoff.
2888 (-gstabs): Delete SDB reference.
2890 (-gcoff@var{level}): Delete.
2891 * doc/passes.texi (Debugging information output): Delete SDB and
2892 sdbout.c references.
2893 * doc/tm.texi: Regenerate.
2894 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2895 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2897 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2898 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2899 (SDB_DEBUGGING_INFO): Delete.
2900 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2901 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2902 * target.def (output_source_filename): Delete COFF reference.
2904 * common.opt (gcoff): Delete.
2905 (gxcoff+): Update Negative chain.
2906 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2908 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2909 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2910 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2912 (expand_function_start): Change sdb reference to past tense.
2913 (expand_function_end): Change sdb reference to past tense.
2914 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2915 * opts.c (debug_type_names): Delete coff entry.
2916 (common_handle_option): Delete OPT_gcoff case.
2917 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2919 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2920 * config/cris/cris.h: Delete SDB reference in comment.
2921 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2922 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2923 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2924 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2926 (ix86_expand_prologue): Likewise.
2927 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2928 * config/ia64/ia64.h: Likewise.
2929 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2930 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2932 * config/mmix/mmix.h: Likewise.
2933 * config/nds32/nds32.c: Likewise.
2934 * config/stormy/storym16.h: Likewise.
2935 * config/visium/visium.h: Likewise.
2936 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2938 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2940 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2941 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2942 FRAME_POINTER_REGNUM point at high end of local var area.
2944 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2946 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2947 Move comment around. Do not reset best_edge for a copiable
2948 destination if the copy would cause a partition change.
2949 (better_edge_p): Remove redundant check.
2951 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2953 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2955 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2957 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2958 for math functions that have _Float<N> and _Float<N>X variants.
2959 (mathfn_built_in_2): Add support for math functions that have
2960 _Float<N> and _Float<N>X variants.
2961 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2962 (expand_builtin_mathfn_ternary): Add support for fma with
2963 _Float<N> and _Float<N>X variants.
2964 (expand_builtin): Likewise.
2965 (fold_builtin_3): Likewise.
2966 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2967 create math function _Float<N> and _Float<N>X variants as external
2969 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2970 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2971 the __builtin_ prefix and if not strict ansi, without the prefix.
2972 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2973 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2974 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2975 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2976 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2977 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2978 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2979 function signatures for fma _Float<N> and _Float<N>X variants.
2980 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2981 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2982 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2983 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2984 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2985 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2986 * gencfn-macros.c (print_case_cfn): Add support for math functions
2987 that have _Float<N> and _Float<N>X variants.
2988 (print_define_operator_list): Likewise.
2989 (fltfn_suffixes): Likewise.
2991 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2992 for math functions that have _Float<N> and _Float<N>X variants.
2993 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2994 and _Float<N>X variants.
2995 (COPYSIGN): Likewise.
2998 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2999 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3001 (integer_valued_read_call_p): Likewise.
3002 * fold-const-call.c (fold_const_call_ss): Likewise.
3003 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3004 _Float<N> and _Float<N>X variants.
3005 (fold_const_call_ssss): Add support for fma _Float<N> and
3006 _Float<N>X variants.
3007 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3008 support for copysign and fma _Float<N> and _Float<N>X variants.
3009 (backprop::process_builtin_call_use): Likewise.
3010 * tree-call-cdce.c (can_test_argument_range); Add support for
3011 sqrt _Float<N> and _Float<N>X variants.
3012 (edom_only_function): Likewise.
3013 (get_no_error_domain): Likewise.
3014 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3015 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3016 copysign _Float<N> and _Float<N>X variants.
3017 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3018 handled by machine independent code.
3019 (FMAF128): Likewise.
3020 * doc/cpp.texi (Common Predefined Macros): Document defining
3021 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3022 fma _Float<N> and _Float<N>X variants.
3024 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3027 * config/i386/i386-modes.def (CCFPU): Remove definition.
3028 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3029 (ix86_cc_modes_compatible): Ditto.
3030 (ix86_expand_carry_flag_compare): Ditto.
3031 (ix86_expand_int_movcc): Ditto.
3032 (ix86_expand_int_addcc): Ditto.
3033 (ix86_reverse_condition): Ditto.
3034 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3035 Return true/false for unordered/ordered fp comparisons.
3036 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3037 (ix86_prepare_fp_compare_args): Update for rename.
3038 (ix86_expand_fp_compare): Update for rename. Generate unordered
3039 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3040 (ix86_expand_sse_compare_and_jump): Ditto.
3041 * config/i386/predicates.md (fcmov_comparison_operator):
3042 Remove CCFPU mode handling.
3043 (ix86_comparison_operator): Ditto.
3044 (ix86_carry_flag_operator): Ditto.
3045 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3046 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3047 (*cmpu<mode>_cc_i387): Ditto.
3048 (FPCMP): Remove mode iterator.
3049 (unord): Remove mode attribute.
3050 (unord_subst): New define_subst transformation
3051 (unord): New define_subst attribute.
3053 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3054 (*cmpi<unord>xf_i387): Ditto.
3055 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3056 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3057 using unord_subst transformation.
3058 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3059 (round_saeonly): Also handle CCFP mode.
3060 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3061 Remove UNSPEC_SAHF unspec handling.
3063 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3065 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3067 2017-10-27 Jeff Law <law@redhat.com>
3069 * gimple-ssa-sprintf.c: Include domwalk.h.
3070 (class sprintf_dom_walker): New class, derived from dom_walker.
3071 (sprintf_dom_walker::before_dom_children): New function.
3072 (struct call_info): Moved into sprintf_dom_walker class
3073 (compute_formath_length, handle_gimple_call): Likewise.
3074 (sprintf_length::execute): Call the dominator walker rather
3075 than walking the statements.
3077 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3078 gimple statement locations.
3079 (check_array_bounds): Corresponding changes. Get the statement's
3080 location directly from wi->stmt.
3082 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3085 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3087 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3089 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3090 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3092 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3095 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3096 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3097 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3098 maybe_get_pool_constant.
3099 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3102 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3104 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3105 2.26 caveat. Update gas and gld versions.
3106 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3109 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3111 * cgraph.h (set_malloc_flag): Declare.
3112 * cgraph.c (set_malloc_flag_1): New function.
3113 (set_malloc_flag): Likewise.
3114 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3115 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3117 (read_ipa_call_summary): Add support for reading is_return_callee.
3118 (write_ipa_call_summary): Stream is_return_callee.
3119 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3120 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3121 ipa-prop.h, ipa-fnsummary.h.
3122 (pure_const_names): Change to static.
3123 (malloc_state_e): Define.
3124 (malloc_state_names): Define.
3125 (funct_state_d): Add field malloc_state.
3126 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3127 (check_retval_uses): New function.
3128 (malloc_candidate_p): Likewise.
3129 (analyze_function): Add support for malloc attribute.
3130 (pure_const_write_summary): Stream malloc_state.
3131 (pure_const_read_summary): Add support for reading malloc_state.
3132 (dump_malloc_lattice): New function.
3133 (propagate_malloc): New function.
3134 (warn_function_malloc): New function.
3135 (ipa_pure_const::execute): Call propagate_malloc and
3136 ipa_free_fn_summary.
3137 (pass_local_pure_const::execute): Add support for malloc attribute.
3138 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3139 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3141 2017-10-27 Martin Liska <mliska@suse.cz>
3143 PR gcov-profile/82457
3144 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3145 for fork-like functions to be properly instrumented.
3147 2017-10-27 Richard Biener <rguenther@suse.de>
3150 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3151 info when we redirected EH.
3153 2017-10-26 Michael Collison <michael.collison@arm.com>
3155 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3157 (<optab>_trunchf<GPI:mode>2: New pattern.
3158 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3159 * config/aarch64/iterators.md (wv): New mode attribute.
3160 (vf, VF): New mode attributes.
3161 (vgp, VGP): New mode attributes.
3162 (s): Update attribute with SImode and DImode prefixes.
3164 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3166 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3167 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3168 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3169 (nios2_option_overide): Initialize it. Don't allow R0-relative
3170 addressing with PIC.
3171 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3172 (nios2_symbolic_constant_p): Likewise.
3173 (nios2_legitimate_address_p): Likewise.
3174 (nios2_r0rel_section_name_p): New.
3175 (nios2_symbol_ref_in_r0rel_data_p): New.
3176 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3177 (r0rel_constant_p): New.
3178 (nios2_print_operand_address): Handle r0rel_constant_p.
3179 (nios2_cdx_narrow_form_p): Likewise.
3180 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3181 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3182 (Nios II Options): Document -mr0rel-sec.
3184 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3186 * config/nios2/nios2.c: Include xregex.h.
3187 (nios2_gprel_sec_regex): New.
3188 (nios2_option_overide): Initialize it. Don't allow GP-relative
3189 addressing with PIC.
3190 (nios2_small_section_name_p): Check for regex match.
3191 * config/nios2/nios2.opt (mgprel-sec=): New option.
3192 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3193 (Nios II Options): Document -mgprel-sec.
3195 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3197 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3199 2017-10-26 Tom de Vries <tom@codesourcery.com>
3201 PR tree-optimization/82707
3202 * gimple.c (gimple_copy): Fix unsharing of
3203 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3205 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3207 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3208 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3209 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3210 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3211 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3212 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3213 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3214 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3215 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3216 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3217 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3218 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3219 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3220 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3221 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3222 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3223 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3225 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3227 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3229 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3230 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3231 warning to rs6000.c. Remove the Undocumented flag, since it has
3233 (-mabi=ibmlongdouble): Likewise.
3234 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3235 already set, set the default format for long double.
3236 (rs6000_debug_reg_global): Print whether long double is IBM or
3238 (rs6000_option_override_internal): Rework setting long double
3239 format. Only warn if the user is changing the long double default
3240 and they did not use -Wno-psabi.
3241 * doc/invoke.texi (PowerPC options): Update the documentation for
3242 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3244 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3245 Alan Hayward <alan.hayward@arm.com>
3246 David Sherwood <david.sherwood@arm.com>
3248 * rtl.h (wider_subreg_mode): New function.
3249 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3250 rather than an unsigned int *.
3251 * ira-color.c (regno_max_ref_width): Replace with...
3252 (regno_max_ref_mode): ...this new variable.
3253 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3254 Use wider_subreg_mode.
3255 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3256 rather than an unsigned int *.
3257 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3258 (process_alt_operands): Likewise.
3259 (invariant_p): Likewise.
3260 * lra-spills.c (assign_mem_slot): Likewise.
3261 (add_pseudo_to_slot): Likewise.
3262 * lra.c (collect_non_operand_hard_regs): Likewise.
3263 (add_regs_to_insn_regno_info): Likewise.
3264 * reload1.c (regno_max_ref_width): Replace with...
3265 (regno_max_ref_mode): ...this new variable.
3266 (reload): Update accordingly. Update call to
3267 ira_sort_regnos_for_alter_reg.
3268 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3269 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3270 (scan_paradoxical_subregs): Likewise.
3272 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3274 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3275 (aarch64_frame): Add emit_frame_chain boolean.
3276 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3277 Move eh_return case to aarch64_layout_frame.
3278 (aarch64_layout_frame): Initialize emit_frame_chain.
3279 (aarch64_expand_prologue): Use emit_frame_chain.
3281 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3283 * config/aarch64/aarch64.c (aarch64_layout_frame):
3284 Ensure LR is always stored at the bottom of the callee-saves.
3285 Remove rarely used frame layout which saves callee-saves at top of
3286 frame, so the store of LR can be used as a valid probe in all cases.
3288 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3290 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3291 Improve unaligned TImode/TFmode base/offset split.
3293 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3294 Alan Hayward <alan.hayward@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3298 * combine.c (find_single_use_1): Likewise.
3299 (expand_field_assignment): Likewise.
3300 (move_deaths): Likewise.
3301 * lra-constraints.c (simplify_operand_subreg): Likewise.
3302 (curr_insn_transform): Likewise.
3303 * lra.c (collect_non_operand_hard_regs): Likewise.
3304 (add_regs_to_insn_regno_info): Likewise.
3305 * rtlanal.c (reg_referenced_p): Likewise.
3306 (covers_regno_no_parallel_p): Likewise.
3308 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3310 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3311 Don't print any bits outside the precision of the value.
3312 * wide-int.cc (test_printing): Add some new tests.
3314 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3316 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3317 supports -xbrace_comment option.
3318 * configure: Regenerate.
3319 * config.in: Regenerate.
3320 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3321 (ASM_CPU_SPEC): Use it.
3323 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3325 * target.def (static_rtx_alignment): New hook.
3326 * targhooks.h (default_static_rtx_alignment): Declare.
3327 * targhooks.c (default_static_rtx_alignment): New function.
3328 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3329 * doc/tm.texi: Regenerate.
3330 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3331 instead of targetm.constant_alignment. Remove call to
3333 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3334 (cris_preferred_mininum_alignment): New function, split out from...
3335 (cris_constant_alignment): ...here.
3336 (cris_static_rtx_alignment): New function.
3337 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3339 (ix86_constant_alignment): ...here.
3340 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3341 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3342 (mmix_static_rtx_alignment): New function.
3343 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3344 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3346 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3349 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3350 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3352 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3354 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3355 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3356 mark as a sign-extending load.
3357 (local_pic_load_u): Define.
3359 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3362 * fold-const.c (operand_equal_for_comparison_p): Also return true
3363 if ARG0 is a simple variant of ARG1 with narrower precision.
3364 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3366 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3368 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3370 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3371 scatter_static, scatter_per_elt.
3372 * x86-tune-costs.h: Add new cost entries.
3374 2017-10-25 Richard Biener <rguenther@suse.de>
3376 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3377 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3378 class pass_fre): Move to ...
3379 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3380 class pass_fre): ... here and adjust for statistics.
3382 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3385 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3386 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3387 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3388 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3391 2017-10-25 Richard Biener <rguenther@suse.de>
3393 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3394 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3395 eliminate_push_avail, eliminate_insert): Move inside...
3396 (class eliminate_dom_walker): ... this class in preparation
3398 (fini_eliminate): Remove by merging with ...
3399 (eliminate): ... this function. Adjust for class changes.
3400 (pass_pre::execute): Remove fini_eliminate call.
3401 (pass_fre::execute): Likewise.
3403 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3406 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3407 (VPERMI2, VPERMI2I): New mode iterators.
3408 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3409 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3411 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3412 mode iterator. Remove 3 old define_insn patterns.
3413 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3414 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3415 VPERMI2 mode iterator, remove the other two expanders.
3416 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3417 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3418 instructions, remove the other two patterns.
3419 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3420 mode iterator, remove the other two patterns.
3421 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3422 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3423 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3424 and adjust argument order accordingly.
3425 (ix86_expand_vec_perm): Adjust caller.
3426 (expand_vec_perm_1): Likewise.
3427 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3428 (expand_vec_perm_vpermt2_vpshub2): ... this.
3429 (ix86_expand_vec_perm_const_1): Adjust caller.
3430 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3433 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3434 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3435 (vec_shl_<mode>): Remove unused expander.
3436 (avx512bw_<shift_insn><mode>3): New define_insn.
3437 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3438 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3440 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3443 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3446 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3448 PR rtl-optimization/82396
3449 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3450 (autopref_multipass_init): Simplify initialization.
3451 (autopref_rank_data): Simplify sort order.
3452 * gcc/sched-int.h (autopref_multipass_data_): Remove
3453 multi_mem_insn_p, min_offset and max_offset.
3455 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3458 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3459 Check special value of flag_omit_frame_pointer.
3460 (aarch64_can_eliminate): Likewise.
3461 (aarch64_override_options_after_change_1): Simplify handling of
3462 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3464 2017-10-24 Richard Biener <rguenther@suse.de>
3466 PR tree-optimization/82697
3467 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3468 zero for conditional load and unconditional store.
3470 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3472 * doc/install.texi: Document bootstrap-cet.
3474 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3477 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3478 ENDBR instruction at function entrance if function is only
3481 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3484 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3485 patterns to better describe from which operation the CF is computed.
3486 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3487 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3488 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3489 is 0, use _0 suffixed expanders instead of emitting a comparison
3492 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3494 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3495 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3497 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3500 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3501 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3502 * loop-iv.c (iv_get_reaching_def): Likewise.
3503 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3504 variable is promoted and the partition contains undefined values.
3506 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3508 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3510 (nios2_address_cost): Define.
3511 (nios2_legitimize_address): Recognize (exp + constant) directly.
3512 (TARGET_ADDRESS_COST): Define.
3514 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3516 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3517 (nios2_symbolic_memory_operand_p): Declare.
3518 (nios2_split_large_constant): Declare.
3519 (nios2_split_symbolic_memory_operand): Declare.
3520 * config/nios2/nios2.c: Adjust includes.
3521 (nios2_symbolic_constant_allowed): New.
3522 (nios2_symbolic_constant_p): New.
3523 (nios2_plus_symbolic_constant_p): New.
3524 (nios2_valid_addr_expr_p): Recognize addresses involving
3526 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3527 (nios2_symbolic_memory_operand_p): New.
3528 (nios2_large_constant_p): New.
3529 (nios2_split_large_constant): New.
3530 (nios2_split_plus_large_constant): New.
3531 (nios2_split_symbolic_memory_operand): New.
3532 (nios2_legitimize_address): Code refactoring. Handle addresses
3533 involving symbolic constants.
3534 (nios2_emit_move_sequence): Likewise.
3535 (nios2_print_operand): Improve error output.
3536 (nios2_print_operand_address): Handle LO_SUM.
3537 (nios2_cdx_narrow_form_p): Likewise.
3538 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3539 operands involving symbolic constants.
3540 (movhi_internal, movsi_internal): Likewise.
3541 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3542 (extendhisi2, extendqi<mode>2): Likewise.
3544 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3546 * tree-pass.h (PROP_rtl_split_insns): Define.
3547 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3549 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3551 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3553 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3556 * target.def (const_not_ok_for_debug_p): Default to
3557 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3558 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3559 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3560 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3561 which targetm.const_not_ok_for_debug_p returned true.
3562 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3564 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3566 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3567 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3568 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3570 (ix86_const_not_ok_for_debug_p): New function.
3571 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3572 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3574 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3577 * system.h: Conditionally include "unique-ptr.h" if
3578 INCLUDE_UNIQUE_PTR is defined.
3579 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3580 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3582 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3584 * config/rl78/rl78.md: New define_expand "subdi3".
3586 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3589 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3590 DF_REF_INSN if DF_REF_INSN_INFO is false.
3592 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3594 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3595 xmm_move instead of sse_move.
3596 (sse_store_index): New function.
3597 (ix86_register_move_cost): Be more sensible about mismatch stall;
3598 model AVX moves correctly; make difference between sse->integer and
3600 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3601 moves; make difference between SSE and AVX.
3602 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3603 and zmm_move. Increase size of sse load and store tables;
3604 add unaligned load and store tables; add ssemmx_to_integer.
3605 * x86-tune-costs.h: Update all entries according to real
3606 move latencies from Agner Fog's manual and chip documentation.
3608 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3611 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3612 * config/i386/constraints.md (Wf): New constraint.
3613 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3614 (cmp<dwi>_doubleword): Removed.
3615 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3616 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3617 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3618 expand with cmp<dwi>_doubleword. For LTU and GEU use
3619 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3621 * common.opt (gcolumn-info): Enable by default.
3622 * doc/invoke.texi (gcolumn-info): Document new default.
3624 2017-10-23 Richard Biener <rguenther@suse.de>
3626 PR tree-optimization/82672
3627 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3628 Fold the stmt if we propagated into it.
3630 2017-10-23 Richard Biener <rguenther@suse.de>
3632 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3633 (bitmap_remove_expr_from_set): ... this. All callers call this
3634 for non-constant values.
3635 (bitmap_set_subtract): Rename to...
3636 (bitmap_set_subtract_expressions): ... this. Adjust and
3638 (bitmap_set_contains_value): Remove superfluous check.
3639 (bitmap_set_replace_value): Inline into single caller ...
3640 (bitmap_value_replace_in_set): ... here and simplify.
3641 (dependent_clean): Merge into ...
3642 (clean): ... this using an overload. Adjust.
3643 (prune_clobbered_mems): Adjust.
3644 (compute_antic_aux): Likewise.
3645 (compute_partial_antic_aux): Likewise.
3647 2017-10-23 Richard Biener <rguenther@suse.de>
3649 PR tree-optimization/82129
3651 2017-08-01 Richard Biener <rguenther@suse.de>
3653 PR tree-optimization/81181
3654 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3655 (compute_antic): ... end of iteration here.
3657 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3659 * target.def (starting_frame_offset): New hook.
3660 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3661 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3662 * doc/tm.texi.in: Regenerate.
3663 * hooks.h (hook_hwi_void_0): Declare.
3664 * hooks.c (hook_hwi_void_0): New function.
3665 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3666 STARTING_FRAME_OFFSET.
3667 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3668 * reload1.c (reload): Likewise.
3669 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3670 instead of STARTING_FRAME_OFFSET.
3671 * function.c (try_fit_stack_local): Likewise.
3672 (assign_stack_local_1): Likewise
3673 (instantiate_virtual_regs): Likewise.
3674 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3675 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3676 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3677 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3678 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3679 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3680 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3681 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3682 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3683 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3684 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3685 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3686 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3687 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3688 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3689 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3690 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3691 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3692 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3693 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3694 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3695 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3696 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3697 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3698 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3699 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3700 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3701 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3702 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3703 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3704 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3705 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3706 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3707 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3708 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3709 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3710 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3711 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3712 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3713 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3714 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3715 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3716 return a HOST_WIDE_INT.
3717 (avr_builtin_setjmp_frame_value): Use it instead of
3718 STARTING_FRAME_OFFSET.
3719 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3720 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3721 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3723 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3724 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3725 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3726 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3727 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3728 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3729 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3730 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3731 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3732 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3733 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3734 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3736 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3737 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3738 * config/mips/mips.c (mips_compute_frame_info): Refer to
3739 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3740 (mips_starting_frame_offset): New function.
3741 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3742 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3743 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3744 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3745 and return a HOST_WIDE_INT.
3746 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3747 (mmix_initial_elimination_offset): Refer to
3748 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3749 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3750 * config/pa/pa.c (pa_starting_frame_offset): New function.
3751 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3752 (pa_expand_prologue): Likewise.
3753 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3754 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3755 !FRAME_GROWS_DOWNWARD handling to...
3756 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3757 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3758 !FRAME_GROWS_DOWNWARD handling to...
3759 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3760 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3761 !FRAME_GROWS_DOWNWARD handling to...
3762 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3763 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3765 (rs6000_starting_frame_offset): New function.
3766 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3767 !FRAME_GROWS_DOWNWARD handling to...
3768 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3769 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3770 !FRAME_GROWS_DOWNWARD handling to...
3771 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3772 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3773 !FRAME_GROWS_DOWNWARD handling to...
3774 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3775 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3776 (rs6000_starting_frame_offset): New function.
3777 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3778 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3779 * config/vax/vax.c (vax_starting_frame_offset): New function.
3780 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3781 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3782 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3783 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3784 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3785 * system.h (STARTING_FRAME_OFFSET): Poison.
3787 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3789 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3790 SCALAR_TYPE_MODE instead of TYPE_MODE.
3792 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3793 Alan Hayward <alan.hayward@arm.com>
3794 David Sherwood <david.sherwood@arm.com>
3796 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3798 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3799 Alan Hayward <alan.hayward@arm.com>
3800 David Sherwood <david.sherwood@arm.com>
3802 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3804 2017-10-23 Richard Biener <rguenther@suse.de>
3806 PR tree-optimization/82129
3807 * tree-ssa-pre.c (bitmap_set_and): Remove.
3808 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3809 canonicalizing expressions in the set to those with lowest
3810 ID rather than taking that from the first edge.
3812 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3814 * combine.c (rtx_equal_for_field_assignment_p): Use
3815 byte_lowpart_offset.
3817 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3818 Alan Hayward <alan.hayward@arm.com>
3819 David Sherwood <david.sherwood@arm.com>
3821 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3822 to a SUBREG_PROMOTED_VAR.
3824 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3825 Alan Hayward <alan.hayward@arm.com>
3826 David Sherwood <david.sherwood@arm.com>
3828 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3829 (expand_debug_source_expr): Likewise.
3830 * combine.c (combine_simplify_rtx): Likewise.
3831 * cse.c (fold_rtx): Likewise.
3832 * optabs.c (expand_float): Likewise.
3833 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3834 (simplify_binary_operation_1): Likewise.
3836 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3837 Alan Hayward <alan.hayward@arm.com>
3838 David Sherwood <david.sherwood@arm.com>
3840 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3841 (record_promoted_value): Likewise.
3842 * expr.c (expand_expr_real_2): Likewise.
3843 * ree.c (update_reg_equal_equiv_notes): Likewise.
3844 (combine_set_extension): Likewise.
3845 * rtlanal.c (low_bitmask_len): Likewise.
3846 * simplify-rtx.c (neg_const_int): Likewise.
3847 (simplify_binary_operation_1): Likewise.
3849 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3850 Alan Hayward <alan.hayward@arm.com>
3851 David Sherwood <david.sherwood@arm.com>
3853 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3854 * regcprop.c (maybe_mode_change): Likewise.
3855 * reload1.c (alter_reg): Likewise.
3857 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3859 * inchash.h (inchash::hash::add_wide_int): New function.
3860 * lto-streamer-out.c (hash_tree): Use it.
3862 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3864 * inchash.h (inchash::hash::add_wide_int): Rename to...
3865 (inchash::hash::add_hwi): ...this.
3866 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3867 (polymorphic_call_target_hasher::hash): Likewise.
3868 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3869 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3870 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3871 * lto-streamer-out.c (hash_tree): Likewise.
3872 * optc-save-gen.awk: Likewise.
3873 * tree.c (add_expr): Likewise.
3875 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3878 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3879 for ordered inequality comparisons even with TARGET_IEEE_FP.
3881 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3884 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3885 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3886 Expand with cmp<dwi>_doubleword.
3888 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3890 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3891 List CET intrinsics.
3892 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3893 specific to -fcf-protection option.
3895 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3897 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3898 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3899 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3900 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3901 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3902 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3903 (extra_objs): Add cet.o for Linux/x86 targets.
3904 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3905 * config/i386/cet.c: New file.
3906 * config/i386/cetintrin.h: Likewise.
3907 * config/i386/t-cet: Likewise.
3908 * config/i386/cpuid.h (bit_SHSTK): New.
3909 (bit_IBT): Likewise.
3910 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3911 pass IBT and SHSTK bits.
3912 * config/i386/i386-builtin-types.def
3913 (VOID_FTYPE_UNSIGNED_PVOID): New.
3914 (VOID_FTYPE_UINT64_PVOID): Likewise.
3915 * config/i386/i386-builtin.def: Add CET intrinsics.
3916 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3917 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3918 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3919 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3921 * config/i386/i386.c (rest_of_insert_endbranch): New.
3922 (pass_data_insert_endbranch): Likewise.
3923 (pass_insert_endbranch): Likewise.
3924 (make_pass_insert_endbranch): Likewise.
3925 (ix86_notrack_prefixed_insn_p): Likewise.
3926 (ix86_target_string): Add -mibt, -mshstk flags.
3927 (ix86_option_override_internal): Add flag_cf_protection
3929 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3930 (ix86_print_operand): Add 'notrack' prefix output.
3931 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3932 (ix86_expand_builtin): Expand CET intrinsics.
3933 (x86_output_mi_thunk): Add 'endbranch' instruction.
3934 * config/i386/i386.h (TARGET_IBT): New.
3935 (TARGET_IBT_P): Likewise.
3936 (TARGET_SHSTK): Likewise.
3937 (TARGET_SHSTK_P): Likewise.
3938 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3939 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3940 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3941 (builtin_setjmp_setup): New pattern.
3942 (builtin_longjmp): Likewise.
3943 (rdssp<mode>): Likewise.
3944 (incssp<mode>): Likewise.
3945 (saveprevssp): Likewise.
3946 (rstorssp): Likewise.
3947 (wrss<mode>): Likewise.
3948 (wruss<mode>): Likewise.
3949 (setssbsy): Likewise.
3950 (clrssbsy): Likewise.
3951 (nop_endbr): Likewise.
3952 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3954 * config/i386/immintrin.h: Include <cetintrin.h>.
3955 * config/i386/linux-common.h
3956 (file_end_indicate_exec_stack_and_cet): New prototype.
3957 (TARGET_ASM_FILE_END): New.
3959 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3961 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3962 latencies instead of having separate table; make difference between
3963 integer and float costs.
3964 * i386.h (processor_costs): Remove scalar_stmt_cost,
3965 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3966 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3968 * x86-tune-costs.h: Remove entries which has been removed in
3969 procesor_costs from all tables; make cond_taken_branch_cost
3970 and cond_not_taken_branch_cost COST_N_INSNS based.
3972 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3974 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3976 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3978 * config/i386/i386.md (isa): Remove fma_avx512f.
3979 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3980 <avx512>_fmadd_<mode>_mask3<round_name>,
3981 <avx512>_fmsub_<mode>_mask<round_name>,
3982 <avx512>_fmsub_<mode>_mask3<round_name>,
3983 <avx512>_fnmadd_<mode>_mask<round_name>,
3984 <avx512>_fnmadd_<mode>_mask3<round_name>,
3985 <avx512>_fnmsub_<mode>_mask<round_name>,
3986 <avx512>_fnmsub_<mode>_mask3<round_name>,
3987 <avx512>_fmaddsub_<mode>_mask<round_name>,
3988 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3989 <avx512>_fmsubadd_<mode>_mask<round_name>,
3990 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3991 (*vec_widen_umult_even_v16si<mask_name>,
3992 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3993 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3995 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3997 * extend.texi: Add 'nocf_check' documentation.
3998 * gimple.texi: Add second parameter to
3999 gimple_build_call_from_tree.
4000 * invoke.texi: Add -fcf-protection documentation.
4001 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4003 2017-10-20 Richard Biener <rguenther@suse.de>
4005 PR tree-optimization/82473
4006 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4007 the largest input type.
4009 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4011 * c-attribs.c (handle_nocf_check_attribute): New function.
4012 (c_common_attribute_table): Add 'nocf_check' handling.
4013 * gimple-parser.c: Add second argument NULL to
4014 gimple_build_call_from_tree.
4015 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4016 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4018 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4019 * common.opt: Add fcf-protection flag.
4020 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4021 * flag-types.h: Add enum cf_protection_level.
4022 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4023 Add 'nocf_check' attribute propagation to gimple call.
4024 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4025 (gimple_build_call_from_tree): Update prototype.
4026 (gimple_call_nocf_check_p): New function.
4027 (gimple_call_set_nocf_check): Likewise.
4028 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4029 * ipa-icf.c: Add nocf_check attribute in statement hash.
4030 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4031 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4032 * toplev.c (process_options): Add flag_cf_protection handling.
4034 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4036 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4038 2017-10-20 Richard Biener <rguenther@suse.de>
4040 PR tree-optimization/82603
4041 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4042 remove false predicated stores.
4044 2017-10-20 Richard Biener <rguenther@suse.de>
4046 * graphite-isl-ast-to-gimple.c
4047 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4048 Remove return value and simplify, dump copied stmt after lhs
4050 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4051 Reduce dump verbosity.
4052 (gsi_insert_earliest): Likewise.
4053 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4054 * graphite.c (print_global_statistics): Adjust dumping.
4055 (print_graphite_scop_statistics): Likewise.
4056 (print_graphite_statistics): Do not dump loops here.
4057 (graphite_transform_loops): But here.
4059 2017-10-20 Nicolas Roche <roche@adacore.com>
4061 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4062 * configure: Regenerate.
4064 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4067 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4068 functions when optimizing replace GIMPLE_RETURN stmts with
4069 calls to __builtin_unreachable ().
4072 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4073 for -fsanitize=thread link of executables.
4074 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4075 link of executables.
4078 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4080 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4081 define_insns for logical vector shifts to use VI248_AVX512BW
4082 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4083 condition, useless isa and prefix attributes. Change the first
4084 2 of these define_insns to ...
4085 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4086 define_insn for avx512vl.
4087 (<shift_insn><mode>3): ... and this, new define_insn without
4088 masking for non-avx512vl.
4091 * config/i386/sse.md (*andnot<mode>3,
4092 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4093 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4094 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4095 not applied use empty suffix even for TARGET_AVX512VL.
4096 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4097 is applied, supply evex,evex or evex,evex,evex instead of just
4100 2017-10-20 Julia Koval <julia.koval@intel.com>
4102 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4103 (OPTION_MASK_ISA_GFNI_UNSET): New.
4104 (ix86_handle_option): Handle OPT_mgfni.
4105 * config/i386/cpuid.h (bit_GFNI): New.
4106 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4107 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4108 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4109 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4110 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4111 * config/i386/i386.opt: Add mgfni.
4113 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4115 * config/msp430/msp430.c (msp430_option_override): Disable
4116 -fdelete-null-pointer-checks.
4117 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4119 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4121 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4122 of x87 and SSE instructions.
4124 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4126 * asan.c (create_cond_insert_point): Do not update edge count.
4127 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4128 (afdo_propagate_circuit): Likewise.
4129 (afdo_calculate_branch_prob): Likewise.
4130 (afdo_annotate_cfg): Likewise.
4131 * basic-block.h (struct edge_def): Remove count.
4132 (edge_def::count): New accessor.
4133 * bb-reorder.c (rotate_loop): Update.
4134 (find_traces_1_round): Update.
4135 (connect_traces): Update.
4136 (sanitize_hot_paths): Update.
4137 * cfg.c (unchecked_make_edge): Update.
4138 (make_single_succ_edge): Update.
4139 (check_bb_profile): Update.
4140 (dump_edge_info): Update.
4141 (update_bb_profile_for_threading): Update.
4142 (scale_bbs_frequencies_int): Update.
4143 (scale_bbs_frequencies_gcov_type): Update.
4144 (scale_bbs_frequencies_profile_count): Update.
4145 (scale_bbs_frequencies): Update.
4146 * cfganal.c (connect_infinite_loops_to_exit): Update.
4147 * cfgbuild.c (compute_outgoing_frequencies): Update.
4148 (find_many_sub_basic_blocks): Update.
4149 * cfgcleanup.c (try_forward_edges): Update.
4150 (try_crossjump_to_edge): Update
4151 * cfgexpand.c (expand_gimple_cond): Update
4152 (expand_gimple_tailcall): Update
4153 (construct_exit_block): Update
4154 * cfghooks.c (verify_flow_info): Update
4155 (redirect_edge_succ_nodup): Update
4156 (split_edge): Update
4157 (make_forwarder_block): Update
4158 (duplicate_block): Update
4159 (account_profile_record): Update
4160 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4161 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4162 * cfgloopmanip.c (scale_loop_profile): Update.
4164 (lv_adjust_loop_entry_edge): Update.
4165 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4166 (force_nonfallthru_and_redirect): Update.
4167 (purge_dead_edges): Update.
4168 (rtl_flow_call_edges_add): Update.
4169 * cgraphunit.c (init_lowered_empty_function): Update.
4170 (cgraph_node::expand_thunk): Update.
4171 * gimple-pretty-print.c (dump_probability): Update.
4172 (dump_edge_probability): Update.
4173 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4174 * haifa-sched.c (sched_create_recovery_edges): Update.
4175 * hsa-gen.c (convert_switch_statements): Update.
4176 * ifcvt.c (dead_or_predicable): Update.
4177 * ipa-inline-transform.c (inline_transform): Update.
4178 * ipa-split.c (split_function): Update.
4179 * ipa-utils.c (ipa_merge_profiles): Update.
4180 * loop-doloop.c (add_test): Update.
4181 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4182 * lto-streamer-in.c (input_cfg): Update.
4183 (input_function): Update.
4184 * lto-streamer-out.c (output_cfg): Update.
4185 * modulo-sched.c (sms_schedule): Update.
4186 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4187 * predict.c (maybe_hot_edge_p): Update.
4188 (unlikely_executed_edge_p): Update.
4189 (probably_never_executed_edge_p): Update.
4190 (dump_prediction): Update.
4191 (drop_profile): Update.
4192 (propagate_unlikely_bbs_forward): Update.
4193 (determine_unlikely_bbs): Update.
4194 (force_edge_cold): Update.
4195 * profile.c (compute_branch_probabilities): Update.
4196 * reg-stack.c (better_edge): Update.
4197 * shrink-wrap.c (handle_simple_exit): Update.
4198 * tracer.c (better_p): Update.
4199 * trans-mem.c (expand_transaction): Update.
4200 (split_bb_make_tm_edge): Update.
4201 * tree-call-cdce.c: Update.
4202 * tree-cfg.c (gimple_find_sub_bbs): Update.
4203 (gimple_split_edge): Update.
4204 (gimple_duplicate_sese_region): Update.
4205 (gimple_duplicate_sese_tail): Update.
4206 (gimple_flow_call_edges_add): Update.
4207 (insert_cond_bb): Update.
4208 (execute_fixup_cfg): Update.
4209 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4210 * tree-complex.c (expand_complex_div_wide): Update.
4211 * tree-eh.c (lower_resx): Update.
4212 (unsplit_eh): Update.
4213 (cleanup_empty_eh_move_lp): Update.
4214 * tree-inline.c (copy_edges_for_bb): Update.
4215 (freqs_to_counts): Update.
4216 (copy_cfg_body): Update.
4217 * tree-ssa-dce.c (remove_dead_stmt): Update.
4218 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4219 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4220 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4221 (unloop_loops): Update.
4222 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4223 * tree-ssa-loop-split.c (connect_loops): Update.
4224 (split_loop): Update.
4225 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4226 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4227 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4228 * tree-ssa-reassoc.c (branch_fixup): Update.
4229 * tree-ssa-tail-merge.c (replace_block_by): Update.
4230 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4231 (compute_path_counts): Update.
4232 (update_profile): Update.
4233 (recompute_probabilities): Update.
4234 (update_joiner_offpath_counts): Update.
4235 (estimated_freqs_path): Update.
4236 (freqs_to_counts_path): Update.
4237 (clear_counts_path): Update.
4238 (ssa_fix_duplicate_block_edges): Update.
4239 (duplicate_thread_path): Update.
4240 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4241 (case_bit_test_cmp): Update.
4242 (collect_switch_conv_info): Update.
4243 (gen_inbound_check): Update.
4244 (do_jump_if_equal): Update.
4245 (emit_cmp_and_jump_insns): Update.
4246 * tree-tailcall.c (decrease_profile): Update.
4247 (eliminate_tail_call): Update.
4248 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4249 (vect_do_peeling): Update.
4250 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4251 * ubsan.c (ubsan_expand_null_ifn): Update.
4252 (ubsan_expand_ptr_ifn): Update.
4253 * value-prof.c (gimple_divmod_fixed_value): Update.
4254 (gimple_mod_pow2): Update.
4255 (gimple_mod_subtract): Update.
4256 (gimple_ic): Update.
4257 (gimple_stringop_fixed_value): Update.
4259 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4262 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4264 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4266 PR rtl-optimization/82395
4267 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4268 based on non_spilled_static_chain_regno_p.
4270 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4272 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4273 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4274 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4276 2017-10-19 Martin Sebor <msebor@redhat.com>
4278 PR tree-optimization/82596
4279 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4281 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4283 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4284 (get_mem_refs_of_builtin_call): Likewise.
4285 * builtins.c (expand_builtin_apply): Adjust call to
4286 allocate_dynamic_stack_space.
4287 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4288 the third argument to allocate_dynamic_stack_space, otherwise -1.
4289 (expand_builtin): Deal with all alloca variants.
4290 (is_inexpensive_builtin): Likewise.
4291 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4292 * calls.c (special_function_p): Deal with all alloca variants.
4293 (initialize_argument_information): Adjust call to
4294 allocate_dynamic_stack_space.
4295 (expand_call): Likewise.
4296 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4297 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4298 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4299 use it for the stack usage computation.
4300 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4301 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4302 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4303 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4304 (in_loop_p): Remove first argument and useless check.
4305 (pass_walloca::execute): Remove useless test and adjust call to above.
4306 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4307 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4308 (gimplify_call_expr): Deal with all alloca variants.
4309 * hsa-gen.c (gen_hsa_alloca): Likewise.
4310 (gen_hsa_insns_for_call): Likewise.
4311 * ipa-pure-const.c (special_builtin_state): Likewise.
4312 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4313 * tree-object-size.c (alloc_object_size): Likewise.
4314 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4315 (call_may_clobber_ref_p_1): Likewise.
4316 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4317 (ccp_fold_stmt): Likewise.
4318 (optimize_stack_restore): Likewise.
4319 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4320 (mark_all_reaching_defs_necessary_1): Likewise.
4321 (propagate_necessity): Likewise.
4322 (eliminate_unnecessary_stmts): Likewise.
4323 * tree.c (build_common_builtin_nodes): Build
4324 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4325 (build_alloca_call_expr): New function.
4326 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4327 (CASE_BUILT_IN_ALLOCA): Likewise.
4328 (build_alloca_call_expr): Declare.
4329 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4331 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4334 * dwarf2out.c (new_die_raw): New static inline function.
4335 (new_die): Use it to create the DIE.
4336 (add_AT_external_die_ref): Likewise.
4337 (clone_die): Likewise.
4338 (clone_as_declaration): Likewise.
4339 (dwarf2out_vms_debug_main_pointer): Likewise.
4340 (base_type_die): Likewise. Remove early return for corner cases.
4341 Do not call add_pubtype on the DIE here.
4342 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4343 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4344 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4345 native order exists for base types, attach the DIE manually and call
4346 add_pubtype on it. Do not equate a reverse order DIE to the type.
4348 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4350 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4351 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4353 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4355 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4357 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4358 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4359 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4360 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4361 * opts.c (sanitizer_opts): Add builtin.
4362 * ubsan.c (instrument_builtin): New function.
4363 (pass_ubsan::execute): Call it.
4364 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4365 * doc/invoke.texi: Document -fsanitize=builtin.
4367 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4368 builtins, store max (log2 (align), 0) into uchar field instead of
4369 align into uptr field.
4370 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4371 store uchar 0 field instead of uptr 0 field.
4372 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4373 instead of passing one address of struct with 2 locations pass
4374 two addresses of structs with 1 location each.
4375 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4376 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4377 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4378 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4379 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4380 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4381 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4382 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4384 2017-10-19 Martin Liska <mliska@suse.cz>
4387 * file-find.c (remove_prefix): Remove.
4388 * file-find.h (remove_prefix): Likewise.
4389 * gcc-ar.c: Remove smartness of lookup.
4391 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4393 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4394 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4395 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4397 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4400 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4401 (setcc + and to xor + setcc): New peephole2.
4403 2017-10-19 Tom de Vries <tom@codesourcery.com>
4405 * doc/sourcebuild.texi (Test Directives, Variants of
4406 dg-require-support): Add dg-require-stack-size.
4408 2017-10-19 Martin Liska <mliska@suse.cz>
4411 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4412 that have a large alignment.
4413 (gimplify_target_expr): Likewise.
4415 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4417 PR rtl-optimization/82602
4418 * ira.c (rtx_moveable_p): Return false for volatile asm.
4420 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4423 * config/i386/i386-modes.def (CCGZ): New CC mode.
4424 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4425 * config/i386/predicates.md (ix86_comparison_operator):
4427 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4428 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4429 with double-word subtraction.
4430 (put_condition_code): Handle CCGZmode.
4432 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4434 * wide-int.cc (debug (const wide_int &)): New.
4435 (debug (const wide_int *)): New.
4436 (debug (const widest_int &)): New.
4437 (debug (const widest_int *)): New.
4439 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4442 * lra-constraints.c (curr_insn_transform): Use non-input operand
4443 instead of output one for matched reload.
4445 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4447 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4448 (tree-ssa-loop-ivopts.h): New header file.
4449 (struct builtin_info): New fields.
4450 (classify_builtin_1): Compute and record base and offset parts for
4451 memset builtin partition by calling strip_offset.
4452 (offset_cmp, fuse_memset_builtins): New functions.
4453 (finalize_partitions): Fuse adjacent memset partitions by calling
4455 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4456 Expose the interface.
4457 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4459 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4461 PR tree-optimization/82574
4462 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4463 that data reference must be executed exactly once per iteration
4464 against the outermost loop in nest.
4465 (classify_partition): Update call to above function.
4467 2017-10-18 Richard Biener <rguenther@suse.de>
4469 PR tree-optimization/82591
4470 * graphite.c (graphite_transform_loops): Move code gen message
4472 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4473 Here. Handle scop_to_isl_ast failing.
4474 (scop_to_isl_ast): Limit the number of ISL operations.
4476 2017-10-18 Richard Biener <rguenther@suse.de>
4478 * graphite-isl-ast-to-gimple.c
4479 (translate_isl_ast_to_gimple::set_rename): Simplify.
4480 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4481 (graphite_copy_stmts_from_block): ... here.
4482 (copy_bb_and_scalar_dependences): Simplify.
4483 (add_parameters_to_ivs_params): Canonicalize.
4484 (generate_entry_out_of_ssa_copies): Simplify.
4485 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4486 by passing in ISL dimension.
4487 (parameter_index_in_region_1): Rename to ...
4488 (parameter_index_in_region): ... this.
4489 (extract_affine): Adjust assert, pass down parameter index.
4490 (add_param_constraints): Use range-info when available.
4491 (build_scop_context): Adjust.
4492 * sese.c (new_sese_info): Adjust.
4493 (free_sese_info): Likewise.
4494 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4495 Remove unused typedefs.
4496 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4498 2017-10-18 Martin Liska <mliska@suse.cz>
4500 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4502 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4504 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4505 (s390_sched_init): Do not reset s390_sched_state if we entered the
4506 current basic block via a fallthru edge and all others are unlikely.
4508 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4510 * config/s390/s390.c (NUM_SIDES): New variable.
4511 (LONGRUNNING_THRESHOLD): New variable.
4512 (LATENCY_FACTOR): New variable.
4513 (s390_sched_score): Decrease score for long-running instructions on
4515 (s390_sched_variable_issue): Perform bookkeeping for long-running
4518 2017-10-18 Richard Biener <rguenther@suse.de>
4520 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4521 Simplify with removal of the parameter rename map.
4522 (set_rename): Likewise.
4523 (should_copy_to_new_region): Likewise.
4524 (graphite_copy_stmts_from_block): Likewise.
4525 (copy_bb_and_scalar_dependences): Remove initialization of
4526 unused copied_bb_map.
4528 (copy_internal_parameters): Likewise.
4529 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4530 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4531 Use INTEGRAL_TYPE_P.
4532 (parameter_index_in_region_1): Rename to ...
4533 (assign_parameter_index_in_region): ... this. Assert we have
4534 a parameter we handle.
4535 (scan_tree_for_params): Adjust.
4536 * sese.h (parameter_rename_map_t): Remove.
4537 (struct sese_info_t): Remove unused parameter_rename_map and
4538 copied_bb_map members.
4539 * sese.c (new_sese_info): Adjust.
4540 (free_sese_info): Likewise.
4542 2017-10-18 Martin Liska <mliska@suse.cz>
4545 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4546 on an abnormal edge.
4548 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4550 * doc/invoke.texi (ffunction-sections and fdata-sections):
4553 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4555 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4556 the use statement can throw internally.
4558 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4560 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4561 any RTX present on the RHS of a SET.
4562 * compare-elim.c (try_eliminate_compare): Restore comment.
4564 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4566 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4567 may be also called on tcc_type nodes.
4568 * langhooks.c (lhd_tree_size): Likewise.
4570 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4572 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4573 format_warning_at_substring.
4574 (maybe_warn): Convert source_range * param to a location_t. Pass
4575 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4576 (format_directive): Remove code to extract source_ranges and
4577 source_range * in favor of just a location_t.
4578 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4580 * substring-locations.c (format_warning_va): Convert
4581 source_range * param to a location_t.
4582 (format_warning_at_substring): Likewise.
4583 * substring-locations.h (format_warning_va): Likewise.
4584 (format_warning_at_substring): Likewise.
4586 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4588 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4590 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4591 and scatter/gather ops.
4593 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4594 vec_gather_load and vec_scatter_store.
4595 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4596 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4597 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4598 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4599 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4600 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4602 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4604 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4605 Detect FTST instruction and handle its register pops. Only pop
4606 second operand if can_pop_second_op is true.
4607 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4608 set can_pop_second_op to false in the compare_for_stack_reg call.
4610 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4611 output_fp_compare for stack register operands.
4612 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4613 instructions here. Do not emit stack register pops here. Assert
4614 that FCOMPP pops next to top stack register. Rewrite function.
4616 2017-10-17 Nathan Sidwell <nathan@acm.org>
4619 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4620 use DECL_ASSEMBLER_NAME_RAW.
4623 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4626 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4627 Wilco Dijkstra <wilco.dijkstra@arm.com>
4629 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4630 converstion to Pmode from the buf_addr.
4632 2017-10-17 Richard Biener <rguenther@suse.de>
4634 * graphite-dependences.c (scop_get_reads_and_writes): Change
4635 output parameters to references.
4637 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4639 PR 71026/tree-optimization
4640 * fold-const.c (distribute_real_division): Removed.
4641 (fold_binary_loc): Remove calls to distribute_real_divison.
4643 2017-10-17 Richard Biener <rguenther@suse.de>
4645 * graphite-scop-detection.c
4646 (scop_detection::stmt_has_simple_data_refs_p): Always use
4647 the full nest as region.
4648 (try_generate_gimple_bb): Likewise.
4649 * sese.c (scalar_evolution_in_region): Simplify now that
4650 SCEV can handle instantiation in regions.
4651 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4652 in the non-loop part of a function if requested.
4654 2017-10-17 Richard Biener <rguenther@suse.de>
4656 PR tree-optimization/82563
4657 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4659 (graphite_regenerate_ast_isl): Call it.
4660 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4662 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4664 PR tree-optimization/82549
4665 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4666 Formatting fixes. Instead of calling make_bit_field_ref with negative
4669 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4671 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4672 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4673 _mm_maskz_reduce_ss): New.
4674 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4675 __builtin_ia32_reducess_mask): Ditto..
4676 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4677 * config/i386/sse.md (reduces<mode>): Renamed to ...
4678 (reduces<mode><mask_scalar_name>): ... this.
4679 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4681 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4682 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4684 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4686 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4687 * selftest-run-tests.c (selftest::run_tests): Call
4688 selftest::unique_ptr_tests_cc_tests.
4689 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4690 * unique-ptr-tests.cc: New file.
4692 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4695 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4697 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4698 (make_hard_regno_born, make_hard_regno_dead): Update
4699 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4701 2017-10-16 Jeff Law <law@redhat.com>
4703 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4705 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4707 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4709 2017-10-16 Olivier Hainque <hainque@adacore.com>
4711 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4712 with_cpu if we were configured for an e500v2 target cpu name.
4714 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4716 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4717 * doc/invoke.texi: Document +nodsp as a valid extension for
4720 2017-10-16 Martin Liska <mliska@suse.cz>
4722 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4723 (test_set_range): Likewise.
4724 (test_range_functions): Rename to ...
4725 (test_bit_in_range): ... this.
4726 (sbitmap_c_tests): Add new test.
4728 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4730 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4732 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4733 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4735 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4737 * config/aarch64/aarch64-builtins.c
4738 (aarch64_types_quadopu_lane_qualifiers): New.
4739 (TYPES_QUADOPU_LANE): New.
4740 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4741 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4742 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4743 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4744 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4745 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4746 (Vdottype, DOTPROD): New.
4747 (sur): Add SDOT and UDOT.
4749 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4751 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4752 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4753 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4755 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4756 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4757 Enable TARGET_DOTPROD.
4758 (cortex-a75.cortex-a55): Likewise.
4759 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4761 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4763 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4764 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4766 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4768 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4769 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4770 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4771 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4772 * config/arm/types.md (neon_dot, neon_dot_q): New.
4773 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4775 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4777 * config/arm/arm.h (TARGET_DOTPROD): New.
4778 * config/arm/arm.c (arm_arch_dotprod): New.
4779 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4780 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4781 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4782 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4783 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4784 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4785 * doc/invoke.texi (armv8.2-a): Document dotprod
4787 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4789 * i386.c (ix86_vec_cost): New function.
4790 (ix86_rtx_costs): Handle vector operations better.
4791 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4792 * x86-tune-costs.h: Add new costs to all tables.
4794 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4796 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4798 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4799 divsd, sqrtss and sqrtsd
4800 * x86-tune-costs.h: Add new entries to all costs.
4801 (znver1_cost): Fix to match real instruction latencies.
4803 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4804 Michael Collison <michael.collison@arm.com>
4806 * compare-elim.c: Include emit-rtl.h.
4807 (can_merge_compare_into_arith): New function.
4808 (try_validate_parallel): Likewise.
4809 (try_merge_compare): Likewise.
4810 (try_eliminate_compare): Call the above when no previous clobber
4812 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4815 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4819 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4820 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4824 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4825 to be any operand_equal_p operands. For & (B - 1) require
4826 B to be power of 2. Recognize
4827 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4829 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4832 * optabs.c (expand_memory_blockage): Fix call of
4833 targetm.have_memory_blockage.
4835 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4838 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4839 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4840 objects to extra_objs instead of overwriting it.
4842 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4844 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4845 Use any_fp_register_operand as operand[3] predicate. Simplify
4846 equality test for operands[2] and operands[4] memory location.
4847 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4848 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4849 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4850 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4851 any_fp_register_operand as operand[1] predicate. Simplify
4852 equality test for operands[0] and operands[3] memory location.
4853 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4854 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4855 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4857 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4859 * target-insns.def: Add memory_blockage.
4860 * optabs.c (expand_memory_blockage): New function.
4861 (expand_asm_memory_barrier): Rename ...
4862 (expand_asm_memory_blockage): ... to this.
4863 (expand_mem_thread_fence): Call expand_memory_blockage
4864 instead of expand_asm_memory_barrier.
4865 (expand_mem_singnal_fence): Ditto.
4866 (expand_atomic_load): Ditto.
4867 (expand_atomic_store): Ditto.
4868 * doc/md.texi (Standard Pattern Names For Generation):
4869 Document memory_blockage instruction pattern.
4871 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4873 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4874 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4875 * config/rl78/rl78.md: New define_expand "adddi3".
4877 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4879 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4882 2017-10-13 Jeff Law <law@redhat.com>
4884 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4886 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4889 * internal-fn.c (expand_mul_overflow): If both operands have
4890 the same highpart of -1 or 0 and the topmost bit of lowpart
4891 is different, overflow is if res <= 0 rather than res < 0.
4893 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4895 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4896 TARGET_P9_VECTOR code for unaligned_load case.
4898 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4900 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4902 2017-10-13 Nathan Sidwell <nathan@acm.org>
4904 * tree-core.h (tree_contains_struct): Make bool.
4905 * tree.c (tree_contains_struct): Likewise.
4906 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4907 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4908 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4909 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4910 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4912 2017-10-13 Richard Biener <rguenther@suse.de>
4914 * graphite-isl-ast-to-gimple.c
4915 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4916 parameters and dominance check.
4917 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4918 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4919 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4920 Do not update SSA form here or do intermediate IL verification.
4921 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4922 (graphite_initialize): Remove check on the number of loops in
4923 the function and inline into graphite_transform_loops.
4924 (graphite_finalize): Inline into graphite_transform_loops.
4925 (graphite_transform_loops): Perform SSA update and IL verification
4927 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4929 2017-10-13 Richard Biener <rguenther@suse.de>
4931 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4932 graphite_expression_type_precision): Avoid global constructor
4934 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4935 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4936 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4937 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4938 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4939 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4941 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4944 * config/i386/i386.h (ix86_red_zone_size): New.
4945 * config/i386/i386.md (push peephole2s): Replace
4946 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4948 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4949 Alan Hayward <alan.hayward@arm.com>
4950 David Sherwood <david.sherwood@arm.com>
4952 * combine.c (can_change_dest_mode): Reject changes in
4953 REGMODE_NATURAL_SIZE.
4955 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4956 Alan Hayward <alan.hayward@arm.com>
4957 David Sherwood <david.sherwood@arm.com>
4959 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4960 (expand_debug_source_expr): Likewise.
4961 * combine.c (combine_simplify_rtx): Likewise.
4962 * cse.c (fold_rtx): Likewise.
4963 * fwprop.c (canonicalize_address): Likewise.
4964 * targhooks.c (default_shift_truncation_mask): Likewise.
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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4971 (widened_mode): Likewise.
4972 (expand_unop): Likewise.
4973 * ree.c (transform_ifelse): Likewise.
4974 (merge_def_and_ext): Likewise.
4975 (combine_reaching_defs): Likewise.
4976 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4978 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4979 Alan Hayward <alan.hayward@arm.com>
4980 David Sherwood <david.sherwood@arm.com>
4982 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4983 * combine.c (gen_lowpart_for_combine): Likewise.
4984 * dwarf2out.c (rtl_for_decl_location): Likewise.
4985 * final.c (alter_subreg): Likewise.
4986 * rtlhooks.c (gen_lowpart_general): Likewise.
4987 (gen_lowpart_if_possible): 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 * calls.c (expand_call): Use subreg_lowpart_offset.
4994 * cse.c (cse_insn): Likewise.
4995 * regcprop.c (copy_value): Likewise.
4996 (copyprop_hardreg_forward_1): Likewise.
4998 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5001 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5002 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5003 =Q constraints to +Q and into insn condition add check
5004 that operands[0] and operands[1] are equal.
5005 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5006 =Q constraints to +Q and into insn condition add check
5007 that operands[0] is equal to either operands[1] or operands[2].
5010 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5011 instead of handling MINUS_EXPR twice (once for each argument),
5012 canonicalize operand order and handle just once, use rtype where
5013 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5016 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5017 any values of __C while still being pattern recognizable as a simple
5020 2017-10-13 Richard Biener <rguenther@suse.de>
5022 PR tree-optimization/82451
5024 2017-10-02 Richard Biener <rguenther@suse.de>
5026 PR tree-optimization/82355
5027 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5028 a mapping for the enclosing loop but avoid generating one for
5030 (copy_bb_and_scalar_dependences): Remove premature codegen
5031 error on PHIs in blocks duplicated into multiple places.
5032 * graphite-scop-detection.c
5033 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5034 in the region use it as loop and nest to analyze the DR in.
5035 (try_generate_gimple_bb): Likewise.
5036 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5037 (add_loop_constraints): For blocks in a loop not in the region
5038 create a dimension with a single iteration.
5039 * sese.h (gbb_loop_at_index): Remove assert.
5041 * cfgloop.c (loop_preheader_edge): For the loop tree root
5042 return the single successor of the entry block.
5043 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5044 Reset the SCEV hashtable and niters.
5045 * graphite-scop-detection.c
5046 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5047 assert that we only have POLYNOMIAL_CHREC that vary in loops
5048 contained in the region.
5049 (scop_detection::graphite_can_represent_expr): Adjust.
5050 (scop_detection::stmt_has_simple_data_refs_p): For loops
5051 not in the region set loop to NULL. The nest is now the
5052 entry edge to the region.
5053 (try_generate_gimple_bb): Likewise.
5054 * sese.c (scalar_evolution_in_region): Adjust for
5055 instantiate_scev change.
5056 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5057 Make nest parameter the edge into the region.
5058 (create_data_ref): Likewise.
5059 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5060 entry edge into a region and adjust instantiate_scev calls.
5061 (create_data_ref): Likewise.
5062 (graphite_find_data_references_in_stmt): Likewise.
5063 (find_data_references_in_stmt): Pass the loop preheader edge
5064 from the nest argument.
5065 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5066 parameter the edge into the region.
5067 (instantiate_parameters): Use the loop preheader edge as entry.
5068 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5070 (get_instantiated_value_entry): Make instantiate_below parameter
5071 the edge into the region.
5072 (instantiate_scev_name): Likewise. Adjust dominance checks,
5073 when we cannot use loop-based instantiation instantiate by
5074 walking use-def chains.
5075 (instantiate_scev_poly): Adjust.
5076 (instantiate_scev_binary): Likewise.
5077 (instantiate_scev_convert): Likewise.
5078 (instantiate_scev_not): Likewise.
5079 (instantiate_array_ref): Remove.
5080 (instantiate_scev_3): Likewise.
5081 (instantiate_scev_2): Likewise.
5082 (instantiate_scev_1): Likewise.
5083 (instantiate_scev_r): Do not blindly handle N-operand trees.
5084 Do not instantiate array-refs. Handle all constants and invariants.
5085 (instantiate_scev): Make instantiate_below parameter
5086 the edge into the region.
5087 (resolve_mixers): Use the loop preheader edge for the region
5088 parameter to instantiate_scev_r.
5089 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5091 2017-10-13 Richard Biener <rguenther@suse.de>
5093 PR tree-optimization/82525
5094 * graphite-isl-ast-to-gimple.c
5095 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5097 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5098 Fail code generation when we cannot represent the isl integer.
5099 (binary_op_to_tree): Elide modulo operations that are no-ops
5100 in the type we code generate. Remove now superfluous code
5103 2017-10-13 Richard Biener <rguenther@suse.de>
5105 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5106 (scop_detection::harmful_loop_in_region): Remove premature
5107 IV type restriction.
5108 (scop_detection::graphite_can_represent_scev): We can handle
5109 pointer IVs just fine.
5111 2017-10-13 Alan Modra <amodra@gmail.com>
5113 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5114 "Clobbers and Scratch Registers". Add paragraph on
5115 alternative to clobbers for scratch registers and OpenBLAS
5118 2017-10-13 Alan Modra <amodra@gmail.com>
5120 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5121 example of a memory input for a string of known length. Move
5122 commentary out of table. Add a number of new examples
5123 covering array memory inputs.
5125 2017-10-12 Martin Liska <mliska@suse.cz>
5127 PR tree-optimization/82493
5128 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5129 (test_range_functions): New function.
5130 (sbitmap_c_tests): Likewise.
5131 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5132 * selftest.h (sbitmap_c_tests): New function.
5134 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5136 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5138 * config/rs6000/amo.h: Fix spacing issue.
5140 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5143 * config/i386/i386.md (*ashl<mode>3_mask_1,
5144 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5145 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5148 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5150 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5151 (profile_probability): Set max_probability
5152 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5154 * profile-count.c (profile_probability::differs_from_p): Do not
5155 rely on max_probaiblity == 10000
5157 2017-10-12 Jeff Law <law@redhat.com>
5159 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5162 2017-10-12 Martin Sebor <msebor@redhat.com>
5166 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5167 (handle_alias_pairs): Call it.
5168 * common.opt (-Wattribute-alias): New option.
5169 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5170 * doc/invoke.texi (-Wattribute-alias): Document.
5172 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5175 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5177 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5179 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5180 (make_hard_regno_born, make_hard_regno_dead): Update
5181 bb_killed_pseudos and bb_gen_pseudos.
5183 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5185 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5187 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5189 * config/alpha/alpha.c (alpha_split_conditional_move):
5190 Use std::swap instead of manually swapping.
5191 (alpha_stdarg_optimize_hook): Ditto.
5192 (alpha_canonicalize_comparison): Ditto.
5194 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5196 * tree-loop-distribution.c (struct builtin_info): New struct.
5197 (struct partition): Refactor fields into struct builtin_info.
5198 (partition_free): Free struct builtin_info.
5199 (build_size_arg_loc, build_addr_arg_loc): Delete.
5200 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5201 information from struct builtin_info.
5202 (find_single_drs): New function refactored from classify_partition.
5203 Also moved builtin validity checks to this function.
5204 (compute_access_range, alloc_builtin): New functions.
5205 (classify_builtin_st, classify_builtin_ldst): New functions.
5206 (classify_partition): Refactor code into functions find_single_drs,
5207 classify_builtin_st and classify_builtin_ldst.
5208 (distribute_loop): Don't do runtime alias check when distributing
5210 (find_seed_stmts_for_distribution): New function.
5211 (pass_loop_distribution::execute): Refactor code finding seed
5212 stmts into above function. Support distribution for the innermost
5213 two-level loop nest. Adjust dump information.
5215 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5217 * tree-loop-distribution.c: Adjust the general comment.
5218 (NUM_PARTITION_THRESHOLD): New macro.
5219 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5220 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5221 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5222 in call to build_partition_graph.
5223 (finalize_partitions): New parameter. Make loop distribution more
5224 conservative by fusing more partitions.
5225 (distribute_loop): Don't do runtime alias check in case of loop nest
5227 (find_seed_stmts_for_distribution): New function.
5228 (prepare_perfect_loop_nest): New function.
5229 (pass_loop_distribution::execute): Refactor code finding seed stmts
5230 and loop nest into above functions. Support loop nest distribution.
5231 Adjust dump information accordingly.
5233 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5235 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5236 and set PTYPE_SEQUENTIAL for merged partition.
5238 2017-10-12 Richard Biener <rguenther@suse.de>
5240 PR tree-optimization/69728
5242 2017-09-19 Richard Biener <rguenther@suse.de>
5244 PR tree-optimization/69728
5245 * graphite-sese-to-poly.c (schedule_error): New global.
5246 (add_loop_schedule): Handle empty domain by failing the
5248 (build_original_schedule): Handle schedule_error.
5250 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5251 domain by returning an unchanged schedule.
5253 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5255 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5256 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5258 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5260 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5263 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5266 * expr.c (store_field): Don't optimize away bitsize == 0 store
5267 from CALL_EXPR with addressable return type.
5269 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5271 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5272 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5273 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5274 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5275 TARGET_ISEL instead of TARGET_ISEL<sel>.
5277 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5279 * config/rs6000/rs6000.c
5280 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5282 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5284 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5286 (reg_or_cint_operand): Fix comment.
5287 (reg_or_zero_operand): New predicate.
5288 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5289 * config/rs6000/rs6000.c (output_isel): Delete.
5290 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5291 instead of reg_or_cint_operand. Output instruction directly (not via
5293 (isel_unsigned_<mode>): Ditto.
5294 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5295 gpc_reg_operand. Add an instruction alternative for this. Output
5296 instruction directly.
5297 (*isel_reversed_unsigned_<mode>): Ditto.
5299 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5301 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5302 (TARGET_CANONICALIZE_COMPARISON): Define.
5304 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5307 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5308 Check whether the dest is REG before adding REG_EQUIV note.
5310 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5313 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5315 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5316 (make_hard_regno_born, make_hard_regno_dead): Update
5317 bb_killed_pseudos and bb_gen_pseudos.
5319 2017-10-11 Nathan Sidwell <nathan@acm.org>
5321 * incpath.h (enum incpath_kind): Name enum, prefix values.
5322 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5323 * incpath.c (heads, tails): Use INC_MAX.
5324 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5325 (merge_include_chains, split_quote_chain,
5326 register_include_chains): Update incpath_kind names.
5327 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5328 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5330 (add_framework_path, darwin_register_objc_includes): Likewise.
5331 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5333 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5335 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5336 Do not use float_operator operator predicate.
5337 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5338 * config/i386/predicates.md (float_operator): Remove predicate.
5340 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5342 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5343 (*jccxf_i387): Ditto.
5344 (*jcc<mode>_i387): Ditto.
5345 (*jccu<mode>_i387): Ditto.
5346 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5347 (*jcc_*_i387 splitters): Remove.
5348 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5349 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5350 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5353 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5355 * profile-count.h (slow_safe_scale_64bit): New function.
5356 (safe_scale_64bit): New inline.
5357 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5358 * profile-count.c: Include wide-int.h
5359 (slow_safe_scale_64bit): New.
5361 2017-10-11 Nathan Sidwell <nathan@acm.org>
5363 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5364 HAS_DECL_ASSEMBLER_NAME_P.
5365 * gimple-expr.c (gimple_decl_printable_name: Check
5366 HAS_DECL_ASSEMBLER_NAME_P too.
5367 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5368 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5369 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5370 * passes.c (rest_of_decl_compilation): Check
5371 HAS_DECL_ASSEMBLER_NAME_P too.
5372 * recog.c (verify_changes): Likewise.
5373 * tree-pretty-print.c (dump_decl_name): Likewise.
5374 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5376 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5377 (SET_DECL_ASSEMBLER_NAME): Use it.
5378 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5379 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5380 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5382 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5384 * config.gcc (i386, x86_64): Add extra objects.
5385 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5386 (ix86_min_insn_size): Declare.
5387 (ix86_issue_rate): Declare.
5388 (ix86_adjust_cost): Declare.
5389 (ia32_multipass_dfa_lookahead): Declare.
5390 (ix86_macro_fusion_p): Declare.
5391 (ix86_macro_fusion_pair_p): Declare.
5392 (ix86_bd_has_dispatch): Declare.
5393 (ix86_bd_do_dispatch): Declare.
5394 (ix86_core2i7_init_hooks): Declare.
5395 (ix86_atom_sched_reorder): Declare.
5396 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5397 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5398 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5399 (rip_relative_addr_p): Rename to ...
5400 (ix86_rip_relative_addr_p): ... this one; export.
5401 (memory_address_length): Update.
5402 (ix86_issue_rate): Move to x86-tune-sched.c.
5403 (ix86_flags_dependent): Move to x86-tune-sched.c.
5404 (ix86_agi_dependent): Move to x86-tune-sched.c.
5405 (exact_dependency_1): Move to x86-tune-sched.c.
5406 (exact_store_load_dependency): Move to x86-tune-sched.c.
5407 (ix86_adjust_cost): Move to x86-tune-sched.c.
5408 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5409 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5410 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5411 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5412 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5413 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5414 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5415 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5416 (min_insn_size): Rename to ...
5417 (ix86_min_insn_size): ... this one; export.
5418 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5419 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5420 (core2i7_first_cycle_multipass_backtrack): Move to
5421 x86-tune-sched-core.c.
5422 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5423 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5424 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5425 (ix86_avoid_jump_mispredicts): Update.
5426 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5427 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5428 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5429 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5430 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5431 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5432 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5433 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5434 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5435 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5436 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5437 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5438 (BIG): Move to ix86-tune-sched-bd.c.
5439 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5440 (enum insn_path): Move to ix86-tune-sched-bd.c.
5441 (get_mem_group): Move to ix86-tune-sched-bd.c.
5442 (is_cmp): Move to ix86-tune-sched-bd.c.
5443 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5444 (is_branch): Move to ix86-tune-sched-bd.c.
5445 (is_prefetch): Move to ix86-tune-sched-bd.c.
5446 (init_window): Move to ix86-tune-sched-bd.c.
5447 (allocate_window): Move to ix86-tune-sched-bd.c.
5448 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5449 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5450 (process_end_window): Move to ix86-tune-sched-bd.c.
5451 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5452 (find_constant): Move to ix86-tune-sched-bd.c.
5453 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5454 (has_immediate): Move to ix86-tune-sched-bd.c.
5455 (get_insn_path): Move to ix86-tune-sched-bd.c.
5456 (get_insn_group): Move to ix86-tune-sched-bd.c.
5457 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5458 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5459 (add_insn_window): Move to ix86-tune-sched-bd.c.
5460 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5461 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5462 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5463 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5464 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5465 (do_dispatch): Move to ix86-tune-sched-bd.c.
5466 (has_dispatch): Move to ix86-tune-sched-bd.c.
5467 * config/i386/t-i386: Add new object files.
5468 * config/i386/x86-tune-costs.h: New file.
5469 * config/i386/x86-tune-sched-atom.c: New file.
5470 * config/i386/x86-tune-sched-bd.c: New file.
5471 * config/i386/x86-tune-sched-core.c: New file.
5472 * config/i386/x86-tune-sched.c: New file.
5474 2017-10-11 Liu Hao <lh_mouse@126.com>
5476 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5477 the generic version below instead.
5478 (should_colorize): Recognize Windows consoles as terminals
5480 * pretty-print.c [__MINGW32__] (write_all): New function.
5481 [__MINGW32__] (find_esc_head): Likewise.
5482 [__MINGW32__] (find_esc_terminator): Likewise.
5483 [__MINGW32__] (eat_esc_sequence): Likewise.
5484 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5486 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5489 2017-10-11 Richard Biener <rguenther@suse.de>
5491 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5492 Properly call analyze_scalar_evolution with the loop of the stmt.
5494 2017-10-11 Richard Biener <rguenther@suse.de>
5496 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5497 * tree-core.h (tree_base): Add chrec_var union member.
5498 * tree.h (CHREC_VAR): Remove.
5499 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5500 * tree-chrec.h (build_polynomial_chrec): Adjust.
5501 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5502 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5504 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5506 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5507 * match.pd: ... here.
5508 ((T) X == (T) Y): Relax condition.
5510 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5512 PR tree-optimization/82472
5513 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5515 (break_alias_scc_partitions): Update postorder number.
5517 2017-10-11 Martin Liska <mliska@suse.cz>
5520 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5522 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5524 2017-10-11 Martin Liska <mliska@suse.cz>
5526 * print-rtl.c (print_insn): Move declaration of idbuf
5527 to same scope as name.
5529 2017-10-11 Martin Liska <mliska@suse.cz>
5534 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5537 2017-10-11 Martin Liska <mliska@suse.cz>
5540 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5543 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5545 * config.gcc (default_gnu_indirect_function): Default to yes for
5546 arm*-*-linux* with glibc.
5548 2017-10-11 Richard Biener <rguenther@suse.de>
5550 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5551 default-defs and types we do not want to analyze.
5552 (interpret_loop_phi): Replace unreachable code with an assert.
5553 (compute_scalar_evolution_in_loop): Remove and inline ...
5554 (analyze_scalar_evolution_1): ... here, replacing condition with
5555 what makes the intent clearer. Remove handling of cases
5556 get_scalar_evolution now handles.
5558 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5560 PR rtl-optimization/81434
5561 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5562 comment for main loop. In sched_group_found if, also add checks for
5563 pass and min_cost_group.
5565 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5567 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5568 (rs6000_insn_cost): New function.
5569 * config/rs6000/rs6000.md (cost): New attribute.
5571 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5572 H.J. Lu <hongjiu.lu@intel.com>
5576 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5577 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5578 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5579 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5580 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5581 OPTION_MASK_ISA_MMX and some other bit set require both
5582 mmx and the ISAs without the mmx bit.
5583 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5584 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5585 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5586 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5587 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5588 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5589 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5590 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5591 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5592 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5593 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5594 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5595 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5596 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5597 Add OPTION_MASK_ISA_MMX.
5599 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5601 * config.gcc (armv7*-*-freebsd*): New target.
5602 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5604 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5606 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5607 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5610 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5612 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5613 a template parameter.
5614 (WIDE_INT_REF_FOR): Update accordingly.
5615 * tree.h (wi::int_traits <const_tree>): Delete.
5616 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5617 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5618 (wi::tree_to_wide_ref): New typedef.
5619 (wi::to_wide): New function.
5620 * calls.c (get_size_range): Use wi::to_wide when operating on
5622 * cgraph.c (cgraph_node::create_thunk): Likewise.
5623 * config/i386/i386.c (ix86_data_alignment): Likewise.
5624 (ix86_local_alignment): Likewise.
5625 * dbxout.c (stabstr_O): Likewise.
5626 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5627 * expr.c (const_vector_from_tree): Likewise.
5628 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5629 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5630 (fold_negate_expr_1, int_const_binop_1, const_binop)
5631 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5632 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5633 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5634 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5635 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5636 (fold_not_const, round_up_loc): Likewise.
5637 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5638 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5639 (alloca_call_type): Likewise.
5640 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5641 * godump.c (go_output_typedef): Likewise.
5642 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5643 * internal-fn.c (get_min_precision): Likewise.
5644 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5645 * ipa-polymorphic-call.c
5646 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5647 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5648 (ipa_modify_call_arguments): Likewise.
5649 * match.pd: Likewise.
5650 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5651 * print-tree.c (print_node_brief, print_node): Likewise.
5652 * stmt.c (expand_case): Likewise.
5653 * stor-layout.c (layout_type): Likewise.
5654 * tree-affine.c (tree_to_aff_combination): Likewise.
5655 * tree-cfg.c (group_case_labels_stmt): Likewise.
5656 * tree-data-ref.c (dr_analyze_indices): Likewise.
5657 (prune_runtime_alias_test_list): Likewise.
5658 * tree-dump.c (dequeue_and_dump): Likewise.
5659 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5660 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5661 * tree-pretty-print.c (dump_generic_node): Likewise.
5662 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5663 (simple_iv_with_niters): Likewise.
5664 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5665 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5666 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5667 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5668 (refine_value_range_using_guard, number_of_iterations_ne_max)
5669 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5670 (get_cst_init_from_scev, record_nonwrapping_iv)
5671 (scev_var_range_cant_overflow): Likewise.
5672 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5673 * tree-ssa-pre.c (compute_avail): Likewise.
5674 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5675 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5676 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5677 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5678 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5679 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5680 (dump_case_nodes, try_switch_expansion): Likewise.
5681 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5682 (vect_do_peeling): Likewise.
5683 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5684 * tree-vect-stmts.c (vectorizable_load): Likewise.
5685 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5686 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5687 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5688 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5689 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5690 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5691 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5692 (evrp_dom_walker::before_dom_children): Likewise.
5693 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5694 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5695 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5696 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5697 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5698 (get_range_pos_neg): Likewise.
5699 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5700 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5701 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5702 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5703 * config/avr/avr.c (avr_fold_builtin): Likewise.
5704 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5705 * config/msp430/msp430.c (msp430_attr): Likewise.
5706 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5707 * config/powerpcspe/powerpcspe-c.c
5708 (altivec_resolve_overloaded_builtin): Likewise.
5709 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5710 (rs6000_expand_ternop_builtin): Likewise.
5711 * config/rs6000/rs6000-c.c
5712 (altivec_resolve_overloaded_builtin): Likewise.
5713 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5714 (rs6000_expand_ternop_builtin): Likewise.
5715 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5717 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5719 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5720 when copying loop nest with only one inner loop.
5722 2017-10-10 Richard Biener <rguenther@suse.de>
5724 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5725 blocks if SCEV is active.
5726 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5728 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5729 (scev_initialize): Assert we are not yet initialized.
5731 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5733 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5734 inner loop's exit stmt by making it always exit the loop, otherwise
5735 we would generate an infinite empty loop.
5737 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5739 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5740 renaming variables in new preheader if it's deleted.
5742 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5744 * tree-loop-distribution.c (struct partition): Remove unused field
5745 loops of the structure.
5746 (partition_alloc, partition_free): Ditto.
5747 (build_rdg_partition_for_vertex): Ditto.
5749 2017-10-09 Jeff Law <law@redhat.com>
5751 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5752 return type to match prototype and documentation.
5754 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5756 * config/rs6000/rs6000.c (processor_costs): Move to ...
5757 * config/rs6000/rs6000.h: ... here.
5758 (rs6000_cost): Declare.
5760 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5762 * except.c (setjmp_fn): New global variable.
5763 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5764 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5765 if DONT_USE_BUILTIN_SETJMP is defined.
5767 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5769 * target.def (insn_cost): New hook.
5770 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5771 * doc/tm.texi: Regenerate.
5772 * rtlanal.c (insn_cost): Use the new hook.
5774 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5776 * combine.c (combine_validate_cost): Compute the new insn_cost,
5777 not just pattern_cost.
5778 (try_combine): Adjust comment.
5780 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5782 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5784 * combine.c (uid_insn_cost): Adjust comment.
5785 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5787 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5788 * dse.c (find_shift_sequence): Ditto.
5789 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5790 (bb_valid_for_noce_process_p): Use pattern_cost.
5791 * rtl.h (insn_rtx_cost): Delete.
5792 (pattern_cost): New prototype.
5793 (insn_cost): New prototype.
5794 * rtlanal.c (insn_rtx_cost): Rename to...
5795 (pattern_cost): ... this.
5798 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5800 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5801 (*jcc<mode>_0_r_i387): Ditto.
5802 (*jccxf_r_i387): Ditto.
5803 (*jcc<mode>_r_i387): Ditto.
5804 (*jccu<mode>_r_i387): Ditto.
5805 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5806 (*jcc): Rename from *jcc_1.
5808 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5810 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5811 deferred rescans after the lvx/stvx recombination pre-pass.
5813 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5815 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5816 memory operation instruction support.
5817 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5818 (rs6000-ibm-aix[789]*): Likewise.
5819 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5820 Document new functions.
5822 2017-10-09 Richard Biener <rguenther@suse.de>
5824 PR tree-optimization/82397
5825 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5826 equality only for semantically equal trees.
5828 2017-10-09 Richard Biener <rguenther@suse.de>
5830 PR tree-optimization/82449
5831 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5832 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5833 allow constant addresses.
5834 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5837 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5839 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5842 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5845 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5848 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5851 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5853 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5856 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5857 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5859 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5861 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5862 (WI_BINARY_PREDICATE_RESULT): Likewise.
5863 (wi::binary_traits::operator_result): New type.
5864 (wi::binary_traits::predicate_result): Likewise.
5865 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5866 (generic_wide_int::operator==, generic_wide_int::operator!=)
5867 (generic_wide_int::operator&, generic_wide_int::and_not)
5868 (generic_wide_int::operator|, generic_wide_int::or_not)
5869 (generic_wide_int::operator^, generic_wide_int::operator+
5870 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5872 (operator~, unary operator-, operator==, operator!=, operator&)
5873 (operator|, operator^, operator+, binary operator-, operator*): New
5875 * expr.c (get_inner_reference): Use wi::bit_and_not.
5876 * fold-const.c (fold_binary_loc): Likewise.
5877 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5878 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5879 (bit_value_binop): Likewise.
5880 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5881 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5882 (extract_range_from_binary_expr_1): Likewise.
5883 (masked_increment): Likewise.
5884 (simplify_bit_ops_using_ranges): Likewise.
5886 2017-10-09 Martin Jambor <mjambor@suse.cz>
5889 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5890 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5891 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5893 (hsa_fixup_mov_insn_type): New function.
5894 (hsa_op_with_type::get_in_type): Use it.
5895 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5896 immediates in an assert.
5897 (hsa_op_with_type::extend_int_to_32bit): New method.
5898 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5899 types. Convert to dest type if necessary.
5900 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5901 (reg_for_gimple_ssa): Pass false as min32int to
5902 hsa_type_for_scalar_tree_type.
5903 (gen_hsa_addr): Fixup type when creating addresable temporary.
5904 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5905 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5906 necessary. Call hsa_fixup_mov_insn_type.
5907 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5908 extend operands and convert to dest type if necessary.
5909 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5910 to dest type if necessary.
5911 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5912 if conversion nt necessary and size matches.
5913 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5914 to dest type if necessary.
5915 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5916 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5918 (gen_hsa_clrsb): Likewise.
5919 (gen_hsa_ffs): Likewise.
5920 (gen_hsa_divmod): Extend operands and convert to dest type if
5922 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5924 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5926 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5927 Remove empty default arguments. Use a brace block as output
5929 (conditional return): Ditto.
5931 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5932 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5934 (group_ending_nop): Ditto.
5935 (doloop_end): Ditto.
5936 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5937 (splitters for those): Ditto.
5939 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5941 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5942 a conditional jump (and the compare for it) so that pc_rtx is the
5944 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5945 for the deleted and renamed ctr<mode>_internal[234] patterns.
5946 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5947 Delete second conditional return pattern.
5948 (ctr<mode>_internal2): Delete this second bdnz pattern.
5949 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5950 (ctr<mode>_internal4): Delete this second bdz pattern.
5952 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5954 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5955 (always_initialized_rtx_for_ssa_name_p): New predicate.
5956 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5957 (finish_out_of_ssa): Free new field of SA.
5958 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5959 * tree-ssa-coalesce.c: Include tree-ssa.h.
5960 (get_parm_default_def_partitions): Remove extern keyword.
5961 (get_undefined_value_partitions): New function.
5962 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5963 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5965 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5967 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5969 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5971 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5973 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5974 for targets that preffer 128bit.
5976 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5978 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5980 2017-10-08 Olivier Hainque <hainque@adacore.com>
5982 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5983 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5985 (thumb_set_return_address): Likewise.
5987 2017-10-08 Olivier Hainque <hainque@adacore.com>
5989 * common/config/arm/arm-common.c (arm_except_unwind_info):
5990 Handle DWARF2_UNWIND_INFO.
5992 2017-10-07 Michael Collison <michael.collison@arm.com>
5994 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5997 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5999 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6000 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6001 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6002 defined, force the creation of a new block for a dispatch label.
6004 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6006 * invoke.texi (Wsuggest-attribute=cold): Document.
6007 * common.opt (Wsuggest-attribute=cold): New
6008 * ipa-pure-const.c (warn_function_cold): New function.
6009 * predict.c (compute_function_frequency): Use it.
6010 * predict.h (warn_function_cold): Declare.
6012 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6014 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6017 2017-10-06 Martin Liska <mliska@suse.cz>
6019 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6020 keyword for member functions.
6021 (struct sanopt_tree_couple): New struct.
6022 (struct sanopt_tree_couple_hash): New function.
6023 (struct sanopt_ctx): Add new hash_map.
6024 (has_dominating_ubsan_ptr_check): New function.
6025 (record_ubsan_ptr_check_stmt): Likewise.
6026 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6027 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6028 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6030 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6033 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6034 aarch64_simd_valid_immediate on CONST_VECTORs.
6035 (aarch64_reg_or_bic_imm): Likewise.
6037 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6039 PR rtl-optimization/82396
6040 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6042 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6044 * graphite-dependences.c (scop_get_reads): Move code to...
6045 (scop_get_must_writes): Move code to...
6046 (scop_get_may_writes): Move code to...
6047 (scop_get_reads_and_writes): ... here.
6048 (scop_get_dependences): Call scop_get_reads_and_writes.
6050 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6052 PR tree-optimization/82434
6053 * fold-const.h (can_native_encode_type_p,
6054 can_native_encode_string_p): Remove.
6055 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6056 don't encode anything, just return what would be otherwise returned.
6057 (native_encode_fixed, native_encode_complex, native_encode_vector):
6059 (native_encode_string): Likewise. Inline by hand
6060 can_native_encode_string_p.
6061 (can_native_encode_type_p): Remove.
6062 (can_native_encode_string_p): Remove.
6063 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6064 STRING_CSTs using can_native_encode_string_p, test all
6065 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6066 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6067 argument from native_encode_expr.
6068 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6069 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6072 2017-10-06 Richard Biener <rguenther@suse.de>
6074 PR tree-optimization/82397
6075 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6076 operand_equal_p but rely on data_ref_compare_tree for detecting
6078 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6079 to match up with dr_group_sort_cmp.
6081 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6084 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6086 * config/s390/s390-builtin-types.def: Regenerate.
6088 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6091 * config/s390/s390-builtin-types.def: Regenerate.
6092 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6093 Change flag from B_VXE to B_VX.
6094 (s390_vec_min_dbl): Remove B_VXE flag.
6096 2017-10-06 Richard Biener <rguenther@suse.de>
6098 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6099 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6100 translate_isl_ast_to_gimple::is_valid_rename,
6101 translate_isl_ast_to_gimple::get_rename,
6102 translate_isl_ast_to_gimple::get_def_bb_for_const,
6103 translate_isl_ast_to_gimple::get_new_name,
6104 translate_isl_ast_to_gimple::collect_all_ssa_names,
6105 translate_isl_ast_to_gimple::copy_loop_phi_args,
6106 translate_isl_ast_to_gimple::collect_all_ssa_names,
6107 translate_isl_ast_to_gimple::copy_loop_phi_args,
6108 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6109 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6110 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6111 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6112 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6113 translate_isl_ast_to_gimple::copy_cond_phi_args,
6114 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6115 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6116 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6117 translate_isl_ast_to_gimple::rename_uses,
6118 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6119 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6120 (set_rename_for_each_def): Likewise.
6121 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6122 here. Handle rewriting SCEV analyzable uses here.
6123 (copy_bb_and_scalar_dependences): Generate code for PHI
6125 (graphite_regenerate_ast_isl): Adjust.
6126 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6127 (add_write, add_read): New functions.
6128 (build_cross_bb_scalars_def): Use it and simplify.
6129 (build_cross_bb_scalars_use): Likewise.
6130 (graphite_find_cross_bb_scalar_vars): Inline into...
6131 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6132 simulating out-of-SSA. Compute liveout and add dependencies.
6133 (build_scops): Force an empty entry block.
6134 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6136 (sese_build_liveouts): Declare.
6137 (sese_trivially_empty_bb_p): Likewise.
6138 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6139 compute liveout and debug_liveout.
6140 (sese_bad_liveouts_use): Remove.
6141 (sese_reset_debug_liveouts_bb): Likewise.
6142 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6143 (sese_build_liveouts): Build liveout and debug_liveout and store
6145 (new_sese_info): Adjust.
6146 (free_sese_info): Likewise.
6147 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6148 do not build liveout here.
6149 (move_sese_in_condition): Adjust region entry.
6150 (scev_analyzable_p): Match up with chrec_apply requirements.
6151 (sese_trivially_empty_bb_p): New.
6152 * tree-into-ssa.c (get_reaching_def): Properly support generating
6153 default-defs for incremental rewrite of anonymous names.
6155 2017-10-06 Richard Biener <rguenther@suse.de>
6157 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6158 precision do not perform modulo reduction.
6160 2017-10-06 Richard Biener <rguenther@suse.de>
6162 PR tree-optimization/82436
6163 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6164 conservatively choose the vectorization factor when checking
6165 whether we can perform the required load permutation.
6166 (vect_transform_slp_perm_load): Assert when we may not fail.
6168 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6170 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6171 message for incompatible -msdata=* and -mcall-* options.
6173 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6175 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6176 rate for post-reload scheduling.
6178 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6180 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6182 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6184 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6185 to improve monte carlo in scimark.
6187 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6189 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6190 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6191 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6192 pentium4_cost, nocona_cost): Set reassociation width to 1.
6193 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6194 width to 2 for fp operations and 1 otherwise.
6195 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6197 (atom_cost): Set reassociation width to 2.
6198 (slm_cost, generic_cost): Set fp reassociation width
6199 to 2 and 1 otherwise.
6200 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6201 (core_cost): Set fp reassociation width to 4 and vector to 2.
6202 (ix86_reassociation_width): Rewrite using cost table; special case
6203 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6204 and TARGET_AVX128_OPTIMAL.
6205 * config/i386/i386.h (processor_costs): Add
6206 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6207 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6208 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6209 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6210 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6211 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6213 2017-10-05 Nathan Sidwell <nathan@acm.org>
6215 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6217 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6219 * config/arm/arm.c (arm_test_fpu_data): New.
6220 (arm_run_selftests): Call arm_test_fpu_data.
6222 2017-10-04 Nathan Sidwell <nathan@acm.org>
6224 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6226 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6228 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6230 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6231 check type for aarch64_simd_valid_immediate.
6232 (aarch64_output_simd_mov_immediate): Update prototype.
6233 (aarch64_simd_valid_immediate): Update prototype.
6234 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6235 support for ORR-immediate.
6236 (and<mode>3): modified pattern to add support for BIC-immediate.
6237 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6238 now checks for valid immediate for BIC and ORR based on new enum
6240 (aarch64_output_simd_mov_immediate): Function now used to output
6241 BIC/ORR imm as well based on new enum argument.
6242 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6244 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6245 (aarch64_reg_or_bic_imm): Likewise.
6247 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6249 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6250 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6252 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6256 PR rtl-optimization/82396
6257 * haifa-sched.c (autopref_multipass_init): Simplify
6259 (autopref_rank_data): Simplify sort order.
6260 * sched-int.h (autopref_multipass_data_): Remove
6261 multi_mem_insn_p, min_offset and max_offset.
6263 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6265 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6267 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6269 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6272 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6274 * doc/sourcebuild.texi: Document vect_long_mult.
6276 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6278 PR tree-optimization/82413
6279 * fold-const.c (build_range_check): Use widest_int when comparing
6280 the maximum ETYPE value with HIGH.
6282 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6284 PR rtl-optimization/82396
6285 * haifa-sched.c (autopref_multipass_init): Simplify
6287 (autopref_rank_data): Simplify sort order.
6288 * sched-int.h (autopref_multipass_data_): Remove
6289 multi_mem_insn_p, min_offset and max_offset.
6291 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6293 PR tree-optimization/82381
6294 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6295 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6298 PR tree-optimization/82374
6299 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6300 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6301 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6302 current_function_decl to the new decl.
6304 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6306 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6307 helper macro for IEEE float128 hardware built-in functions.
6308 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6310 (TRUNCF128_ODD): Likewise.
6311 (ADDF128_ODD): Likewise.
6312 (SUBF128_ODD): Likewise.
6313 (MULF128_ODD): Likewise.
6314 (DIVF128_ODD): Likewise.
6315 (FMAF128_ODD): Likewise.
6316 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6317 UNSPEC_TRUNC_ROUND_TO_ODD.
6318 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6319 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6320 floating point round to odd instructions.
6321 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6322 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6323 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6324 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6325 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6326 (trunc<mode>sf2_hw): Change the truncate with round to odd
6327 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6328 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6329 to odd hardware instructions.
6330 (sub<mode>3_odd): Likewise.
6331 (mul<mode>3_odd): Likewise.
6332 (div<mode>3_odd): Likewise.
6333 (sqrt<mode>2_odd): Likewise.
6334 (fma<mode>4_odd): Likewise.
6335 (fms<mode>4_odd): Likewise.
6336 (nfma<mode>4_odd): Likewise.
6337 (nfms<mode>4_odd): Likewise.
6338 (trunc<mode>df2_odd): Change the truncate with round to odd
6339 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6341 * doc/extend.texi (PowerPC built-in functions): Update documentation
6342 for existing IEEE float128-bit built-in functions. Add built-in
6343 functions that generate the IEEE 128-bit floating point round to
6346 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6348 PR rtl-optimization/77729
6349 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6350 to (X&(C1&~C2))|C2 transformations.
6352 2017-10-03 Martin Jambor <mjambor@suse.cz>
6354 PR tree-optimization/82363
6355 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6356 mismatch, mark lacc written regardless of racc.
6358 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6360 PR tree-optimization/82381
6361 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6362 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6363 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6364 fallthrough into reassoc_stmt_dominates_stmt_p.
6367 * combine.c (combine_instructions): Don't combine in unreachable
6370 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6374 function to not use the have_cpu variable. Do not set cpu_index,
6375 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6377 (rs6000_valid_attribute_p): Remove duplicate initializations of
6378 old_optimize and func_optimize.
6379 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6380 (rs6000_activate_target_options): Make global.
6381 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6384 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6386 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6387 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6388 Set *poffset to 0 and *psize and *pmax_size to -1 if
6389 *poffset + *psize overflows in HOST_WIDE_INT.
6391 PR tree-optimization/82387
6392 PR tree-optimization/82388
6393 PR tree-optimization/82389
6394 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6395 instead of live_bytes non-NULL.
6397 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6400 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6403 2017-10-02 Richard Biener <rguenther@suse.de>
6405 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6406 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6407 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6409 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6411 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6412 requested precision matches the type's.
6413 * calls.c (alloc_max_size): Calculate the new candidate size as
6414 a widest_int and use wi::to_widest when comparing it with the
6415 current candidate size.
6416 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6417 zero rather than integer_zero_node.
6418 * match.pd: Check for a no-op conversion before using wi::add
6419 rather than after. Use tree_to_uhwi when summing small shift
6420 counts into an unsigned int.
6422 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6423 Alan Hayward <alan.hayward@arm.com>
6424 David Sherwood <david.sherwood@arm.com>
6427 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6428 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6429 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6430 POINTER_AND_FP_REGS.
6432 2017-10-02 Richard Biener <rguenther@suse.de>
6434 PR tree-optimization/82355
6435 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6436 a mapping for the enclosing loop but avoid generating one for
6438 (copy_bb_and_scalar_dependences): Remove premature codegen
6439 error on PHIs in blocks duplicated into multiple places.
6440 * graphite-scop-detection.c
6441 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6442 in the region use it as loop and nest to analyze the DR in.
6443 (try_generate_gimple_bb): Likewise.
6444 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6445 (add_loop_constraints): For blocks in a loop not in the region
6446 create a dimension with a single iteration.
6447 * sese.h (gbb_loop_at_index): Remove assert.
6449 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6451 * omp-expand.c (adjust_context_scope): New function.
6452 (expand_parallel_call): Call adjust_context_scope.
6454 2017-10-01 Jeff Law <law@redhat.com>
6456 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6457 dom_opt_dom_walker class with direct access to private members.
6458 Add comments. Call test_for_singularity.
6459 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6460 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6461 m_dummy_cond anymore.
6462 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6464 (pass_dominator:execute): Build the dummy_cond here and pass it
6465 to the dom_opt_dom_walker ctor.
6466 (test_for_singularity): New function.
6468 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6469 Maya Rashish <coypu@sdf.org>
6471 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6472 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6473 (alpha*-*-netbsd*) Use nbsd_tm_file.
6474 (arm*-*-netbsdelf*) Likewise.
6475 (i[34567]86-*-netbsdelf*) Likewise.
6476 (x86_64-*-netbsd*) Likewise.
6477 (mips*-*-netbsd*) Likewise.
6478 (powerpc-*-netbsd*) Likewise.
6479 (sh*-*-netbsd*) Likewise.
6480 (sparc-*-netbsdelf*) Likewise.
6481 (sparc64-*-netbsd*) Likewise.
6482 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6484 (vax-*-netbsdelf*) Likewise.
6485 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6486 (UINT_FAST8_TYPE) Likewise.
6487 (INT_FAST16_TYPE) Check CHAR_FAST16.
6488 (UINT_FAST16_TYPE) Likewise.
6490 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6493 * config/i386/i386.md
6494 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6495 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6496 *divmodsi4_zext_2): New define_insn_and_split.
6497 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6498 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6499 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6500 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6501 New define_insn_and_split.
6502 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6503 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6504 operands[1] having DImode when mode is SImode.
6506 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6507 always SImode for DIV and MOD in REG_EQUAL notes.
6509 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6512 * match.pd: Fix handling of NaNs in pattern.
6514 2017-09-29 Jeff Law <law@redhat.com>
6516 * sbitmap.c (bitmap_bit_in_range_p): New function.
6517 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6518 * tree-ssa-dse.c (live_bytes_read): New function.
6519 (dse_classify_store): Ignore reads of dead bytes.
6521 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6522 typos and whitespace errors.
6523 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6524 * config/s390/s390.c (s390_emit_prologue): Likewise.
6526 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6529 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6530 with a symbol for LRA.
6532 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6534 PR rtl-optimization/82338
6535 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6537 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6539 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6540 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6541 (qsort_chk): Declare.
6542 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6543 (qsort_chk): New function.
6545 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6547 PR tree-optimization/82337
6548 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6549 phi definition if the PHI result appears in an abnormal PHI.
6550 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6551 basis appears in an abnormal PHI.
6553 2017-09-29 Richard Biener <rguenther@suse.de>
6555 * graphite-isl-ast-to-gimple.c
6556 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6557 (binary_op_to_tree): Use it.
6558 (get_rename_from_scev): Likewise.
6559 (copy_loop_phi_nodes): Likewise.
6560 (copy_bb_and_scalar_dependences): Likewise.
6561 (translate_pending_phi_nodes): Likewise.
6563 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6566 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6567 for movabsq $(i32 << shift), r64.
6569 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6571 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6572 index when encoding %esp as %rsp to avoid 0x67 prefix.
6574 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6576 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6577 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6579 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6581 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6582 Extensions with more than 16 double VFP registers.
6583 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6584 to_clear_mask and all code related to it. Replace the remaining
6585 entry by a sbitmap and adapt code accordingly.
6587 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6589 * brig-builtins.def: Change pure attributes to const.
6591 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6593 * config.gcc (default_gnu_indirect_function): Default to yes for
6594 sparc*-*-linux* with glibc.
6596 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6598 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6599 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6600 when creating .init_array and .fini_array sections with priority
6603 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6606 * config/aarch64/aarch64.c
6607 (aarch64_builtin_support_vector_misalignment): Always return false
6608 when misalignment is unknown.
6610 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6612 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6613 this function to return false if the definition used by the swap
6614 instruction is artificial, or if the memory address from which the
6615 constant value is loaded is not represented by a base address held
6616 in a register or if the base address register is a frame or stack
6617 pointer. Additionally, return false if the base address of the
6618 loaded constant is a SYMBOL_REF but is not considered to be a
6620 (replace_swapped_load_constant): New function.
6621 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6622 loaded constant vector with a load of a swapped constant vector.
6624 2017-09-27 Carl Love <cel@us.ibm.com>
6626 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6627 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6628 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6631 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6633 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6634 first, always call autopref_rank_data otherwise.
6636 2017-09-27 Richard Biener <rguenther@suse.de>
6638 * graphite-scop-detection.c (find_scop_parameters): Move
6639 loop bound handling ...
6640 (gather_bbs::before_dom_children): ... here, avoiding the need
6641 to build scop_info->loop_nest.
6642 (record_loop_in_sese): Remove.
6643 * sese.h (sese_info_t::loop_nest): Remove.
6644 * sese.c (new_sese_info): Do not allocate loop_nest.
6645 (free_sese_info): Do not free loop_nest.
6647 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6650 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6651 lhs from calls if the lhs has addressable type.
6653 2017-09-27 Richard Biener <rguenther@suse.de>
6655 * graphite.h (scop::max_alias_set): New member.
6656 * graphite-scop-detection.c: Remove references to non-existing
6657 --param in comments.
6658 (build_alias_sets): Record the maximum alias set used for drs.
6659 (build_scops): Support zero as unlimited for
6660 --param graphite-max-arrays-per-scop.
6661 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6663 (build_poly_sr_1): ... here. Compute alias set based on the
6664 maximum alias set used for drs rather than
6665 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6667 2017-09-27 Richard Biener <rguenther@suse.de>
6669 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6670 --param loop-block-tile-size=0 to disable tiling.
6672 2017-09-27 Richard Biener <rguenther@suse.de>
6674 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6675 (graphite-max-nb-scop-params): Document special value zero.
6676 * domwalk.h (dom_walker::STOP): New symbolical constant.
6677 (dom_walker::dom_walker): Add optional parameter for bb to
6679 (dom_walker::~dom_walker): Declare.
6680 (dom_walker::before_dom_children): Document STOP return value.
6681 (dom_walker::m_user_bb_to_rpo): New member.
6682 (dom_walker::m_bb_to_rpo): Likewise.
6683 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6684 mapping here if not provided by the user.
6685 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6686 provided by the user.
6687 (dom_walker::STOP): Define.
6688 (dom_walker::walk): Do not compute bb to RPO mapping here.
6689 Support STOP return value from before_dom_children to stop
6691 * graphite-optimize-isl.c (optimize_isl): If the schedule
6692 is the same still generate code if -fgraphite-identity
6693 or -floop-parallelize-all are given.
6694 * graphite-scop-detection.c: Include cfganal.h.
6695 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6697 (gather_bbs::before_dom_children): Return STOP for BBs
6699 (build_scops): Compute bb to RPO mapping and pass it to
6700 the domwalk. Treat --param graphite-max-nb-scop-params=0
6701 as not limiting the number of params.
6702 * graphite.c (graphite_initialize): Remove limit on the
6703 number of basic-blocks in a function.
6704 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6705 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6706 default value of 10.
6708 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6710 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6711 Adjust code to eliminate needing to do the shift right 32-bits
6712 operation after XSCVDPSPN.
6714 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6716 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6717 ((X / Y) != 0 -> X >= Y): Likewise.
6719 2017-09-26 Carl Love <cel@us.ibm.com>
6721 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6722 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6723 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6724 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6725 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6726 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6727 definitions and overloading.
6728 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6729 statement for P9V_BUILTIN_XST_LEN_R.
6730 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6731 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6732 define_expand and define_insn for the instructions and builtins.
6733 * doc/extend.texi: Update the built-in documentation file for the new
6735 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6736 define_insn for the instructions
6738 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6741 * gcc/config/netbsd-protos.h: New file.
6742 * gcc/config/netbsd.c: New file.
6743 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6744 * gcc/config/t-netbsd: New file.
6745 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6746 (tmake_file) Add t-netbsd.
6747 (extra_objs) Add netbsd.o.
6749 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6753 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6755 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6757 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6758 sign extension from a vector register to a GPR by doing a 32-bit
6759 direct move and then an EXTSW.
6760 (extendsi<mode>2 splitter): Likewise.
6761 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6762 right or vector extract after doing XSCVDPSPN. Use
6763 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6765 (movdi_from_sf_zero_ext): Likewise.
6766 (reload_gpr_from_vsxsf): Likewise.
6767 (p8_mfvsrd_4_disf): Delete, no longer used.
6768 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6769 then needing to move the SFmode to a GPR to use the XSCVDPSP
6770 instruction instead of FRSP and XSCVDPSPN.
6771 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6772 it is adjacent to the other XSCVSPDP insns.
6773 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6774 SFmode to be in traditional Altivec registers.
6775 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6776 (vsx_xscvspdpn): Likewise.
6777 (vsx_xscvspdpn_scalar): Likewise.
6779 2017-09-26 Martin Jambor <mjambor@suse.cz>
6781 * tree-sra.c (compare_access_positions): Put integral types first,
6782 stabilize sorting of integral types, remove conditions putting
6783 non-full-precision integers last.
6784 (sort_and_splice_var_accesses): Disable scalarization if a
6785 non-integert would be represented by a non-full-precision integer.
6787 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6789 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6790 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6791 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6792 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6793 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6794 conditionals inside the function instead of around it. Call
6795 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6796 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6798 2017-09-26 Richard Biener <rguenther@suse.de>
6800 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6802 (scop_detection::build_scop_breadth): ... this. Removed.
6803 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6804 (scop_detection::harmful_stmt_in_bb): Likewise.
6805 (scop_detection::graphite_can_represent_stmt): Likewise.
6806 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6807 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6808 (scop_detection::can_represent_loop_1): ... this. Removed.
6809 (scop_detection::harmful_loop_in_region): Simplify after inlining
6810 the above and remove more quadraticness.
6811 (build_scops): Adjust.
6812 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6815 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6818 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6819 REGNO (base) == SP_REG if base is a REG.
6822 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6823 if it is different SSA_NAME.
6824 (optimize_range_tests_cmp_bitwise): New function.
6825 (optimize_range_tests): Call it.
6827 2017-09-26 Richard Biener <rguenther@suse.de>
6829 PR tree-optimization/82321
6830 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6831 for the def being inside the loop.
6833 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6835 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6837 * config/s390/s390-builtins.def: Fix constraint on op4.
6839 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6841 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6842 independent expanders.
6843 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6844 ("vec_ordered", "vec_unordered"): New expanders.
6846 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6848 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6851 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6853 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6854 vec_unpacks_lo_v16qi.
6855 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6859 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6860 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6861 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6863 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6865 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6867 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6869 ("*vec_slb<mode>"): New insn pattern.
6870 ("vec_shr_<mode>"): New expander.
6871 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6872 and force the shift count operand to V16QImode.
6873 ("vec_srb<mode>"): Set shift count mode to V16QI.
6875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6877 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6878 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6879 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6881 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6884 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6886 2017-09-26 Richard Biener <rguenther@suse.de>
6888 PR tree-optimization/82320
6889 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6892 2017-09-25 Jeff Law <law@redhat.com>
6894 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6895 prototype for new argument.
6896 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6897 mostly extracted from rs6000_emit_allocate_stack.
6898 (rs6000_emit_probe_stack_range_stack_clash): New function.
6899 (rs6000_emit_allocate_stack): Call
6900 rs6000_emit_probe_stack_range_stack_clash as needed.
6901 (rs6000_emit_probe_stack_range): Add additional argument
6902 to call to gen_probe_stack_range{si,di}.
6903 (output_probe_stack_range): New.
6904 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6905 (output_probe_stack_range_stack_clash): New.
6906 (rs6000_emit_prologue): Emit notes into dump file as requested.
6907 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6908 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6909 Add additional operand and pass it to output_probe_stack_range.
6911 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6913 PR tree-optimization/82163
6914 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6915 (checking_verify_loop_closed_ssa): New parameter.
6916 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6917 (check_loop_closed_ssa_stmt): Delete.
6918 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6919 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6920 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6923 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6925 * brig-builtins.def: Treat HSAIL barrier builtins as
6926 setjmp/longjump style functions.
6928 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6930 * target.def (constant_alignment): New hook.
6931 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6932 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6933 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6934 * doc/tm.texi: Regenerate.
6935 * targhooks.h (default_constant_alignment): Declare.
6936 (constant_alignment_word_strings): Likewise.
6937 * targhooks.c (default_constant_alignment): New function.
6938 (constant_alignment_word_strings): Likewise.
6939 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6940 instead of CONSTANT_ALIGNMENT.
6941 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6942 (force_const_mem): Likewise.
6943 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6944 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6945 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6946 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6947 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6949 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6950 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6951 constant_alignment_word_strings.
6952 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6953 (CONSTANT_ALIGNMENT): Likewise.
6954 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6955 (arm_constant_alignment): New function.
6956 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6957 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6958 constant_alignment_word_strings.
6959 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6960 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6961 constant_alignment_word_strings.
6962 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6963 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6964 (cris_constant_alignment): New function.
6965 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6966 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6967 (epiphany_constant_alignment): New function.
6968 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6969 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6970 constant_alignment_word_strings.
6971 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6972 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6973 constant_alignment_word_strings.
6974 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6975 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6976 constant_alignment_word_strings.
6977 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6978 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6979 * config/i386/i386.c (ix86_constant_alignment): Make static.
6980 Use the same interface as the target hook.
6981 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6982 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6983 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6984 constant_alignment_word_strings.
6985 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6986 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6987 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6988 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6989 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6990 constant_alignment_word_strings.
6991 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6992 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6993 constant_alignment_word_strings.
6994 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6995 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6996 constant_alignment_word_strings.
6997 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6998 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7000 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7001 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7002 * config/mips/mips.c (mips_constant_alignment): New function.
7003 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7004 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7005 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7006 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7007 (mmix_constant_alignment): Make static. Use the same interface
7009 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7010 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7011 constant_alignment_word_strings.
7012 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7013 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7014 constant_alignment_word_strings.
7015 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7016 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7017 constant_alignment_word_strings.
7018 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7019 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7020 (rs6000_constant_alignment): New function.
7021 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7022 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7023 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7024 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7025 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7026 (rs6000_constant_alignment): New function.
7027 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7028 * config/s390/s390.c (s390_constant_alignment): New function.
7029 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7030 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7031 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7032 constant_alignment_word_strings.
7033 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7034 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7035 (sparc_constant_alignment): New function.
7036 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7037 * config/spu/spu.c (spu_constant_alignment): New function.
7038 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7039 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7040 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7041 constant_alignment_word_strings.
7042 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7043 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7044 constant_alignment_word_strings.
7045 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7046 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7047 constant_alignment_word_strings.
7048 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7049 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7050 (visium_constant_alignment): New function.
7051 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7052 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7053 (xtensa_constant_alignment): New function.
7054 * system.h (CONSTANT_ALIGNMENT): Poison.
7056 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7059 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7060 (rs6000_builtin_valid_without_lhs): New helper function.
7061 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7062 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7064 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7066 * target.h (vec_perm_indices): Use unsigned short rather than
7068 (auto_vec_perm_indices): Likewise.
7069 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7070 Use unsigned int rather than unsigned char.
7071 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7073 2017-09-25 Richard Biener <rguenther@suse.de>
7075 * cfgloop.h (sort_sibling_loops): Declare.
7076 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7077 (sort_sibling_loops): New function sorting the sibling loop list
7079 * graphite.c (graphite_transform_loops): Sort sibling loops.
7081 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7083 * target.def (vec_perm_const_ok): Change sel parameter to
7085 * optabs-query.c (can_vec_perm_p): Update accordingly.
7086 * doc/tm.texi: Regenerate.
7087 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7088 auto_vec_perm_indices and remove separate nelt field.
7089 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7090 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7091 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7092 (aarch64_expand_vec_perm_const): Update accordingly.
7093 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7094 to vec_perm_indices.
7095 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7096 auto_vec_perm_indices and remove separate nelt field.
7097 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7098 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7099 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7101 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7102 to vec_perm_indices.
7103 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7104 sel to vec_perm_indices.
7105 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7106 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7107 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7109 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7112 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7115 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7116 on the FUNCTION_DECL function context if it has a DIE that is a
7119 2017-09-25 Richard Biener <rguenther@suse.de>
7121 PR tree-optimization/82285
7122 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7125 2017-09-25 Tom de Vries <tom@codesourcery.com>
7129 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7132 2017-09-25 Richard Biener <rguenther@suse.de>
7134 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7135 ISL errors other than isl_error_quota happen. Dump if the
7136 schedule is the same.
7137 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7138 errors instead of aborting inside ISL.
7140 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7143 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7144 of libgcc_eh for m64.
7145 * config/i386/darwin64.h: Likewise.
7147 2017-09-25 Richard Biener <rguenther@suse.de>
7150 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7151 attribute for incomplete types nor twice for complete ones.
7153 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7156 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7157 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7160 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7163 * config/i386/i386.opt (mprefer-avx256): Use
7164 ix86_target_flags variable.
7165 * config/i386/i386.c (ix86_target_string): Move
7166 -mprefer-avx256 to flag2_opts.
7168 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7171 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7172 and x != -1 | y != -1 into (x & y) != -1.
7174 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7176 * config.gcc: Add new case statement to set
7177 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7178 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7179 s390x-*-linux* case statements. Added aarch64 to the list of
7180 supported architectures.
7182 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7184 PR tree-optimization/82289
7185 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7186 STMT_VINFO_RELEVANT_P.
7188 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7189 Alan Hayward <alan.hayward@arm.com>
7190 David Sherwood <david.sherwood@arm.com>
7192 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7193 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7194 (extract_range_from_binary_expr_1): Don't call
7195 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7197 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7198 Alan Hayward <alan.hayward@arm.com>
7199 David Sherwood <david.sherwood@arm.com>
7201 * target.def (preferred_vector_alignment): New hook.
7202 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7204 * doc/tm.texi: Regenerate.
7205 * targhooks.h (default_preferred_vector_alignment): Declare.
7206 * targhooks.c (default_preferred_vector_alignment): New function.
7207 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7209 (DR_TARGET_ALIGNMENT): New macro.
7210 (aligned_access_p): Update commentary.
7211 (vect_known_alignment_in_bytes): New function.
7212 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7214 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7215 Calculate the misalignment based on the target alignment rather than
7217 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7218 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7219 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7220 the target alignment, rather than masking the element misalignment
7221 with the number of elements in a vector. Also use the target
7222 alignment when calculating the maximum number of peels.
7223 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7224 instead of TYPE_ALIGN_UNIT.
7225 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7226 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7227 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7228 (vect_create_data_ref_ptr): Likewise.
7229 (vect_setup_realignment): Realign by ANDing with
7230 -DR_TARGET_MISALIGNMENT.
7231 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7232 the number of peels based on DR_TARGET_ALIGNMENT.
7233 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7234 with the guaranteed alignment boundary when deciding whether
7236 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7237 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7238 (ensure_base_align): Remove stmt_info parameter. Get the
7239 target base alignment from DR_TARGET_ALIGNMENT.
7240 (vectorizable_store): Update call accordingly. Interpret
7241 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7243 (vectorizable_load): Likewise.
7245 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7246 Alan Hayward <alan.hayward@arm.com>
7247 David Sherwood <david.sherwood@arm.com>
7249 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7250 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7251 (vect_enhance_data_refs_alignment): Likewise.
7253 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7255 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7256 error. Only quit immediately if parsing is complete.
7257 (BEGIN): Initialize fatal_err and parse_done.
7258 (begin fpu, end fpu): Check number of arguments.
7259 (begin arch, end arch): Likewise.
7260 (begin cpu, end cpu): Likewise.
7261 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7262 (optalias): Likewise.
7264 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7266 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7267 * config/arm/arm-isa.h: Delete. Move definitions to ...
7268 * arm-cpus.in: ... here. Use new feature and fgroup values.
7269 * config/arm/arm.c (arm_option_override): Use lower case for feature
7271 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7272 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7273 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7275 (print_isa_bits_for): New function.
7276 (gen_isa): New function.
7277 (gen_comm_data): Use print_isa_bits_for.
7278 (define feature): New keyword.
7279 (define fgroup): New keyword.
7280 * config/arm/t-arm (TM_H): Remove.
7281 (GTM_H): Add arm-isa.h.
7282 (arm-isa.h): Add rule to generate file.
7283 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7284 case for feature bit names.
7286 2017-09-22 Richard Biener <rguenther@suse.de>
7288 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7290 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7291 print of no dependency loops ...
7292 * graphite.c (graphite_transform_loops): ... here.
7293 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7295 (same_close_phi_node, remove_duplicate_close_phi,
7296 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7297 (canonicalize_loop_closed_ssa): ... here and simplify.
7298 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7299 (optimize_isl): Use dump_printf_loc to tell when we stopped
7300 optimizing because of an ISL timeout.
7302 2017-09-22 Richard Biener <rguenther@suse.de>
7304 PR tree-optimization/82291
7305 * tree-if-conv.c (predicate_mem_writes): Make sure to
7306 remove writes in blocks predicated with false.
7308 2017-09-22 Richard Biener <rguenther@suse.de>
7310 * sese.c: Include cfganal.h.
7311 (if_region_set_false_region): Remove.
7312 (create_if_region_on_edge): Likewise.
7313 (move_sese_in_condition): Re-implement without destroying
7316 2017-09-22 Richard Biener <rguenther@suse.de>
7318 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7319 Verify both BBs contain loop PHI nodes before dispatching to
7321 (graphite_regenerate_ast_isl): Do not recompute dominators,
7322 do not verify three times. Restructure for clarity.
7323 * graphite-scop-detection.c (same_close_phi_node,
7324 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7325 defined_in_loop_p, canonicalize_loop_closed_ssa,
7326 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7327 checking and SSA rewrite, move to ...
7328 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7329 (graphite_initialize): Do not pass in ctx, do not reset the
7330 SCEV cache, compute only dominators.
7331 (graphite_transform_loops): Allocate ISL ctx after
7332 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7333 Maintain post-dominators only around build_scops.
7334 * sese.c (if_region_set_false_region): Make static. Free
7335 and recompute dominators.
7336 (move_sese_in_condition): Assert we don't get called with
7337 post-dominators computed.
7338 * sese.h (if_region_set_false_region): Remove.
7340 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7342 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7343 mode attribute for TARGET_AVX512VL.
7345 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7347 * config/i386/i386.opt (mprefer-avx256): New option.
7348 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7350 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7351 for TARGET_PREFER_AVX256.
7352 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7354 2017-09-21 Jeff Law <law@redhat.com>
7356 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7357 Fix dump output if the only stack space is for pushed registers.
7359 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7361 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7364 2017-09-21 Martin Sebor <msebor@redhat.com>
7367 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7368 code (in C++) or code that triggers warnings.
7370 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7372 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7374 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7376 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7377 * sched-rgn.c: Ditto.
7378 * sel-sched-ir.c: Ditto.
7380 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7382 * toplev.h (set_random_seed): Adjust return type.
7383 * toplev.c (init_local_tick): Move eager initialization of random_seed
7384 to get_random_seed. Adjust comment.
7385 (init_random_seed): Inline to get_random_seed, delete.
7386 (get_random_seed): Initialize random_seed lazily.
7387 (set_random_seed): Do not return previous value.
7388 (print_switch_value): Do not call get_random_seed.
7390 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7392 * cgraph.c (delete_function_version): New, broken out from...
7393 (cgraph_node::delete_function_version): ...here. Rename to
7394 cgraph_node::delete_function_version_by_decl. Update all uses.
7395 (cgraph_node::remove): Call delete_function_version.
7397 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7400 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7401 VAR_DECLs to which addressable non-volatile parameters are mapped
7402 and for id->retvar after the return value assignment. Clear
7403 id->retval and id->retbnd after inlining.
7405 2017-09-21 Richard Biener <rguenther@suse.de>
7407 PR tree-optimization/82276
7408 PR tree-optimization/82244
7409 * tree-vrp.c (build_assert_expr_for): Set
7410 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7412 (remove_range_assertions): Revert earlier change.
7414 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7417 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7419 2017-09-21 Richard Biener <rguenther@suse.de>
7421 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7422 Restore valid IL after code generation errors.
7423 * graphite.c (graphite_transform_loops): Diagnose code
7424 generation issues as MSG_MISSED_OPTIMIZATION and continue
7425 with processing SCOPs.
7427 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7428 Alan Hayward <alan.hayward@arm.com>
7429 David Sherwood <david.sherwood@arm.com>
7431 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7432 rather than choosing between plus_constant and gen_rtx_<CODE>.
7433 * expr.c (emit_push_insn): Likewise.
7434 (expand_expr_real_2): Likewise.
7436 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7437 Alan Hayward <alan.hayward@arm.com>
7438 David Sherwood <david.sherwood@arm.com>
7440 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7442 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7443 Alan Hayward <alan.hayward@arm.com>
7444 David Sherwood <david.sherwood@arm.com>
7446 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7447 calling tree_to_uhwi.
7449 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7450 Alan Hayward <alan.hayward@arm.com>
7451 David Sherwood <david.sherwood@arm.com>
7453 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7454 INTEGER_CST rather than a negative test for ADDR_EXPR.
7456 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7457 Alan Hayward <alan.hayward@arm.com>
7458 David Sherwood <david.sherwood@arm.com>
7460 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7461 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7463 2017-09-21 Richard Biener <rguenther@suse.de>
7465 PR tree-optimization/71351
7466 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7467 graphite_create_new_loop_guard): Remove, fold remaining parts
7469 (translate_isl_ast_node_for): ... here and simplify.
7471 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7474 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7475 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7476 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7477 alternative always use QI mode, for -Os imov (=R,R) alternative
7478 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7481 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7482 Jeff Law <law@redhat.com>
7484 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7485 (allocate_stack_space): New function, partially extracted from
7487 (s390_emit_prologue): Track offset to most recent stack probe.
7488 Code to allocate space moved into allocate_stack_space.
7489 Dump actions when no stack is allocated.
7490 (s390_prologue_plus_offset): New function.
7491 (s390_emit_stack_probe): Likewise.
7493 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7495 * common.opt (Wa, Wl, Wp, g, gz=): Add
7497 (gno-column-info): Remove.
7498 (gcolumn-info): Drop RejectNegative.
7500 (gno-record-gcc-switches): Remove.
7501 (grecord-gcc-switches): Drop RejectNegative.
7502 (gno-split-dwarf): Remove.
7503 (gsplit-dwarf): Drop RejectNegative.
7504 (gno-strict-dwarf): Remove.
7505 (gstrict-dwarf): Drop RejectNegative.
7506 * config/darwin.opt (gfull, gused): Add RejectNegative.
7507 * dwarf2out.c (gen_producer_string): Drop
7508 gno-record-gcc-switches handler.
7509 * optc-gen.awk: Add g to prefixes with negative forms.
7510 * opts-common.c (remapping_prefix_p): New.
7511 (find_opt): Check it.
7512 (generate_canonical_option): Test g prefix.
7513 (option_map): Add -gno- mapping.
7514 (add_misspelling_candidates): Check remapping_prefix_p.
7516 2017-09-20 Jeff Law <law@redhat.com>
7518 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7519 thinko in stack clash protection support.
7521 * explow.c (compute_stack_clash_protection_loop_data): Use
7522 CONST_INT_P instead of explicit test. Verify object is a
7523 CONST_INT_P before looking at INTVAL.
7524 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7525 instead of explicit test.
7527 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7530 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7531 address instead of to r1 and r11.
7533 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7535 * config.gcc: Support "knm".
7536 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7537 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7539 * config/i386/i386.c (m_KNM): Define.
7540 (processor_target_table): Add "knm".
7542 (ix86_option_override_internal): Add "knm".
7543 (ix86_issue_rate): Add PROCESSOR_KNM.
7544 (ix86_adjust_cost): Ditto.
7545 (ia32_multipass_dfa_lookahead): Ditto.
7546 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7547 (fold_builtin_cpu): Add M_INTEL_KNM.
7548 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7549 (processor_type): Add PROCESSOR_KNM.
7550 * config/i386/x86-tune.def: Add m_KNM.
7551 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7553 2017-09-20 Richard Biener <rguenther@suse.de>
7555 PR tree-optimization/80213
7556 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7557 are allowed in empty BBs as well.
7558 (canonicalize_loop_closed_ssa): Also look for other complex
7560 (scop_detection::get_sese): Include the loop-closed PHI block
7562 (scop_detection::merge_sese): Remove code adding extra blocks.
7563 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7564 (build_scops): Assert the final returned scop is invalid.
7566 2017-09-20 Richard Biener <rguenther@suse.de>
7568 PR tree-optimization/82264
7569 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7571 (vn_phi_lookup): Likewise.
7572 (vn_phi_insert): Likewise.
7574 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7576 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7577 that fits into uhwi or shwi, add DW_AT_const_value regardless
7578 of early_dwarf without going through RTL, using add_AT_unsigned
7581 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7582 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7583 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7584 (DEBUG_MACRO_SECTION): Likewise.
7585 (DEBUG_LTO_MACRO_SECTION): Likewise.
7586 (DEBUG_STR_DWO_SECTION): Likewise.
7587 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7588 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7589 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7590 (DEBUG_LTO_LINE_STR_SECTION): Define.
7591 (init_sections_and_labels): Initialize debug_line_str_section
7592 variable. Initialize debug_loc_section for -gdwarf-5 to
7593 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7595 2017-09-20 Richard Biener <rguenther@suse.de>
7597 * graphite-sese-to-poly.c (extract_affine): Properly handle
7598 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7600 2017-09-20 Richard Biener <rguenther@suse.de>
7602 PR tree-optimization/81373
7603 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7604 Force SESE live-out defs to be handled even if they are
7607 2017-09-19 Jeff Law <law@redhat.com>
7609 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7610 nothing for stack adjustments with REG_STACK_CHECK.
7611 * sched-deps.c (parse_add_or_inc): Reject insns with
7612 REG_STACK_CHECK from dependency breaking.
7613 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7614 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7615 * reg-notes.def (STACK_CHECK): New note.
7617 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7618 (ix86_expand_prologue): Dump stack clash info as needed.
7619 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7621 * function.c (dump_stack_clash_frame_info): New function.
7622 * function.h (dump_stack_clash_frame_info): Prototype.
7623 (enum stack_clash_probes): New enum.
7625 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7626 flag_stack_clash_protection.
7627 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7628 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7629 (arm_frame_pointer_required): Likewise.
7630 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7631 (ia64_expand_prologue): Likewise.
7632 * config/mips/mips.c (mips_expand_prologue): Likewise.
7633 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7634 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7635 (sparc_flat_expand_prologue): Likewise.
7636 * config/spu/spu.c (spu_expand_prologue): Likewise.
7638 * explow.c: Include "params.h".
7639 (anti_adjust_stack_and_probe_stack_clash): New function.
7640 (get_stack_check_protect): Likewise.
7641 (compute_stack_clash_protection_loop_data): Likewise.
7642 (emit_stack_clash_protection_loop_start): Likewise.
7643 (emit_stack_clash_protection_loop_end): Likewise.
7644 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7645 Use anti_adjust_stack_and_probe_stack_clash.
7646 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7647 (emit_stack_clash_protection_loop_start): Likewise.
7648 (emit_stack_clash_protection_loop_end): Likewise.
7649 * rtl.h (get_stack_check_protect): Prototype.
7650 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7651 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7652 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7654 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7656 * doc/tm.texi: Rebuilt.
7657 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7658 get_stack_check_protect.
7659 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7660 * config/arm/arm.c (arm_expand_prologue): Likewise.
7661 (arm_frame_pointer_required): Likewise.
7662 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7663 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7664 * config/mips/mips.c (mips_expand_prologue): Likewise.
7665 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7666 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7667 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7668 (sparc_flat_expand_prologue): Likewise.
7670 * common.opt (-fstack-clash-protection): New option.
7671 * flag-types.h (enum stack_check_type): Note difference between
7672 -fstack-check= and -fstack-clash-protection.
7673 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7674 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7675 * toplev.c (process_options): Issue warnings/errors for cases
7676 not handled with -fstack-clash-protection.
7677 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7678 (-fstack-check): Note additional problem with -fstack-check=generic.
7679 Note that -fstack-check is primarily for Ada and refer users
7680 to -fstack-clash-protection for stack-clash-protection.
7681 Document new params for stack clash protection.
7683 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7685 * config/i386/i386.c (ix86_split_long_move): Do not handle
7686 address used for LEA in a special way.
7688 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7690 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7692 2017-09-19 Martin Sebor <msebor@redhat.com>
7695 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7696 of incompatible types.
7698 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7700 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7701 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7702 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7703 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7705 2017-09-19 Richard Biener <rguenther@suse.de>
7707 PR tree-optimization/82244
7708 * tree-vrp.c (remove_range_assertions): Do not propagate
7709 a constant to abnormals but replace the assert with a copy.
7711 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7713 PR rtl-optimization/57878
7714 PR rtl-optimization/68988
7715 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7716 avoidance test involving non_reload_pseudos. Move frequency test
7717 below the general fragmentation avoidance test.
7719 2017-09-19 Richard Biener <rguenther@suse.de>
7721 PR tree-optimization/69728
7722 * graphite-sese-to-poly.c (schedule_error): New global.
7723 (add_loop_schedule): Handle empty domain by failing the
7725 (build_original_schedule): Handle schedule_error.
7727 2017-09-19 Richard Biener <rguenther@suse.de>
7729 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7730 Do not iterate to sibling loops but only to siblings of inner
7733 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7736 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7737 that the registers are different.
7739 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7741 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7742 to processor_model and "amdfam17h" to arch_names_table.
7743 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7745 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7748 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7750 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7751 Alan Hayward <alan.hayward@arm.com>
7752 David Sherwood <david.sherwood@arm.com>
7754 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7756 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7758 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7759 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7760 vect_schedule_slp_instance.
7761 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7762 Update call to vect_slp_analyze_node_operations. Simplify return
7764 (vect_slp_analyze_bb_1): Update call accordingly.
7765 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7766 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7767 (vect_schedule_slp): Update call accordingly.
7769 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7770 Alan Hayward <alan.hayward@arm.com>
7771 David Sherwood <david.sherwood@arm.com>
7773 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7774 with types that aren't in fact scalar.
7776 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7778 * tree-vect-slp.c (vect_record_max_nunits): New function,
7780 (vect_build_slp_tree_1): ...here.
7781 (vect_build_slp_tree_2): Call it for phis too.
7783 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7785 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7786 to vect_get_vec_def_for_operand when getting the mask operand.
7788 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7789 Alan Hayward <alan.hayward@arm.com>
7790 David Sherwood <david.sherwood@arm.com>
7792 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7795 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7796 Alan Hayward <alan.hayward@arm.com>
7797 David Sherwood <david.sherwood@arm.com>
7799 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7800 calculation for vector booleans.
7802 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7803 Alan Hayward <alan.hayward@arm.com>
7804 David Sherwood <david.sherwood@arm.com>
7806 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7808 (vect_transform_stmt): ...here.
7809 (vect_analyze_stmt): Use it instead of calling
7810 vectorizable_live_operation directly.
7812 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7814 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7815 non-SIMT targets in acc vector loops.
7817 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7819 * configure.ac: Add arc and check if assembler supports gdwarf2.
7820 * configure: Regenerate.
7822 2017-09-18 Richard Biener <rguenther@suse.de>
7824 PR tree-optimization/82220
7825 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7826 epilogue niters from the min_profitable_iters compute.
7828 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7831 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7832 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7833 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7835 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7838 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7839 switching to a new text section.
7841 2017-09-18 Richard Biener <rguenther@suse.de>
7843 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7845 (build_alias_set): Reject aliases with no access function.
7847 2017-09-18 Richard Biener <rguenther@suse.de>
7849 PR tree-optimization/79622
7850 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7852 (build_cross_bb_scalars_use): Likewise.
7854 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7856 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7858 2017-09-18 Alan Modra <amodra@gmail.com>
7861 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7862 stack_pointer_rtx for count 0. Update comments. Break up
7863 large rtl expression.
7865 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7867 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7868 Increase to 20 bytes.
7869 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7870 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7871 or avx version of the stub.
7873 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7876 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7877 compute the minimum stack alignment. Also update preferred stack
7878 boundary for leaf functions.
7880 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7882 PR tree-optimization/82228
7883 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7886 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7888 * common/config/nds32/nds32-common.c
7889 (nds32_option_optimization_table): Refine formatting.
7890 (nds32_option_optimization_table): Use -fsched-pressure and
7891 -fomit-frame-pointer for specific optimization level.
7893 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7895 * config/nds32/nds32.c: Refine formatting and comments.
7896 * config/nds32/nds32.h: Likewise.
7897 * config/nds32/nds32.md: Likewise.
7898 * config/nds32/nds32-cost.c: Likewise.
7899 * config/nds32/nds32-isr.c: Likewise.
7900 * config/nds32/nds32-md-auxiliary.c: Likewise.
7901 * config/nds32/nds32-multiple.md: Likewise.
7902 * config/nds32/nds32-predicates.c: Likewise.
7904 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7905 Jakub Jelinek <jakub@redhat.com>
7907 Add support for -std=c++2a.
7908 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7910 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7912 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7915 * doc/extend.texi (Common Function Attributes): Add
7916 references to ARM, AArch64, and S/390 specific attributes.
7917 (Function Specific Option Pragmas): Add AArch64 and S/390
7918 to list of back ends that support the target pragma.
7920 2017-09-15 Nathan Sidwell <nathan@acm.org>
7922 * doc/standards.texi: Fix C++17 description. Update URLs for
7925 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7927 * common.opt (Wcast-align=strict): New warning option.
7928 * doc/invoke.texi: Document -Wcast-align=strict.
7930 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7932 * cgraph.h (cgraph_thunk_info): Add comments.
7933 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7934 assert for VIRTUAL_* arguments stricter.
7936 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7938 PR tree-optimization/71026
7939 * match.pd: Move RDIV patterns from fold-const.c
7940 * fold-const.c (distribute_real_division): Removed.
7941 (fold_binary_loc): Remove calls to distribute_real_divison.
7943 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7945 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7946 c++1z and gnu++1z as deprecated. Change other references to
7947 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7948 Change -Wc++1z-compat to -Wc++17-compat.
7949 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7950 * dwarf2out.c (highest_c_language): Handle C++17.
7951 (gen_compile_unit_die): Likewise.
7953 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7955 PR rtl-optimization/82192
7956 * combine.c (make_extraction): Don't look through non-paradoxical
7957 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7960 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7961 Alan Hayward <alan.hayward@arm.com>
7962 David Sherwood <david.sherwood@arm.com>
7964 * target.def (function_arg_offset): New hook.
7965 * targhooks.h (default_function_arg_offset): Declare.
7966 * targhooks.c (default_function_arg_offset): New function.
7967 * function.c (locate_and_pad_parm): Use
7968 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7969 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7970 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7971 * doc/tm.texi: Regenerate.
7972 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7973 * config/spu/spu.c (spu_function_arg_offset): New function.
7974 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7975 * system.h (FUNCTION_ARG_OFFSET): Poison.
7977 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7978 Alan Hayard <alan.hayward@arm.com>
7979 David Sherwood <david.sherwood@arm.com>
7981 * target.def (truly_noop_truncation): New hook.
7982 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7983 than TRULY_NOOP_TRUNCATION.
7984 * hooks.h (hook_bool_uint_uint_true): Declare.
7985 * hooks.c (hook_bool_uint_uint_true): New function.
7986 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7987 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7988 * doc/tm.texi: Regenerate.
7989 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7990 rather than TRULY_NOOP_TRUNCATION in comments.
7991 (simplify_comparison): Likewise.
7992 (record_truncated_value): Likewise.
7993 * expmed.c (extract_bit_field_1): Likewise.
7994 (extract_split_bit_field): Likewise.
7995 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7996 instead of TRULY_NOOP_TRUNCATION.
7997 * function.c (assign_parm_setup_block): Likewise.
7998 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7999 * rtlhooks.c: Include target.h.
8000 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8001 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8002 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8003 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8004 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8005 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8006 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8007 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8008 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8009 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8010 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8011 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8012 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8013 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8014 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8015 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8016 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8017 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8018 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8019 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8020 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8021 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8022 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8023 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8024 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8025 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8026 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8027 rather than TRULY_NOOP_TRUNCATION in comments.
8028 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8029 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8030 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8031 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8032 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8033 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8034 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8035 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8036 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8037 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8038 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8039 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8040 rather than TRULY_NOOP_TRUNCATION in comments.
8041 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8042 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8043 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8044 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8045 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8046 TRULY_NOOP_TRUNCATION condition.
8047 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8048 (TRULY_NOOP_TRUNCATION): Delete.
8049 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8050 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8051 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8052 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8053 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8054 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8055 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8056 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8057 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8058 rather than TRULY_NOOP_TRUNCATION in comments.
8059 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8060 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8061 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8062 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8063 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8064 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8066 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8069 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8070 (*cmp_ior): Likewise.
8071 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8072 (*ior_scc_scc_cmp): Likewise.
8073 (*and_scc_scc): Likewise.
8074 (*and_scc_scc_cmp): Likewise.
8076 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8077 Alan Hayard <alan.hayward@arm.com>
8078 David Sherwood <david.sherwood@arm.com>
8080 * target.def (can_change_mode_class): New hook.
8081 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8082 (hard_regno_nregs): Likewise.
8083 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8084 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8085 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8086 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8087 (LOAD_EXTEND_OP): Update accordingly.
8088 * doc/tm.texi: Regenerate.
8089 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8090 CANNOT_CHANGE_MODE_CLASS.
8091 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8092 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8093 * combine.c (simplify_set): Update accordingly.
8094 * emit-rtl.c (validate_subreg): Likewise.
8095 * recog.c (general_operand): Likewise.
8096 * regcprop.c (mode_change_ok): Likewise.
8097 * reload1.c (choose_reload_regs): Likewise.
8098 (inherit_piecemeal_p): Likewise.
8099 * rtlanal.c (simplify_subreg_regno): Likewise.
8100 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8101 instead of CANNOT_CHANGE_MODE_CLASS.
8102 (reload_cse_simplify_operands): Likewise.
8103 * reload.c (push_reload): Use targetm.can_change_mode_class
8104 instead of CANNOT_CHANGE_MODE_CLASS.
8105 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8106 REG_CANNOT_CHANGE_MODE_P.
8107 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8108 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8109 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8110 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8111 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8112 (arm_can_change_mode_class): New function.
8113 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8114 than CANNOT_CHANGE_MODE_CLASS in comments.
8115 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8116 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8117 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8118 (ix86_can_change_mode_class): ...this new function, inverting the
8119 sense of the return value.
8120 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8121 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8122 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8123 (ia64_can_change_mode_class): New function.
8124 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8125 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8126 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8127 (m32c_can_change_mode_class): ...this new function, inverting the
8128 sense of the return value.
8129 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8130 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8131 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8132 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8133 (mips_can_change_mode_class): ...this new function, inverting the
8134 sense of the return value.
8135 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8136 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8137 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8138 (msp430_can_change_mode_class): New function.
8139 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8140 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8141 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8142 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8143 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8144 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8145 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8146 (pa_cannot_change_mode_class): Replace with...
8147 (pa_can_change_mode_class): ...this new function, inverting the
8148 sense of the return value.
8149 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8150 than CANNOT_CHANGE_MODE_CLASS in comments.
8151 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8152 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8153 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8154 (pdp11_cannot_change_mode_class): Replace with...
8155 (pdp11_can_change_mode_class): ...this new function, inverting the
8156 sense of the return value.
8157 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8158 * config/powerpcspe/powerpcspe-protos.h
8159 (rs6000_cannot_change_mode_class_ptr): Delete.
8160 * config/powerpcspe/powerpcspe.c
8161 (rs6000_cannot_change_mode_class_ptr): Delete.
8162 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8163 (rs6000_option_override_internal): Assign to
8164 targetm.can_change_mode_class instead of
8165 rs6000_cannot_change_mode_class_ptr.
8166 (rs6000_cannot_change_mode_class): Replace with...
8167 (rs6000_can_change_mode_class): ...this new function, inverting the
8168 sense of the return value.
8169 (rs6000_debug_cannot_change_mode_class): Replace with...
8170 (rs6000_debug_can_change_mode_class): ...this new function.
8171 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8172 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8173 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8174 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8175 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8177 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8178 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8179 (rs6000_option_override_internal): Assign to
8180 targetm.can_change_mode_class instead of
8181 rs6000_cannot_change_mode_class_ptr.
8182 (rs6000_cannot_change_mode_class): Replace with...
8183 (rs6000_can_change_mode_class): ...this new function, inverting the
8184 sense of the return value.
8185 (rs6000_debug_cannot_change_mode_class): Replace with...
8186 (rs6000_debug_can_change_mode_class): ...this new function.
8187 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8188 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8189 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8190 (s390_can_change_mode_class): ...this new function, inverting the
8191 sense of the return value.
8192 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8193 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8194 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8195 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8196 (sh_cannot_change_mode_class): Replace with...
8197 (sh_can_change_mode_class): ...this new function, inverting the
8198 sense of the return value.
8199 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8200 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8201 (sparc_can_change_mode_class): New function.
8202 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8203 * config/spu/spu.c (spu_can_change_mode_class): New function.
8204 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8205 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8206 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8207 (visium_can_change_mode_class): New function.
8208 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8210 2017-09-15 Richard Biener <rguenther@suse.de>
8212 PR tree-optimization/82217
8213 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8214 but not undefined case.
8216 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8219 * postreload.c (reload_cse_simplify_operands): Skip
8220 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8222 2017-09-15 Richard Biener <rguenther@suse.de>
8224 PR tree-optimization/68823
8225 * graphite-scop-detection.c (build_alias_set): If we have a
8226 possible dependence check whether we can handle them by just
8227 looking at the DRs DR_ACCESS_FNs.
8228 (build_scops): If build_alias_set fails, fail the SCOP.
8230 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8232 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8233 to support float128 built-in functions that require the ISA 3.0
8235 (BU_FLOAT128_3_HW): Likewise.
8236 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8238 (FMAF128): Likewise.
8240 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8241 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8242 floating point instructions.
8243 (rs6000_invalid_builtin): Likewise.
8244 (rs6000_builtin_mask_names): Likewise.
8245 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8246 (RS6000_BTM_FLOAT128_HW): Likewise.
8247 (RS6000_BTM_COMMON): Likewise.
8248 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8250 * doc/extend.texi (RS/6000 built-in functions): Document the
8251 IEEE 128-bit floating point square root and fused multiply-add
8254 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8256 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8257 reg (r2) isn't in the set of registers defined in the prologue.
8259 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8264 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8265 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8266 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8267 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8269 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8270 max_vectorization_factor.
8271 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8273 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8274 Alan Hayward <alan.hayward@arm.com>
8275 David Sherwood <david.sherwood@arm.com>
8277 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8278 (vect_worthwhile_without_simd_p): Declare.
8279 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8280 (vectorizable_reduction): Use it.
8281 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8282 (vectorizable_operation): Likewise.
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_get_num_copies): New function.
8289 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8290 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8291 (vectorizable_induction): Likewise.
8292 (vectorizable_live_operation): Likewise.
8293 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8294 (vectorizable_bswap): Likewise.
8295 (vectorizable_call): Likewise.
8296 (vectorizable_conversion): Likewise.
8297 (vectorizable_assignment): Likewise.
8298 (vectorizable_shift): Likewise.
8299 (vectorizable_operation): Likewise.
8300 (vectorizable_store): Likewise.
8301 (vectorizable_load): Likewise.
8302 (vectorizable_condition): Likewise.
8303 (vectorizable_comparison): Likewise.
8304 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8306 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8307 Alan Hayward <alan.hayward@arm.com>
8308 David Sherwood <david.sherwood@arm.com>
8310 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8311 of vect_init_vector.
8313 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8314 Alan Hayward <alan.hayward@arm.com>
8315 David Sherwood <david.sherwood@arm.com>
8317 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8318 an inline wrapper that provides a location.
8319 (gimple_build_vector): Likewise.
8320 * gimple-fold.c (gimple_build_vector_from_val): New function.
8321 (gimple_build_vector): Likewise.
8322 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8323 functions to build the initial value. Always return a gimple value.
8324 (get_initial_defs_for_reduction): Likewise. Only compute
8326 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8327 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8328 (vectorizable_induction): Use gimple_build_vector rather than
8331 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8332 Alan Hayward <alan.hayward@arm.com>
8333 David Sherwood <david.sherwood@arm.com>
8335 * target.h (vec_perm_indices): New typedef.
8336 (auto_vec_perm_indices): Likewise.
8337 * optabs-query.h: Include target.h
8338 (can_vec_perm_p): Take a vec_perm_indices *.
8339 * optabs-query.c (can_vec_perm_p): Likewise.
8340 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8341 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8342 * tree-vect-generic.c (lower_vec_perm): Likewise.
8343 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8344 (vect_grouped_load_supported): Likewise.
8345 (vect_shift_permute_load_chain): Likewise.
8346 (vect_permute_store_chain): Use auto_vec_perm_indices.
8347 (vect_permute_load_chain): Likewise.
8348 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8349 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8350 Update uses of can_vec_perm_p.
8351 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8352 mode with a number of elements. Take a vec_perm_indices *.
8353 (vect_create_epilog_for_reduction): Update accordingly.
8354 Use auto_vec_perm_indices.
8355 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8356 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8357 (vect_transform_slp_perm_load): Likewise.
8358 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8359 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8360 (vect_gen_perm_mask_checked): Likewise.
8361 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8362 (vect_gen_perm_mask_checked): Likewise.
8363 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8364 (vectorizable_store): Likewise.
8365 (vectorizable_load): Likewise.
8366 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8367 (vectorizable_bswap): Likewise.
8369 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8370 Alan Hayward <alan.hayward@arm.com>
8371 David Sherwood <david.sherwood@arm.com>
8373 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8374 * tree.c (build_vector): Likewise.
8375 (build_vector_from_ctor): Update accordingly.
8376 (build_vector_from_val): Likewise.
8377 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8378 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8379 * tree-vect-generic.c (add_rshift): Likewise.
8380 (expand_vector_divmod): Likewise.
8381 (optimize_vector_constructor): Likewise.
8382 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8383 (vect_transform_slp_perm_load): Likewise.
8384 (vect_schedule_slp_instance): Likewise.
8385 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8386 (vectorizable_call): Likewise.
8387 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8388 * expmed.c (make_tree): Likewise.
8389 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8390 a vector passed to build_vector.
8391 (fold_convert_const): Likewise.
8392 (exact_inverse): Likewise.
8393 (fold_ternary_loc): Likewise.
8394 (fold_relational_const): Likewise.
8395 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8396 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8397 (const_unop): Likewise. Store the reduction accumulator in a
8398 variable rather than an array.
8399 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8400 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8401 the new vector, rather than constructing it after the input arrays.
8402 (native_interpret_vector): Use auto_vec<tree> when building
8403 a vector passed to build_vector. Add elements in order.
8404 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8405 auto_vec<tree> when building a vector passed to build_vector.
8406 (vect_create_epilog_for_reduction): Likewise.
8407 (vectorizable_induction): Likewise.
8408 (get_initial_def_for_reduction): Likewise. Fix indentation of
8410 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8412 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8413 passed to build_vector.
8415 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8416 Alan Hayward <alan.hayward@arm.com>
8417 David Sherwood <david.sherwood@arm.com>
8419 * tree-core.h (tree_base::u): Add an "nelts" field.
8420 (tree_vector): Use VECTOR_CST_NELTS as the length.
8421 * tree.c (tree_size): Likewise.
8422 (make_vector): Initialize VECTOR_CST_NELTS.
8423 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8424 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8425 TYPE_VECTOR_SUBPARTS.
8426 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8427 as the number of units, setting it from VECTOR_CST_NELTS.
8428 (const_vector_from_tree): Likewise.
8429 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8430 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8431 (fold_negate_expr_1): Likewise.
8432 (fold_convert_const): Likewise.
8433 (const_binop): Likewise. Differentiate the number of output and
8435 (const_unop): Likewise.
8436 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8437 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8438 in cases that did the opposite.
8440 2017-09-14 Richard Biener <rguenther@suse.de>
8442 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8445 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8447 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8449 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8452 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8453 if and where to split a bb, except for splitting before debug insn
8454 sequences followed by non-label real insn. Delete debug insns
8455 in between basic blocks.
8457 * combine.c (make_compound_operation_int): Formatting fixes.
8459 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8460 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8461 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8462 * config/sol2.h (LINK_EH_SPEC): Likewise.
8463 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8464 * config/s390/linux.h (LINK_SPEC): Likewise.
8465 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8466 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8467 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8468 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8469 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8470 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8471 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8472 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8474 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8476 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8478 (ENDFILE_LINUX_SPEC): Likewise.
8479 (LINK_EH_SPEC): Likewise.
8480 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8481 (LINK_OS_LINUX_SPEC32): Likewise.
8482 (LINK_OS_LINUX_SPEC64): Likewise.
8483 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8484 (LINK_OS_LINUX_SPEC): Likewise.
8486 2017-09-13 Martin Liska <mliska@suse.cz>
8489 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8490 CASE_HIGH is NULL_TREE.
8492 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8493 Alan Hayward <alan.hayward@arm.com>
8494 David Sherwood <david.sherwood@arm.com>
8496 * target.def (secondary_memory_needed): New hook.
8497 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8498 instead of SECONDARY_MEMORY_NEEDED.
8499 (secondary_memory_needed_mode): Likewise.
8500 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8501 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8502 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8503 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8504 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8505 * doc/tm.texi: Regenerate.
8506 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8507 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8508 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8509 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8510 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8511 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8512 mode argument first and change the reg_class arguments to reg_class_t.
8513 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8514 Make static. Update the call to inline_secondary_memory_needed.
8515 (ix86_register_move_cost): Update the call to
8516 inline_secondary_memory_needed.
8517 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8518 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8520 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8521 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8523 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8524 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8525 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8526 and match hook interface. Add comment from mips.h.
8527 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8528 * config/mmix/mmix.md (truncdfsf2): Refer to
8529 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8531 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8532 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8533 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8534 (pa_secondary_memory_needed): New function.
8535 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8536 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8537 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8538 (pdp11_secondary_memory_needed): Make static and match hook interface.
8539 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8540 * config/powerpcspe/powerpcspe-protos.h
8541 (rs6000_secondary_memory_needed_ptr): Delete.
8542 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8544 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8545 (rs6000_option_override_internal): Assign to
8546 targetm.secondary_memory_needed rather than
8547 rs6000_secondary_memory_needed_ptr.
8548 (rs6000_secondary_memory_needed): Match hook interface.
8549 (rs6000_debug_secondary_memory_needed): Likewise.
8550 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8551 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8552 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8553 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8554 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8555 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8557 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8558 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8559 (rs6000_option_override_internal): Assign to
8560 targetm.secondary_memory_needed rather than
8561 rs6000_secondary_memory_needed_ptr.
8562 (rs6000_secondary_memory_needed): Match hook interface.
8563 (rs6000_debug_secondary_memory_needed): Likewise.
8564 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8565 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8566 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8567 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8568 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8569 (sparc_secondary_memory_needed): New function.
8570 * lra-constraints.c (check_and_process_move): Refer to
8571 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8573 (curr_insn_transform): Likewise.
8574 (process_alt_operands): Use targetm.secondary_memory_needed
8575 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8576 (check_secondary_memory_needed_p): Likewise.
8577 (choose_split_class): Likewise.
8578 * reload.c: Unconditionally include code that was previously
8579 conditional on SECONDARY_MEMORY_NEEDED.
8580 (push_secondary_reload): Use targetm.secondary_memory_needed
8581 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8582 (push_reload): Likewise.
8583 * reload1.c: Unconditionally include code that was previously
8584 conditional on SECONDARY_MEMORY_NEEDED.
8585 (choose_reload_regs): Use targetm.secondary_memory_needed
8586 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8587 (gen_reload): Likewise.
8588 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8590 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8591 Alan Hayward <alan.hayward@arm.com>
8592 David Sherwood <david.sherwood@arm.com>
8594 * target.def (secondary_memory_needed_mode): New hook:
8595 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8596 * targhooks.h (default_secondary_memory_needed_mode): New function.
8597 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8598 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8599 * doc/tm.texi: Regenerate.
8600 * lra-constraints.c (check_and_process_move): Use
8601 targetm.secondary_memory_needed_mode instead of
8602 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8603 (curr_insn_transform): Likewise.
8604 * reload.c (get_secondary_mem): Likewise.
8605 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8606 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8608 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8609 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8610 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8611 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8612 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8614 * config/powerpcspe/powerpcspe-protos.h
8615 (rs6000_secondary_memory_needed_mode): Delete.
8616 * config/powerpcspe/powerpcspe.c
8617 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8618 (rs6000_secondary_memory_needed_mode): Make static.
8619 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8620 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8622 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8624 (rs6000_secondary_memory_needed_mode): Make static.
8625 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8626 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8627 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8628 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8629 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8631 (sparc_secondary_memory_needed_mode): New function.
8632 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8634 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8636 * config/aarch64/constraints.md (Umq): New constraint.
8637 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8639 (mov<mode>): Update condition.
8641 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8643 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8644 when bitposition is the same.
8646 2017-09-13 Richard Biener <rguenther@suse.de>
8648 * dwarf2out.c (output_die_symbol): Remove.
8649 (output_die): Do not output a DIEs symbol.
8651 2017-09-13 Richard Biener <rguenther@suse.de>
8654 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8655 default-def to avoid breaking iterator update with the weird
8656 interaction with cgraph_update_edges_for_call_stmt_node.
8658 2017-09-13 Richard Biener <rguenther@suse.de>
8660 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8661 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8662 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8663 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8665 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8667 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8668 Disable pc relative literal load irrespective of
8669 TARGET_FIX_ERR_A53_84341 for default.
8671 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8673 * config/sparc/sparc.c (output_return): Output the source location of
8674 the insn in the delay slot, if any.
8675 (output_sibcall): Likewise.
8677 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8680 * common.opt (-static-pie): New alias.
8681 (shared): Negate static-pie.
8682 (-no-pie): Update help text.
8684 (static-pie): New option.
8685 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8686 -static-pie support.
8687 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8688 (LINK_EH_SPEC): Likewise.
8689 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8690 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8691 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8692 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8693 (init_gcc_specs): Likewise.
8694 (init_spec): Likewise.
8695 (display_help): Update help message for -pie.
8696 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8699 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8701 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8702 (movdi_aarch64): Likewise.
8703 (movti_aarch64): Likewise.
8705 2017-09-12 Simon Wright <simon@pushface.org>
8708 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8709 calculation of the minor version, always output as 0.
8711 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8714 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8715 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8716 on it early, rather than manual conversion late. For
8717 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8718 instead of performing manual conversion.
8720 2017-09-12 Carl Love <cel@us.ibm.com>
8722 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8723 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8725 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8726 VMULOSW): Add definitions.
8727 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8728 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8729 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8731 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8733 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8735 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8737 (movti_aarch64): Likewise.
8738 (movdf_aarch64): Likewise.
8739 (movtf_aarch64): Likewise.
8740 (load_pairdi): Likewise.
8741 (store_pairdi): Likewise.
8742 (load_pairdf): Likewise.
8743 (store_pairdf): Likewise.
8744 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8745 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8746 (ldr_got_small_<mode>): Likewise.
8747 (ldr_got_small_28k_<mode>): Likewise.
8748 (ldr_got_tiny): Likewise.
8749 * config/aarch64/iterators.md (ldst_sz): New.
8750 (ldpstp_sz): Likewise.
8751 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8753 (thunderx_load): Split load_8 to load_16.
8754 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8756 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8757 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8758 (xgene1_store_pair): Split store_8 to store_16.
8759 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8760 (falkor_st_0_st_sd): Split store_8 to store_16.
8762 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8764 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8765 and store1/2/3/4 to store_4/8/12/16.
8766 * config/aarch64/aarch64.md: Update for rename.
8767 * config/arm/arm.md: Likewise.: Likewise.
8768 * config/arm/arm.c: Likewise.
8769 * config/arm/thumb1.md: Likewise.
8770 * config/arm/thumb2.md: Likewise.
8771 * config/arm/vfp.md: Likewise.
8772 * config/arm/arm-generic.md: Likewise.
8773 * config/arm/arm1020e.md: Likewise.
8774 * config/arm/arm1026ejs.md: Likewise.
8775 * config/arm/arm1136jfs.md: Likewise.
8776 * config/arm/arm926ejs.md: Likewise.
8777 * config/arm/cortex-a15.md: Likewise.
8778 * config/arm/cortex-a17.md: Likewise.
8779 * config/arm/cortex-a5.md: Likewise.
8780 * config/arm/cortex-a53.md: Likewise.
8781 * config/arm/cortex-a57.md: Likewise.
8782 * config/arm/cortex-a7.md: Likewise.
8783 * config/arm/cortex-a8.md: Likewise.
8784 * config/arm/cortex-a9.md: Likewise.
8785 * config/arm/cortex-m4.md: Likewise.
8786 * config/arm/cortex-m7.md: Likewise.
8787 * config/arm/cortex-r4.md: Likewise.
8788 * config/arm/exynos-m1.md: Likewise.
8789 * config/arm/fa526.md: Likewise.
8790 * config/arm/fa606te.md: Likewise.
8791 * config/arm/fa626te.md: Likewise.
8792 * config/arm/fa726te.md: Likewise.
8793 * config/arm/fmp626.md: Likewise.
8794 * config/arm/iwmmxt.md: Likewise.
8795 * config/arm/ldmstm.md: Likewise.
8796 * config/arm/marvell-pj4.md: Likewise.
8797 * config/arm/xgene1.md: Likewise.
8798 * config/aarch64/thunderx.md: Likewise.
8799 * config/aarch64/thunderx2t99.md: Likewise.
8800 * config/aarch64/falkor.md: Likewise.
8802 2017-09-12 Martin Liska <mliska@suse.cz>
8804 * attribs.c (private_lookup_attribute): New function.
8805 * attribs.h (private_lookup_attribute): Declared here.
8806 (lookup_attribute): Called from this place.
8808 2017-09-12 Richard Biener <rguenther@suse.de>
8810 PR tree-optimization/82157
8811 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8812 stmts with side-effects.
8814 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8815 Alan Hayward <alan.hayward@arm.com>
8816 David Sherwood <david.sherwood@arm.com>
8818 * target.def (hard_regno_nregs): New hook.
8819 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8820 * targhooks.h (default_hard_regno_nregs): Declare.
8821 * targhooks.c (default_hard_regno_nregs): New function.
8822 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8823 (TARGET_HARD_REGNO_NREGS): ...this hook.
8824 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8825 (CLASS_MAX_NREGS): Likewise.
8826 * doc/tm.texi: Regenerate.
8827 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8828 instead of HARD_REGNO_NREGS.
8829 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8830 HARD_REGNO_NREGS in the comment.
8831 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8832 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8833 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8834 Return an unsigned int.
8835 (TARGET_HARD_REGNO_NREGS): Redefine.
8836 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8837 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8838 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8839 (arc_hard_regno_nregs): New function.
8840 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8841 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8842 (arm_hard_regno_nregs): New function.
8843 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8844 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8845 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8846 (TARGET_HARD_REGNO_NREGS): Redefine.
8847 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8848 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8849 (HARD_REGNO_NREGS): Delete.
8850 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8851 (cr16_hard_regno_nregs): New function.
8852 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8853 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8854 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8855 (cris_hard_regno_nregs): New function.
8856 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8857 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8858 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8859 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8860 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8861 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8862 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8863 (frv_hard_regno_nregs): Make static. Take and return an
8865 (frv_class_max_nregs): Remove outdated copy of documentation.
8866 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8867 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8868 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8869 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8870 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8871 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8872 (TARGET_HARD_REGNO_NREGS): Redefine.
8873 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8874 (CLASS_MAX_NREGS): Update comment.
8875 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8876 (ia64_hard_regno_nregs): New function.
8877 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8878 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8879 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8880 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8881 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8883 (m32c_hard_regno_nregs): Likewise. Make static.
8884 (TARGET_HARD_REGNO_NREGS): Redefine.
8885 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8886 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8887 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8888 (m68k_hard_regno_nregs): New function.
8889 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8890 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8891 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8892 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8893 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8894 Take and return an unsigned int.
8895 (TARGET_HARD_REGNO_NREGS): Redefine.
8896 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8897 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8898 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8899 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8900 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8901 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8902 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8903 (msp430_hard_regno_nregs): Make static. Take and return an
8905 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8906 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8907 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8908 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8909 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8910 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8911 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8912 (TARGET_HARD_REGNO_NREGS): Redefine.
8913 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8914 (PA_HARD_REGNO_NREGS): ...this.
8915 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8916 (PA_HARD_REGNO_NREGS): ...this.
8917 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8918 (pa_hard_regno_nregs): New function.
8919 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8920 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8921 (pdp11_hard_regno_nregs): New function.
8922 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8923 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8924 (rs6000_hard_regno_nregs_hook): New function.
8925 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8926 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8927 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8928 Take and return an unsigned int. Move earlier in file.
8929 (TARGET_HARD_REGNO_NREGS): Redefine.
8930 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8931 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8932 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8933 (rl78_hard_regno_nregs): Make static. Take and return an
8935 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8936 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8937 (rs6000_hard_regno_nregs_hook): New function.
8938 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8939 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8940 (TARGET_HARD_REGNO_NREGS): Redefine.
8941 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8942 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8943 instead of HARD_REGNO_NREGS.
8944 (s390_hard_regno_nregs): New function.
8945 (s390_hard_regno_mode_ok): Add comment from s390.h.
8946 (TARGET_HARD_REGNO_NREGS): Redefine.
8947 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8948 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8949 (sh_hard_regno_nregs): New function.
8950 (sh_pass_in_reg_p): Use it.
8951 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8952 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8953 (sparc_hard_regno_nregs): New function.
8954 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8955 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8956 (spu_function_arg_advance): Use it, supplying a valid register number.
8957 (TARGET_HARD_REGNO_NREGS): Redefine.
8958 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8959 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8960 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8961 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8962 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8963 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8964 (CLASS_MAX_NREGS): Remove copy of old documentation.
8965 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8966 (visium_hard_regno_nregs): New function.
8967 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8968 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8969 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8970 xtensa_hard_regno_nregs): New function.
8971 * system.h (HARD_REGNO_NREGS): Poison.
8973 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8975 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8976 hard_regno_nregs instead of HARD_REGNO_NREGS.
8977 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8978 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8979 (c6x_expand_epilogue): Likewise.
8980 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8981 (frv_read_iacc_argument): Likewise.
8982 * config/sh/sh.c: Include regs.h.
8983 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8984 (regs_used): Likewise.
8985 (output_stack_adjust): Likewise.
8986 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8987 * expmed.c: Include regs.h.
8988 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8989 * ree.c: Include regs.h.
8990 (combine_reaching_defs): Use hard_regno_nregs instead of
8992 (add_removable_extension): Likewise.
8994 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8996 * regs.h (hard_regno_nregs): Turn into a function.
8997 (end_hard_regno): Update accordingly.
8998 * caller-save.c (setup_save_areas): Likewise.
8999 (save_call_clobbered_regs): Likewise.
9000 (replace_reg_with_saved_mem): Likewise.
9001 (insert_restore): Likewise.
9002 (insert_save): Likewise.
9003 * combine.c (can_change_dest_mode): Likewise.
9004 (move_deaths): Likewise.
9005 (distribute_notes): Likewise.
9006 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9007 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9008 (rs6000_split_multireg_move): Likewise.
9009 (rs6000_register_move_cost): Likewise.
9010 (rs6000_memory_move_cost): Likewise.
9011 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9012 (rs6000_split_multireg_move): Likewise.
9013 (rs6000_register_move_cost): Likewise.
9014 (rs6000_memory_move_cost): Likewise.
9015 * cselib.c (cselib_reset_table): Likewise.
9016 (cselib_lookup_1): Likewise.
9017 * emit-rtl.c (set_mode_and_regno): Likewise.
9018 * function.c (aggregate_value_p): Likewise.
9019 * ira-color.c (setup_profitable_hard_regs): Likewise.
9020 (check_hard_reg_p): Likewise.
9021 (calculate_saved_nregs): Likewise.
9022 (assign_hard_reg): Likewise.
9023 (improve_allocation): Likewise.
9024 (calculate_spill_cost): Likewise.
9025 * ira-emit.c (modify_move_list): Likewise.
9026 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9027 (ira_hard_reg_in_set_p): Likewise.
9028 * ira.c (setup_reg_mode_hard_regset): Likewise.
9029 (clarify_prohibited_class_mode_regs): Likewise.
9030 (check_allocation): Likewise.
9031 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9032 (lra_setup_reg_renumber): Likewise.
9033 (setup_try_hard_regno_pseudos): Likewise.
9034 (spill_for): Likewise.
9035 (assign_hard_regno): Likewise.
9036 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9037 * lra-constraints.c (in_class_p): Likewise.
9038 (lra_constraint_offset): Likewise.
9039 (simplify_operand_subreg): Likewise.
9040 (lra_constraints): Likewise.
9041 (split_reg): Likewise.
9042 (split_if_necessary): Likewise.
9043 (invariant_p): Likewise.
9044 (inherit_in_ebb): Likewise.
9045 * lra-lives.c (process_bb_lives): Likewise.
9046 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9047 (get_hard_regs): Likewise.
9048 (do_remat): Likewise.
9049 * lra-spills.c (assign_spill_hard_regs): Likewise.
9050 * mode-switching.c (create_pre_exit): Likewise.
9051 * postreload.c (reload_combine_recognize_pattern): Likewise.
9052 * recog.c (peep2_find_free_register): Likewise.
9053 * regcprop.c (kill_value_regno): Likewise.
9054 (set_value_regno): Likewise.
9055 (copy_value): Likewise.
9056 (maybe_mode_change): Likewise.
9057 (find_oldest_value_reg): Likewise.
9058 (copyprop_hardreg_forward_1): Likewise.
9059 * regrename.c (check_new_reg_p): Likewise.
9060 (regrename_do_replace): Likewise.
9061 * reload.c (push_reload): Likewise.
9062 (combine_reloads): Likewise.
9063 (find_dummy_reload): Likewise.
9064 (operands_match_p): Likewise.
9065 (find_reloads): Likewise.
9066 (find_equiv_reg): Likewise.
9067 (reload_adjust_reg_for_mode): Likewise.
9068 * reload1.c (count_pseudo): Likewise.
9069 (count_spilled_pseudo): Likewise.
9070 (find_reg): Likewise.
9071 (clear_reload_reg_in_use): Likewise.
9072 (free_for_value_p): Likewise.
9073 (allocate_reload_reg): Likewise.
9074 (choose_reload_regs): Likewise.
9075 (reload_adjust_reg_for_temp): Likewise.
9076 (emit_reload_insns): Likewise.
9077 (delete_output_reload): Likewise.
9078 * rtlanal.c (subreg_get_info): Likewise.
9079 * sched-deps.c (sched_analyze_reg): Likewise.
9080 * sel-sched.c (init_regs_for_mode): Likewise.
9081 (mark_unavailable_hard_regs): Likewise.
9082 (choose_best_reg_1): Likewise.
9083 (verify_target_availability): Likewise.
9084 * valtrack.c (dead_debug_insert_temp): Likewise.
9085 * var-tracking.c (track_loc_p): Likewise.
9086 (emit_note_insn_var_location): Likewise.
9087 * varasm.c (make_decl_rtl): Likewise.
9088 * reginfo.c (choose_hard_reg_mode): Likewise.
9089 (init_reg_modes_target): Refer directly to
9090 this_target_regs->x_hard_regno_nregs.
9092 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9094 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9095 instead of hard_regno_nregs.
9097 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9099 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9100 end_hard_regno instead of hard_regno_nregs.
9101 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9102 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9103 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9104 * ira-color.c (improve_allocation): Likewise.
9105 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9106 * lra-lives.c (mark_regno_live): Likewise.
9107 (mark_regno_dead): Likewise.
9108 * lra-remat.c (operand_to_remat): Likewise.
9109 * lra.c (collect_non_operand_hard_regs): Likewise.
9110 * postreload.c (reload_combine_note_store): Likewise.
9111 (move2add_valid_value_p): Likewise.
9112 * reload.c (regno_clobbered_p): Likewise.
9114 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9116 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9118 * config/v850/v850.c (v850_reorg): Likewise.
9119 * reload.c (refers_to_regno_for_reload_p): Likewise.
9120 (find_equiv_reg): Likewise.
9121 * reload1.c (reload_reg_reaches_end_p): Likewise.
9123 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9125 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9127 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9128 * config/arm/arm.c (output_move_neon): Likewise.
9129 (arm_attr_length_move_neon): Likewise.
9130 (neon_split_vcombine): Likewise.
9131 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9132 (c6x_mark_reg_written): Likewise.
9133 (c6x_dwarf_register_span): Likewise.
9134 * config/i386/i386.c (ix86_save_reg): Likewise.
9135 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9136 (rws_access_reg): Likewise.
9137 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9138 * mode-switching.c (create_pre_exit): Likewise.
9139 * ree.c (combine_reaching_defs): Likewise.
9140 (add_removable_extension): Likewise.
9141 * regcprop.c (find_oldest_value_reg): Likewise.
9142 (copyprop_hardreg_forward_1): Likewise.
9143 * reload.c (reload_inner_reg_of_subreg): Likewise.
9144 (push_reload): Likewise.
9145 (combine_reloads): Likewise.
9146 (find_dummy_reload): Likewise.
9147 (reload_adjust_reg_for_mode): Likewise.
9148 * reload1.c (find_reload_regs): Likewise.
9149 (forget_old_reloads_1): Likewise.
9150 (reload_reg_free_for_value_p): Likewise.
9151 (reload_adjust_reg_for_temp): Likewise.
9152 (emit_reload_insns): Likewise.
9153 (delete_output_reload): Likewise.
9154 * sel-sched.c (choose_best_reg_1): Likewise.
9155 (choose_best_pseudo_reg): Likewise.
9157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9158 Alan Hayward <alan.hayward@arm.com>
9159 David Sherwood <david.sherwood@arm.com>
9161 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9162 * target.def (slow_unaligned_access): New hook.
9163 * targhooks.h (default_slow_unaligned_access): Declare.
9164 * targhooks.c (default_slow_unaligned_access): New function.
9165 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9166 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9167 * doc/tm.texi: Regenerate.
9168 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9169 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9170 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9172 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9173 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9175 (rs6000_slow_unaligned_access): New function.
9176 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9177 (expand_block_compare): Likewise.
9178 (expand_strn_compare): Likewise.
9179 (rs6000_rtx_costs): Likewise.
9180 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9181 (riscv_slow_unaligned_access): Likewise.
9182 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9183 (riscv_slow_unaligned_access_p): ...this and make static.
9184 (riscv_option_override): Update accordingly.
9185 (riscv_slow_unaligned_access): New function.
9186 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9187 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9188 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9189 (rs6000_slow_unaligned_access): New function.
9190 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9191 (rs6000_rtx_costs): Likewise.
9192 * config/rs6000/rs6000-string.c (expand_block_compare)
9193 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9194 of SLOW_UNALIGNED_ACCESS.
9195 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9196 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9197 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9198 of SLOW_UNALIGNED_ACCESS.
9199 * expmed.c (simple_mem_bitfield_p): Likewise.
9200 * expr.c (alignment_for_piecewise_move): Likewise.
9201 (emit_group_load_1): Likewise.
9202 (emit_group_store): Likewise.
9203 (copy_blkmode_from_reg): Likewise.
9204 (emit_push_insn): Likewise.
9205 (expand_assignment): Likewise.
9206 (store_field): Likewise.
9207 (expand_expr_real_1): Likewise.
9208 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9209 * lra-constraints.c (simplify_operand_subreg): Likewise.
9210 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9211 * gimple-ssa-store-merging.c: Likewise in block comment at start
9213 * tree-ssa-strlen.c: Include target.h.
9214 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9215 of SLOW_UNALIGNED_ACCESS.
9216 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9218 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9220 PR rtl-optimization/82185
9221 * expmed.c (emit_store_flag_int): Only test tem if it has been
9224 2017-09-12 Richard Biener <rguenther@suse.de>
9227 * match.pd ((FTYPE) N CMP CST): Fix typo.
9229 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9231 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9233 (mips_near_type_p): Add 'short_call' attribute as a synonym
9235 * doc/extend.texi (short_call): Document new function attribute.
9237 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9240 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9241 assertion check that in the condition.
9242 (get_atomic_generic_size): Likewise. Before testing if parameter
9243 has pointer type, if it has array type, call for C++
9244 default_conversion to perform array-to-pointer conversion.
9246 2017-09-12 Richard Biener <rguenther@suse.de>
9248 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9249 for operations we cannot scalarize.
9251 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9253 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9254 vectors heap vectors. Clean up comments.
9255 Make visited_bbs a reference.
9256 (profitable_jump_thread_path): Make GC
9257 vectors heap vectors. Clean up comments.
9259 (convert_and_register_jump_thread_path): Make GC vectors heap
9261 (check_subpath_and_update_thread_path): Same. Clean up comments.
9262 Make visited_bbs a reference.
9263 (handle_phi): Abstract common code to to
9264 register_jump_thread_path_if_profitable.
9265 Rename VAR_BB to DEF_BB.
9267 Make GC vectors heap vectors.
9268 Make visited_bbs a reference.
9269 (handle_assignment): Same.
9270 (register_jump_thread_path_if_profitable): New.
9271 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9273 Make GC vectors heap vectors. Clean up comments.
9274 Make visited_bbs a reference.
9275 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9276 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9279 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9282 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9283 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9285 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9287 Revert r251800 and r251799.
9289 2017-09-11 Martin Jambor <mjambor@suse.cz>
9292 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9293 arguments in advance.
9294 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9295 use it to find predecessor edges.
9296 (naive_outof_ssa): Collect vector of predecessors.
9298 2017-09-08 Jason Merrill <jason@redhat.com>
9300 PR c++/70029 - ICE with ref-qualifier and -flto
9301 * langhooks.h (struct lang_hooks_for_types): Add
9302 copy_lang_qualifiers.
9303 * attribs.c (build_type_attribute_qual_variant): Use it.
9304 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9306 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9307 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9309 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9312 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9313 (*mulsi3_sp64): New instruction.
9314 (mulsi3): New expander.
9316 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9318 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9320 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9322 * sancov.c: Include memmodel.h.
9324 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9327 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9330 2017-09-07 Carl Love <cel@us.ibm.com>
9332 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9333 the sldi instruction.
9335 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9337 * sancov.c: Include tm_p.h.
9339 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9342 * output.h (switch_to_other_text_partition): New declaration.
9343 * varasm.c (switch_to_other_text_partition): New function.
9344 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9345 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9346 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9347 to the other text partition before emitting LCL label and switch back
9348 after emitting the word after it.
9350 2017-09-07 Richard Biener <rguenther@suse.de>
9352 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9353 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9354 critical edges here, after loop init.
9355 (pass_data_pre): Remove PROP_no_crit_edges flags.
9356 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9357 for valueization of call args to avoid leaking VN_TOP.
9358 (visit_use): Assert we do not visit default defs.
9359 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9360 Use error_mark_node to more easily detect leaking VN_TOP.
9361 All default-defs are varying, not VN_TOP. Mark them visited.
9362 (run_scc_vn): Make code match comment.
9364 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9366 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9367 OPTION_MASK_FLOAT128_KEYWORD.
9368 (POWERPC_MASKS): Likewise.
9369 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9370 support for the -mfloat128-type option, and make -mfloat128
9371 default on PowerPC Linux systems. Define or undefine
9372 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9373 Define __float128 to be __ieee128 if IEEE 128-bit support is
9374 enabled, or undefine it.
9375 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9376 Delete defining __FLOAT128_TYPE__.
9377 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9378 -mfloat128-type option and make -mfloat128 default on PowerPC
9380 (TARGET_FLOAT128_TYPE): Likewise.
9381 (-mfloat128-type): Likewise.
9382 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9383 Delete the -mfloat128-type option and make -mfloat128 default on
9384 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9385 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9386 128-bit floating point is enabled. Change tests from using
9387 -mfloat128-type to -mfloat128.
9388 (rs6000_mangle_type): Use the correct mangling for the __float128
9389 type even if normal long double is restricted to 64-bits.
9390 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9392 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9393 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9394 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9396 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9397 documentation for -mfloat128.
9399 2017-09-06 Olivier Hainque <hainque@adacore.com>
9401 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9403 2017-09-06 Wish Wu <wishwu007@gmail.com>
9404 Jakub Jelinek <jakub@redhat.com>
9406 * asan.c (initialize_sanitizer_builtins): Add
9407 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9408 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9409 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9410 BT_FN_VOID_UINT64_PTR variables.
9411 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9412 (BT_FN_VOID_UINT16_UINT16): Likewise.
9413 (BT_FN_VOID_UINT32_UINT32): Likewise.
9414 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9415 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9416 (BT_FN_VOID_UINT64_PTR): Likewise.
9417 * common.opt (flag_sanitize_coverage): New variable.
9418 (fsanitize-coverage=trace-pc): Remove.
9419 (fsanitize-coverage=): Add.
9420 * flag-types.h (enum sanitize_coverage_code): New enum.
9421 * fold-const.c (fold_range_test): Disable non-short-circuit
9422 optimization if flag_sanitize_coverage.
9423 (fold_truth_andor): Likewise.
9424 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9425 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9426 (coverage_sanitizer_opts): New array.
9427 (get_closest_sanitizer_option): Add OPTS argument, handle also
9428 OPT_fsanitize_coverage_.
9429 (parse_sanitizer_options): Adjusted to also handle
9430 OPT_fsanitize_coverage_.
9431 (common_handle_option): Add OPT_fsanitize_coverage_.
9432 * sancov.c (instrument_comparison, instrument_switch): New function.
9433 (sancov_pass): Add trace-cmp support.
9434 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9435 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9436 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9437 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9438 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9439 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9440 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9441 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9442 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9443 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9445 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9447 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9448 error. Only quit immediately if parsing is complete.
9449 (BEGIN): Initialize fatal_err and parse_done.
9450 (begin fpu, end fpu): Check number of arguments.
9451 (begin arch, end arch): Likewise.
9452 (begin cpu, end cpu): Likewise.
9453 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9454 (optalias): Likewise.
9456 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9458 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9459 * config/arm/arm-isa.h: Delete. Move definitions to ...
9460 * arm-cpus.in: ... here. Use new feature and fgroup values.
9461 * config/arm/arm.c (arm_option_override): Use lower case for feature
9463 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9464 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9465 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9467 (print_isa_bits_for): New function.
9468 (gen_isa): New function.
9469 (gen_comm_data): Use print_isa_bits_for.
9470 (define feature): New keyword.
9471 (define fgroup): New keyword.
9472 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9473 (arm-isa.h): Add rule to generate file.
9474 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9475 case for feature bit names.
9477 2017-09-06 Richard Biener <rguenther@suse.de>
9479 * tree-ssa-pre.c (NECESSARY): Remove.
9480 (create_expression_by_pieces): Do not touch pass-local flags.
9481 (insert_into_preds_of_block): Likewise.
9482 (do_pre_regular_insertion): Likewise.
9483 (eliminate_insert): Likewise.
9484 (eliminate_dom_walker::before_dom_children): Likewise.
9485 (fini_eliminate): Do not look at inserted_exprs.
9486 (mark_operand_necessary): Remove.
9487 (remove_dead_inserted_code): Replace with simple work-list
9488 algorithm based on inserted_exprs and SSA uses.
9489 (pass_pre::execute): Re-order fini_eliminate and
9490 remove_dead_inserted_code.
9492 2017-09-06 Olivier Hainque <hainque@adacore.com>
9494 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9495 for VxWorks 7. Adjust surrounding comments.
9497 2017-09-06 Richard Biener <rguenther@suse.de>
9499 * gimple-ssa-strength-reduction.c
9500 (find_candidates_dom_walker::before_dom_children): Also allow
9503 2017-09-06 Richard Biener <rguenther@suse.de>
9505 PR tree-optimization/82108
9506 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9507 for gap in the non-permutation SLP case.
9509 2017-09-06 Martin Jambor <mjambor@suse.cz>
9511 PR tree-optimization/82078
9512 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9513 add_access_to_work_queue...
9514 (build_accesses_from_assign): ...here.
9515 (propagate_all_subaccesses): Make sure racc is the group
9516 representative, if there is one.
9518 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9521 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9524 2017-09-06 Richard Biener <rguenther@suse.de>
9526 * gimple-ssa-strength-reduction.c
9527 (find_candidates_dom_walker::before_doom_children): Use a
9528 type and not a mode check.
9530 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9533 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9535 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9537 2017-09-05 Jeff Law <law@redhat.com>
9539 PR tree-optimization/64910
9540 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9541 cases where we have 3 or more operands.
9543 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9546 * omp-low.c (lower_omp_for): Recompute tree invariant if
9547 gimple_omp_for_initial/final is ADDR_EXPR.
9550 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9551 into gimple val before gimplification fo the COND_EXPR.
9553 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9555 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9556 REGION_COPY argument.
9557 (thread_through_all_blocks): Remove unused argument to
9558 duplicate_thread_path.
9560 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9561 Alan Hayward <alan.hayward@arm.com>
9562 David Sherwood <david.sherwood@arm.com>
9564 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9565 Take a scalar_mode rather than a machine_mode.
9566 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9567 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9568 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9569 (aarch64_gen_adjusted_ldpstp): Likewise.
9570 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9572 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9577 Take a scalar_int_mode instead of a machine_mode.
9578 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9579 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9580 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9581 (aarch64_simd_attr_length_rglist): Delete.
9582 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9583 a scalar_int_mode instead of a machine_mode.
9584 (aarch64_add_offset): Likewise.
9585 (aarch64_internal_mov_immediate): Likewise
9586 (aarch64_add_constant_internal): Likewise.
9587 (aarch64_add_constant): Likewise.
9588 (aarch64_movw_imm): Likewise.
9589 (aarch64_rtx_arith_op_extract_p): Likewise.
9590 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9591 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9592 Remove assert that the mode isn't a vector.
9593 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9594 (aarch64_expand_mov_immediate): Update calls after above changes.
9595 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9596 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9597 (aarch64_move_imm): Likewise.
9598 (aarch64_can_const_movi_rtx_p): Likewise.
9599 (aarch64_strip_extend): Likewise.
9600 (aarch64_extr_rtx_p): Likewise.
9601 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9602 a CONST_INT when the mode parameter is VOIDmode.
9603 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9605 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9607 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9608 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9609 (bitwise_type_for_mode): Update accordingly.
9611 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9613 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9614 * stor-layout.c (mode_for_size_tree): Likewise.
9615 (mode_for_array): Update accordingly.
9616 (layout_decl): Likewise.
9617 (compute_record_mode): Likewise. Only set the mode once.
9619 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9621 * target.def (get_mask_mode): Change return type to opt_mode.
9623 * doc/tm.texi: Regenerate.
9624 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9625 * targhooks.c (default_get_mask_mode): Likewise.
9626 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9627 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9628 targetm.get_mask_mode.
9629 * tree.c (build_truth_vector_type): Likewise.
9631 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9633 * machmode.h (mode_for_vector): Return an opt_mode.
9634 * stor-layout.c (mode_for_vector): Likewise.
9635 (mode_for_int_vector): Update accordingly.
9636 (layout_type): Likewise.
9637 * config/i386/i386.c (emit_memmov): Likewise.
9638 (ix86_expand_set_or_movmem): Likewise.
9639 (ix86_expand_vector_init): Likewise.
9640 (ix86_get_mask_mode): Likewise.
9641 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9643 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9644 * expmed.c (extract_bit_field_1): Likewise.
9645 * expr.c (expand_expr_real_2): Likewise.
9646 * optabs-query.c (can_vec_perm_p): Likewise.
9647 (can_vec_mask_load_store_p): Likewise.
9648 * optabs.c (expand_vec_perm): Likewise.
9649 * targhooks.c (default_get_mask_mode): Likewise.
9650 * tree-vect-stmts.c (vectorizable_store): Likewise.
9651 (vectorizable_load): Likewise.
9652 (get_vectype_for_scalar_type_and_size): Likewise.
9654 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9656 * machmode.h (mode_for_int_vector): New function.
9657 * stor-layout.c (mode_for_int_vector): Likewise.
9658 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9659 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9660 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9661 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9662 (s390_expand_vcond): Likewise.
9664 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9666 * machmode.h (opt_machine_mode): New type.
9667 (opt_mode<T>): Allow construction from anything that can be
9669 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9670 (mode_for_size): Return an opt_machine_mode.
9671 * stor-layout.c (mode_for_size): Likewise.
9672 (mode_for_size_tree): Update call accordingly.
9673 (bitwise_mode_for_mode): Likewise.
9674 (make_fract_type): Likewise.
9675 (make_accum_type): Likewise.
9676 * caller-save.c (replace_reg_with_saved_mem): Update call
9678 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9679 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9680 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9681 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9682 * expmed.c (extract_bit_field_1): Likewise.
9683 * reload.c (get_secondary_mem): Likewise.
9684 * varasm.c (assemble_integer): Likewise.
9685 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9690 * machmode.h (decimal_float_mode_for_size): New function.
9691 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9692 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9693 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9694 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9695 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9696 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9698 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9700 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9701 (get_builtin_sync_mode): Likewise.
9702 (expand_ifn_atomic_compare_exchange): Likewise.
9703 (expand_builtin_atomic_clear): Likewise.
9704 (expand_builtin_atomic_test_and_set): Likewise.
9705 (fold_builtin_atomic_always_lock_free): Likewise.
9706 * calls.c (compute_argument_addresses): Likewise.
9707 (emit_library_call_value_1): Likewise.
9708 (store_one_arg): Likewise.
9709 * combine.c (combine_instructions): Likewise.
9710 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9711 * config/arm/arm.c (arm_function_value): Likewise.
9712 (aapcs_allocate_return_reg): Likewise.
9713 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9714 * config/i386/i386.c (construct_container): Likewise.
9715 (ix86_gimplify_va_arg): Likewise.
9716 (ix86_expand_sse_cmp): Likewise.
9717 (emit_memmov): Likewise.
9718 (emit_memset): Likewise.
9719 (expand_small_movmem_or_setmem): Likewise.
9720 (ix86_expand_pextr): Likewise.
9721 (ix86_expand_pinsr): Likewise.
9722 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9723 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9725 * config/mips/mips.c (mips_function_value_1) Likewise.
9726 (mips_block_move_straight): Likewise.
9727 (mips_expand_ins_as_unaligned_store): Likewise.
9728 * config/powerpcspe/powerpcspe.c
9729 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9730 (rs6000_darwin64_record_arg_flush): Likewise.
9731 * config/rs6000/rs6000.c
9732 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9733 (rs6000_darwin64_record_arg_flush): Likewise.
9734 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9735 (sparc_function_value_1): Likewise.
9736 * config/spu/spu.c (adjust_operand): Likewise.
9737 (spu_emit_branch_or_set): Likewise.
9738 (arith_immediate_p): Likewise.
9739 * emit-rtl.c (gen_lowpart_common): Likewise.
9740 * expr.c (expand_expr_real_1): Likewise.
9741 * function.c (assign_parm_setup_block): Likewise.
9742 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9743 * reload1.c (alter_reg): Likewise.
9744 * stor-layout.c (mode_for_vector): Likewise.
9745 (layout_type): Likewise.
9747 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9749 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9750 (spu_convert_move): Likewise.
9751 * lower-subreg.c (resolve_simple_move): Likewise.
9753 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9756 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9757 define_insn to a define_expand.
9758 (altivec_vsum2sws_direct): New define_insn.
9759 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9761 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9763 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9766 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9771 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9772 realign stack if stack alignment needed is less than incoming
9775 2017-09-05 Marek Polacek <polacek@redhat.com>
9778 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9781 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9783 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9785 2017-09-05 Richard Biener <rguenther@suse.de>
9787 PR tree-optimization/82084
9788 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9790 2017-09-05 Richard Biener <rguenther@suse.de>
9792 PR tree-optimization/82102
9793 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9795 2017-09-05 Martin Liska <mliska@suse.cz>
9797 PR tree-optimization/82032
9798 * tree-cfg.c (generate_range_test): New function.
9799 * tree-cfg.h (generate_range_test): Declared here.
9800 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9801 (cleanup_control_expr_graph): Use it.
9802 * tree-switch-conversion.c (try_switch_expansion): Remove
9804 (emit_case_nodes): Use generate_range_test.
9806 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9809 * config/i386/i386.md (*<btsc><mode>_mask): Add
9810 TARGET_USE_BT to insn constraint.
9811 (*btr<mode>_mask): Ditto.
9813 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9815 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9816 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9818 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9821 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9822 TARGET_NEON and TARGET_IWMMXT.
9823 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9824 TARGET_NEON and TARGET_IWMMXT.
9825 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9827 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9829 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9830 (ix86_rewrite_tls_address): Ditto.
9831 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9832 (ix86_rewrite_tls_address_1): Ditto.
9833 (ix86_rewrite_tls_address): Ditto.
9834 * config/i386/predicates.md (tls_address_pattern): New predicate.
9835 * config/i386/i386.md (TLS address splitter): New splitter.
9837 2017-09-04 Richard Biener <rguenther@suse.de>
9839 PR tree-optimization/82084
9840 * fold-const.h (can_native_encode_string_p): Declare.
9841 * fold-const.c (can_native_encode_string_p): Factor out from ...
9842 (native_encode_string): ... here.
9843 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9844 vectorizing stores from constants we later cannot handle.
9846 2017-09-04 Marek Polacek <polacek@redhat.com>
9849 * doc/invoke.texi: Update -Wtautological-compare documentation.
9851 2017-09-04 Jeff Law <law@redhat.com>
9853 PR tree-optimization/64910
9854 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9855 swap the first and last operand if the last is a constant.
9857 2017-09-04 Marek Polacek <polacek@redhat.com>
9860 * convert.c (do_narrow): When sanitizing signed integer overflows,
9861 bail out for signed types.
9862 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9864 2017-09-04 Richard Biener <rguenther@suse.de>
9866 PR tree-optimization/82060
9867 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9868 Move devirtualization after stmt folding and before EH/AB/noreturn
9869 cleanup to get the stmt refs canonicalized. Use a bool instead
9870 of gimple_modified_p since that doesn't work for NOPs. Schedule
9871 NOPs generated by folding for removal.
9873 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9874 Alan Hayward <alan.hayward@arm.com>
9875 David Sherwood <david.sherwood@arm.com>
9877 * coretypes.h (pad_direction): New enum.
9878 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9879 (FUNCTION_ARG_PADDING): Likewise.
9880 * target.def (function_arg_padding): New hook.
9881 * targhooks.h (default_function_arg_padding): Declare.
9882 * targhooks.c (default_function_arg_padding): New function.
9883 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9884 (TARGET_FUNCTION_ARG_PADDING): ...this.
9885 * doc/tm.texi: Regenerate.
9886 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9887 instead of direction.
9888 (compute_argument_addresses): Likewise.
9889 (load_register_parameters): Likewise.
9890 (emit_library_call_value_1): Likewise.
9891 (store_one_arg): Use targetm.calls.function_arg_padding instead
9892 of FUNCTION_ARG_PADDING.
9893 (must_pass_in_stack_var_size_or_pad): Likewise.
9894 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9895 (emit_group_store): Likewise.
9896 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9897 instead of FUNCTION_ARG_PADDING.
9898 (emit_push_insn): Likewise, and propagate enum change throughout
9900 * function.h (direction): Delete.
9901 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9903 * function.c (assign_parm_find_stack_rtl): Likewise.
9904 (assign_parm_setup_block_p): Likewise.
9905 (assign_parm_setup_block): Likewise.
9906 (gimplify_parameters): Likewise.
9907 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9908 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9910 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9911 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9912 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9913 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9914 (aarch64_function_arg_padding): ...this new function.
9915 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9916 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9917 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9918 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9919 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9920 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9921 (arm_pad_arg_upward): Replace with...
9922 (arm_function_arg_padding): ...this new function.
9923 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9925 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9926 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9927 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9928 (ia64_hpux_function_arg_padding): Replace with...
9929 (ia64_function_arg_padding): ...this new function. Use pad_direction
9930 instead of direction. Check for TARGET_HPUX.
9931 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9932 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9933 (iq2000_function_arg_padding): New function.
9934 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9935 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9936 (mips_function_arg_padding): ...this new function.
9937 (mips_pad_reg_upward): Update accordingly.
9938 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9939 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9940 targetm.calls.function_arg_padding.
9941 (FUNCTION_ARG_PADDING): Delete.
9942 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9943 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9944 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9945 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9946 (nios2_block_reg_padding): Return pad_direction instead of direction.
9947 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9948 instead of direction.
9949 (nios2_function_arg_padding): Likewise. Make static.
9950 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9951 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9952 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9953 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9954 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9955 (pa_function_arg_padding): Make static. Return pad_direction instead
9957 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9958 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9959 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9960 instead of direction. Use targetm.calls.function_arg_padding.
9961 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9962 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9963 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9964 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9965 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9967 (function_arg_padding): Rename to...
9968 (rs6000_function_arg_padding): ...this. Make static. Return
9969 pad_direction instead of direction.
9970 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9971 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9972 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9973 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9974 instead of direction. Use targetm.calls.function_arg_padding.
9975 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9976 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9977 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9978 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9979 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9980 (function_arg_padding): Rename to...
9981 (rs6000_function_arg_padding): ...this. Make static. Return
9982 pad_direction instead of direction.
9983 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9984 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9985 * config/s390/s390.c (s390_function_arg_padding): New function.
9986 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9987 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9988 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9989 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9990 (function_arg_padding): Rename to...
9991 (sparc_function_arg_padding): ...this. Make static. Return
9992 pad_direction instead of direction.
9993 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9994 * config/spu/spu.c (spu_function_arg_padding): New function.
9995 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9996 * system.h (FUNCTION_ARG_PADDING): Poison.
9998 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9999 Alan Hayward <alan.hayward@arm.com>
10000 David Sherwood <david.sherwood@arm.com>
10002 * target.def (modes_tieable_p): New hook.
10003 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10004 (TARGET_MODES_TIEABLE_P): ...this.
10005 * doc/tm.texi.in: Regenerate.
10006 * hooks.h (hook_bool_mode_mode_true): Declare.
10007 * hooks.c (hook_bool_mode_mode_true): New function.
10008 * combine.c (subst): Use targetm.modes_tieable_p instead of
10010 * dse.c (find_shift_sequence): Likewise.
10011 * expmed.c (extract_low_bits): Likewise.
10012 * lower-subreg.c: Include target.h.
10013 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10015 * rtlanal.c (rtx_cost): Likewise.
10016 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10017 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10018 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10019 (TARGET_MODES_TIEABLE_P): Redefine.
10020 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10021 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10022 (TARGET_MODES_TIEABLE_P): Redefine.
10023 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10024 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10025 (arc_modes_tieable_p): New function.
10026 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10027 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10028 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10029 (arm_modes_tieable_p): Make static.
10030 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10031 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10032 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10033 (TARGET_MODES_TIEABLE_P): Redefine.
10034 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10035 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10036 (TARGET_MODES_TIEABLE_P): Redefine.
10037 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10038 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10039 (cr16_modes_tieable_p): New function.
10040 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10041 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10042 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10043 (TRULY_NOOP_TRUNCATION): Update comment.
10044 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10045 (TRULY_NOOP_TRUNCATION): Update comment.
10046 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10047 (frv_modes_tieable_p): New function.
10048 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10049 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10050 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10051 (TARGET_MODES_TIEABLE_P): Redefine.
10052 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10053 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10054 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10055 (TARGET_MODES_TIEABLE_P): Redefine.
10056 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10057 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10058 (ia64_modes_tieable_p): New function.
10059 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10060 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10061 (iq2000_modes_tieable_p): New function.
10062 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10063 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10064 (lm32_modes_tieable_p): New function.
10065 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10066 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10067 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10068 (TARGET_MODES_TIEABLE_P): Redefine.
10069 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10070 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10071 (m32r_modes_tieable_p): New function.
10072 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10073 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10074 (m68k_modes_tieable_p): New function.
10075 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10076 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10077 (mcore_modes_tieable_p): New function.
10078 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10079 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10081 (TARGET_MODES_TIEABLE_P): Redefine.
10082 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10083 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10084 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10085 (TARGET_MODES_TIEABLE_P): Redefine.
10086 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10087 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10088 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10089 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10090 (mn10300_modes_tieable_p): ...this and make static.
10091 (TARGET_MODES_TIEABLE_P): Redefine.
10092 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10093 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10094 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10095 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10096 (msp430_modes_tieable_p): Make static.
10097 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10098 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10099 (TARGET_MODES_TIEABLE_P): Redefine.
10100 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10101 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10102 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10103 (TARGET_MODES_TIEABLE_P): Redefine.
10104 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10105 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10106 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10107 (TARGET_MODES_TIEABLE_P): Redefine.
10108 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10109 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10110 (pdp11_modes_tieable_p): New function.
10111 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10112 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10113 (rs6000_modes_tieable_p): New function.
10114 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10115 * config/powerpcspe/powerpcspe.md: Update comment.
10116 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10117 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10118 (TARGET_MODES_TIEABLE_P): Redefine.
10119 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10120 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10121 (rl78_modes_tieable_p): New function.
10122 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10123 * config/rs6000/rs6000.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/rs6000/rs6000.md: Update comment.
10127 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10128 * config/rx/rx.c (rx_modes_tieable_p): New function.
10129 (TARGET_MODES_TIEABLE_P): Redefine.
10130 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10131 * config/s390/s390.c (s390_modes_tieable_p): New function.
10132 (TARGET_MODES_TIEABLE_P): Redefine.
10133 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10134 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10135 (sh_modes_tieable_p): New function.
10136 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10137 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10138 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10139 (sparc_modes_tieable_p): Make static.
10140 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10141 * config/spu/spu.c (spu_modes_tieable_p): New function.
10142 (TARGET_MODES_TIEABLE_P): Redefine.
10143 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10144 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10145 (TARGET_MODES_TIEABLE_P): Redefine.
10146 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10147 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10148 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10149 * config/v850/v850.c (v850_modes_tieable_p): New function.
10150 (TARGET_MODES_TIEABLE_P): Redefine.
10151 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10152 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10153 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10154 (visium_modes_tieable_p): New function.
10155 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10156 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10157 (xtensa_modes_tieable_p): New function.
10158 * system.h (MODES_TIEABLE_P): Poison.
10160 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10161 Alan Hayward <alan.hayward@arm.com>
10162 David Sherwood <david.sherwood@arm.com>
10164 * target.def (hard_regno_mode_ok): New hook.
10165 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10166 (TARGET_HARD_REGNO_MODE_OK): ...this.
10167 * doc/tm.texi.in: Regenerate.
10168 * hooks.h (hook_bool_uint_mode_true): Declare.
10169 * hooks.c (hook_bool_uint_mode_true): New function.
10170 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10171 HARD_REGNO_MODE_OK.
10172 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10173 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10174 instead of HARD_REGNO_MODE_OK.
10175 * caller-save.c: Include target.h.
10176 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10177 HARD_REGNO_MODE_OK.
10178 * combine.c (can_combine_p): Likewise.
10179 (combinable_i3pat): Likewise.
10180 (can_change_dest_mode): Likewise.
10181 * expr.c (init_expr_target): Likewise.
10182 (convert_move): Likewise.
10183 (convert_modes): Likewise.
10184 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10185 (setup_prohibited_mode_move_regs): Likewise.
10186 * ira.h (target_ira): Likewise.
10187 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10188 * lra-constraints.c (process_alt_operands): Likewise.
10189 (split_reg): Likewise.
10190 * recog.c (peep2_find_free_register): Likewise.
10191 * ree.c (combine_reaching_defs): Likewise.
10192 * regcprop.c (maybe_mode_change): Likewise.
10193 * reginfo.c (init_reg_sets_1): Likewise.
10194 (choose_hard_reg_mode): Likewise.
10195 (simplifiable_subregs): Likewise.
10196 * regrename.c (check_new_reg_p): Likewise.
10197 * reload.c (find_valid_class): Likewise.
10198 (find_valid_class_1): Likewise.
10199 (reload_inner_reg_of_subreg): Likewise.
10200 (push_reload): Likewise.
10201 (combine_reloads): Likewise.
10202 (find_dummy_reload): Likewise.
10203 (find_reloads): Likewise.
10204 * reload1.c (find_reg): Likewise.
10205 (set_reload_reg): Likewise.
10206 (allocate_reload_reg): Likewise.
10207 (choose_reload_regs): Likewise.
10208 (reload_adjust_reg_for_temp): Likewise.
10209 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10210 (simplify_subreg_regno): Likewise.
10211 * sel-sched.c (init_regs_for_mode): Likewise.
10212 * varasm.c (make_decl_rtl): Likewise.
10213 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10214 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10215 HARD_REGNO_MODE_OK.
10216 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10217 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10218 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10219 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10220 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10221 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10222 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10223 (arc_mode_class): Delete.
10224 (HARD_REGNO_MODE_OK): Delete.
10225 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10226 (arc_hard_regno_mode_ok): Rename old array to...
10227 (arc_hard_regno_mode_ok_modes): ...this.
10228 (arc_conditional_register_usage): Update accordingly.
10229 (arc_mode_class): Make static.
10230 (arc_hard_regno_mode_ok): New function.
10231 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10232 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10233 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10234 (arm_hard_regno_mode_ok): Make static.
10235 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10236 HARD_REGNO_MODE_OK.
10237 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10238 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10239 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10242 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10243 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10244 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10245 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10247 * config/bfin/predicates.md (valid_reg_operand): Use
10248 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10249 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10250 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10251 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10252 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10253 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10254 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10255 (cr16_hard_regno_mode_ok): Make static and return a bool.
10256 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10257 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10258 (cris_hard_regno_mode_ok): New function.
10259 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10260 (epiphany_mode_class): Delete.
10261 (HARD_REGNO_MODE_OK): Delete.
10262 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10263 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10264 (hard_regno_mode_ok): Rename to...
10265 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10266 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10267 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10268 HARD_REGNO_MODE_OK.
10269 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10270 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10271 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10272 (frv_hard_regno_mode_ok): Make static and return a bool.
10273 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10274 HARD_REGNO_MODE_OK.
10275 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10276 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10277 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10278 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10281 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10282 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10283 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10285 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10286 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10287 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10288 (ia64_hard_regno_mode_ok): New function.
10289 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10290 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10291 (iq2000_hard_regno_mode_ok): New function.
10292 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10293 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10294 (lm32_hard_regno_mode_ok): New function.
10295 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10296 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10297 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10298 instead of HARD_REGNO_MODE_OK.
10299 (m32c_hard_regno_ok): Rename to...
10300 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10301 (m32c_cannot_change_mode_class): Update accordingly.
10302 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10303 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10304 (m32r_mode_class): Delete.
10305 (HARD_REGNO_MODE_OK): Delete.
10306 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10307 (m32r_hard_regno_mode_ok): Rename to...
10308 (m32r_hard_regno_modes): ...this.
10309 (m32r_mode_class): Make static.
10310 (m32r_hard_regno_mode_ok): New function.
10311 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10312 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10313 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10314 (m68k_hard_regno_mode_ok): Make static.
10315 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10316 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10317 (mcore_hard_regno_mode_ok): New function.
10318 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10319 (HARD_REGNO_MODE_OK): Delete.
10320 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10322 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10323 (microblaze_hard_regno_mode_ok): New function.
10324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10325 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10326 (mips_hard_regno_mode_ok): Delete.
10327 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10328 (mips_hard_regno_mode_ok_p): ...this and make static.
10329 (mips_hard_regno_mode_ok_p): Rename to...
10330 (mips_hard_regno_mode_ok_uncached): ...this.
10331 (mips_hard_regno_mode_ok): New function.
10332 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10333 of HARD_REGNO_MODE_OK.
10334 (mips_option_override): Update after above name changes.
10335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10336 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10337 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10338 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10339 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10340 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10341 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10342 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10343 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10344 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10345 (msp430_hard_regno_mode_ok): Make static and return a bool.
10346 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10347 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10348 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10351 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10352 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10353 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10354 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10355 (PA_HARD_REGNO_MODE_OK): ...this
10356 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10357 (PA_HARD_REGNO_MODE_OK): ...this.
10358 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10359 (pa_hard_regno_mode_ok): New function.
10360 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10361 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10362 (pdp11_hard_regno_mode_ok): New function.
10363 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10364 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10366 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10369 (rs6000_hard_regno_mode_ok): Rename to...
10370 (rs6000_hard_regno_mode_ok_uncached): ...this.
10371 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10372 (rs6000_hard_regno_mode_ok): New function.
10373 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10374 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10375 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10376 (riscv_hard_regno_mode_ok): ...this and make static.
10377 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10378 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10379 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10380 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10381 (rl78_hard_regno_mode_ok): Make static and return bool.
10382 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10383 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10385 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10386 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10387 (rs6000_hard_regno_mode_ok): Rename to...
10388 (rs6000_hard_regno_mode_ok_uncached): ...this.
10389 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10390 (rs6000_hard_regno_mode_ok): New function.
10391 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10392 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10394 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10395 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10396 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10398 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10399 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10400 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10401 (sh_hard_regno_mode_ok): Make static.
10402 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10403 instead of HARD_REGNO_MODE_OK.
10404 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10405 (sparc_mode_class): Delete.
10406 (HARD_REGNO_MODE_OK): Delete.
10407 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10408 (hard_regno_mode_classes): Make static.
10409 (sparc_mode_class): Likewise.
10410 (sparc_hard_regno_mode_ok): New function.
10411 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10412 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10413 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10415 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10416 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10417 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10418 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10419 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10421 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10422 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10423 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10424 (visium_hard_regno_mode_ok): New function.
10425 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10426 instead of HARD_REGNO_MODE_OK.
10427 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10428 (HARD_REGNO_MODE_OK): Delete.
10429 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10430 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10431 (xtensa_option_override): Update accordingly.
10432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10433 (xtensa_hard_regno_mode_ok): New function.
10434 * system.h (HARD_REGNO_MODE_OK): Poison.
10436 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10437 Alan Hayward <alan.hayward@arm.com>
10438 David Sherwood <david.sherwood@arm.com>
10440 * target.def (hard_regno_call_part_clobbered): New hook.
10441 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10442 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10443 * doc/tm.texi: Regenerate.
10444 * hooks.h (hook_bool_uint_mode_false): Declare.
10445 * hooks.c (hook_bool_uint_mode_false): New function.
10446 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10447 * cselib.c (cselib_process_insn): Use
10448 targetm.hard_regno_call_part_clobbered instead of
10449 HARD_REGNO_CALL_PART_CLOBBERED.
10450 * ira-conflicts.c (ira_build_conflicts): Likewise.
10451 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10452 * lra-constraints.c (need_for_call_save_p): Likewise.
10453 * lra-lives.c: Include target.h.
10454 (check_pseudos_live_through_calls): Use
10455 targetm.hard_regno_call_part_clobbered instead of
10456 HARD_REGNO_CALL_PART_CLOBBERED.
10457 * regcprop.c: Include target.h.
10458 (copyprop_hardreg_forward_1): Use
10459 targetm.hard_regno_call_part_clobbered instead of
10460 HARD_REGNO_CALL_PART_CLOBBERED.
10461 * reginfo.c (choose_hard_reg_mode): Likewise.
10462 * regrename.c (check_new_reg_p): Likewise.
10463 * reload.c (find_equiv_reg): Likewise.
10464 * reload1.c (emit_reload_insns): Likewise.
10465 * sched-deps.c (deps_analyze_insn): Likewise.
10466 * sel-sched.c (init_regs_for_mode): Likewise.
10467 (mark_unavailable_hard_regs): Likewise.
10468 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10469 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10470 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10472 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10473 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10474 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10476 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10478 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10479 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10480 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10482 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10483 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10484 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10486 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10487 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10489 * config/powerpcspe/powerpcspe.c
10490 (rs6000_hard_regno_call_part_clobbered): New function.
10491 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10492 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10493 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10495 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10496 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10497 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10499 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10500 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10501 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10503 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10504 Alan Hayward <alan.hayward@arm.com>
10505 David Sherwood <david.sherwood@arm.com>
10507 * rtl.h (subreg_memory_offset): Declare.
10508 * emit-rtl.c (subreg_memory_offset): New function.
10509 * expmed.c (store_bit_field_1): Use it.
10510 * expr.c (undefined_operand_subword_p): Likewise.
10511 * simplify-rtx.c (simplify_subreg): Likewise.
10513 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10515 PR rtl-optimization/57448
10518 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10519 using atomic_load pattern.
10520 (expand_atomic_store): Likewise.
10522 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10525 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10526 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10527 handling. Use replace_call_with_value with NULL instead of
10528 gsi_replace, unlink_stmt_vdef and release_defs.
10530 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10533 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10535 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10538 * rtl.h (emit_library_call_value_1): Declare.
10539 (emit_library_call): Replace declaration with a series of overloads.
10540 Remove the parameter count argument.
10541 (emit_library_call_value): Likewise.
10542 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10543 with an "rtx_mode_t *".
10544 (emit_library_call_value): Delete.
10545 (emit_library_call): Likewise.
10546 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10547 (asan_emit_allocas_unpoison): Likewise.
10548 * builtins.c (expand_builtin_powi): Likewise.
10549 (expand_asan_emit_allocas_unpoison): Likewise.
10550 * cfgexpand.c (expand_main_function): Likewise.
10551 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10552 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10553 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10554 * config/arm/arm.c (arm_trampoline_init): Likewise.
10555 (arm_call_tls_get_addr): Likewise.
10556 (arm_expand_divmod_libfunc): Likewise.
10557 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10558 (smulsi3_highpart): Likewise.
10559 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10560 (c6x_expand_compare): Likewise.
10561 (c6x_expand_movmem): Likewise.
10562 * config/frv/frv.c (frv_trampoline_init): Likewise.
10563 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10564 (ix86_expand_divmod_libfunc): Likewise.
10565 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10566 (ia64_expand_compare): Likewise.
10567 (ia64_profile_hook): Likewise.
10568 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10569 (nonlocal_goto): Likewise.
10570 (restore_stack_nonlocal): Likewise.
10571 * config/m32r/m32r.c (block_move_call): Likewise.
10572 (m32r_trampoline_init): Likewise.
10573 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10574 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10575 (m68k_call_m68k_read_tp): Likewise.
10576 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10577 (microblaze_expand_divide): Likewise.
10578 * config/mips/mips.h (mips_args): Likewise.
10579 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10580 (MIPS_ICACHE_SYNC): Likewise.
10581 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10582 (nios2_trampoline_init): Likewise.
10583 * config/pa/pa.c (hppa_tls_call): Likewise.
10584 (pa_trampoline_init): Likewise.
10585 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10586 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10587 (expand_strn_compare): Likewise.
10588 (rs6000_generate_compare): Likewise.
10589 (rs6000_expand_float128_convert): Likewise.
10590 (output_profile_hook): Likewise.
10591 (rs6000_trampoline_init): Likewise.
10592 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10593 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10594 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10595 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10596 (rs6000_generate_compare): Likewise.
10597 (rs6000_expand_float128_convert): Likewise.
10598 (output_profile_hook): Likewise.
10599 (rs6000_trampoline_init): Likewise.
10600 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10601 * config/sh/sh.c (sh_trampoline_init): Likewise.
10602 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10603 (sparc_emit_float_lib_cmp): Likewise.
10604 (sparc32_initialize_trampoline): Likewise.
10605 (sparc64_initialize_trampoline): Likewise.
10606 (sparc_profile_hook): Likewise.
10607 * config/spu/spu.c (ea_load_store): Likewise.
10608 * config/spu/spu.md (floatunssidf2): Likewise.
10609 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10610 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10611 * config/visium/visium.c (expand_block_move_4): Likewise.
10612 (expand_block_move_2): Likewise.
10613 (expand_block_move_1): Likewise.
10614 (expand_block_set_4): Likewise.
10615 (expand_block_set_2): Likewise.
10616 (expand_block_set_1): Likewise.
10617 (visium_trampoline_init): Likewise.
10618 (visium_profile_hook): Likewise.
10619 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10620 (xtensa_setup_frame_addresses): Likewise.
10621 (xtensa_trampoline_init): Likewise.
10622 * except.c (sjlj_emit_function_enter): Likewise.
10623 (sjlj_emit_function_exit): Likewise.
10624 * explow.c (allocate_dynamic_stack_space): Likewise.
10625 (probe_stack_range): Likewise.
10626 * expr.c (convert_mode_scalar): Likewise.
10627 * optabs.c (expand_binop): Likewise.
10628 (expand_twoval_binop_libfunc): Likewise.
10629 (expand_unop): Likewise.
10630 (prepare_cmp_insn): Likewise.
10631 (prepare_float_lib_cmp): Likewise.
10632 (expand_float): Likewise.
10633 (expand_fix): Likewise.
10634 (expand_fixed_convert): Likewise.
10635 (maybe_emit_sync_lock_test_and_set): Likewise.
10636 (expand_atomic_compare_and_swap): Likewise.
10637 (expand_mem_thread_fence): Likewise.
10638 (expand_atomic_fetch_op): Likewise.
10640 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10642 * doc/generic.texi (OpenACC): Adjust URL.
10643 * doc/invoke.texi (C Dialect Options): Ditto.
10645 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10647 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10648 predicate for operand 1. Add (m,<S>) constraint.
10649 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10650 Prevent memory operand 1 with register operand 2.
10652 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10654 PR rtl-optimization/82024
10655 * combine.c (try_combine): If the combination result is a PARALLEL,
10656 and we only need to retain the SET in there that would be placed
10657 at I2, check that we can place that at I3 instead, before doing so.
10659 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10662 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10664 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10665 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10668 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10669 Jeff Law <law@redhat.com>
10671 * varasm.c (bss_initializer_p): Do not put constants into .bss
10672 (categorize_decl_for_section): Handle bss_initializer_p returning
10673 false when DECL_INITIAL is NULL.
10675 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10678 * config/s390/s390.c (s390_can_inline_p): New function.
10680 2017-09-01 Jeff Law <law@redhat.com>
10682 PR tree-optimization/82052
10683 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10684 Always initialize the returned slot after a hash table miss
10685 when INSERT is true.
10687 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10689 * config/s390/s390.md (mem_signal_fence): Remove.
10690 * doc/md.texi (mem_signal_fence): Remove.
10691 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10693 * target-insns.def (mem_signal_fence): Remove.
10695 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10698 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10701 * asan.c (create_odr_indicator): Strip name encoding from assembler
10702 name before appending it after __odr_asan_.
10704 2017-09-01 Martin Liska <mliska@suse.cz>
10706 PR tree-optimization/82059
10707 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10708 frequency only when an edge is redirected.
10710 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10712 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10713 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10714 (arc_conditional_register_usage): Remove ARC600 lp_count
10716 (arc_file_start): Emit Tag_ARC_CPU_variation.
10717 (arc_can_use_doloop_p): New conditions to use ZOLs.
10718 (hwloop_fail): New function.
10719 (hwloop_optimize): Likewise.
10720 (hwloop_pattern_reg): Likewise.
10721 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10722 (arc_reorg_loops): New function, calls reorg_loops.
10723 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10724 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10726 (arc_loop_hazard): Remove function, functionality moved into
10728 (arc_hazard): Remove arc_loop_hazard call.
10729 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10730 into hwloop_optimize.
10731 (arc_label_align): Remove ZOL handling.
10732 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10733 * config/arc/arc.md (doloop_begin): Remove pattern.
10734 (doloop_begin_i): Likewise.
10735 (doloop_end_i): Likewise.
10736 (doloop_fallback): Likewise.
10737 (doloop_fallback_m): Likewise.
10738 (doloop_end): Reimplement expand.
10739 (arc_lp): New pattern for LP instruction.
10740 (loop_end): New pattern.
10741 (loop_fail): Likewise.
10742 (decrement_and_branch_until_zero): Likewise.
10743 * config/arc/arc.opt (mlpc-width): New option.
10744 * doc/invoke.texi (mlpc-width): Document option.
10746 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10748 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10749 (arc_ccfsm_advance): Fix checking for delay slots.
10750 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10752 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10754 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10756 (movhi_insn): Update store instruction constraint which are saving
10757 6-bit short immediates.
10758 (movsi_insn): Consider also short scaled load operations.
10759 (zero_extendhisi2_i): Use Usd constraint instead of T.
10760 (extendhisi2_i): Add q constraint.
10761 (arc_clzsi2): Add type and length attributes.
10762 (arc_ctzsi2): Likewise.
10763 * config/arc/constraints.md (Usc): Update constraint, the
10764 assembler can parse two relocations for a single instruction.
10766 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10768 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10769 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10771 2017-08-31 Olivier Hainque <hainque@adacore.com>
10773 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10774 match as powerpc-wrs-vxworks*.
10776 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10778 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10779 register constraint for by-element operand.
10780 (aarch64_mls_elt_merge<mode>): Likewise.
10782 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10784 * config/arc/arc.c (arc_can_follow_jump): Check for short
10787 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10789 * config.gcc: Use g.opt for arc.
10790 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10791 functionality moved to ...
10792 (legitimate_scaled_address_p): New function, ...here.
10793 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10794 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10795 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10797 (arc_override_options): Handle G option.
10798 (arc_output_pic_addr_const): Correct function definition.
10799 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10800 (arc_decl_anon_ns_mem_p): Delete.
10801 (arc_in_small_data_p): Overhaul this function to take into
10802 consideration the value given via G option.
10803 (arc_rewrite_small_data_1): Renamed and corrected old
10804 arc_rewrite_small_data function.
10805 (arc_rewrite_small_data): New function.
10806 (small_data_pattern): Don't use pic_offset_table_rtx.
10807 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10808 * config/arc/simdext.md (movmisalignv2hi): Use
10809 prepare_move_operands function.
10811 (movmisalign*): Likewise.
10812 * doc/invoke.texi (ARC options): Document -G option.
10814 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10816 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10818 * config/arc/arc.c (arc_print_operand): Output scalled address for
10819 sdata whenever is possible.
10820 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10821 load/stores are available.
10822 (compact_sda_memory_operand): Check for the alignment required by
10823 code density instructions.
10824 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10826 * config/arc/constraints.md (Usd): Update constraint.
10827 (Us0): New constraint.
10828 (Usc): Update constraint.
10830 2017-08-31 Richard Biener <rguenther@suse.de>
10832 PR middle-end/82054
10833 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10834 function only once.
10836 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10838 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10839 Resize type_signature.
10841 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10842 Alan Hayward <alan.hayward@arm.com>
10843 David Sherwood <david.sherwood@arm.com>
10845 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10846 subregs whose inner modes can be stored in GPRs.
10847 (aarch64_classify_index): Likewise.
10849 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10850 Alan Hayward <alan.hayward@arm.com>
10851 David Sherwood <david.sherwood@arm.com>
10853 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10854 (V_INT_EQUIV): ...this.
10855 (v_cmp_result): Rename to...
10856 (v_int_equiv): ...this.
10857 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10858 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10859 (copysign<mode>3): Likewise.
10860 (aarch64_simd_bsl<mode>_internal): Likewise.
10861 (aarch64_simd_bsl<mode>): Likewise.
10862 (vec_cmp<mode><mode>): Likewise.
10863 (vcond<mode><mode>): Likewise.
10864 (vcond<v_cmp_mixed><mode>): Likewise.
10865 (vcondu<mode><v_cmp_mixed>): Likewise.
10866 (aarch64_cm<optab><mode>): Likewise.
10867 (aarch64_cmtst<mode>): Likewise.
10868 (aarch64_fac<optab><mode>): Likewise.
10869 (vec_perm_const<mode>): Likewise.
10870 (vcond_mask_<mode><v_cmp_result>): Rename to...
10871 (vcond_mask_<mode><v_int_equiv>): ...this.
10872 (vec_cmp<mode><v_cmp_result>): Rename to...
10873 (vec_cmp<mode><v_int_equiv>): ...this.
10875 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10876 Alan Hayward <alan.hayward@arm.com>
10877 David Sherwood <david.sherwood@arm.com>
10879 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10881 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10882 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10883 (UNSPEC_LD4_DREG): New unspecs.
10884 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10885 (aarch64_ld2<mode>_dreg_be): Replace with...
10886 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10888 (aarch64_ld3<mode>_dreg_le)
10889 (aarch64_ld3<mode>_dreg_be): Replace with...
10890 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10892 (aarch64_ld4<mode>_dreg_le)
10893 (aarch64_ld4<mode>_dreg_be): Replace with...
10894 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10897 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10899 PR tree-optimization/81987
10900 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10901 insert an initializer in a location not dominated by the stride
10904 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10906 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10907 on the entire header of the finally block in the fallthru case.
10909 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10911 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10913 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10915 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10916 rs6000_emit_move_from_cr and call renamed function.
10917 (rs6000_emit_prologue): Call renamed functions.
10918 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10919 movesi_from_cr, remove volatile CRs.
10921 2017-08-30 Jon Beniston <jon@beniston.com>
10922 Richard Biener <rguenther@suse.de>
10924 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10925 of VECTOR_MODE_P check.
10926 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10927 element vector types.
10929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10931 * df.h (df_read_modify_subreg_p): Remove in favor of...
10932 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10933 const_rtx instead of an rtx.
10934 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10935 * df-problems.c (df_word_lr_mark_ref): Likewise.
10936 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10937 (mark_pseudo_reg_dead): Likewise.
10938 (mark_ref_dead): Likewise.
10939 * reginfo.c (init_subregs_of_mode): Likewise.
10940 * sched-deps.c (sched_analyze_1): Likewise.
10941 * df-scan.c (df_def_record_1): Likewise.
10942 (df_uses_record): Likewise.
10943 (df_read_modify_subreg_p): Remove in favor of...
10944 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10945 const_rtx instead of an rtx.
10947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10948 Alan Hayward <alan.hayward@arm.com>
10949 David Sherwood <david.sherwood@arm.com>
10951 * rtl.h (partial_subreg_p): New function.
10952 * caller-save.c (save_call_clobbered_regs): Use it.
10953 * calls.c (expand_call): Likewise.
10954 * combine.c (combinable_i3pat): Likewise.
10955 (simplify_set): Likewise.
10956 (make_extraction): Likewise.
10957 (make_compound_operation_int): Likewise.
10958 (gen_lowpart_or_truncate): Likewise.
10959 (force_to_mode): Likewise.
10960 (make_field_assignment): Likewise.
10961 (reg_truncated_to_mode): Likewise.
10962 (record_truncated_value): Likewise.
10963 (move_deaths): Likewise.
10964 * cse.c (record_jump_cond): Likewise.
10965 (cse_insn): Likewise.
10966 * cselib.c (cselib_lookup_1): Likewise.
10967 * expmed.c (extract_bit_field_using_extv): Likewise.
10968 * function.c (assign_parm_setup_reg): Likewise.
10969 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10970 * ira-build.c (create_insn_allocnos): Likewise.
10971 * lra-coalesce.c (merge_pseudos): Likewise.
10972 * lra-constraints.c (match_reload): Likewise.
10973 (simplify_operand_subreg): Likewise.
10974 (curr_insn_transform): Likewise.
10975 * lra-lives.c (process_bb_lives): Likewise.
10976 * lra.c (new_insn_reg): Likewise.
10977 (lra_substitute_pseudo): Likewise.
10978 * regcprop.c (mode_change_ok): Likewise.
10979 (maybe_mode_change): Likewise.
10980 (copyprop_hardreg_forward_1): Likewise.
10981 * reload.c (push_reload): Likewise.
10982 (find_reloads): Likewise.
10983 (find_reloads_subreg_address): Likewise.
10984 * reload1.c (alter_reg): Likewise.
10985 (eliminate_regs_1): Likewise.
10986 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10988 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10990 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10991 back to if statements, including unpack.
10993 2017-08-30 Martin Liska <mliska@suse.cz>
10995 PR inline-asm/82001
10996 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10998 (func_checker::compare_asm_inputs_outputs): ... this function.
10999 (func_checker::compare_gimple_asm): Use the function to compare
11000 also ASM constrains.
11001 * ipa-icf-gimple.h: Rename the function.
11003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11004 Alan Hayward <alan.hayward@arm.com>
11005 David Sherwood <david.sherwood@arm.com>
11007 * coretypes.h (complex_mode): New type.
11008 * gdbhooks.py (build_pretty_printer): Handle it.
11009 * machmode.h (complex_mode): New class.
11010 (complex_mode::includes_p): New function.
11011 (is_complex_int_mode): Likewise.
11012 (is_complex_float_mode): Likewise.
11013 * genmodes.c (get_mode_class): Handle complex mode classes.
11014 * function.c (expand_function_end): Use is_complex_int_mode.
11016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11017 Alan Hayward <alan.hayward@arm.com>
11018 David Sherwood <david.sherwood@arm.com>
11020 * coretypes.h (scalar_mode_pod): New typedef.
11021 * gdbhooks.py (build_pretty_printer): Handle it.
11022 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11023 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11024 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11025 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11026 as_a <scalar_mode>.
11028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11029 Alan Hayward <alan.hayward@arm.com>
11030 David Sherwood <david.sherwood@arm.com>
11032 * machmode.h (mode_for_vector): Take a scalar_mode instead
11034 * stor-layout.c (mode_for_vector): Likewise.
11035 * explow.c (promote_mode): Use as_a <scalar_mode>.
11036 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11039 Alan Hayward <alan.hayward@arm.com>
11040 David Sherwood <david.sherwood@arm.com>
11042 * target.def (preferred_simd_mode): Take a scalar_mode
11043 instead of a machine_mode.
11044 * targhooks.h (default_preferred_simd_mode): Likewise.
11045 * targhooks.c (default_preferred_simd_mode): Likewise.
11046 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11047 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11048 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11049 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11050 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11051 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11052 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11053 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11055 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11056 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11057 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11058 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11059 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11060 * doc/tm.texi: Regenerate.
11061 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11065 Alan Hayward <alan.hayward@arm.com>
11066 David Sherwood <david.sherwood@arm.com>
11068 * target.def (scalar_mode_supported_p): Take a scalar_mode
11069 instead of a machine_mode.
11070 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11071 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11072 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11073 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11074 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11075 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11076 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11077 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11078 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11079 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11080 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11081 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11082 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11083 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11085 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11086 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11087 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11088 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11089 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11091 * doc/tm.texi: Regenerate.
11093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11094 Alan Hayward <alan.hayward@arm.com>
11095 David Sherwood <david.sherwood@arm.com>
11097 * coretypes.h (opt_scalar_mode): New typedef.
11098 * gdbhooks.py (build_pretty_printers): Handle it.
11099 * machmode.h (mode_iterator::get_2xwider): Add overload for
11101 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11103 * expr.c (convert_mode_scalar): Likewise.
11104 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11105 * optabs.c (expand_float): Likewise.
11106 (expand_fix): Likewise.
11107 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11110 Alan Hayward <alan.hayward@arm.com>
11111 David Sherwood <david.sherwood@arm.com>
11113 * optabs.c (expand_float): Explicitly check for scalars before
11114 using a branching expansion.
11115 (expand_fix): Likewise.
11117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11118 Alan Hayward <alan.hayward@arm.com>
11119 David Sherwood <david.sherwood@arm.com>
11121 * expr.c (convert_mode): Split scalar handling out into...
11122 (convert_mode_scalar): ...this new function. Treat the modes
11125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11126 Alan Hayward <alan.hayward@arm.com>
11127 David Sherwood <david.sherwood@arm.com>
11129 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11131 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11134 Alan Hayward <alan.hayward@arm.com>
11135 David Sherwood <david.sherwood@arm.com>
11137 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11138 rather than a machine_mode.
11139 (fixed_from_string): Likewise.
11140 (fixed_convert): Likewise.
11141 (fixed_convert_from_int): Likewise.
11142 (fixed_convert_from_real): Likewise.
11143 (real_convert_from_fixed): Likewise.
11144 * fixed-value.c (fixed_from_double_int): Likewise.
11145 (fixed_from_string): Likewise.
11146 (fixed_convert): Likewise.
11147 (fixed_convert_from_int): Likewise.
11148 (fixed_convert_from_real): Likewise.
11149 (real_convert_from_fixed): Likewise.
11150 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11153 Alan Hayward <alan.hayward@arm.com>
11154 David Sherwood <david.sherwood@arm.com>
11156 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11157 of separate mode class checks. Do not allow vector modes here.
11158 (immed_wide_int_const): Use as_a <scalar_mode>.
11159 * explow.c (trunc_int_for_mode): Likewise.
11160 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11161 (wi::shwi): Likewise.
11162 (wi::min_value): Likewise.
11163 (wi::max_value): Likewise.
11164 * dwarf2out.c (loc_descriptor): Likewise.
11165 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11166 for CONST_WIDE_INT.
11168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11169 Alan Hayward <alan.hayward@arm.com>
11170 David Sherwood <david.sherwood@arm.com>
11172 * tree.h (SCALAR_TYPE_MODE): New macro.
11173 * expr.c (expand_expr_addr_expr_1): Use it.
11174 (expand_expr_real_2): Likewise.
11175 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11176 (fold_convert_const_fixed_from_int): Likewise.
11177 (fold_convert_const_fixed_from_real): Likewise.
11178 (native_encode_fixed): Likewise
11179 (native_encode_complex): Likewise
11180 (native_encode_vector): Likewise.
11181 (native_interpret_fixed): Likewise.
11182 (native_interpret_real): Likewise.
11183 (native_interpret_complex): Likewise.
11184 (native_interpret_vector): Likewise.
11185 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11186 (simd_clone_adjust_argument_types): Likewise.
11187 (simd_clone_init_simd_arrays): Likewise.
11188 (simd_clone_adjust): Likewise.
11189 * stor-layout.c (layout_type): Likewise.
11190 * tree.c (build_minus_one_cst): Likewise.
11191 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11192 * tree-inline.c (estimate_move_cost): Likewise.
11193 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11194 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11195 (vectorizable_reduction): Likewise.
11196 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11197 (vect_recog_mixed_size_cond_pattern): Likewise.
11198 (check_bool_pattern): Likewise.
11199 (adjust_bool_pattern): Likewise.
11200 (search_type_for_mask_1): Likewise.
11201 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11202 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11203 (vectorizable_load): Likewise.
11204 (vectorizable_store): Likewise.
11205 * ubsan.c (ubsan_encode_value): Likewise.
11206 * varasm.c (output_constant): Likewise.
11208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11209 Alan Hayward <alan.hayward@arm.com>
11210 David Sherwood <david.sherwood@arm.com>
11212 * coretypes.h (scalar_mode): New class.
11213 * machmode.h (scalar_mode): Likewise.
11214 (scalar_mode::includes_p): New function.
11215 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11216 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11217 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11218 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11219 * expmed.c (store_bit_field_1): Likewise.
11220 (extract_bit_field_1): Likewise.
11221 * expr.c (write_complex_part): Likewise.
11222 (read_complex_part): Likewise.
11223 (emit_move_complex_push): Likewise.
11224 (expand_expr_real_2): Likewise.
11225 * function.c (assign_parm_setup_reg): Likewise.
11226 (assign_parms_unsplit_complex): Likewise.
11227 * optabs.c (expand_binop): Likewise.
11228 * rtlanal.c (subreg_get_info): Likewise.
11229 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11230 * varasm.c (output_constant_pool_2): Likewise.
11232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11238 (expmed_mult_highpart_optab): Likewise.
11239 (expmed_mult_highpart): Likewise.
11241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11242 Alan Hayward <alan.hayward@arm.com>
11243 David Sherwood <david.sherwood@arm.com>
11245 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11246 instead of a machine_mode.
11247 (builtin_memset_read_str): Likewise.
11248 * builtins.c (c_readstr): Likewise.
11249 (builtin_memcpy_read_str): Likewise.
11250 (builtin_strncpy_read_str): Likewise.
11251 (builtin_memset_read_str): Likewise.
11252 (builtin_memset_gen_str): Likewise.
11253 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11254 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11255 instead of a machine_mode.
11256 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11258 (make_extraction): Likewise.
11259 (try_widen_shift_mode): Take and return scalar_int_modes instead
11261 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11262 a scalar_int_mode instead of a machine_mode.
11263 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11264 (avr_addr_space_pointer_mode): Likewise.
11265 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11266 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11267 (msp430_unwind_word_mode): Likewise.
11268 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11269 (spu_addr_space_pointer_mode): Likewise.
11270 (spu_addr_space_address_mode): Likewise.
11271 (spu_libgcc_cmp_return_mode): Likewise.
11272 (spu_libgcc_shift_count_mode): Likewise.
11273 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11274 (rl78_addr_space_pointer_mode): Likewise.
11275 (fl78_unwind_word_mode): Likewise.
11276 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11278 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11279 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11280 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11281 (mips_valid_pointer_mode): Likewise.
11282 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11283 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11284 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11286 (ft32_addr_space_address_mode): Likewise.
11287 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11288 scalar_int_mode instead of a machine_mode.
11289 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11291 (m32c_addr_space_address_mode): Likewise.
11292 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11293 (rs6000_eh_return_filter_mode): Likewise.
11294 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11295 (rs6000_eh_return_filter_mode): Likewise.
11296 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11297 (s390_libgcc_shift_count_mode): Likewise.
11298 (s390_unwind_word_mode): Likewise.
11299 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11301 * target.def (mode_rep_extended): Likewise.
11302 (valid_pointer_mode): Likewise.
11303 (addr_space.valid_pointer_mode): Likewise.
11304 (eh_return_filter_mode): Return a scalar_int_mode rather than
11306 (libgcc_cmp_return_mode): Likewise.
11307 (libgcc_shift_count_mode): Likewise.
11308 (unwind_word_mode): Likewise.
11309 (addr_space.pointer_mode): Likewise.
11310 (addr_space.address_mode): Likewise.
11311 * doc/tm.texi: Regenerate.
11312 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11314 (do_jump): Use scalar_int_mode for local variables.
11315 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11316 rather than a machine_mode.
11317 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11318 (scompare_loc_descriptor_wide): Likewise.
11319 (scompare_loc_descriptor_narrow): Likewise.
11320 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11322 * except.c (sjlj_emit_dispatch_table): Likewise.
11323 (expand_builtin_eh_copy_values): Likewise.
11324 * explow.c (convert_memory_address_addr_space_1): Likewise.
11325 Take a scalar_int_mode rather than a machine_mode.
11326 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11327 than a machine_mode.
11328 (memory_address_addr_space): Use scalar_int_mode for local variables.
11329 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11330 rather than a machine_mode.
11331 * expmed.c (mask_rtx): Likewise.
11332 (init_expmed_one_conv): Likewise.
11333 (expand_mult_highpart_adjust): Likewise.
11334 (extract_high_half): Likewise.
11335 (expmed_mult_highpart_optab): Likewise.
11336 (expmed_mult_highpart): Likewise.
11337 (expand_smod_pow2): Likewise.
11338 (expand_sdiv_pow2): Likewise.
11339 (emit_store_flag_int): Likewise.
11340 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11342 (extract_low_bits): Likewise.
11343 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11345 * expr.c (pieces_addr::adjust): Likewise.
11346 (can_store_by_pieces): Likewise.
11347 (store_by_pieces): Likewise.
11348 (clear_by_pieces_1): Likewise.
11349 (expand_expr_addr_expr_1): Likewise.
11350 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11351 (expand_expr_real_1): Likewise.
11352 (try_casesi): Likewise.
11353 * final.c (shorten_branches): Likewise.
11354 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11355 type of "mode" to machine_mode.
11356 * internal-fn.c (expand_arith_overflow_result_store): Take a
11357 scalar_int_mode rather than a machine_mode.
11358 (expand_mul_overflow): Use scalar_int_mode for local variables.
11359 * loop-doloop.c (doloop_modify): Likewise.
11360 (doloop_optimize): Likewise.
11361 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11362 than a machine_mode.
11363 (expand_doubleword_shift_condmove): Likewise.
11364 (expand_doubleword_shift): Likewise.
11365 (expand_doubleword_clz): Likewise.
11366 (expand_doubleword_popcount): Likewise.
11367 (expand_doubleword_parity): Likewise.
11368 (expand_absneg_bit): Use scalar_int_mode for local variables.
11369 (prepare_float_lib_cmp): Likewise.
11370 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11371 rather than a machine_mode.
11372 (convert_memory_address_addr_space): Likewise.
11373 (get_mode_bounds): Likewise.
11374 (get_address_mode): Return a scalar_int_mode rather than a
11376 * rtlanal.c (get_address_mode): Likewise.
11377 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11378 than a machine_mode.
11379 * targhooks.c (default_mode_rep_extended): Likewise.
11380 (default_valid_pointer_mode): Likewise.
11381 (default_addr_space_valid_pointer_mode): Likewise.
11382 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11383 than a machine_mode.
11384 (default_libgcc_cmp_return_mode): Likewise.
11385 (default_libgcc_shift_count_mode): Likewise.
11386 (default_unwind_word_mode): Likewise.
11387 (default_addr_space_pointer_mode): Likewise.
11388 (default_addr_space_address_mode): Likewise.
11389 * targhooks.h (default_eh_return_filter_mode): Likewise.
11390 (default_libgcc_cmp_return_mode): Likewise.
11391 (default_libgcc_shift_count_mode): Likewise.
11392 (default_unwind_word_mode): Likewise.
11393 (default_addr_space_pointer_mode): Likewise.
11394 (default_addr_space_address_mode): Likewise.
11395 (default_mode_rep_extended): Take a scalar_int_mode rather than
11397 (default_valid_pointer_mode): Likewise.
11398 (default_addr_space_valid_pointer_mode): Likewise.
11399 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11401 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11402 rather than a machine_mode.
11403 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11404 for local variables.
11405 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11406 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11407 than a machine_mode.
11409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11410 Alan Hayward <alan.hayward@arm.com>
11411 David Sherwood <david.sherwood@arm.com>
11413 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11414 the mode argument to scalar_int_mode.
11415 (do_jump_by_parts_zero_rtx): Likewise.
11416 (do_jump_by_parts_equality_rtx): Likewise.
11417 (do_jump_by_parts_greater): Take a mode argument.
11418 (do_jump_by_parts_equality): Likewise.
11419 (do_jump_1): Update calls accordingly.
11421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * is-a.h (safe_dyn_cast): New function.
11426 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11427 (jump_table_for_label): Likewise.
11428 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11429 instead of an rtx_insn *.
11430 (shorten_branches): Use dyn_cast instead of LABEL_P and
11431 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11432 rtx_jump_table_data::get_data_mode.
11433 (final_scan_insn): Likewise.
11435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11440 trying to combine a full-register integer set with a subreg
11443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11444 Alan Hayward <alan.hayward@arm.com>
11445 David Sherwood <david.sherwood@arm.com>
11447 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11448 that is always either address_mode or pointer_mode.
11450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11451 Alan Hayward <alan.hayward@arm.com>
11452 David Sherwood <david.sherwood@arm.com>
11454 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11455 when the two are known to be equal.
11457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11458 Alan Hayward <alan.hayward@arm.com>
11459 David Sherwood <david.sherwood@arm.com>
11461 * simplify-rtx.c (simplify_const_unary_operation): Use
11462 is_a <scalar_int_mode> instead of checking for a nonzero
11463 precision. Forcibly convert op_mode to a scalar_int_mode
11464 in that case. More clearly differentiate the operand and
11465 result modes and use the former when deciding what the value
11466 of a count-bits operation should be. Use is_int_mode instead
11467 of checking for a MODE_INT. Remove redundant check for whether
11468 this mode has a zero precision.
11470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11471 Alan Hayward <alan.hayward@arm.com>
11472 David Sherwood <david.sherwood@arm.com>
11474 * optabs.c (widen_leading): Change the type of the mode argument
11475 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11476 (widen_bswap): Likewise.
11477 (expand_parity): Likewise.
11478 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11479 (expand_ffs): Likewise.
11480 (epand_unop): Check for scalar integer modes before calling the
11483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11484 Alan Hayward <alan.hayward@arm.com>
11485 David Sherwood <david.sherwood@arm.com>
11487 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11489 (expand_expr_real_1): Update call accordingly.
11491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11492 Alan Hayward <alan.hayward@arm.com>
11493 David Sherwood <david.sherwood@arm.com>
11495 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11496 value_mode arguments. Use scalar_int_mode internally.
11497 (store_bit_field_1): Rename the new integer mode from imode
11498 to op0_mode and use it instead of GET_MODE (op0). Update calls
11499 to store_split_bit_field, store_bit_field_using_insv and
11500 store_fixed_bit_field.
11501 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11502 Use scalar_int_mode internally. Use a bit count rather than a mode
11503 when calculating the largest bit size for get_best_mode.
11504 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11505 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11506 Remove assertion that OP0 has a scalar integer mode.
11507 (store_split_bit_field): Add op0_mode and value_mode arguments.
11508 Update calls to extract_fixed_bit_field.
11509 (extract_bit_field_using_extv): Add an op0_mode argument.
11510 Use scalar_int_mode internally.
11511 (extract_bit_field_1): Rename the new integer mode from imode to
11512 op0_mode and use it instead of GET_MODE (op0). Update calls to
11513 extract_split_bit_field, extract_bit_field_using_extv and
11514 extract_fixed_bit_field.
11515 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11516 to extract_split_bit_field and extract_fixed_bit_field_1.
11517 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11518 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11519 on the target mode.
11520 (extract_split_bit_field): Add an op0_mode argument. Update call
11521 to extract_fixed_bit_field.
11523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11524 Alan Hayward <alan.hayward@arm.com>
11525 David Sherwood <david.sherwood@arm.com>
11527 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11528 * explow.c (hard_function_value): Likewise.
11529 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11530 convert_to_mode call outside the loop.
11531 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11532 for the mode iterator. Require the mode specified by max_pieces
11534 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11536 (copy_blkmode_to_reg): Likewise.
11537 (set_storage_via_setmem): Likewise.
11538 * optabs.c (prepare_cmp_insn): Likewise.
11539 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11540 * stor-layout.c (finish_bitfield_representative): Likewise.
11542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11543 Alan Hayward <alan.hayward@arm.com>
11544 David Sherwood <david.sherwood@arm.com>
11546 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11547 * expr.c (convert_move): Use them.
11548 (convert_modes): Likewise.
11549 (store_expr_with_bounds): Likewise.
11551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11552 Alan Hayward <alan.hayward@arm.com>
11553 David Sherwood <david.sherwood@arm.com>
11555 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11556 parameter for the mode of "x". Remove the "known_x", "known_mode"
11557 and "known_ret" arguments. Change the type of the mode argument
11558 to scalar_int_mode.
11559 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11560 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11561 (reg_num_sign_bit_copies_for_combine): Likewise.
11562 * rtlanal.c (nonzero_bits1): Likewise.
11563 (num_sign_bit_copies1): Likewise.
11564 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11565 (reg_num_sign_bit_copies_general): Likewise.
11566 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11567 (reg_nonzero_bits_general): Likewise.
11569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11570 Alan Hayward <alan.hayward@arm.com>
11571 David Sherwood <david.sherwood@arm.com>
11573 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11574 than in subroutines. Return 1 for non-integer modes.
11575 (cached_num_sign_bit_copies): Change the type of the mode parameter
11576 to scalar_int_mode.
11577 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11578 classes. Handle CONST_INT_P first and then check whether X also
11579 has a scalar integer mode. Check the same thing for inner registers
11580 of a SUBREG and for values that are being extended or truncated.
11582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11583 Alan Hayward <alan.hayward@arm.com>
11584 David Sherwood <david.sherwood@arm.com>
11586 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11587 in subroutines. Return the mode mask for non-integer modes.
11588 (cached_nonzero_bits): Change the type of the mode parameter
11589 to scalar_int_mode.
11590 (nonzero_bits1): Likewise. Remove early exit for other mode
11591 classes. Handle CONST_INT_P first and then check whether X
11592 also has a scalar integer mode.
11594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11595 Alan Hayward <alan.hayward@arm.com>
11596 David Sherwood <david.sherwood@arm.com>
11598 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11599 Return a scalar_int_mode and assert that the size is greater than
11601 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11602 (op_by_pieces_d::op_by_pieces_d): Likewise.
11603 (op_by_pieces_d::run): Likewise.
11604 (can_store_by_pieces): Likewise.
11606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11607 Alan Hayward <alan.hayward@arm.com>
11608 David Sherwood <david.sherwood@arm.com>
11610 * combine.c (extract_left_shift): Add a mode argument and update
11612 (make_compound_operation_int): Change the type of the mode parameter
11613 to scalar_int_mode and update the call to extract_left_shift.
11615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11616 Alan Hayward <alan.hayward@arm.com>
11617 David Sherwood <david.sherwood@arm.com>
11619 * combine.c (simplify_and_const_int): Change the type of the mode
11620 parameter to scalar_int_mode.
11621 (simplify_and_const_int_1): Likewise. Update recursive call.
11623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11624 Alan Hayward <alan.hayward@arm.com>
11625 David Sherwood <david.sherwood@arm.com>
11627 * combine.c (simplify_compare_const): Check that the mode is a
11628 scalar_int_mode (rather than VOIDmode) before testing its
11630 (simplify_comparison): Move COMPARISON_P handling out of the
11631 loop and restrict the latter part of the loop to scalar_int_modes.
11632 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11633 and when considering SUBREG_REGs. Use is_int_mode instead of
11634 checking GET_MODE_CLASS against MODE_INT.
11636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11641 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11642 shift_unit_mode and for modes involved in scalar shifts.
11644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11645 Alan Hayward <alan.hayward@arm.com>
11646 David Sherwood <david.sherwood@arm.com>
11648 * combine.c (force_int_to_mode): New function, split out from...
11649 (force_to_mode): ...here. Keep xmode up-to-date and use it
11650 instead of GET_MODE (x).
11652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11653 Alan Hayward <alan.hayward@arm.com>
11654 David Sherwood <david.sherwood@arm.com>
11656 * optabs-query.h (extraction_insn::struct_mode): Change type to
11657 opt_scalar_int_mode and update comment.
11658 (extraction_insn::field_mode): Change type to scalar_int_mode.
11659 (extraction_insn::pos_mode): Likewise.
11660 * combine.c (make_extraction): Update accordingly.
11661 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11662 (get_optab_extraction_insn): Likewise.
11663 * recog.c (simplify_while_replacing): Likewise.
11664 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11665 parameter to opt_scalar_int_mode.
11667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11668 Alan Hayward <alan.hayward@arm.com>
11669 David Sherwood <david.sherwood@arm.com>
11671 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11672 to a scalar_int_mode instead of a machine_mode.
11673 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11674 (get_best_mode): Return a boolean and use a pointer argument to store
11675 the selected mode. Replace the limit mode parameter with a bit limit.
11676 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11677 for the values returned by bit_field_mode_iterator::next_mode.
11678 (store_bit_field): Update call to get_best_mode.
11679 (store_fixed_bit_field): Likewise.
11680 (extract_fixed_bit_field): Likewise.
11681 * expr.c (optimize_bitfield_assignment_op): Likewise.
11682 * fold-const.c (optimize_bit_field_compare): Likewise.
11683 (fold_truth_andor_1): Likewise.
11684 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11685 Update for new type of m_mode.
11686 (get_best_mode): As above.
11688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11689 Alan Hayward <alan.hayward@arm.com>
11690 David Sherwood <david.sherwood@arm.com>
11692 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11693 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11694 (store_bit_field): Check is_a <scalar_int_mode> before calling
11695 strict_volatile_bitfield_p.
11696 (extract_bit_field): Likewise.
11698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11699 Alan Hayward <alan.hayward@arm.com>
11700 David Sherwood <david.sherwood@arm.com>
11702 * target.def (cstore_mode): Return a scalar_int_mode.
11703 * doc/tm.texi: Regenerate.
11704 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11705 * targhooks.h (default_cstore_mode): Likewise.
11706 * targhooks.c (default_cstore_mode): Likewise, using a forced
11708 * expmed.c (emit_cstore): Expect the target of the cstore to be
11711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11712 Alan Hayward <alan.hayward@arm.com>
11713 David Sherwood <david.sherwood@arm.com>
11715 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11717 (niter_desc): Likewise mode.
11718 (iv_analyze): Add a mode parameter.
11720 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11721 and change its type to scalar_int_mode.
11722 * loop-iv.c: Update commentary at head of file.
11723 (iv_constant): Pass the mode paraeter before the rtx it describes
11724 and change its type to scalar_int_mode. Remove VOIDmode handling.
11725 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11726 (iv_extend): Likewise.
11727 (shorten_into_mode): Likewise.
11728 (iv_add): Use scalar_int_mode.
11729 (iv_mult): Likewise.
11730 (iv_shift): Likewise.
11731 (canonicalize_iv_subregs): Likewise.
11732 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11733 it describes and change its mode to scalar_int_mode. Also change
11734 the type of the returned inner_mode to scalar_int_mode.
11735 (get_biv_step): Likewise, turning outer_mode from a pointer
11736 into a direct parameter. Update call to get_biv_step_1.
11737 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11738 iv_constant and get_biv_step.
11739 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11740 and change its type to scalar_int_mode. Don't initialise iv->mode
11741 to VOIDmode and remove later checks for its still being VOIDmode.
11742 Update calls to iv_analyze_op and iv_analyze_expr. Check
11743 is_a <scalar_int_mode> when changing the mode under consideration.
11744 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11745 Update call to iv_analyze_expr.
11746 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11747 inner register is not also a scalar_int_mode. Update call to
11749 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11750 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11751 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11752 separate mode class checks. Update calls to iv_analyze. Remove
11753 fix-up of VOIDmodes after iv_analyze_biv.
11754 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11755 don't have a scalar_int_mode. Update call to biv_p.
11757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11758 Alan Hayward <alan.hayward@arm.com>
11759 David Sherwood <david.sherwood@arm.com>
11761 * cfgexpand.c (convert_debug_memory_address): Use
11762 as_a <scalar_int_mode>.
11763 * combine.c (expand_compound_operation): Likewise.
11764 (make_extraction): Likewise.
11765 (change_zero_ext): Likewise.
11766 (simplify_comparison): Likewise.
11767 * cse.c (cse_insn): Likewise.
11768 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11769 (mem_loc_descriptor): Likewise.
11770 (loc_descriptor): Likewise.
11771 * expmed.c (init_expmed_one_mode): Likewise.
11772 (synth_mult): Likewise.
11773 (emit_store_flag_1): Likewise.
11774 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11775 of a comparison with size.
11776 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11777 (reduce_to_bit_field_precision): Likewise.
11778 * function.c (expand_function_end): Likewise.
11779 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11780 * loop-doloop.c (doloop_modify): Likewise.
11781 * optabs.c (expand_binop): Likewise.
11782 (expand_unop): Likewise.
11783 (expand_copysign_absneg): Likewise.
11784 (prepare_cmp_insn): Likewise.
11785 (maybe_legitimize_operand): Likewise.
11786 * recog.c (const_scalar_int_operand): Likewise.
11787 * rtlanal.c (get_address_mode): Likewise.
11788 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11789 (simplify_cond_clz_ctz): Likewise.
11790 * tree-nested.c (get_nl_goto_field): Likewise.
11791 * tree.c (build_vector_type_for_mode): Likewise.
11792 * var-tracking.c (use_narrower_mode): Likewise.
11794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11795 Alan Hayward <alan.hayward@arm.com>
11796 David Sherwood <david.sherwood@arm.com>
11798 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11799 * builtins.c (expand_builtin_signbit): Use it.
11800 * cfgexpand.c (expand_debug_expr): Likewise.
11801 * dojump.c (do_jump): Likewise.
11802 (do_compare_and_jump): Likewise.
11803 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11804 * expmed.c (make_tree): Likewise.
11805 * expr.c (expand_expr_real_2): Likewise.
11806 (expand_expr_real_1): Likewise.
11807 (try_casesi): Likewise.
11808 * fold-const-call.c (fold_const_call_ss): Likewise.
11809 * fold-const.c (unextend): Likewise.
11810 (extract_muldiv_1): Likewise.
11811 (fold_single_bit_test): Likewise.
11812 (native_encode_int): Likewise.
11813 (native_encode_string): Likewise.
11814 (native_interpret_int): Likewise.
11815 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11816 * internal-fn.c (expand_addsub_overflow): Likewise.
11817 (expand_neg_overflow): Likewise.
11818 (expand_mul_overflow): Likewise.
11819 (expand_arith_overflow): Likewise.
11820 * match.pd: Likewise.
11821 * stor-layout.c (layout_type): Likewise.
11822 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11823 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11824 * tree-ssanames.c (get_range_info): Likewise.
11825 * tree-switch-conversion.c (array_value_type) Likewise.
11826 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11827 (vect_recog_divmod_pattern): Likewise.
11828 (vect_recog_mixed_size_cond_pattern): Likewise.
11829 * tree-vrp.c (extract_range_basic): Likewise.
11830 (simplify_float_conversion_using_ranges): Likewise.
11831 * tree.c (int_fits_type_p): Likewise.
11832 * ubsan.c (instrument_bool_enum_load): Likewise.
11833 * varasm.c (mergeable_string_section): Likewise.
11834 (narrowing_initializer_constant_valid_p): Likewise.
11835 (output_constant): Likewise.
11837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11838 Alan Hayward <alan.hayward@arm.com>
11839 David Sherwood <david.sherwood@arm.com>
11841 * machmode.h (NARROWEST_INT_MODE): New macro.
11842 * expr.c (alignment_for_piecewise_move): Use it instead of
11843 GET_CLASS_NARROWEST_MODE (MODE_INT).
11844 (push_block): Likewise.
11845 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11847 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11850 Alan Hayward <alan.hayward@arm.com>
11851 David Sherwood <david.sherwood@arm.com>
11853 * postreload.c (move2add_valid_value_p): Change the type of the
11854 mode parameter to scalar_int_mode.
11855 (move2add_use_add2_insn): Add a mode parameter and use it instead
11857 (move2add_use_add3_insn): Likewise.
11858 (reload_cse_move2add): Update accordingly.
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 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11866 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11867 * optabs.c (expand_unop): Likewise.
11869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11870 Alan Hayward <alan.hayward@arm.com>
11871 David Sherwood <david.sherwood@arm.com>
11873 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11874 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11875 (popcount_loc_descriptor): Likewise.
11876 (bswap_loc_descriptor): Likewise.
11877 (rotate_loc_descriptor): Likewise.
11878 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11879 calling the functions above.
11881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11882 Alan Hayward <alan.hayward@arm.com>
11883 David Sherwood <david.sherwood@arm.com>
11885 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11887 (try_combine): Likewise.
11888 (simplify_if_then_else): Likewise.
11889 * cse.c (cse_insn): Likewise.
11890 * dwarf2out.c (mem_loc_descriptor): Likewise.
11891 * emit-rtl.c (gen_lowpart_common): Likewise.
11892 * simplify-rtx.c (simplify_truncation): Likewise.
11893 (simplify_binary_operation_1): Likewise.
11894 (simplify_const_relational_operation): Likewise.
11895 (simplify_ternary_operation): Likewise.
11896 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11899 Alan Hayward <alan.hayward@arm.com>
11900 David Sherwood <david.sherwood@arm.com>
11902 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11903 * reload.c (push_reload): Likewise.
11904 (find_reloads): Likewise.
11906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11907 Alan Hayward <alan.hayward@arm.com>
11908 David Sherwood <david.sherwood@arm.com>
11910 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11911 (make_compound_operation_int): Likewise.
11912 (change_zero_ext): Likewise.
11913 * expr.c (convert_move): Likewise.
11914 (convert_modes): Likewise.
11915 * fwprop.c (forward_propagate_subreg): Likewise.
11916 * loop-iv.c (get_biv_step_1): Likewise.
11917 * optabs.c (widen_operand): Likewise.
11918 * postreload.c (move2add_valid_value_p): Likewise.
11919 * recog.c (simplify_while_replacing): Likewise.
11920 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11921 (simplify_binary_operation_1): Likewise. Remove redundant
11922 mode equality check.
11924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11925 Alan Hayward <alan.hayward@arm.com>
11926 David Sherwood <david.sherwood@arm.com>
11928 * combine.c (combine_simplify_rtx): Add checks for
11929 is_a <scalar_int_mode>.
11930 (simplify_if_then_else): Likewise.
11931 (make_field_assignment): Likewise.
11932 (simplify_comparison): Likewise.
11933 * ifcvt.c (noce_try_bitop): Likewise.
11934 * loop-invariant.c (canonicalize_address_mult): Likewise.
11935 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11938 Alan Hayward <alan.hayward@arm.com>
11939 David Sherwood <david.sherwood@arm.com>
11941 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11942 is_a <scalar_int_mode> instead of != BLKmode.
11944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11945 Alan Hayward <alan.hayward@arm.com>
11946 David Sherwood <david.sherwood@arm.com>
11948 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11949 instead of != VOIDmode.
11950 * combine.c (if_then_else_cond): Likewise.
11951 (change_zero_ext): Likewise.
11952 * dwarf2out.c (mem_loc_descriptor): Likewise.
11953 (loc_descriptor): Likewise.
11954 * rtlanal.c (canonicalize_condition): Likewise.
11955 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11958 Alan Hayward <alan.hayward@arm.com>
11959 David Sherwood <david.sherwood@arm.com>
11961 * simplify-rtx.c (simplify_binary_operation_1): Use
11962 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11965 Alan Hayward <alan.hayward@arm.com>
11966 David Sherwood <david.sherwood@arm.com>
11968 * wide-int.h (int_traits<unsigned char>) New class.
11969 (int_traits<unsigned short>) Likewise.
11970 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11971 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11973 * combine.c (set_nonzero_bits_and_sign_copies): Use
11974 is_a <scalar_int_mode>.
11975 (find_split_point): Likewise.
11976 (combine_simplify_rtx): Likewise.
11977 (simplify_logical): Likewise.
11978 (expand_compound_operation): Likewise.
11979 (expand_field_assignment): Likewise.
11980 (make_compound_operation): Likewise.
11981 (extended_count): Likewise.
11982 (change_zero_ext): Likewise.
11983 (simplify_comparison): Likewise.
11984 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11985 (ucompare_loc_descriptor): Likewise.
11986 (minmax_loc_descriptor): Likewise.
11987 (mem_loc_descriptor): Likewise.
11988 (loc_descriptor): Likewise.
11989 * expmed.c (init_expmed_one_mode): Likewise.
11990 * lra-constraints.c (lra_constraint_offset): Likewise.
11991 * optabs.c (prepare_libcall_arg): Likewise.
11992 * postreload.c (move2add_note_store): Likewise.
11993 * reload.c (operands_match_p): Likewise.
11994 * rtl.h (load_extend_op): Likewise.
11995 * rtlhooks.c (gen_lowpart_general): Likewise.
11996 * simplify-rtx.c (simplify_truncation): Likewise.
11997 (simplify_unary_operation_1): Likewise.
11998 (simplify_binary_operation_1): Likewise.
11999 (simplify_const_binary_operation): Likewise.
12000 (simplify_const_relational_operation): Likewise.
12001 (simplify_subreg): Likewise.
12002 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12003 * var-tracking.c (adjust_mems): Likewise.
12004 (prepare_call_arguments): Likewise.
12006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12007 Alan Hayward <alan.hayward@arm.com>
12008 David Sherwood <david.sherwood@arm.com>
12010 * machmode.h (is_int_mode): New fuction.
12011 * combine.c (find_split_point): Use it.
12012 (combine_simplify_rtx): Likewise.
12013 (simplify_if_then_else): Likewise.
12014 (simplify_set): Likewise.
12015 (simplify_shift_const_1): Likewise.
12016 (simplify_comparison): Likewise.
12017 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12018 * cse.c (notreg_cost): Likewise.
12019 (cse_insn): Likewise.
12020 * cselib.c (cselib_lookup_1): Likewise.
12021 * dojump.c (do_jump_1): Likewise.
12022 (do_compare_rtx_and_jump): Likewise.
12023 * dse.c (get_call_args): Likewise.
12024 * dwarf2out.c (rtl_for_decl_init): Likewise.
12025 (native_encode_initializer): Likewise.
12026 * expmed.c (emit_store_flag_1): Likewise.
12027 (emit_store_flag): Likewise.
12028 * expr.c (convert_modes): Likewise.
12029 (store_field): Likewise.
12030 (expand_expr_real_1): Likewise.
12031 * fold-const.c (fold_read_from_constant_string): Likewise.
12032 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12033 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12034 * optabs.c (expand_binop): Likewise.
12035 (expand_unop): Likewise.
12036 (expand_abs_nojump): Likewise.
12037 (expand_one_cmpl_abs_nojump): Likewise.
12038 * simplify-rtx.c (mode_signbit_p): Likewise.
12039 (val_signbit_p): Likewise.
12040 (val_signbit_known_set_p): Likewise.
12041 (val_signbit_known_clear_p): Likewise.
12042 (simplify_relational_operation_1): Likewise.
12043 * tree.c (vector_type_mode): Likewise.
12045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12046 Alan Hayward <alan.hayward@arm.com>
12047 David Sherwood <david.sherwood@arm.com>
12049 * machmode.h (smallest_mode_for_size): Fix formatting.
12050 (smallest_int_mode_for_size): New function.
12051 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12052 instead of smallest_mode_for_size.
12053 * combine.c (make_extraction): Likewise.
12054 * config/arc/arc.c (arc_expand_movmem): Likewise.
12055 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12056 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12057 * config/s390/s390.c (s390_expand_insv): Likewise.
12058 * config/sparc/sparc.c (assign_int_registers): Likewise.
12059 * config/spu/spu.c (spu_function_value): Likewise.
12060 (spu_function_arg): Likewise.
12061 * coverage.c (get_gcov_type): Likewise.
12062 (get_gcov_unsigned_t): Likewise.
12063 * dse.c (find_shift_sequence): Likewise.
12064 * expmed.c (store_bit_field_1): Likewise.
12065 * expr.c (convert_move): Likewise.
12066 (store_field): Likewise.
12067 * internal-fn.c (expand_arith_overflow): Likewise.
12068 * optabs-query.c (get_best_extraction_insn): Likewise.
12069 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12070 * stor-layout.c (layout_type): Likewise.
12071 (initialize_sizetypes): Likewise.
12072 * targhooks.c (default_get_mask_mode): Likewise.
12073 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12076 Alan Hayward <alan.hayward@arm.com>
12077 David Sherwood <david.sherwood@arm.com>
12079 * machmode.h (opt_mode::else_blk): New function.
12080 (int_mode_for_mode): Declare.
12081 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12082 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12084 * cfgexpand.c (expand_debug_expr): Likewise.
12085 * combine.c (gen_lowpart_or_truncate): Likewise.
12086 (gen_lowpart_for_combine): Likewise.
12087 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12088 * config/avr/avr.c (avr_to_int_mode): Likewise.
12089 (avr_out_plus_1): Likewise.
12090 (avr_out_plus): Likewise.
12091 (avr_out_round): Likewise.
12092 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12093 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12094 (s390_expand_vcond): Likewise.
12095 * config/spu/spu.c (spu_split_immediate): Likewise.
12096 (spu_expand_mov): Likewise.
12097 * dse.c (get_stored_val): Likewise.
12098 * expmed.c (store_bit_field_1): Likewise.
12099 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12101 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12102 (extract_low_bits): Likewise.
12103 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12104 handling rather than repeating the check.
12105 (emit_group_store): Likewise.
12106 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12107 * optabs.c (expand_absneg_bit): Likewise.
12108 (expand_copysign_absneg): Likewise.
12109 (expand_copysign_bit): Likewise.
12110 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12111 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12112 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12113 * var-tracking.c (prepare_call_arguments): Likewise.
12114 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12115 int_mode_for_mode instead of mode_for_size.
12116 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * machmode.h (int_mode_for_size): New function.
12123 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12124 instead of mode_for_size.
12125 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12127 * combine.c (expand_field_assignment): Use int_mode_for_size
12128 instead of mode_for_size.
12129 (make_extraction): Likewise.
12130 (simplify_shift_const_1): Likewise.
12131 (simplify_comparison): Likewise.
12132 * dojump.c (do_jump): Likewise.
12133 * dwarf2out.c (mem_loc_descriptor): Likewise.
12134 * emit-rtl.c (init_derived_machine_modes): Likewise.
12135 * expmed.c (flip_storage_order): Likewise.
12136 (convert_extracted_bit_field): Likewise.
12137 * expr.c (copy_blkmode_from_reg): Likewise.
12138 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12139 * internal-fn.c (expand_mul_overflow): Likewise.
12140 * lower-subreg.c (simple_move): Likewise.
12141 * optabs-libfuncs.c (init_optabs): Likewise.
12142 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12143 * tree.c (vector_type_mode): Likewise.
12144 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12145 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12146 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12147 * tree-vect-stmts.c (vectorizable_load): Likewise.
12148 (vectorizable_store): Likewise.
12150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12151 Alan Hayward <alan.hayward@arm.com>
12152 David Sherwood <david.sherwood@arm.com>
12154 * coretypes.h (pod_mode): New type.
12155 (scalar_int_mode_pod): New typedef.
12156 * machmode.h (pod_mode): New class.
12157 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12158 * genmodes.c (emit_mode_int_n): Update accordingly.
12159 * lower-subreg.h (target_lower_subreg): Change type to
12160 scalar_int_mode_pod.
12161 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12162 scalar_int_mode_pod.
12164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12165 Alan Hayward <alan.hayward@arm.com>
12166 David Sherwood <david.sherwood@arm.com>
12168 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12169 machine_mode to scalar_int_mode.
12170 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12171 (rs6000_option_override_internal): Remove cast to int.
12172 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12173 machine_mode to scalar_int_mode.
12174 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12175 (rs6000_option_override_internal): Remove cast to int.
12176 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12177 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12179 * config/s390/s390.c (s390_expand_builtin): Likewise.
12180 * coretypes.h (scalar_int_mode): New type.
12181 (opt_scalar_int_mode): New typedef.
12182 * machmode.h (scalar_int_mode): New class.
12183 (scalar_int_mode::includes_p): New function.
12184 (byte_mode): Change type to scalar_int_mode.
12185 (word_mode): Likewise.
12186 (ptr_mode): Likewise.
12187 * emit-rtl.c (byte_mode): Likewise.
12188 (word_mode): Likewise.
12189 (ptr_mode): Likewise.
12190 (init_derived_machine_modes): Update accordingly.
12191 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12192 and MODE_PARTIAL_INT.
12193 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12194 opt_scalar_int_mode.
12196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12197 Alan Hayward <alan.hayward@arm.com>
12198 David Sherwood <david.sherwood@arm.com>
12200 * target.def (libgcc_floating_mode_supported_p): Take a
12202 * doc/tm.texi: Regenerate.
12203 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12205 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12206 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12210 Alan Hayward <alan.hayward@arm.com>
12211 David Sherwood <david.sherwood@arm.com>
12213 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12214 * doc/tm.texi: Regenerate.
12215 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12216 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12217 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12218 * targhooks.h (default_floatn_mode): Likewise.
12219 * targhooks.c (default_floatn_mode): Likewise.
12220 * tree.c (build_common_tree_nodes): Update accordingly.
12222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12223 Alan Hayward <alan.hayward@arm.com>
12224 David Sherwood <david.sherwood@arm.com>
12226 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12227 (mode_iterator::iterate_p): Likewise.
12228 (mode_iterator::get_wider): Likewise.
12229 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12232 Alan Hayward <alan.hayward@arm.com>
12233 David Sherwood <david.sherwood@arm.com>
12235 * coretypes.h (opt_scalar_float_mode): New typedef.
12236 * machmode.h (float_mode_for_size): New function.
12237 * emit-rtl.c (double_mode): Delete.
12238 (init_emit_once): Use float_mode_for_size.
12239 * stor-layout.c (layout_type): Likewise.
12240 * gdbhooks.py (build_pretty_printer): Handle 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 * output.h (assemble_real): Take a scalar_float_mode.
12247 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12248 * config/arm/arm.md (consttable_4): Likewise.
12249 (consttable_8): Likewise.
12250 (consttable_16): Likewise.
12251 * config/mips/mips.md (consttable_float): Likewise.
12252 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12253 * varasm.c (assemble_real): Take a scalar_float_mode.
12254 (output_constant_pool_2): Update accordingly.
12255 (output_constant): Likewise.
12257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12258 Alan Hayward <alan.hayward@arm.com>
12259 David Sherwood <david.sherwood@arm.com>
12261 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12262 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12263 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12264 (native_encode_real): Likewise.
12265 (native_interpret_real): Likewise.
12266 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12267 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12270 Alan Hayward <alan.hayward@arm.com>
12271 David Sherwood <david.sherwood@arm.com>
12273 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12274 <scalar_float_mode>. Simplify.
12275 (gen_extend_conv_libfunc): Likewise.
12277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12278 Alan Hayward <alan.hayward@arm.com>
12279 David Sherwood <david.sherwood@arm.com>
12281 * coretypes.h (scalar_float_mode): New type.
12282 * machmode.h (mode_traits::from_int): Use machine_mode if
12283 USE_ENUM_MODES is defined.
12284 (is_a): New function.
12286 (dyn_cast): Likewise.
12287 (scalar_float_mode): New class.
12288 (scalar_float_mode::includes_p): New function.
12289 (is_float_mode): Likewise.
12290 * gdbhooks.py (MachineModePrinter): New class.
12291 (build_pretty_printer): Use it for scalar_float_mode.
12292 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12293 (format_helper::format_helper): Turn into a template.
12294 * genmodes.c (get_mode_class): New function.
12295 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12296 or machine_mode if none.
12297 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12298 as_a <scalar_float_mode>.
12299 * dwarf2out.c (mem_loc_descriptor): Likewise.
12300 (insert_float): Likewise.
12301 (add_const_value_attribute): Likewise.
12302 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12303 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12304 (expand_unop): Update accordingly.
12305 (expand_abs_nojump): Likewise.
12306 (expand_copysign_absneg): Take a scalar_float_mode.
12307 (expand_copysign_bit): Likewise.
12308 (expand_copysign): Update accordingly.
12310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12311 Alan Hayward <alan.hayward@arm.com>
12312 David Sherwood <david.sherwood@arm.com>
12314 * coretypes.h (opt_mode): New class.
12315 * machmode.h (opt_mode): Likewise.
12316 (opt_mode::else_void): New function.
12317 (opt_mode::require): Likewise.
12318 (opt_mode::exists): Likewise.
12319 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12320 (GET_MODE_2XWIDER_MODE): Likewise.
12321 (mode_iterator::get_wider): Update accordingly.
12322 (mode_iterator::get_2xwider): Likewise.
12323 (mode_iterator::get_known_wider): Likewise, turning into a template.
12324 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12325 forcing a wider mode to exist.
12326 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12327 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12328 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12329 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12330 * lower-subreg.c (init_lower_subreg): Likewise.
12331 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12332 on the final iteration.
12333 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12334 a wider mode exists before asking for a move pattern.
12335 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12336 forcing a wider mode to exist.
12337 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12338 returning false if no such mode exists.
12339 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12340 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12341 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12342 Avoid checking for a MODE_INT if we already know the mode is not a
12344 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12345 forcing a wider mode to exist.
12346 (expmed_mult_highpart_optab): Likewise.
12347 (expmed_mult_highpart): Likewise.
12348 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12350 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12351 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12352 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12353 * internal-fn.c (expand_mul_overflow): Update use of
12354 GET_MODE_2XWIDER_MODE.
12355 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12356 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12357 GET_MODE_WIDER_MODE.
12358 (convert_plusminus_to_widen): Likewise.
12359 * tree-switch-conversion.c (array_value_type): Likewise.
12360 * var-tracking.c (emit_note_insn_var_location): Likewise.
12361 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12362 Return false inside rather than outside the loop if no wider mode
12364 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12365 and GET_MODE_2XWIDER_MODE
12366 (can_compare_p): Use else_void.
12367 * gdbhooks.py (OptMachineModePrinter): New class.
12368 (build_pretty_printer): Use it for opt_mode.
12370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12371 Alan Hayward <alan.hayward@arm.com>
12372 David Sherwood <david.sherwood@arm.com>
12374 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12375 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12378 Alan Hayward <alan.hayward@arm.com>
12379 David Sherwood <david.sherwood@arm.com>
12381 * machmode.h (mode_traits): New structure.
12382 (get_narrowest_mode): New function.
12383 (mode_iterator::start): Likewise.
12384 (mode_iterator::iterate_p): Likewise.
12385 (mode_iterator::get_wider): Likewise.
12386 (mode_iterator::get_known_wider): Likewise.
12387 (mode_iterator::get_2xwider): Likewise.
12388 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12389 (FOR_EACH_MODE): Likewise.
12390 (FOR_EACH_MODE_FROM): Likewise.
12391 (FOR_EACH_MODE_UNTIL): Likewise.
12392 (FOR_EACH_WIDER_MODE): Likewise.
12393 (FOR_EACH_2XWIDER_MODE): Likewise.
12394 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12395 * combine.c (simplify_comparison): Likewise
12396 * config/i386/i386.c (type_natural_mode): Likewise.
12397 * cse.c (cse_insn): Likewise.
12398 * dse.c (find_shift_sequence): Likewise.
12399 * emit-rtl.c (init_derived_machine_modes): Likewise.
12400 (init_emit_once): Likewise.
12401 * explow.c (hard_function_value): Likewise.
12402 * expmed.c (extract_fixed_bit_field_1): Likewise.
12403 (extract_bit_field_1): Likewise.
12404 (expand_divmod): Likewise.
12405 (emit_store_flag_1): Likewise.
12406 * expr.c (init_expr_target): Likewise.
12407 (convert_move): Likewise.
12408 (alignment_for_piecewise_move): Likewise.
12409 (widest_int_mode_for_size): Likewise.
12410 (emit_block_move_via_movmem): Likewise.
12411 (copy_blkmode_to_reg): Likewise.
12412 (set_storage_via_setmem): Likewise.
12413 (compress_float_constant): Likewise.
12414 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12415 * optabs-query.c (get_best_extraction_insn): Likewise.
12416 * optabs.c (expand_binop): Likewise.
12417 (expand_twoval_unop): Likewise.
12418 (expand_twoval_binop): Likewise.
12419 (widen_leading): Likewise.
12420 (widen_bswap): Likewise.
12421 (expand_parity): Likewise.
12422 (expand_unop): Likewise.
12423 (prepare_cmp_insn): Likewise.
12424 (prepare_float_lib_cmp): Likewise.
12425 (expand_float): Likewise.
12426 (expand_fix): Likewise.
12427 (expand_sfix_optab): Likewise.
12428 * postreload.c (move2add_use_add2_insn): Likewise.
12429 * reg-stack.c (reg_to_stack): Likewise.
12430 * reginfo.c (choose_hard_reg_mode): Likewise.
12431 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12432 * stor-layout.c (mode_for_size): Likewise.
12433 (smallest_mode_for_size): Likewise.
12434 (mode_for_vector): Likewise.
12435 (finish_bitfield_representative): Likewise.
12436 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12437 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12438 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12439 * var-tracking.c (prepare_call_arguments): Likewise.
12441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12442 Alan Hayward <alan.hayward@arm.com>
12443 David Sherwood <david.sherwood@arm.com>
12445 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12446 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12447 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12448 * machmode.h (mode_size): Move earlier in file.
12449 (mode_precision): Likewise.
12450 (mode_inner): Likewise.
12451 (mode_nunits): Likewise.
12452 (mode_unit_size): Likewise.
12453 (unit_unit_precision): Likewise.
12454 (mode_wider): Likewise.
12455 (mode_2xwider): Likewise.
12456 (machine_mode): New class.
12457 (mode_to_bytes): New function.
12458 (mode_to_bits): Likewise.
12459 (mode_to_precision): Likewise.
12460 (mode_to_inner): Likewise.
12461 (mode_to_unit_size): Likewise.
12462 (mode_to_unit_precision): Likewise.
12463 (mode_to_nunits): Likewise.
12464 (GET_MODE_SIZE): Use mode_to_bytes.
12465 (GET_MODE_BITSIZE): Use mode_to_bits.
12466 (GET_MODE_PRECISION): Use mode_to_precision.
12467 (GET_MODE_INNER): Use mode_to_inner.
12468 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12469 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12470 (GET_MODE_NUNITS): Use mode_to_nunits.
12471 * system.h (ALWAYS_INLINE): New macro.
12472 * config/powerpcspe/powerpcspe-c.c
12473 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12474 int for arg1_mode and arg2_mode.
12476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12477 Alan Hayward <alan.hayward@arm.com>
12478 David Sherwood <david.sherwood@arm.com>
12480 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12481 Prefix mode names with E_ in case statements.
12482 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12483 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12484 (aarch64_split_simd_move): Likewise.
12485 (aarch64_gen_storewb_pair): Likewise.
12486 (aarch64_gen_loadwb_pair): Likewise.
12487 (aarch64_gen_store_pair): Likewise.
12488 (aarch64_gen_load_pair): Likewise.
12489 (aarch64_get_condition_code_1): Likewise.
12490 (aarch64_constant_pool_reload_icode): Likewise.
12491 (get_rsqrte_type): Likewise.
12492 (get_rsqrts_type): Likewise.
12493 (get_recpe_type): Likewise.
12494 (get_recps_type): Likewise.
12495 (aarch64_gimplify_va_arg_expr): Likewise.
12496 (aarch64_simd_container_mode): Likewise.
12497 (aarch64_emit_load_exclusive): Likewise.
12498 (aarch64_emit_store_exclusive): Likewise.
12499 (aarch64_expand_compare_and_swap): Likewise.
12500 (aarch64_gen_atomic_cas): Likewise.
12501 (aarch64_emit_bic): Likewise.
12502 (aarch64_emit_atomic_swap): Likewise.
12503 (aarch64_emit_atomic_load_op): Likewise.
12504 (aarch64_evpc_trn): Likewise.
12505 (aarch64_evpc_uzp): Likewise.
12506 (aarch64_evpc_zip): Likewise.
12507 (aarch64_evpc_ext): Likewise.
12508 (aarch64_evpc_rev): Likewise.
12509 (aarch64_evpc_dup): Likewise.
12510 (aarch64_gen_ccmp_first): Likewise.
12511 (aarch64_gen_ccmp_next): Likewise.
12512 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12513 (alpha_emit_xfloating_libcall): Likewise.
12514 (emit_insxl): Likewise.
12515 (alpha_arg_type): Likewise.
12516 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12517 (arc_preferred_simd_mode): Likewise.
12518 (arc_secondary_reload): Likewise.
12519 (get_arc_condition_code): Likewise.
12520 (arc_print_operand): Likewise.
12521 (arc_legitimate_constant_p): Likewise.
12522 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12523 * config/arc/arc.md (casesi_load): Likewise.
12524 (casesi_compact_jump): Likewise.
12525 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12526 (cc_use_register): Likewise.
12527 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12528 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12529 (arm_init_iwmmxt_builtins): Likewise.
12530 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12531 (neon_expand_vector_init): Likewise.
12532 (arm_attr_length_move_neon): Likewise.
12533 (maybe_get_arm_condition_code): Likewise.
12534 (arm_emit_vector_const): Likewise.
12535 (arm_preferred_simd_mode): Likewise.
12536 (arm_output_iwmmxt_tinsr): Likewise.
12537 (thumb1_output_casesi): Likewise.
12538 (thumb2_output_casesi): Likewise.
12539 (arm_emit_load_exclusive): Likewise.
12540 (arm_emit_store_exclusive): Likewise.
12541 (arm_expand_compare_and_swap): Likewise.
12542 (arm_evpc_neon_vuzp): Likewise.
12543 (arm_evpc_neon_vzip): Likewise.
12544 (arm_evpc_neon_vrev): Likewise.
12545 (arm_evpc_neon_vtrn): Likewise.
12546 (arm_evpc_neon_vext): Likewise.
12547 (arm_validize_comparison): Likewise.
12548 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12549 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12550 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12551 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12552 (c6x_preferred_simd_mode): Likewise.
12553 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12554 (epiphany_rtx_costs): Likewise.
12555 * config/epiphany/predicates.md (proper_comparison_operator):
12557 * config/frv/frv.c (condexec_memory_operand): Likewise.
12558 (frv_emit_move): Likewise.
12559 (output_move_single): Likewise.
12560 (output_condmove_single): Likewise.
12561 (frv_hard_regno_mode_ok): Likewise.
12562 (frv_matching_accg_mode): Likewise.
12563 * config/h8300/h8300.c (split_adds_subs): Likewise.
12564 (h8300_rtx_costs): Likewise.
12565 (h8300_print_operand): Likewise.
12566 (compute_mov_length): Likewise.
12567 (output_logical_op): Likewise.
12568 (compute_logical_op_length): Likewise.
12569 (compute_logical_op_cc): Likewise.
12570 (h8300_shift_needs_scratch_p): Likewise.
12571 (output_a_shift): Likewise.
12572 (compute_a_shift_length): Likewise.
12573 (compute_a_shift_cc): Likewise.
12574 (expand_a_rotate): Likewise.
12575 (output_a_rotate): Likewise.
12576 * config/i386/i386.c (classify_argument): Likewise.
12577 (function_arg_advance_32): Likewise.
12578 (function_arg_32): Likewise.
12579 (function_arg_64): Likewise.
12580 (function_value_64): Likewise.
12581 (ix86_gimplify_va_arg): Likewise.
12582 (ix86_legitimate_constant_p): Likewise.
12583 (put_condition_code): Likewise.
12584 (split_double_mode): Likewise.
12585 (ix86_avx256_split_vector_move_misalign): Likewise.
12586 (ix86_expand_vector_logical_operator): Likewise.
12587 (ix86_split_idivmod): Likewise.
12588 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12589 (ix86_build_const_vector): Likewise.
12590 (ix86_build_signbit_mask): Likewise.
12591 (ix86_match_ccmode): Likewise.
12592 (ix86_cc_modes_compatible): Likewise.
12593 (ix86_expand_branch): Likewise.
12594 (ix86_expand_sse_cmp): Likewise.
12595 (ix86_expand_sse_movcc): Likewise.
12596 (ix86_expand_int_sse_cmp): Likewise.
12597 (ix86_expand_vec_perm_vpermi2): Likewise.
12598 (ix86_expand_vec_perm): Likewise.
12599 (ix86_expand_sse_unpack): Likewise.
12600 (ix86_expand_int_addcc): Likewise.
12601 (ix86_split_to_parts): Likewise.
12602 (ix86_vectorize_builtin_gather): Likewise.
12603 (ix86_vectorize_builtin_scatter): Likewise.
12604 (avx_vpermilp_parallel): Likewise.
12605 (inline_memory_move_cost): Likewise.
12606 (ix86_tieable_integer_mode_p): Likewise.
12607 (x86_maybe_negate_const_int): Likewise.
12608 (ix86_expand_vector_init_duplicate): Likewise.
12609 (ix86_expand_vector_init_one_nonzero): Likewise.
12610 (ix86_expand_vector_init_one_var): Likewise.
12611 (ix86_expand_vector_init_concat): Likewise.
12612 (ix86_expand_vector_init_interleave): Likewise.
12613 (ix86_expand_vector_init_general): Likewise.
12614 (ix86_expand_vector_set): Likewise.
12615 (ix86_expand_vector_extract): Likewise.
12616 (emit_reduc_half): Likewise.
12617 (ix86_emit_i387_round): Likewise.
12618 (ix86_mangle_type): Likewise.
12619 (ix86_expand_round_sse4): Likewise.
12620 (expand_vec_perm_blend): Likewise.
12621 (canonicalize_vector_int_perm): Likewise.
12622 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12623 (expand_vec_perm_1): Likewise.
12624 (expand_vec_perm_interleave3): Likewise.
12625 (expand_vec_perm_even_odd_pack): Likewise.
12626 (expand_vec_perm_even_odd_1): Likewise.
12627 (expand_vec_perm_broadcast_1): Likewise.
12628 (ix86_vectorize_vec_perm_const_ok): Likewise.
12629 (ix86_expand_vecop_qihi): Likewise.
12630 (ix86_expand_mul_widen_hilo): Likewise.
12631 (ix86_expand_sse2_abs): Likewise.
12632 (ix86_expand_pextr): Likewise.
12633 (ix86_expand_pinsr): Likewise.
12634 (ix86_preferred_simd_mode): Likewise.
12635 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12636 * config/i386/sse.md (*andnot<mode>3): Likewise.
12637 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12638 (*<code><mode>3): Likewise.
12639 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12640 (ia64_expand_atomic_op): Likewise.
12641 (ia64_arg_type): Likewise.
12642 (ia64_mode_to_int): Likewise.
12643 (ia64_scalar_mode_supported_p): Likewise.
12644 (ia64_vector_mode_supported_p): Likewise.
12645 (expand_vec_perm_broadcast): Likewise.
12646 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12647 (iq2000_function_arg_advance): Likewise.
12648 (iq2000_function_arg): Likewise.
12649 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12650 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12651 (m68k_libcall_value): Likewise.
12652 (m68k_function_value): Likewise.
12653 (sched_attr_op_type): Likewise.
12654 * config/mcore/mcore.c (mcore_output_move): Likewise.
12655 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12657 (microblaze_function_arg): Likewise.
12658 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12659 (mips_print_operand): Likewise.
12660 (mips_mode_ok_for_mov_fmt_p): Likewise.
12661 (mips_vector_mode_supported_p): Likewise.
12662 (mips_preferred_simd_mode): Likewise.
12663 (mips_expand_vpc_loongson_even_odd): Likewise.
12664 (mips_expand_vec_unpack): Likewise.
12665 (mips_expand_vi_broadcast): Likewise.
12666 (mips_expand_vector_init): Likewise.
12667 (mips_expand_vec_reduc): Likewise.
12668 (mips_expand_msa_cmp): Likewise.
12669 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12670 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12671 (cc_flags_for_mode): Likewise.
12672 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12673 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12674 (nds32_output_casesi_pc_relative): Likewise.
12675 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12676 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12677 (nvptx_gen_unpack): Likewise.
12678 (nvptx_gen_pack): Likewise.
12679 (nvptx_gen_shuffle): Likewise.
12680 (nvptx_gen_wcast): Likewise.
12681 (nvptx_preferred_simd_mode): Likewise.
12682 * config/pa/pa.c (pa_secondary_reload): Likewise.
12683 * config/pa/predicates.md (base14_operand): Likewise.
12684 * config/powerpcspe/powerpcspe-c.c
12685 (altivec_resolve_overloaded_builtin): Likewise.
12686 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12688 (rs6000_preferred_simd_mode): Likewise.
12689 (output_vec_const_move): Likewise.
12690 (rs6000_expand_vector_extract): Likewise.
12691 (rs6000_split_vec_extract_var): Likewise.
12692 (reg_offset_addressing_ok_p): Likewise.
12693 (rs6000_legitimate_offset_address_p): Likewise.
12694 (rs6000_legitimize_address): Likewise.
12695 (rs6000_emit_set_const): Likewise.
12696 (rs6000_const_vec): Likewise.
12697 (rs6000_emit_move): Likewise.
12698 (spe_build_register_parallel): Likewise.
12699 (rs6000_darwin64_record_arg_recurse): Likewise.
12700 (swap_selector_for_mode): Likewise.
12701 (spe_init_builtins): Likewise.
12702 (paired_init_builtins): Likewise.
12703 (altivec_init_builtins): Likewise.
12704 (do_load_for_compare): Likewise.
12705 (rs6000_generate_compare): Likewise.
12706 (rs6000_expand_float128_convert): Likewise.
12707 (emit_load_locked): Likewise.
12708 (emit_store_conditional): Likewise.
12709 (rs6000_output_function_epilogue): Likewise.
12710 (rs6000_handle_altivec_attribute): Likewise.
12711 (rs6000_function_value): Likewise.
12712 (emit_fusion_gpr_load): Likewise.
12713 (emit_fusion_p9_load): Likewise.
12714 (emit_fusion_p9_store): Likewise.
12715 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12716 (fusion_gpr_mem_load): Likewise.
12717 (fusion_addis_mem_combo_load): Likewise.
12718 (fusion_addis_mem_combo_store): Likewise.
12719 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12720 (fusion_gpr_mem_load): Likewise.
12721 (fusion_addis_mem_combo_load): Likewise.
12722 (fusion_addis_mem_combo_store): Likewise.
12723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12725 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12726 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12727 (rs6000_preferred_simd_mode): Likewise.
12728 (output_vec_const_move): Likewise.
12729 (rs6000_expand_vector_extract): Likewise.
12730 (rs6000_split_vec_extract_var): Likewise.
12731 (reg_offset_addressing_ok_p): Likewise.
12732 (rs6000_legitimate_offset_address_p): Likewise.
12733 (rs6000_legitimize_address): Likewise.
12734 (rs6000_emit_set_const): Likewise.
12735 (rs6000_const_vec): Likewise.
12736 (rs6000_emit_move): Likewise.
12737 (rs6000_darwin64_record_arg_recurse): Likewise.
12738 (swap_selector_for_mode): Likewise.
12739 (paired_init_builtins): Likewise.
12740 (altivec_init_builtins): Likewise.
12741 (rs6000_expand_float128_convert): Likewise.
12742 (emit_load_locked): Likewise.
12743 (emit_store_conditional): Likewise.
12744 (rs6000_output_function_epilogue): Likewise.
12745 (rs6000_handle_altivec_attribute): Likewise.
12746 (rs6000_function_value): Likewise.
12747 (emit_fusion_gpr_load): Likewise.
12748 (emit_fusion_p9_load): Likewise.
12749 (emit_fusion_p9_store): Likewise.
12750 * config/rx/rx.c (rx_gen_move_template): Likewise.
12751 (flags_from_mode): Likewise.
12752 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12753 (s390_slb_comparison): Likewise.
12754 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12755 (s390_vector_mode_supported_p): Likewise.
12756 (s390_cc_modes_compatible): Likewise.
12757 (s390_match_ccmode_set): Likewise.
12758 (s390_canonicalize_comparison): Likewise.
12759 (s390_emit_compare_and_swap): Likewise.
12760 (s390_branch_condition_mask): Likewise.
12761 (s390_rtx_costs): Likewise.
12762 (s390_secondary_reload): Likewise.
12763 (__SECONDARY_RELOAD_CASE): Likewise.
12764 (s390_expand_cs): Likewise.
12765 (s390_preferred_simd_mode): Likewise.
12766 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12767 * config/sh/sh.c (sh_print_operand): Likewise.
12768 (dump_table): Likewise.
12769 (sh_secondary_reload): Likewise.
12770 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12771 * config/sh/sh.md (casesi_worker_1): Likewise.
12772 (casesi_worker_2): Likewise.
12773 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12774 (fcc_comparison_operator): Likewise.
12775 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12776 (emit_soft_tfmode_cvt): Likewise.
12777 (sparc_preferred_simd_mode): Likewise.
12778 (output_cbranch): Likewise.
12779 (sparc_print_operand): Likewise.
12780 (sparc_expand_vec_perm_bmask): Likewise.
12781 (vector_init_bshuffle): Likewise.
12782 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12783 (spu_vector_mode_supported_p): Likewise.
12784 (spu_expand_insv): Likewise.
12785 (spu_emit_branch_or_set): Likewise.
12786 (spu_handle_vector_attribute): Likewise.
12787 (spu_builtin_splats): Likewise.
12788 (spu_builtin_extract): Likewise.
12789 (spu_builtin_promote): Likewise.
12790 (spu_expand_sign_extend): Likewise.
12791 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12792 (tilegx_simd_int): Likewise.
12793 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12794 (tilepro_simd_int): Likewise.
12795 * config/v850/v850.c (const_double_split): Likewise.
12796 (v850_print_operand): Likewise.
12797 (ep_memory_offset): Likewise.
12798 * config/vax/vax.c (vax_rtx_costs): Likewise.
12799 (vax_output_int_move): Likewise.
12800 (vax_output_int_add): Likewise.
12801 (vax_output_int_subtract): Likewise.
12802 * config/visium/predicates.md (visium_branch_operator): Likewise.
12803 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12804 (visium_print_operand_address): Likewise.
12805 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12806 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12807 (xtensa_expand_conditional_branch): Likewise.
12808 (xtensa_copy_incoming_a7): Likewise.
12809 (xtensa_output_literal): Likewise.
12810 * dfp.c (decimal_real_maxval): Likewise.
12811 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12814 Alan Hayward <alan.hayward@arm.com>
12815 David Sherwood <david.sherwood@arm.com>
12817 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12818 (mode_nunits_inline): Likewise.
12819 (mode_inner_inline): Likewise.
12820 (mode_unit_size_inline): Likewise.
12821 (mode_unit_precision_inline): Likewise.
12822 (emit_insn_modes_h): Likewise. Also emit a #define of the
12824 (emit_mode_wider): Add an E_ prefix to mode names.
12825 (emit_mode_complex): Likewise.
12826 (emit_mode_inner): Likewise.
12827 (emit_mode_adjustments): Likewise.
12828 (emit_mode_int_n): Likewise.
12829 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12830 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12831 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12832 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12833 (CRC32_BUILTIN, ENTRY): Likewise.
12834 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12835 (aarch64_pop_regs): Likewise.
12836 (aarch64_process_components): Likewise.
12837 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12838 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12839 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12840 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12841 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12842 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12844 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12845 (pcmpistr): Likewise.
12846 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12847 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12848 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12850 * config/rl78/rl78.c (mduc_regs): Likewise.
12851 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12852 (htm_expand_builtin): Likewise.
12853 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12854 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12855 * config/xtensa/xtensa.c (print_operand): Likewise.
12856 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12857 (NUM_MODE_VECTOR_INT): Likewise.
12858 * genoutput.c (null_operand): Likewise.
12859 (output_operand_data): Likewise.
12860 * genrecog.c (print_parameter_value): Likewise.
12861 * lra.c (debug_operand_data): Likewise.
12863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12865 * dwarf2out.c (scompare_loc_descriptor_wide)
12866 (scompare_loc_descriptor_narrow): New functions, split out from...
12867 (scompare_loc_descriptor): ...here.
12868 * expmed.c (emit_store_flag_int): New function, split out from...
12869 (emit_store_flag): ...here.
12871 2017-08-30 Richard Biener <rguenther@suse.de>
12873 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12874 (dwarf2out_early_finish): Move setting of AT_pubnames from
12875 early debug output to early finish.
12877 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12879 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12880 and -mdata-region to the assembler.
12882 2017-08-30 Richard Biener <rguenther@suse.de>
12884 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12886 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12887 (dwarf2out_early_global_decl): Only generate a DIE for the
12888 abstract origin if it doesn't already exist or is a declaration DIE.
12889 (resolve_addr): Do not add the linkage name twice when
12890 generating a stub DIE for the DW_TAG_GNU_call_site target.
12892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12894 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12895 Use machine_mode rather than int for arg1_mode.
12897 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12900 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12901 that the second argument of the built-in functions to unpack
12902 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12903 switch statement instead a lot of if statements.
12904 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12905 Allow 64-bit values to be in Altivec registers as well as
12906 traditional floating point registers.
12907 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12909 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12911 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12912 MAX_REGS_PER_ADDRESS == 1.
12914 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12916 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12917 * config/i386/i386.c (ix86_option_override_internal): Simplify
12918 setting of opts->x_flag_entry.
12920 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12921 Jakub Jelinek <jakub@redhat.com>
12922 Richard Biener <rguenther@suse.de>
12924 PR tree-optimization/81503
12925 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12926 folded constant fits in the target type; reorder tests for clarity.
12928 2017-08-29 Martin Liska <mliska@suse.cz>
12930 * passes.def: Include pass_lower_switch.
12931 * stmt.c (dump_case_nodes): Remove and move to
12932 tree-switch-conversion.
12933 (case_values_threshold): Likewise.
12934 (expand_switch_as_decision_tree_p): Likewise.
12935 (emit_case_decision_tree): Likewise.
12936 (expand_case): Likewise.
12937 (balance_case_nodes): Likewise.
12938 (node_has_low_bound): Likewise.
12939 (node_has_high_bound): Likewise.
12940 (node_is_bounded): Likewise.
12941 (emit_case_nodes): Likewise.
12942 (struct simple_case_node): New struct.
12943 (add_case_node): Remove.
12944 (emit_case_dispatch_table): Use vector instead of case_list.
12945 (reset_out_edges_aux): Remove.
12946 (compute_cases_per_edge): Likewise.
12947 (expand_case): Build list of simple_case_node.
12948 (expand_sjlj_dispatch_table): Use it.
12949 * tree-switch-conversion.c (struct case_node): Moved from
12950 stmt.c and adjusted.
12951 (emit_case_nodes): Likewise.
12952 (node_has_low_bound): Likewise.
12953 (node_has_high_bound): Likewise.
12954 (node_is_bounded): Likewise.
12955 (case_values_threshold): Likewise.
12956 (reset_out_edges_aux): Likewise.
12957 (compute_cases_per_edge): Likewise.
12958 (add_case_node): Likewise.
12959 (dump_case_nodes): Likewise.
12960 (balance_case_nodes): Likewise.
12961 (expand_switch_as_decision_tree_p): Likewise.
12962 (emit_jump): Likewise.
12963 (emit_case_decision_tree): Likewise.
12964 (try_switch_expansion): Likewise.
12965 (do_jump_if_equal): Likewise.
12966 (emit_cmp_and_jump_insns): Likewise.
12967 (fix_phi_operands_for_edge): New function.
12968 (record_phi_operand_mapping): Likewise.
12969 (class pass_lower_switch): New pass.
12970 (pass_lower_switch::execute): New function.
12971 (make_pass_lower_switch): Likewise.
12972 (conditional_probability):
12973 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12974 * tree-pass.h: Add make_pass_lower_switch.
12976 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12979 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12982 2017-08-29 Richard Biener <rguenther@suse.de>
12984 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12985 we do not add a DW_AT_inline attribute twice.
12986 (gen_subprogram_die): Remove code setting DW_AT_inline on
12987 DECL_ABSTRACT_P nodes.
12989 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12991 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12992 calls to internal functions.
12993 (gimplify_modify_expr): Likewise.
12994 * tree-call-cdce.c (use_internal_fn): Likewise.
12995 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12996 (convert_to_divmod): Set the nothrow flag.
12997 * tree-if-conv.c (predicate_mem_writes): Likewise.
12998 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12999 (vectorizable_call): Likewise.
13000 (vectorizable_store): Likewise.
13001 (vectorizable_load): Likewise.
13002 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13003 (vect_recog_mask_conversion_pattern): Likewise.
13005 2017-08-29 Martin Liska <mliska@suse.cz>
13008 * gcc.c (driver_handle_option): Add new argument.
13009 * opts-common.c (handle_option): Pass
13010 target_option_override_hook.
13011 * opts-global.c (lang_handle_option): Add new option.
13012 (set_default_handlers): Add new argument.
13013 (decode_options): Likewise.
13014 * opts.c (target_handle_option): Likewise.
13015 (common_handle_option): Call target_option_override_hook.
13016 * opts.h (struct cl_option_handler_func): Add hook for
13017 target option override.
13018 (struct cl_option_handlers): Likewise.
13019 (set_default_handlers): Add new argument.
13020 (decode_options): Likewise.
13021 (common_handle_option): Likewise.
13022 (target_handle_option): Likewise.
13023 * toplev.c (toplev::main): Pass targetm.target_option.override
13026 2017-08-29 Richard Biener <rguenther@suse.de>
13027 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13029 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13030 life to the active subtree.
13032 2017-08-28 Jeff Law <law@redhat.com>
13034 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13035 derive_equivalences.
13036 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13037 Code moved into....
13038 (edge_info::derive_equivalences): New private member function
13040 * tree-ssa-dom.c (class edge_info): Changed from a struct
13041 to a class. Add ctor/dtor, methods and data members.
13042 (edge_info::edge_info): Renamed from allocate_edge_info.
13043 Initialize additional members.
13044 (edge_info::~edge_info): New.
13045 (free_dom_edge_info): Delete the edge info.
13046 (record_edge_info): Use new class & associated member functions.
13047 Tighten forms for testing for edge equivalences.
13048 (record_temporary_equivalences): Iterate over the simple
13049 equivalences rather than assuming there's only one per edge.
13050 (cprop_into_successor_phis): Iterate over the simple
13051 equivalences rather than assuming there's only one per edge.
13052 (optimize_stmt): Use operand_equal_p rather than pointer
13053 equality for mini-DSE code.
13055 2017-08-28 Nathan Sidwell <nathan@acm.org>
13057 * gcc.c (execute): Fold SIGPIPE handling into switch
13058 statement. Adjust internal error message.
13060 2017-08-28 Richard Biener <rguenther@suse.de>
13063 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13064 Do nothing for removed DIEs.
13066 2017-08-28 Richard Biener <rguenther@suse.de>
13068 PR tree-optimization/81977
13069 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13072 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13075 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13076 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13077 using targetm.gen_mem_thread_fence.
13079 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13082 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13083 predicate to register_operand. Reorder operands.
13084 (*btr<mode>): Ditto.
13085 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13086 (*btr<mode>_mask): Ditto.
13088 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13090 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13091 * config/rs6000/xmmintrin.h: New file.
13092 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13094 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13097 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13098 parameter and_insn and return it.
13099 (recombine_lvx_pattern): Insert a copy to ensure availability of
13100 the base register of the copied masking operation at the point of
13101 the instruction replacement.
13102 (recombine_stvx_pattern): Likewise.
13104 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13106 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13107 undocumented switches.
13108 (-mpower9-dform-vector): Likewise.
13109 (-mpower9-dform): Likewise.
13110 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13111 comments to delete references to -mpower9-dform* switches.
13112 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13113 Delete reference to -mpower9-dform* switches, test for
13114 -mpower9-vector instead.
13115 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13116 (OTHER_P9_VECTOR_MASKS): Likewise.
13117 (POWERPC_MASKS): Likewise.
13118 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13119 tests against -mpower9-dform* to -mpower9-vector. Delete code
13120 that checked for -mpower9-dform* consistancy with other options.
13121 Add test for -mpower9-misc to enable other power9 switches.
13122 (rs6000_init_hard_regno_mode_ok): Likewise.
13123 (rs6000_option_override_internal): Likewise.
13124 (rs6000_emit_prologue): Likewise.
13125 (rs6000_emit_epilogue): Likewise.
13126 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13127 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13128 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13130 (emit_fusion_p9_store): Likewise.
13131 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13132 resetting these macros if the assembler does not support ISA 3.0
13134 (TARGET_P9_DFORM_VECTOR): Likewise.
13135 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13136 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13138 2017-08-25 Alan Modra <amodra@gmail.com>
13140 PR rtl-optimization/81747
13141 * cse.c (cse_extended_basic_block): Don't attempt to record
13142 equivalences for degenerate conditional branches that branch
13143 to their fall-through.
13145 2017-08-24 Martin Sebor <msebor@redhat.com>
13147 PR middle-end/81908
13148 * gimple-fold.c (size_must_be_zero_p): New function.
13149 (gimple_fold_builtin_memory_op): Call it.
13151 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13153 * config/rs6000/mm_malloc.h: New file.
13155 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13157 PR tree-optimization/81913
13158 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13159 analysis when either IVs in condition can wrap.
13161 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13163 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13164 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13166 2017-08-24 Richard Biener <rguenther@suse.de>
13169 * targhooks.c (default_target_can_inline_p): Properly
13170 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13171 is present and always compare.
13172 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13173 infer -mfpmath=sse from TARGET_SSE_P.
13174 (ix86_can_inline_p): Properly use target_option_default_node when
13175 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13177 2017-08-24 Richard Biener <rguenther@suse.de>
13180 * dwarf2out.c (output_die): Handle flag_generate_offload like
13182 (output_comp_unit): Likewise.
13183 (gen_array_type_die): Likewise.
13184 (dwarf2out_early_finish): Likewise.
13185 (note_variable_value_in_expr): Likewise.
13186 (dwarf2out_finish): Likewise. Adjust assert.
13187 * cgraphunit.c (symbol_table::compile): Move setting of
13188 flag_generate_offload earlier ...
13189 (symbol_table::finalize_compilation_unit): ... here, before
13190 early debug finalization.
13192 2017-08-24 Richard Biener <rguenther@suse.de>
13194 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13195 and ipa-fnsummary.h.
13196 (ix86_can_inline_p): When ix86_fpmath flags do not match
13197 check whether the callee uses FP math at all.
13199 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13201 PR middle-end/81931
13202 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13203 instead of TYPE_PRECISION.
13205 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13206 Alan Hayward <alan.hayward@arm.com>
13207 David Sherwood <david.sherwood@arm.com>
13209 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13211 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13213 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13215 (ix86_frame::stack_realign_allocate): New field.
13216 (struct machine_frame_state): Modify comments.
13217 (machine_frame_state::sp_realigned_fp_end): New field.
13218 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13219 layout calculation.
13220 (sp_valid_at): Add assertion to assure no attempt to access invalid
13221 offset of a realigned stack.
13222 (fp_valid_at): Likewise.
13223 (choose_baseaddr): Modify comments.
13224 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13225 ix86_expand_prologue.
13226 (ix86_expand_prologue): Modify stack realignment and allocation.
13227 (ix86_expand_epilogue): Modify comments.
13228 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13229 avx2_runtime, avx512f, and avx512f_runtime.
13231 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13233 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13234 (mstackrealign): Do not init to -1.
13235 * config/i386/i386.c (ix86_option_override_internal):
13236 Check opts_set, not opts when setting default value of
13237 opts->x_ix86_force_align_arg_pointer.
13239 2017-08-23 Richard Biener <rguenther@suse.de>
13241 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13242 lang_hooks.decl_printable_name.
13243 * print-rtl-function.c (print_rtx_function): Likewise.
13244 * tree-pretty-print.c (dump_function_header): Likewise.
13246 2017-08-23 Richard Biener <rguenther@suse.de>
13249 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13250 -g0 at compile-time.
13252 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13254 PR middle-end/19706
13255 * doc/sourcebuild.texi (Other hardware attributes):
13258 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13260 PR middle-end/19706
13261 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13262 Fix single-use check.
13264 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13266 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13268 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13270 * doc/install.texi: Modify to add more details on running selected
13273 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13275 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13276 is combined with -mabi=ms.
13277 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13280 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13282 PR tree-optimization/81488
13283 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13284 and cached_basis fields.
13285 (MAX_SPREAD): New constant.
13286 (alloc_cand_and_find_basis): Initialize new fields.
13287 (clear_visited): New function.
13288 (create_phi_basis_1): Rename from create_phi_basis, set visited
13289 and cached_basis fields.
13290 (create_phi_basis): New wrapper function.
13291 (phi_add_costs_1): Rename from phi_add_costs, add spread
13292 parameter, set visited field, short-circuit when limits reached.
13293 (phi_add_costs): New wrapper function.
13294 (record_phi_increments_1): Rename from record_phi_increments, set
13296 (record_phi_increments): New wrapper function.
13297 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13298 (phi_incr_cost): New wrapper function.
13299 (all_phi_incrs_profitable_1): Rename from
13300 all_phi_incrs_profitable, set visited field.
13301 (all_phi_incrs_profitable): New wrapper function.
13303 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13304 Alan Hayward <alan.hayward@arm.com>
13305 David Sherwood <david.sherwood@arm.com>
13307 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13308 that takes the outer and inner modes.
13309 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13310 comparison as the canonical test for a paradoxical subreg.
13311 * combine.c (simplify_set): Use paradoxical_subreg_p.
13312 (make_extraction): Likewise.
13313 (force_to_mode): Likewise.
13314 (rtx_equal_for_field_assignment_p): Likewise.
13315 (gen_lowpart_for_combine): Likewise.
13316 (simplify_comparison): Likewise.
13317 * cse.c (equiv_constant): Likewise.
13318 * expmed.c (store_bit_field_1): Likewise.
13319 * final.c (alter_subreg): Likewise.
13320 * fwprop.c (propagate_rtx): Likewise.
13321 (forward_propagate_subreg): Likewise.
13322 * ira-conflicts.c (ira_build_conflicts): Likewise.
13323 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13324 * lra-constraints.c (curr_insn_transform): Likewise.
13325 (split_reg): Likewise.
13326 * lra-eliminations.c (move_plus_up): Likewise.
13327 (lra_eliminate_regs_1): Likewise.
13328 * recog.c (general_operand): Likewise.
13329 * ree.c (combine_reaching_defs): Likewise.
13330 * reload.c (push_reload): Likewise.
13331 (find_reloads): Likewise.
13332 * reload1.c (elimination_effects): Likewise.
13333 (compute_reload_subreg_offset): Likewise.
13334 (choose_reload_regs): Likewise.
13335 * rtlanal.c (subreg_lsb_1): Likewise.
13336 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13337 (simplify_subreg): Likewise.
13338 * var-tracking.c (track_loc_p): Likewise.
13339 * emit-rtl.c (byte_lowpart_offset): Likewise.
13340 (paradoxical_subreg_p): Delete out-of-line definition.
13342 2017-08-22 Jeff Law <law@redhat.com>
13344 PR tree-optimization/81741
13345 PR tree-optimization/71947
13346 * tree-ssa-dom.c: Include tree-inline.h.
13347 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13348 equivalences if one is more expensive to compute than the other.
13349 * tree-ssa-scopedtables.h (class const_or_copies): Make
13350 record_const_or_copy_raw method private.
13351 (class avail_exprs_stack): New method simplify_binary_operation.
13352 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13353 avail_exprs_stack::simplify_binary_operation as needed.
13354 (avail_exprs_stack::simplify_binary_operation): New function.
13356 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13358 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13359 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13360 (DOT_SYMBOLS): Likewise.
13361 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13362 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13363 (RS6000_ABI_NAME): Likewise.
13364 (TARGET_CMODEL): Likewise.
13365 (TOC_SECTION_ASM_OP): Likewise.
13366 (SET_CMODEL): New macro.
13367 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13369 2017-08-22 Richard Biener <rguenther@suse.de>
13371 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13372 to allow for free-lang-data replacements similar to verify_type_variant.
13374 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13375 Alan Hayward <alan.hayward@arm.com>
13376 David Sherwood <david.sherwood@arm.com>
13378 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13379 VOIDmode for the LABEL_REF.
13381 2017-08-22 Richard Biener <rguenther@suse.de>
13383 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13385 2017-08-22 Richard Biener <rguenther@suse.de>
13387 * common.opt (feliminate-dwarf2-dups): Ignore.
13388 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13389 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13390 same_die_p_wrap, compute_section_prefix,
13391 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13392 (comdat_symbol_id, comdat_symbol_number): Likewise.
13393 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13395 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13396 (output_die): Mark unreachable path unreachable.
13397 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13398 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13399 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13400 (dwarf2out_early_finish): Likewise.
13402 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13404 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13406 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13409 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13410 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13411 (avr_attribute_table) <io, io_low, address>: Initialize
13412 .decl_required with true.
13414 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13416 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13417 undocumented debugging options.
13418 (-mvsx-scalar-double): Likewise.
13419 (-mallow-df-permute): Likewise.
13420 (-mvectorize-builtins): Likewise.
13421 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13422 (rs6000_builtin_vectorized_function): Likewise.
13423 (rs6000_builtin_md_vectorized_function): Likewise.
13424 (rs6000_opt_vars): Likewise.
13426 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13429 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13430 (*btrq_imm): Rename from *btrq.
13431 (*btcq_imm): Rename from *btcq.
13432 (btsc): New code attribute.
13433 (*<btsc><mode>): New insn pattern.
13434 (*btr<mode>): Ditto.
13435 (*<btsc><mode>_mask): New insn_and_split pattern.
13436 (*btr<mode>_mask): Ditto.
13438 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13439 Alan Hayward <alan.hayward@arm.com>
13440 David Sherwood <david.sherwood@arm.com>
13442 * function.c (pad_below): Simplify padding calculation.
13444 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13445 Alan Hayward <alan.hayward@arm.com>
13446 David Sherwood <david.sherwood@arm.com>
13448 * target.def (function_prologue): Remove frame size argument.
13449 (function_epilogue): Likewise.
13450 * doc/tm.texi: Regenerate.
13451 * final.c (final_start_function): Update call to function_prologue.
13452 (final_end_function): Update call to function_epilogue.
13453 (default_function_pro_epilogue): Remove frame size argument.
13454 * output.h (default_function_pro_epilogue): Likewise.
13455 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13456 (arm_output_function_prologue): Likewise.
13457 * config/frv/frv.c (frv_function_prologue): Likewise.
13458 (frv_function_epilogue): Likewise.
13459 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13460 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13461 (ia64_output_function_epilogue): Likewise.
13462 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13463 (m32r_output_function_epilogue): Likewise.
13464 * config/microblaze/microblaze.c (microblaze_function_prologue)
13465 (microblaze_function_epilogue): Likewise.
13466 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13467 (mips_output_function_epilogue): Likewise.
13468 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13469 (mmix_target_asm_function_epilogue): Likewise.
13470 * config/msp430/msp430.c (msp430_start_function): Likewise.
13471 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13472 (nds32_asm_function_epilogue): Likewise.
13473 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13474 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13475 (pa_output_function_epilogue): Likewise.
13476 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13477 (rs6000_output_function_epilogue): Likewise.
13478 * config/rl78/rl78.c (rl78_start_function): Likewise.
13479 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13480 (rs6000_output_function_epilogue): Likewise.
13481 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13482 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13483 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13484 (sparc_asm_function_epilogue): Likewise.
13486 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13488 * tree.h (type_has_mode_precision_p): New function.
13489 * convert.c (convert_to_integer_1): Use it.
13490 * expr.c (expand_expr_real_2): Likewise.
13491 (expand_expr_real_1): Likewise.
13492 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13493 * match.pd: Likewise.
13494 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13495 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13496 * tree-tailcall.c (process_assignment): Likewise.
13497 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13498 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13499 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13500 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13501 (vectorizable_assignment): Likewise.
13502 (vectorizable_shift): Likewise.
13503 (vectorizable_operation): Likewise.
13504 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13506 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13508 * match.pd: Add pow (C, x) simplification.
13510 2017-08-21 Richard Biener <rguenther@suse.de>
13512 PR tree-optimization/81900
13513 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13514 for blocks with abnormal predecessors.
13515 (compute_antic): Do not set visited flag prematurely.
13517 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13520 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13522 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13524 * stor-layout.h (vector_type_mode): Move to...
13525 * tree.h (vector_type_mode): ...here.
13526 * stor-layout.c (vector_type_mode): Move to...
13527 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13529 2017-08-21 Richard Biener <rguenther@suse.de>
13531 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13532 register_external_die hooks.
13533 (debug_false_tree_charstarstar_uhwistar): Declare.
13534 (debug_nothing_tree_charstar_uhwi): Likewise.
13535 * debug.c (do_nothing_debug_hooks): Adjust.
13536 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13537 (debug_nothing_tree_charstar_uhwi): Likewise.
13538 * dbxout.c (dbx_debug_hooks): Adjust.
13539 (xcoff_debug_hooks): Likewise.
13540 * sdbout.c (sdb_debug_hooks): Likewise.
13541 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13542 * dwarf2out.c (macinfo_label_base): New global.
13543 (dwarf2out_register_external_die): New function for the
13544 register_external_die hook.
13545 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13546 (dwarf2_debug_hooks): Use them.
13547 (dwarf2_lineno_debug_hooks): Adjust.
13548 (struct die_struct): Add with_offset flag.
13549 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13550 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13551 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13552 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13553 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13554 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13555 defining section names for the early LTO debug variants.
13556 (reset_indirect_string): New helper.
13557 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13558 (print_dw_val): Add support for offsetted symbol references.
13559 (get_ultimate_context): Split out from is_cxx.
13560 (is_cxx): Use get_ultimate_context.
13561 (is_fortran): Add decl overload.
13562 (compute_comp_unit_symbol): Split out worker from
13563 compute_section_prefix.
13564 (compute_section_prefix): Call compute_comp_unit_symbol and
13565 set comdat_type_p here.
13566 (output_die): Skip DIE symbol output for the LTO added one.
13567 Handle DIE symbol references with offset.
13568 (output_comp_unit): Guard section name mangling properly.
13569 For LTO debug sections emit a symbol at the section beginning
13570 which we use to refer to its DIEs.
13571 (add_abstract_origin_attribute): For DIEs registered via
13572 dwarf2out_register_external_die directly refer to the early
13573 DIE rather than indirectly through the shadow one we created.
13574 Remove obsolete call to dwarf2out_abstract_function for
13575 non-function/block origins.
13576 (gen_array_type_die): When generating early LTO debug do
13577 not emit DW_AT_string_length.
13578 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13579 late when in LTO. As suggested place a gcc_unreachable for
13580 the DECL_ABSTRACT_P case.
13581 (gen_subprogram_die): Avoid another specification DIE
13582 for early built declarations/definitions for the late LTO case.
13583 (gen_variable_die): Add type references for late duplicated VLA dies
13585 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13586 we have the abstract instance already.
13587 (process_scope_var): Adjust decl DIE contexts in LTO which
13588 first puts them in limbo.
13589 (gen_decl_die): Do not generate type DIEs late apart from
13590 types for VLAs or for decls we do not yet have a DIE. Do not
13591 call dwarf2out_abstract_function late.
13592 (dwarf2out_early_global_decl): Make sure to create DIEs
13593 for abstract instances of a decl first.
13594 (dwarf2out_late_global_decl): Adjust comment.
13595 (output_macinfo_op): With multiple macro sections use
13596 macinfo_label_base to distinguish labels.
13597 (output_macinfo): Likewise. Update macinfo_label_base.
13598 Pass in the line info label.
13599 (note_variable_value_in_expr): When generating LTO resolve
13600 all variable values here by generating DIEs as needed.
13601 (init_sections_and_labels): Add early LTO debug flag parameter
13602 and generate different sections and names if set. Add generation
13603 counter for the labels so we can have multiple of them.
13604 (reset_dies): Helper to allow DIEs to be output multiple times.
13605 (dwarf2out_finish): When outputting DIEs to the fat part of an
13606 LTO object first reset DIEs.
13607 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13608 (modified_type_die): Check for decl_ultimate_origin being self
13610 (gen_type_die_with_usage): Likewise.
13611 (gen_typedef_die): Allow decl_ultimate_origin being self.
13612 (set_decl_abstract_flags): Remove.
13613 (set_block_abstract_flags): Likewise.
13614 (dwarf2out_abstract_function): Treat the early generated DIEs
13615 as the abstract copy and only add DW_AT_inline and
13616 DW_AT_artificial here and call set_decl_origin_self.
13617 If the DIE has an abstract origin don't do anything.
13618 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13619 if we have none yet (Go fails to build one, PR78628).
13620 (variably_modified_type_p): Prevent endless recursion for Ada
13621 cyclic pointer types.
13622 * lto-streamer-in.c: Include debug.h.
13623 (dref_queue): New global.
13624 (lto_read_tree_1): Stream in DIE references.
13625 (lto_input_tree): Register DIE references.
13626 (input_function): Stream DECL_DEBUG_ARGS.
13627 * lto-streamer-out.c: Include debug.h.
13628 (lto_write_tree_1): Output DIE references.
13629 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13630 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13631 (output_function): Stream DECL_DEBUG_ARGS.
13632 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13633 Stream DECL_ABSTRACT_ORIGIN.
13634 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13635 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13636 DECL_CONTEXT for file-scope decls.
13637 * lto-streamer.h (struct dref_entry): Declare.
13638 (dref_queue): Likewise.
13639 * cfgexpand.c (pass_expand::execute): Do not call the
13640 outlining_inline_function hook here.
13641 * lto-wrapper.c (debug_obj): New global.
13642 (tool_cleanup): Unlink it if required.
13643 (debug_objcopy): New function.
13644 (run_gcc): Handle early debug sections in the IL files by
13645 extracting them to separate files, partially linkin them and
13646 feeding the result back as result to the linker.
13647 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13648 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13649 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13650 sections into a separate segment.
13651 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13653 (darwin_asm_dwarf_section): Likewise.
13654 (darwin_asm_output_dwarf_offset): Likewise.
13655 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13657 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13658 Alan Hayward <alan.hayward@arm.com>
13659 David Sherwood <david.sherwood@arm.com>
13661 * read-md.h (md_reader::record_potential_iterator_use): Replace
13662 pointer argument with an rtx and an index.
13663 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13664 (apply_mode_iterator): Likewise.
13665 (apply_code_iterator): Likewise.
13666 (apply_int_iterator): Likewise.
13667 (apply_subst_iterator): Likewise.
13668 (record_iterator_use): Likewise.
13669 (record_attribute_use): Likewise.
13670 (md_reader::record_potential_iterator_use): Likewise. Update calls
13671 to record_iterator_use and apply_iterator.
13672 (iterator_use): Replace ptr with x and index.
13673 (attribute_use): Likewise.
13674 (apply_attribute_uses): Update calls to apply_iterator.
13675 (apply_iterators): Likewise. Update initialization of iterator_use.
13676 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13677 and record_potential_iterator_use.
13678 (rtx_reader::read_rtx_operand): Likewise.
13680 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13681 Alan Hayward <alan.hayward@arm.com>
13682 David Sherwood <david.sherwood@arm.com>
13684 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13687 2017-08-21 Richard Biener <rguenther@suse.de>
13689 PR middle-end/81884
13690 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13691 at struct end conservatively when comparing common bases.
13693 2017-08-21 Richard Biener <rguenther@suse.de>
13695 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13696 (mem_ref_in_stmt): Remove.
13697 (determine_max_movement): Use ref index to get at the reference.
13698 (invariantness_dom_walker::before_dom_children): Deal with
13699 lim data already initialized.
13700 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13702 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13704 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13705 (TARGET_ISA_ROUND): Ditto.
13706 (TARGET_ROUND): Ditto.
13707 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13708 * config/i386/i386.md: Ditto.
13709 * config/i386/sse.md: Ditto.
13710 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13711 with OPTION_MASK_ISA_SSE4_1.
13713 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13716 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13717 __builtin_ia32_lzcnt_u16.
13719 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13722 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13723 (rs6000_set_current_function): Rewrite function to use it.
13725 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13728 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13729 and TYPE_WARN_IF_NOT_ALIGN.
13730 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13731 (handle_warn_if_not_align): New.
13732 (place_union_field): Call handle_warn_if_not_align.
13733 (place_field): Call handle_warn_if_not_align.
13734 Copy TYPE_WARN_IF_NOT_ALIGN.
13735 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13736 (layout_type): Likewise.
13737 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13739 (tree_decl_common): Add warn_if_not_align.
13740 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13741 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13742 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13743 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13744 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13745 * doc/extend.texi: Document warn_if_not_aligned attribute.
13746 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13748 2017-08-17 Martin Liska <mliska@suse.cz>
13751 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13752 (get_data_dependence): Use it as pointer type.
13753 (distribute_loop): Likewise.
13755 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13757 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13758 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13759 (p8_vmrgow_<mode>_direct): New define_insn.
13760 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13761 handle endianness for vmrgew and vmrgow permute patterns.
13763 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13765 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13766 * config/rs6000/rs6000-cpus.def: Remove comment.
13767 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13768 (POWERPC_MASKS): Likewise.
13769 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13770 use of TARGET_VSX_TIMODE.
13771 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13772 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13773 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13774 (rs6000_option_override_internal): Remove dead code.
13775 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13776 (rs6000_legitimize_reload_address): Likewise.
13777 (rs6000_legitimate_address_p): Likewise.
13778 (rs6000_opt_masks): Delete "vsx-timode".
13779 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13780 from function comment.
13781 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13782 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13783 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13785 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13786 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13787 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13788 (VSX_TI): Likewise.
13790 (define_peephole2): Likewise.
13792 2017-08-17 Martin Sebor <msebor@redhat.com>
13795 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13796 past the end of an array.
13797 (test_pp_format): Add test cases.
13799 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13801 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13802 missing ECF_NOTHROW flags.
13804 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13807 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13808 operands residing in integer registers.
13809 (*vsx_le_perm_load_<mode>): Likewise.
13810 (*vsx_le_perm_store_<mode>): Likewise.
13811 (define_peephole2): Add peepholes to optimize the above.
13813 2017-08-17 Marek Polacek <polacek@redhat.com>
13815 PR middle-end/81814
13816 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13817 to mimic what shorten_compare did. Change the return type to bool.
13818 (fold_cond_expr_with_comparison): Update call to
13819 operand_equal_for_comparison_p.
13820 (fold_ternary_loc): Likewise.
13822 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13824 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13826 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13828 2017-08-17 Richard Biener <rguenther@suse.de>
13830 * tree-ssa-structalias.c (solve_graph): When propagating
13831 to successors update the graphs succ edges and avoid duplicate work.
13833 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13836 * config/i386/i386.c (ix86_option_override_internal): Save target
13837 specific options after ix86_stack_protector_guard_reg was changed.
13839 2017-08-17 Richard Biener <rguenther@suse.de>
13841 PR tree-optimization/81827
13842 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13843 (new_var_info): Initialize it conservatively.
13844 (get_call_vi): Mark register vars.
13845 (new_scalar_tmp_constraint_exp): Likewise.
13846 (handle_rhs_call): Likewise.
13847 (handle_const_call): Likewise.
13848 (create_function_info_for): Likewise.
13849 (solve_constraints): Sort varinfos to separate register from
13850 non-register vars to pack points-to solution bitmaps during
13853 2017-08-17 Marek Polacek <polacek@redhat.com>
13855 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13857 2017-08-17 Richard Biener <rguenther@suse.de>
13859 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13860 to true when overflow is undefined and we saturated the result.
13862 2017-08-17 Alan Modra <amodra@gmail.com>
13865 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13866 Don't use store multiple if only one reg needs saving.
13867 (interesting_frame_related_regno): New function.
13868 (rs6000_frame_related): Don't emit frame info for regs that
13870 (rs6000_emit_epilogue): Likewise.
13872 2017-08-16 Nathan Sidwell <nathan@acm.org>
13874 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13875 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13876 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13877 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13878 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13879 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13880 (verify_type): Adjust for TYPE_BINFO move.
13881 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13882 process TYPE_BINFO directly.
13883 (hash_tree): Likewise.
13884 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13886 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13889 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13891 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13893 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13896 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13897 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13898 Add ix86_binary_operator_ok to insn constraint.
13899 (*iordi_1_bts): Ditto.
13900 (*xordi_1_btc): Ditto.
13901 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13902 Update corresponding peephole2 pattern.
13906 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13908 PR tree-optimization/81832
13909 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13910 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13912 2017-08-16 Marek Polacek <polacek@redhat.com>
13915 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13916 regarding size_zero_node.
13918 2017-08-16 Martin Liska <mliska@suse.cz>
13921 * config.gcc: Respect previously set extra_objs in case
13924 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13926 PR tree-optimization/81835
13927 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13928 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13929 not depend on the phi.
13931 2017-08-16 Alan Modra <amodra@gmail.com>
13933 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13936 2017-08-16 Alan Modra <amodra@gmail.com>
13938 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13939 (save_reg_p): ..into this. Update all callers.
13940 (first_reg_to_save): Simplify.
13942 2017-08-16 Alan Modra <amodra@gmail.com>
13944 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13947 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13951 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13952 constant count if that count is less than 32.
13954 2017-08-15 Nathan Sidwell <nathan@acm.org>
13956 * gcc.c (execute): Emit friendlier message if inferior is killed
13957 by an external cause.
13959 2017-08-15 Richard Biener <rguenther@suse.de>
13961 PR tree-optimization/81790
13962 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13963 CONSTRUCTORs from simplifying and VN.
13965 2017-08-14 Martin Sebor <msebor@redhat.com>
13967 * builtin-attrs.def: Add comments.
13969 2017-08-14 Martin Sebor <msebor@redhat.com>
13972 * doc/extend.texi (attribute nonstring): Document new attribute.
13974 2017-08-14 Martin Sebor <msebor@redhat.com>
13977 * tree-diagnostic.c (default_tree_printer): Handle %G.
13978 * gimple-pretty-print.h (percent_G_format): Declare new function.
13979 * gimple-pretty-print.c (percent_G_format): Define.
13980 * tree-pretty-print.c (percent_K_format): Add argument.
13982 2017-08-14 Martin Sebor <msebor@redhat.com>
13984 PR translation/79998
13985 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13986 Remove a stray space.
13988 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13991 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13992 (*iordi_1_bts): Ditto.
13993 (*xordi_1_btc): Ditto.
13995 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13998 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13999 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14001 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14002 quoted strings, and make more translator-friendly.
14003 (darwin_rs6000_override_options): Likewise.
14004 (rs6000_option_override_internal): Likewise.
14005 (rs6000_return_in_memory): Fix overlong line.
14006 (init_cmulative_args): Use quoted strings, and make more
14007 translator-friendly.
14008 (rs6000_pass_by_reference): Fix overlong line.
14009 (def_builtin): Use quoted strings.
14010 (altivec_expand_predicate_builtin): Use quoted strings, and make
14011 more translator-friendly.
14012 (htm_expand_builtin): Use quoted strings.
14013 (cpu_expand_builtin): Use quoted strings, and make more
14014 translator-friendly.
14015 (altivec_expand_builtin): Likewise.
14016 (paired_expand_predicate_builtin): Likewise.
14017 (rs6000_invalid_builtin): Likewise.
14018 (builtin_function_type): Use quoted strings.
14019 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14020 more translator-friendly.
14021 (rs6000_trampoline_init): Likewise.
14022 (rs6000_handle_altivec_attribute): Likewise.
14023 (rs6000_inner_target_options): Use quoted strings.
14024 (rs6000_disable_incompatible_switches): Likewise.
14025 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14026 strings, and make more translator-friendly.
14027 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14029 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14031 PR tree-optimization/81799
14032 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14033 cond_expr to simple gimple operand.
14035 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14037 PR middle-end/46932
14038 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14040 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14044 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14045 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14046 TARGET_GASISR_PROLOGUES.
14047 * config/avr/avr.c (avr_option_override): Same.
14048 (avr_pass_pre_proep::execute): Same.
14050 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14053 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14054 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14056 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14058 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14059 define_insn_and_split. Split to a memory load from 0 in
14060 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14061 using PTR mode iterator.
14062 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14063 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14064 (*add_tp_<mode>): Redefine as define_insn_and_split.
14065 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14066 address space. Merge with *add_tp_x32 using PTR mode iterator.
14067 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14068 Split to an add with a memory load from 0 in
14069 DEFAULT_TLS_SEG_REG address space.
14071 2017-08-12 Andrew Pinski <apinski@cavium.com>
14073 * config/aarch64/aarch64-option-extensions.def (rdma):
14074 Fix feature string to what Linux prints out in /proc/cpuinfo.
14076 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14079 * dwarf2out.c (modified_type_die): For C typedef types that have
14080 an ultimate origin, process the ultimate origin instead of the
14082 (gen_typedef_die): Assert that input DECLs have no ultimate
14084 (gen_type_die_with_usage): For typedef variants that have an
14085 ultimate origin, just call gen_decl_die on the original DECL.
14086 (process_scope_var): Avoid creating DIEs for local typedefs and
14087 concrete static variables.
14089 2017-08-12 Alan Modra <amodra@gmail.com>
14093 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14094 match gnu-user.h startfile.
14095 (ENDFILE_LINUX_SPEC): Similarly.
14097 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14100 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14102 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14104 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14105 * config/aarch64/aarch64.md (mov<mode>): Change.
14106 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14107 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14108 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14110 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14112 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14113 for storage order barriers.
14115 2017-08-11 Martin Liska <mliska@suse.cz>
14118 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14119 variables of void type.
14121 2017-08-11 Martin Liska <mliska@suse.cz>
14123 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14124 TARGET_SUPPORTS_ALIASES.
14125 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14126 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14127 (optimize_weakref): Likewise.
14128 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14129 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14130 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14132 2017-08-11 Martin Liska <mliska@suse.cz>
14135 * config/i386/i386.c (make_resolver_func): Do complete
14136 refactoring of the function.
14138 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14141 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14142 * config/i386/i386.c (ix86_stack_protect_guard): Use
14143 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14144 * doc/invoke.texi (x86 Options): Document
14145 -mstack-protector-guard-symbol= option.
14147 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14149 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14150 * config/i386/i386.c (ix86_split_stack_guard): New function.
14151 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14152 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14153 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14154 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14155 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14156 (split_stack_space_check): Call ix86_split_stack_guard.
14158 2017-08-10 Martin Sebor <msebor@redhat.com>
14160 * print-tree.c (print_node): Print location using the established
14162 Replace spaces with colons.
14163 (debug_raw, debug): Ditto.
14165 2017-08-10 Martin Sebor <msebor@redhat.com>
14168 * pretty-print.c (pp_format): Correct the handling of %s precision.
14170 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14173 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14175 (ix86_finalize_stack_frame_flags): This. Also clear
14176 frame_pointer_needed if -fno-omit-frame-pointer is used without
14178 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14179 with ix86_finalize_stack_frame_flags.
14180 (ix86_expand_epilogue): Likewise.
14181 (ix86_expand_split_stack_prologue): Likewise.
14182 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14184 2017-08-10 Martin Liska <mliska@suse.cz>
14187 * c-attribs.c (handle_target_attribute):
14188 Report warning for an empty string argument of target attribute.
14190 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14193 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14195 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14196 or DECL_NONLOCAL labels.
14197 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14198 or DECL_NONLOCAL labels here.
14200 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14202 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14203 to indicate when early gimple folding has been disabled.
14204 (rs6000_gimple_fold_builtin): Add debug content.
14205 (rs6000_invalid_builtin): Fix whitespace.
14206 (rs6000_expand_builtin): Fix whitespace.
14207 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14209 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14212 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14213 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14215 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14217 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14218 (qdf24xx): Likewise.
14219 * config/aarch64/aarch64.md: Include falkor.md.
14220 * config/aarch64/falkor.md: New.
14222 2017-08-09 Marek Polacek <polacek@redhat.com>
14225 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14226 * diagnostic.c (emit_diagnostic): Add a comment.
14227 (emit_diagnostic_valist): New function.
14229 2017-08-09 Marek Polacek <polacek@redhat.com>
14232 * input.c (make_location): New overload.
14233 * input.h (make_location): Declare.
14235 2017-08-08 Alan Modra <amodra@gmail.com>
14236 H.J. Lu <hongjiu.lu@intel.com>
14239 * gcc.c (NO_PIE_SPEC): Delete.
14240 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14242 (LINK_PIE_SPEC): ..to here.
14243 (LINK_COMMAND_SPEC): Support -no-pie.
14244 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14245 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14247 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14248 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14249 (ENDFILE_CRTEND_SPEC): Similarly.
14251 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14254 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14255 (mstack-protector-guard-offset=): Ditto.
14256 * config/i386/i386.c (ix86_option_override): Handle
14257 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14259 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14260 ix86_stack_protect_guard_offset variables.
14261 (TARGET_STACK_PROTECT_GUARD): Always define.
14262 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14263 and -mstack-protector-guard-offset= options.
14265 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14267 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14268 boundary case for the last candidate.
14270 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14272 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14273 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14275 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14277 PR middle-end/19706
14278 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14279 * config/aarch64/aarch64-builtins.c
14280 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14281 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14282 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14284 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14285 Andrew Pinski <pinskia@gmail.com>
14287 PR middle-end/19706
14288 * internal-fn.def (XORSIGN): New.
14289 * optabs.def (xorsign_optab): New.
14290 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14291 (convert_expand_mult_copysign): New.
14292 (pass_optimize_widening_mul::execute): Call
14293 convert_expand_mult_copysign.
14295 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14297 PR tree-optimization/81354
14298 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14299 Insert on edges rather than explicitly creating landing pads.
14300 (analyze_candidates_and_replace): Commit edge inserts.
14302 2017-08-08 Richard Biener <rguenther@suse.de>
14304 PR middle-end/81719
14305 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14306 (expand_simple_operations): Also look through ADDR_EXPRs with
14307 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14309 2017-08-08 Richard Biener <rguenther@suse.de>
14311 PR tree-optimization/81723
14312 * tree-vect-slp.c (struct bst_traits): New hash traits.
14313 (bst_fail): New global.
14314 (vect_build_slp_tree_2): New worker, split out from ...
14315 (vect_build_slp_tree): ... this now wrapping it with using
14316 bst_fail set to cache SLP tree build fails. Properly handle
14318 (vect_analyze_slp_instance): Allocate and free bst_fail.
14320 2017-08-08 Martin Liska <mliska@suse.cz>
14323 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14324 LABEL_DECLs that can be from a different function.
14326 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14328 PR tree-optimization/81744
14329 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14330 loop's number of iterations.
14332 2017-08-08 Martin Liska <mliska@suse.cz>
14334 * asan.c: Include header files.
14335 * attribs.c (build_decl_attribute_variant): New function moved
14337 (build_type_attribute_qual_variant): Likewise.
14338 (cmp_attrib_identifiers): Likewise.
14339 (simple_cst_list_equal): Likewise.
14340 (omp_declare_simd_clauses_equal): Likewise.
14341 (attribute_value_equal): Likewise.
14342 (comp_type_attributes): Likewise.
14343 (build_type_attribute_variant): Likewise.
14344 (lookup_ident_attribute): Likewise.
14345 (remove_attribute): Likewise.
14346 (merge_attributes): Likewise.
14347 (merge_type_attributes): Likewise.
14348 (merge_decl_attributes): Likewise.
14349 (merge_dllimport_decl_attributes): Likewise.
14350 (handle_dll_attribute): Likewise.
14351 (attribute_list_equal): Likewise.
14352 (attribute_list_contained): Likewise.
14353 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14354 (lookup_attribute_by_prefix): Likewise.
14355 * bb-reorder.c: Include header files.
14356 * builtins.c: Likewise.
14357 * calls.c: Likewise.
14358 * cfgexpand.c: Likewise.
14359 * cgraph.c: Likewise.
14360 * cgraphunit.c: Likewise.
14361 * convert.c: Likewise.
14362 * dwarf2out.c: Likewise.
14363 * final.c: Likewise.
14364 * fold-const.c: Likewise.
14365 * function.c: Likewise.
14366 * gimple-expr.c: Likewise.
14367 * gimple-fold.c: Likewise.
14368 * gimple-pretty-print.c: Likewise.
14369 * gimple.c: Likewise.
14370 * gimplify.c: Likewise.
14371 * hsa-common.c: Likewise.
14372 * hsa-gen.c: Likewise.
14373 * internal-fn.c: Likewise.
14374 * ipa-chkp.c: Likewise.
14375 * ipa-cp.c: Likewise.
14376 * ipa-devirt.c: Likewise.
14377 * ipa-fnsummary.c: Likewise.
14378 * ipa-inline.c: Likewise.
14379 * ipa-visibility.c: Likewise.
14381 * lto-cgraph.c: Likewise.
14382 * omp-expand.c: Likewise.
14383 * omp-general.c: Likewise.
14384 * omp-low.c: Likewise.
14385 * omp-offload.c: Likewise.
14386 * omp-simd-clone.c: Likewise.
14387 * opts-global.c: Likewise.
14388 * passes.c: Likewise.
14389 * predict.c: Likewise.
14390 * sancov.c: Likewise.
14391 * sanopt.c: Likewise.
14392 * symtab.c: Likewise.
14393 * toplev.c: Likewise.
14394 * trans-mem.c: Likewise.
14395 * tree-chkp.c: Likewise.
14396 * tree-eh.c: Likewise.
14397 * tree-into-ssa.c: Likewise.
14398 * tree-object-size.c: Likewise.
14399 * tree-parloops.c: Likewise.
14400 * tree-profile.c: Likewise.
14401 * tree-ssa-ccp.c: Likewise.
14402 * tree-ssa-live.c: Likewise.
14403 * tree-ssa-loop.c: Likewise.
14404 * tree-ssa-sccvn.c: Likewise.
14405 * tree-ssa-structalias.c: Likewise.
14406 * tree-ssa.c: Likewise.
14407 * tree-streamer-in.c: Likewise.
14408 * tree-vectorizer.c: Likewise.
14409 * tree-vrp.c: Likewise.
14410 * tsan.c: Likewise.
14411 * ubsan.c: Likewise.
14412 * varasm.c: Likewise.
14413 * varpool.c: Likewise.
14414 * tree.c: Remove functions moved to attribs.[ch].
14415 * tree.h: Likewise.
14416 * config/aarch64/aarch64.c: Add attrs.h header file.
14417 * config/alpha/alpha.c: Likewise.
14418 * config/arc/arc.c: Likewise.
14419 * config/arm/arm.c: Likewise.
14420 * config/avr/avr.c: Likewise.
14421 * config/bfin/bfin.c: Likewise.
14422 * config/c6x/c6x.c: Likewise.
14423 * config/cr16/cr16.c: Likewise.
14424 * config/cris/cris.c: Likewise.
14425 * config/darwin.c: Likewise.
14426 * config/epiphany/epiphany.c: Likewise.
14427 * config/fr30/fr30.c: Likewise.
14428 * config/frv/frv.c: Likewise.
14429 * config/ft32/ft32.c: Likewise.
14430 * config/h8300/h8300.c: Likewise.
14431 * config/i386/winnt.c: Likewise.
14432 * config/ia64/ia64.c: Likewise.
14433 * config/iq2000/iq2000.c: Likewise.
14434 * config/lm32/lm32.c: Likewise.
14435 * config/m32c/m32c.c: Likewise.
14436 * config/m32r/m32r.c: Likewise.
14437 * config/m68k/m68k.c: Likewise.
14438 * config/mcore/mcore.c: Likewise.
14439 * config/microblaze/microblaze.c: Likewise.
14440 * config/mips/mips.c: Likewise.
14441 * config/mmix/mmix.c: Likewise.
14442 * config/mn10300/mn10300.c: Likewise.
14443 * config/moxie/moxie.c: Likewise.
14444 * config/msp430/msp430.c: Likewise.
14445 * config/nds32/nds32-isr.c: Likewise.
14446 * config/nds32/nds32.c: Likewise.
14447 * config/nios2/nios2.c: Likewise.
14448 * config/nvptx/nvptx.c: Likewise.
14449 * config/pa/pa.c: Likewise.
14450 * config/pdp11/pdp11.c: Likewise.
14451 * config/powerpcspe/powerpcspe.c: Likewise.
14452 * config/riscv/riscv.c: Likewise.
14453 * config/rl78/rl78.c: Likewise.
14454 * config/rx/rx.c: Likewise.
14455 * config/s390/s390.c: Likewise.
14456 * config/sh/sh.c: Likewise.
14457 * config/sol2.c: Likewise.
14458 * config/sparc/sparc.c: Likewise.
14459 * config/spu/spu.c: Likewise.
14460 * config/stormy16/stormy16.c: Likewise.
14461 * config/tilegx/tilegx.c: Likewise.
14462 * config/tilepro/tilepro.c: Likewise.
14463 * config/v850/v850.c: Likewise.
14464 * config/vax/vax.c: Likewise.
14465 * config/visium/visium.c: Likewise.
14466 * config/xtensa/xtensa.c: Likewise.
14468 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14471 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14472 constraints since the -mupper-regs-* switches have been
14474 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14475 into a vector from a double word element that was extracted from
14476 another vector, and eliminate extra XXPERMDI instructions.
14477 (vsx_concat_<mode>_2): Likewise.
14478 (vsx_concat_<mode>_3): Likewise.
14479 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14480 concat to allow optimizing inserts from previous extracts.
14482 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14484 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14485 memory reference to a SSP offset in TLS address space.
14486 (ix86_print_operand) <case '@'>: Remove.
14487 (ix86_print_operand_punct_valid_p): Remove '@' code.
14488 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14489 UNSPEC_SP_TLS_TEST.
14490 (stack_tls_protect_set_<mode>): Remove.
14491 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14492 (stack_tls_protect_test_<mode>): Remove.
14493 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14495 2017-08-07 Olivier Hainque <hainque@adacore.com>
14498 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14500 2017-08-07 Douglas Rupp <rupp@adacore.com>
14502 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14503 variable was referenced as multidir in command.
14505 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14508 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14511 2017-08-07 Martin Liska <mliska@suse.cz>
14513 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14514 * config/rl78/rl78.c: Add include of attribs.h.
14515 * config/sh/sh.c: Likewise.
14516 * config/v850/v850.c: Likewise.
14518 2017-08-07 Tom de Vries <tom@codesourcery.com>
14520 PR middle-end/78266
14521 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14523 2017-08-07 Martin Liska <mliska@suse.cz>
14525 * config/mips/mips.c: Include attribs.h.
14527 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14530 * doc/invoke.texi: Document change in behvaior for -Ofast for
14533 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14535 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14537 (aarch64_pop_regs): Likewise.
14538 (aarch64_gen_load_pair): Likewise.
14539 (aarch64_save_callee_saves): Likewise.
14540 (aarch64_restore_callee_saves): Likewise.
14542 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14544 * config/i386/i386.c: Revert the last change.
14546 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14549 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14551 (ix86_finalize_stack_frame_flags): This. Also clear
14552 frame_pointer_needed if -fno-omit-frame-pointer is used without
14554 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14555 with ix86_finalize_stack_frame_flags.
14556 (ix86_expand_epilogue): Likewise.
14557 (ix86_expand_split_stack_prologue): Likewise.
14559 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14562 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14563 to P_AES for Westmere.
14565 2017-08-07 Jonathan Yong <10walls@gmail.com>
14567 * config/i386/mingw.opt (fset-stack-executable): Removed.
14568 * config/i386/cygming.opt (fset-stack-executable): Moved
14570 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14572 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14574 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14576 2017-08-07 Marek Polacek <polacek@redhat.com>
14578 PR middle-end/81737
14579 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14581 2017-08-07 Martin Liska <mliska@suse.cz>
14583 * attribs.h (canonicalize_attr_name): New function.
14584 (cmp_attribs): Move from c-format.c and adjusted.
14585 (is_attribute_p): Moved from tree.h.
14586 * tree-inline.c: Add new includes.
14587 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14588 (private_is_attribute_p): Remove.
14589 (private_lookup_attribute): Likewise.
14590 (private_lookup_attribute_by_prefix): Simplify.
14591 (remove_attribute): Use is_attribute_p.
14592 * tree.h: Remove removed declarations.
14594 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14596 PR middle-end/81698
14597 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14598 instead of computing it in the function. Formatting fix.
14599 (expand_case): Don't rely on default_edge being the first edge,
14600 clear it if removing it, pass default_edge to
14601 emit_case_dispatch_table.
14602 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14605 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14607 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14608 insn in the function, emit NOP after the insn.
14610 2017-08-06 Tom de Vries <tom@codesourcery.com>
14612 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14615 2017-08-06 Tom de Vries <tom@codesourcery.com>
14617 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14620 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14622 PR tree-optimization/57371
14623 * match.pd: New pattern.
14625 2017-08-04 Marek Polacek <polacek@redhat.com>
14627 PR middle-end/81695
14628 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14629 perform the computation in offset_int.
14631 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14633 PR tree-optimization/81136
14634 * tree-vectorizer.h: Include tree-hash-traits.h.
14635 (vec_base_alignments): New typedef.
14636 (vec_info): Add a base_alignments field.
14637 (vect_record_base_alignments): Declare.
14638 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14640 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14641 (create_data_ref): Add an is_conditional_in_stmt argument.
14642 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14643 the is_conditional_in_stmt field.
14644 (data_ref_loc): Add an is_conditional_in_stmt field.
14645 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14646 (find_data_references_in_stmt): Update call to create_data_ref.
14647 (graphite_find_data_references_in_stmt): Likewise.
14648 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14649 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14650 (vect_record_base_alignment): New function.
14651 (vect_record_base_alignments): Likewise.
14652 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14653 for nested statements even if we fail to compute a misalignment.
14654 Use pooled base alignments for unconditional references.
14655 (vect_find_same_alignment_drs): Compare base addresses instead
14657 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14658 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14660 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14662 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14663 Add an explicit name for the enum. Use auto_vec for slp_instances
14664 and grouped_stores.
14665 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14667 (_bb_vec_info): Add a constructor and destructor.
14668 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14669 (destroy_loop_vec_info): Delete.
14670 (vect_destroy_datarefs): Likewise.
14671 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14672 (vec_info::vec_info): New function.
14673 (vec_info::~vec_info): Likewise.
14674 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14675 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14676 destroy_loop_vec_info.
14677 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14678 (_loop_vec_info::_loop_vec_info): ...this.
14679 (destroy_loop_vec_info): Replace with...
14680 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14681 the stmt_vec_infos. Leave handling of vec_info information to its
14682 destructor. Remove explicit vector releases.
14683 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14684 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14685 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14686 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14687 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14688 (destroy_bb_vec_info): Replace with...
14689 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14690 information to its destructor.
14691 (vect_slp_analyze_bb_1): Use new and delete instead of
14692 new_bb_vec_info and destroy_bb_vec_info.
14693 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14696 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14698 * tree-data-ref.h (subscript): Add access_fn field.
14699 (data_dependence_relation): Add could_be_independent_p.
14700 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14701 (same_access_functions): Move to tree-data-ref.c.
14702 * tree-data-ref.c (ref_contains_union_access_p): New function.
14703 (access_fn_component_p): Likewise.
14704 (access_fn_components_comparable_p): Likewise.
14705 (dr_analyze_indices): Add a reference to access_fn_component_p.
14706 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14708 (constant_access_functions): Likewise.
14709 (add_other_self_distances): Likewise.
14710 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14711 (initialize_data_dependence_relation): Use XCNEW and remove
14712 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14713 of access functions that have the same type. Allow the
14714 subsequence to end with different bases in some circumstances.
14715 Record the chosen access functions in SUB_ACCESS_FN.
14716 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14717 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14718 (subscript_dependence_tester_1): Likewise dra and drb.
14719 (build_classic_dist_vector): Update calls accordingly.
14720 (subscript_dependence_tester): Likewise.
14721 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14722 DDR_COULD_BE_INDEPENDENT_P.
14723 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14724 comp_alias_ddrs instead of may_alias_ddrs.
14725 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14727 (vect_analyze_data_ref_dependence): Use it if
14728 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14729 distance vectors if that fails.
14730 (dependence_distance_ge_vf): New function.
14731 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14732 LOOP_VINFO_MAY_ALIAS_DDRS.
14734 2017-08-04 Richard Biener <rguenther@suse.de>
14736 PR middle-end/81705
14737 * fold-const.c (fold_binary_loc): Properly restrict
14738 minus_var0 && minus_var1 case when associating undefined overflow
14741 2017-08-04 Tom de Vries <tom@codesourcery.com>
14743 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14745 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14748 Don't start diagnostic messages with a capital letter.
14749 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14751 (rs6000_invalid_builtin): Likewise.
14752 (rs6000_trampoline_init): Likewise.
14754 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14757 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14758 after setting changeable df flags.
14760 2017-08-03 Richard Biener <rguenther@suse.de>
14762 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14763 up if the use is in USE - X.
14765 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14767 * toplev.c (dumpfile.h): New include.
14768 (internal_error_reentered): New static function. Use it...
14769 (internal_error_function): ...here to handle reentered internal_error.
14771 2017-08-03 Richard Biener <rguenther@suse.de>
14773 PR middle-end/81148
14774 * fold-const.c (split_tree): Add minus_var and minus_con
14775 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14776 here but always use minus_*.
14777 (associate_trees): Assert we never associate with MINUS_EXPR
14778 and NULL first operand. Do not recurse for PLUS_EXPR operands
14779 when associating as MINUS_EXPR either.
14780 (fold_binary_loc): Track minus_var and minus_con.
14782 2017-08-03 Tom de Vries <tom@codesourcery.com>
14785 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14786 ACCEL_COMPILER, apply finish_options on
14787 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14789 2017-08-03 Tom de Vries <tom@codesourcery.com>
14792 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14793 function_entry_patch_area_size > 0.
14795 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14798 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14799 instead of 10??LU, perform unit multiplication in wide_int,
14800 don't change alloc_object_size_limit if the limit is larger
14803 PR tree-optimization/81655
14804 PR tree-optimization/81588
14805 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14806 the case when ranges[i].low and high are 1 for unsigned type with
14809 PR middle-end/81052
14810 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14811 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14813 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14815 * tree-vrp.h: Add include guard.
14817 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14820 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14821 (ud2): New insn pattern.
14822 * config/i386/i386.c (ix86_expand_epilogue):
14823 For naked functions, generate ud2 instead of trap insn.
14825 2017-08-02 Marek Polacek <polacek@redhat.com>
14828 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14830 2017-08-02 Tom de Vries <tom@codesourcery.com>
14831 Cesar Philippidis <cesar@codesourcery.com>
14833 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14834 Add missing edge probabilities.
14836 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14838 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14839 Correct endianness.
14841 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14843 PR middle-end/79499
14844 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14845 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14846 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14847 prologue_seq sequences - if any.
14849 2017-08-02 Richard Biener <rguenther@suse.de>
14851 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14852 via vectors if supported, integer extracts via punning if supported
14853 or otherwise vector extracts.
14855 2017-08-02 Richard Biener <rguenther@suse.de>
14857 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14859 (bitmap_insert_into_set): ... this.
14861 2017-08-02 Richard Biener <rguenther@suse.de>
14863 PR tree-optimization/81633
14865 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14867 PR tree-optimization/71752
14868 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14870 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14872 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14873 (machine_function::call_ms2sysv_pad_out): Remove field.
14874 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14875 (ix86_compute_frame_layout): Likewise.
14877 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14880 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14881 attribute with interrupt attribute.
14883 2017-08-01 Andrew Pinski <apinski@cavium.com>
14885 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14886 BIT_INSERT_EXPR's operand 1
14887 to see if the types precision matches.
14889 2017-08-01 Martin Liska <mliska@suse.cz>
14891 PR middle-end/70140
14892 * builtins.c (expand_builtin_memcpy_args): Remove.
14893 (expand_builtin_memcpy): Call newly added function
14894 expand_builtin_memory_copy_args.
14895 (expand_builtin_memcpy_with_bounds): Likewise.
14896 (expand_builtin_mempcpy): Remove last argument.
14897 (expand_builtin_mempcpy_with_bounds): Likewise.
14898 (expand_builtin_memory_copy_args): New function created from
14899 expand_builtin_mempcpy_args with small modifications.
14900 (expand_builtin_mempcpy_args): Remove.
14901 (expand_builtin_stpcpy): Remove unused argument.
14902 (expand_builtin): Likewise.
14903 (expand_builtin_with_bounds): Likewise.
14905 2017-08-01 Martin Liska <mliska@suse.cz>
14908 Make mempcpy more optimal (PR middle-end/70140).
14910 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14914 __builtin_vec_cmpne verify both arguments are compatible vectors
14915 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14916 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14917 move computation of aligned to after checking the argument types.
14921 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14922 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14925 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14926 Doug Rupp <rupp@adacore.com>
14927 Olivier Hainque <hainque@adacore.com>
14929 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14930 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14932 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14933 for TARGET_OS_CPP_BUILTIN.
14934 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14935 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14937 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14938 passing required abi options to the assembler for EABI configurations.
14939 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14940 of .text.hot and .text.unlikely sections for kernel modules when
14941 using ARM style exceptions.
14942 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14943 options. Add EXTRA_CC1_SPEC.
14944 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14946 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14948 (ARM_TARGET2_DWARF_FORMAT): Define.
14949 * config/arm/t-vxworks: Adjust multilib control to removal of the
14950 Diab command line options.
14952 2017-08-01 Martin Liska <mliska@suse.cz>
14954 PR gcov-profile/81561
14955 * gcov.c (unblock): Make unblocking safe as we need to preserve
14956 index correspondence of blocks and block_lists.
14958 2017-08-01 Richard Biener <rguenther@suse.de>
14960 PR tree-optimization/81181
14961 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14962 (compute_antic): ... end of iteration here.
14964 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14966 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14967 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14968 (ftree-slp-vectorize): Likewise.
14969 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14970 can no longer be set independent of flag_tree_loop_vectorize.
14971 * omp-general.c (emp_max_vf): Likewise.
14972 * opts.c (enable_fdo_optimizations): Remove references to
14973 flag_tree_vectorize, these are now implicit.
14974 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14975 and leave it for the options machinery.
14977 2017-08-01 Martin Liska <mliska@suse.cz>
14979 PR middle-end/70140
14980 * builtins.c (expand_builtin_memcpy_args): Remove.
14981 (expand_builtin_memcpy): Call newly added function
14982 expand_builtin_memory_copy_args.
14983 (expand_builtin_memcpy_with_bounds): Likewise.
14984 (expand_builtin_mempcpy): Remove last argument.
14985 (expand_builtin_mempcpy_with_bounds): Likewise.
14986 (expand_builtin_memory_copy_args): New function created from
14987 expand_builtin_mempcpy_args with small modifications.
14988 (expand_builtin_mempcpy_args): Remove.
14989 (expand_builtin_stpcpy): Remove unused argument.
14990 (expand_builtin): Likewise.
14991 (expand_builtin_with_bounds): Likewise.
14993 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14996 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14997 print "ds:" only for immediates in generic address space.
14999 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15002 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15003 (ix86_function_ok_for_sibcall): Return false for naked functions.
15005 2017-08-01 Richard Biener <rguenther@suse.de>
15007 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15008 (compute_antic): Seed worklist with exit block predecessors.
15009 * cfganal.c (dfs_find_deadend): For a cycle return the source
15010 of the edge closing it.
15012 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15014 * config/aarch64/aarch64.c
15015 (aarch64_can_const_movi_rtx_p): Move 0 check.
15017 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15019 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15020 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15022 * match.pd: Ditto in address comparison pattern.
15024 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15026 PR tree-optimization/81627
15027 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15028 closed ssa form for store-store chain.
15030 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15032 PR tree-optimization/81620
15033 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15034 for store-store chain.
15036 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15038 PR tree-optimization/81588
15039 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15040 ranges[i].in_p, invert comparison code ccode. For >/>=,
15041 swap rhs1 and rhs2 and comparison code unconditionally,
15042 for </<= don't do that. Don't swap rhs1/rhs2 again if
15043 ranges[i].in_p, instead invert comparison code ccode if
15044 opcode or oe->rank is BIT_IOR_EXPR.
15047 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15048 a direct optab to conversion optab.
15049 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15050 with GET_MODE_INNER as last argument instead of optab_handler.
15051 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15052 vector extraction if possible and optab is available.
15053 * expr.c (store_constructor): Use convert_optab_handler instead
15054 of optab_handler. Use vector initialization from smaller
15055 vectors if possible and optab is available.
15056 * tree-vect-stmts.c (vectorizable_load): Likewise.
15057 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15058 now have two modes.
15059 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15060 of vec_init from half-sized vectors with the same element mode.
15061 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15062 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15063 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15064 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15065 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15066 after mode in gen_vec_extract* calls.
15067 (vec_extract<mode>): Renamed to ...
15068 (vec_extract<mode><ssescalarmodelower>): ... this.
15069 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15070 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15071 element mode after mode in gen_vec_init* calls.
15072 (VEC_INIT_HALF_MODE): New mode iterator.
15073 (vec_init<mode>): Renamed to ...
15074 (vec_init<mode><ssescalarmodelower>): ... this.
15075 (vec_init<mode><ssehalfvecmodelower>): New expander.
15076 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15077 (vec_extractv2sfsf): ... this.
15078 (vec_initv2sf): Renamed to ...
15079 (vec_initv2sfsf): ... this.
15080 (vec_extractv2si): Renamed to ...
15081 (vec_extractv2sisi): ... this.
15082 (vec_initv2si): Renamed to ...
15083 (vec_initv2sisi): ... this.
15084 (vec_extractv4hi): Renamed to ...
15085 (vec_extractv4hihi): ... this.
15086 (vec_initv4hi): Renamed to ...
15087 (vec_initv4hihi): ... this.
15088 (vec_extractv8qi): Renamed to ...
15089 (vec_extractv8qiqi): ... this.
15090 (vec_initv8qi): Renamed to ...
15091 (vec_initv8qiqi): ... this.
15092 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15093 (vec_init<mode>): Renamed to ...
15094 (vec_init<mode><VEC_base_l>): ... this.
15095 (vec_extract<mode>): Renamed to ...
15096 (vec_extract<mode><VEC_base_l>): ... this.
15097 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15098 (vec_initv2sfsf): ... this.
15099 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15100 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15101 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15102 element mode after mode in gen_vec_init* calls.
15103 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15104 (vec_init<mode><Vel>): ... this.
15105 (vec_extract<mode>): Renamed to ...
15106 (vec_extract<mode><Vel>): ... this.
15107 * config/aarch64/iterators.md (Vel): New mode attribute.
15108 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15109 Add element mode after mode in gen_vec_extract* calls.
15110 * config/s390/vector.md (non_vec_l): New mode attribute.
15111 (vec_extract<mode>): Renamed to ...
15112 (vec_extract<mode><non_vec_l>): ... this.
15113 (vec_init<mode>): Renamed to ...
15114 (vec_init<mode><non_vec_l>): ... this.
15115 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15116 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15118 * config/arm/iterators.md (V_elem_l): New mode attribute.
15119 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15120 (vec_extract<mode><V_elem_l>): ... this.
15121 (vec_extractv2di): Renamed to ...
15122 (vec_extractv2didi): ... this.
15123 (vec_init<mode>): Renamed to ...
15124 (vec_init<mode><V_elem_l>): ... this.
15125 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15126 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15127 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15128 Add element mode after gen_vec_extract* calls.
15129 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15130 (vec_init<mode><unitmode>): ... this.
15131 (vec_extract<mode>): Renamed to ...
15132 (vec_extract<mode><unitmode>): ... this.
15133 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15134 (vec_init<mode><unitmode>): ... this.
15135 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15136 (vec_initv2sfsf): ... this.
15137 (vec_extractv2sf): Renamed to ...
15138 (vec_extractv2sfsf): ... this.
15139 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15140 Add element mode after gen_vec_extract* calls.
15141 * config/mips/mips.md (unitmode): New mode iterator.
15142 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15143 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15144 * config/spu/spu.md (inner_l): New mode attribute.
15145 (vec_init<mode>): Renamed to ...
15146 (vec_init<mode><inner_l>): ... this.
15147 (vec_extract<mode>): Renamed to ...
15148 (vec_extract<mode><inner_l>): ... this.
15149 * config/sparc/sparc.md (veltmode): New mode iterator.
15150 (vec_init<VMALL:mode>): Renamed to ...
15151 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15152 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15153 (vec_initv2sisi): ... this.
15154 (vec_initv2sf): Renamed to ...
15155 (vec_initv2sfsf): ... this.
15156 (vec_extractv2sf): Renamed to ...
15157 (vec_extractv2sfsf): ... this.
15158 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15159 (vec_init<mode>): Renamed to ...
15160 (vec_init<mode><VEC_base_l>): ... this.
15161 (vec_extract<mode>): Renamed to ...
15162 (vec_extract<mode><VEC_base_l>): ... this.
15163 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15164 (vec_initv2sfsf): ... this.
15165 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15166 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15167 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15168 gen_vec_init* calls.
15170 2017-08-01 Richard Biener <rguenther@suse.de>
15172 PR tree-optimization/81297
15173 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15174 TREE_OVERFLOW from INTEGER_CSTs.
15176 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15178 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15180 2017-07-31 Carl Love <cel@us.ibm.com>
15182 * config/rs6000/rs6000-c: Add support for built-in functions
15183 vector signed char vec_xl_be (signed long long, signed char *);
15184 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15185 vector signed int vec_xl_be (signed long long, signed int *);
15186 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15187 vector signed long long vec_xl_be (signed long long, signed long long *);
15188 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15189 vector signed short vec_xl_be (signed long long, signed short *);
15190 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15191 vector double vec_xl_be (signed long long, double *);
15192 vector float vec_xl_be (signed long long, float *);
15193 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15194 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15195 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15197 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15198 (altivec_expand_builtin): Add switch statement to call
15199 altivec_expand_xl_be for each builtin.
15200 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15201 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15202 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15203 __builtin_vsx_le_be_v16qi.
15204 * doc/extend.texi: Update the built-in documentation file for the
15205 new built-in functions.
15207 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15210 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15212 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15214 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15216 * config.gcc: Add z14.
15217 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15218 CPU model numbers for z13s and z14.
15219 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15221 * config/s390/s390-opts.h (enum processor_type): Rename
15222 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15223 * config/s390/s390.c (processor_table): Add field for CPU name to
15224 be passed to Binutils.
15225 (s390_asm_output_machine_for_arch): Use the new field in
15226 processor_table for Binutils.
15227 (s390_expand_builtin): Replace arch12 with z14.
15228 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15229 (s390_get_sched_attrmask): Likewise.
15230 (s390_get_unit_mask): Likewise.
15231 * config/s390/s390.opt: Add z14 to processor_type enum.
15233 2017-07-31 Martin Jambor <mjambor@suse.cz>
15236 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15237 regardless of optimization level.
15239 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15240 Martin Liska <mliska@suse.cz>
15242 * predict.def: Remove old comment and adjust probability.
15243 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15244 PREDICT statements.
15246 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15249 * config/i386/i386.c (ix86_function_naked): New function.
15250 (ix86_can_use_return_insn_p): Return false for naked functions.
15251 (ix86_expand_prologue): Skip prologue for naked functions.
15252 (ix86_expand_epilogue): Skip epilogue for naked functions
15253 and emit trap instruction.
15254 (ix86_warn_func_return): New function.
15255 (ix86_attribute_table): Add "naked" attribute specification.
15256 (TARGET_WARN_FUNC_RETURN): Define.
15257 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15259 2017-07-31 Martin Liska <mliska@suse.cz>
15261 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15262 (dump_gimple_bb_header): Always dump BB info.
15263 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15265 2017-07-31 Martin Liska <mliska@suse.cz>
15268 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15269 also with current_function_decl non-null equality.
15271 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15274 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15275 change type to the element type, instead add eltype variable and
15276 use it where we are interested in the element type.
15278 PR tree-optimization/81603
15279 * ipa-polymorphic-call.c
15280 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15281 offset arithmetic in offset_int, bail out if the resulting bit offset
15282 doesn't fit into shwi.
15284 2017-07-31 Martin Liska <mliska@suse.cz>
15286 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15287 (gimplify_save_expr): Fix comment.
15289 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15292 * config/i386/i386.c (ix86_function_arg): Update arguments for
15294 (ix86_compute_frame_layout): Set the initial stack offset to
15295 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15296 INCOMING_FRAME_SP_OFFSET.
15297 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15298 stack before exception handler returns.
15299 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15300 the 'ERROR_CODE' for exception handler.
15302 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15304 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15305 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15306 (ASM_OUTPUT_REG_POP): Ditto.
15307 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15308 instead of asm_fprintf to output pure string.
15310 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15312 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15313 to imported_module_or_decl hook.
15314 (debug_nothing_tree_tree_tree_bool): Remove.
15315 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15316 * debug.c (do_nothing_debug_hooks): Use
15317 debug_nothing_tree_tree_tree_bool_bool instead of
15318 debug_nothing_tree_tree_tree_bool.
15319 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15320 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15321 * sdbout.c (sdb_debug_hooks): Likewise.
15322 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15323 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15325 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15326 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15327 attribute, don't add anything.
15329 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15331 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15332 (fold_build2_stat_loc): Likewise.
15333 (fold_build3_stat_loc): Likewise.
15334 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15335 (fold_build1_loc): Remove macro.
15336 (fold_build2_loc): Likewise.
15337 (fold_build3_loc): Likewise.
15339 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15341 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15342 (gimple_build_debug_bind_source_stat): Likewise.
15343 * gimple.h (gimple_build_debug_bind): Remove macro.
15344 (gimple_build_debug_bind_source): Likewise.
15346 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15348 * bitmap.c (bitmap_alloc): Adjust.
15349 (bitmap_gc_alloc): Likewise.
15350 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15352 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15354 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15355 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15356 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15357 (bitmap_gc_alloc_stat): Likewise.
15358 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15362 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15363 * rtl.h (shallow_copy_rtx): Remove macro.
15365 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15367 * emit-rtl.c (gen_raw_REG): Adjust.
15368 * gengenrtl.c (gendef): Likewise.
15369 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15370 * rtl.h (rtx_alloc): Remove macro.
15372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15374 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15375 (build_tree_list_stat): Likewise.
15376 * tree.h (build_tree_list): Remove macro.
15377 (build_tree_list_vec): Likewise.
15379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15381 * tree.c (make_vector_stat): Remove _stat from name.
15382 (build_vector_stat): Likewise.
15383 * tree.h (make_vector_stat): Remove macro.
15384 (build_vector_stat): Likewise.
15386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15388 * tree.h (build_var_debug_value): Remove prototype.
15390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15392 * tree.c (tree_cons_stat): Remove _stat from name.
15393 * tree.h (tree_cons): Remove macro.
15395 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15397 * tree.c (build_vl_exp_stat): Remove _stat from name.
15398 * tree.h (build_vl_exp): Remove macro.
15400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15402 * tree.c (build_decl_stat): Remove _stat from name.
15403 * tree.h (build_decl): Remove macro.
15405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15407 * gimple.c (gimple_build_with_ops_stat): Adjust.
15408 (gimple_alloc_stat): Remove _stat from name.
15409 * gimple.h (gimple_alloc): Remove macro.
15411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15413 * tree.c (make_tree_vec_stat): Remove _stat from name.
15414 (grow_tree_vec_stat): Likewise.
15415 * tree.h (make_tree_vec_stat): Adjust prototype.
15416 (grow_tree_vec_stat): Likewise.
15417 (make_tree_vec): Remove macro.
15418 (grow_tree_vec): Likewise.
15420 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15422 * fold-const.c (fold_build1_stat_loc): Adjust.
15423 (fold_build2_stat_loc): Likewise.
15424 (fold_build3_stat_loc): Likewise.
15425 * tree.c (build0_stat): Remove _stat from name.
15426 (build1_stat): Likewise.
15427 (build2_stat): Likewise.
15428 (build3_stat): Likewise.
15429 (build4_stat): Likewise.
15430 (build5_stat): Likewise.
15431 * tree.h (build1_loc): Remove macro, and rename _stat function
15433 (build2_loc): Likewise.
15434 (build3_loc): Likewise.
15435 (build4_loc): Likewise.
15436 (build5_loc): Likewise.
15438 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15440 * tree.c (make_int_cst_stat): Remove _stat from name.
15441 * tree.h (make_int_cst_stat): Adjust prototype.
15442 (make_int_cst): Remove macro.
15444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15446 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15447 * tree.h (make_tree_binfo_stat): Adjust prototype.
15448 (make_tree_binfo): Remove.
15450 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15452 * tree.c (copy_node_stat): Rename to copy_node.
15453 (build_distinct_type_copy): Adjust.
15454 * tree.h (copy_node_stat): Adjust prototype.
15455 (copy_node): Remove macro.
15457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15459 * tree.c (make_node_stat): rename to make_node.
15460 (build_tree_list_stat): Adjust.
15461 (build0_stat): Likewise.
15462 (build2_stat): Likewise.
15463 (build3_stat): Likewise.
15464 (build4_stat): Likewise.
15465 (build5_stat): Likewise.
15466 (build_decl_stat): Likewise.
15467 * tree.h (make_node_stat): Adjust prototype.
15468 (make_node): remove macro.
15470 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15472 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15473 (PPC_FEATURE2_SCV): Likewise.
15474 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15476 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15478 * config/aarch64/aarch64.c
15479 (aarch64_internal_mov_immediate): Add new special pattern.
15480 * config/aarch64/aarch64.md (*movdi_aarch64):
15481 Add reg/32bit const mov case.
15483 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15484 Richard Sandiford <richard.sandiford@linaro.org>
15486 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15487 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15488 Add integer and movi cases.
15489 (movi-split-hf-df-sf split, fp16): New.
15490 (enabled): Added TARGET_FP_F16INST.
15491 * config/aarch64/iterators.md (GPF_HF): New.
15492 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15494 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15496 * config/aarch64/aarch64.c
15497 (aarch64_simd_container_mode): Add prototype.
15498 (aarch64_expand_mov_immediate): Add HI support.
15499 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15500 (aarch64_can_const_movi_rtx_p): New.
15501 (aarch64_preferred_reload_class):
15502 Remove restrictions of using FP registers for certain SIMD operations.
15503 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15504 (aarch64_valid_floating_const): Add integer move validation.
15505 (aarch64_simd_imm_scalar_p): Remove.
15506 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15507 (aarch64_legitimate_constant_p): Expand list of supported cases.
15508 * config/aarch64/aarch64-protos.h
15509 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15510 (aarch64_reinterpret_float_as_int): New.
15511 (aarch64_simd_imm_scalar_p): Remove.
15512 * config/aarch64/constraints.md (Uvi): New.
15513 (Dd): Split into Ds and new Dd.
15514 * config/aarch64/aarch64.md (*movsi_aarch64):
15516 (*movdi_aarch64): Add SIMD mov case.
15518 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15520 * tree-predcom.c: (struct chain): Handle store-store chain in which
15521 stores for elimination only store loop invariant values.
15522 (execute_pred_commoning_chain): Ditto.
15523 (prepare_initializers_chain_store_elim): Ditto.
15524 (prepare_finalizers): Ditto.
15525 (is_inv_store_elimination_chain): New function.
15526 (initialize_root_vars_store_elim_1): New function.
15528 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15530 * tree-predcom.c: Revise general description of the pass.
15531 (enum chain_type): New enum type for store elimination.
15532 (struct chain): New field supporting store elimination.
15533 (struct component): Ditto.
15534 (dump_chain): Dump store-stores chain.
15535 (release_chain): Release resources.
15536 (split_data_refs_to_components): Compute and create component
15537 contains only stores for elimination.
15538 (get_chain_last_ref_at): New function.
15539 (make_invariant_chain): Initialization.
15540 (make_rooted_chain): Specify chain type in parameter and record it.
15541 (add_looparound_copies): Skip for store-stores chain.
15542 (determine_roots_comp): Compute type of chain and pass it to
15544 (initialize_root_vars_store_elim_2): New function.
15545 (finalize_eliminated_stores): New function.
15546 (remove_stmt): Handle store for elimination.
15547 (execute_pred_commoning_chain): Execute predictive commoning on
15548 store-store chains.
15549 (determine_unroll_factor): Skip unroll for store-stores chain.
15550 (prepare_initializers_chain_store_elim): New function.
15551 (prepare_initializers_chain): Hanlde store-store chain.
15552 (prepare_finalizers_chain, prepare_finalizers): New function.
15553 (tree_predictive_commoning_loop): Return integer value indicating
15554 if loop is unrolled or lcssa form is corrupted.
15555 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15557 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15559 * tree-predcom.c (initialize_root): Delete.
15560 (execute_pred_commoning_chain): Initialize root vars and replace
15561 reference of non-combined chain directly, rather than call above
15564 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15566 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15567 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15569 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15571 * tree-predcom.c (struct chain): New field init_seq.
15572 (release_chain): Release init_seq.
15573 (prepare_initializers_chain): Record intialization stmts in above
15575 (insert_init_seqs): New function.
15576 (tree_predictive_commoning_loop): Call insert_init_seqs.
15578 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15580 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15582 2017-07-28 Richard Biener <rguenther@suse.de>
15584 * match.pd: Remove superfluous :c.
15585 * genmatch.c (simplify::id): Add member.
15586 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15588 (current_id): New global.
15589 (dt_node::parent): Move from ...
15590 (dt_operand::parent): ... here. Add for_id member.
15591 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15592 (decision_tree::find_node): Relax order requirement when
15593 merging DT_TRUE nodes to ones inbetween the current simplify
15594 and the one we try to merge with. Add diagnostic whenever
15595 we need to enforce pattern order by not merging.
15596 (decision_tree::insert): Set current_id.
15597 (decision_tree::print_node): Dump parent node and for_id.
15598 (parser::last_id): Add member.
15599 (parser::push_simplify): Assign unique id.
15600 (parser::parser): Initialize last_id.
15602 2017-07-28 Martin Liska <mliska@suse.cz>
15605 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15606 gimple_build_debug_bind.
15608 2017-07-28 Richard Biener <rguenther@suse.de>
15610 PR tree-optimization/81502
15611 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15613 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15615 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15616 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15617 for BIT_FIELD_REF args.
15618 * fold-const.c (make_bit_field_ref): Likewise.
15619 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15621 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15624 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15625 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15626 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15627 Or it into SANITIZER_UNDEFINED.
15628 * ubsan.c: Include gimple-fold.h and varasm.h.
15629 (ubsan_expand_ptr_ifn): New function.
15630 (instrument_pointer_overflow): New function.
15631 (maybe_instrument_pointer_overflow): New function.
15632 (instrument_object_size): Formatting fix.
15633 (pass_ubsan::execute): Call instrument_pointer_overflow
15634 and maybe_instrument_pointer_overflow.
15635 * internal-fn.c (expand_UBSAN_PTR): New function.
15636 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15637 * sanitizer.def (__ubsan_handle_pointer_overflow,
15638 __ubsan_handle_pointer_overflow_abort): New builtins.
15639 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15640 * internal-fn.def (UBSAN_PTR): New internal function.
15641 * opts.c (sanitizer_opts): Add pointer-overflow.
15642 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15643 * fold-const.c (build_range_check): Compute pointer range check in
15644 integral type if pointer arithmetics would be needed. Formatting
15647 2017-07-28 Martin Liska <mliska@suse.cz>
15650 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15651 parameters that are of a variable-length.
15653 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15655 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15657 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15658 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15659 (CRT_CALL_STATIC_FUNCTION): Likewise.
15660 (ASM_DEFAULT_SPEC): New define.
15661 (ASM_SPEC32): Likewise.
15662 (ASM_SPEC64): Likewise.
15663 (ASM_SPEC_COMMON): Likewise.
15664 (ASM_SPEC): Likewise.
15665 (INVALID_64BIT): Likewise.
15666 (LINK_OS_DEFAULT_SPEC): Likewise.
15667 (LINK_OS_SPEC32): Likewise.
15668 (LINK_OS_SPEC64): Likewise.
15669 (POWERPC_LINUX): Likewise.
15670 (PTRDIFF_TYPE): Likewise.
15671 (RESTORE_FP_PREFIX): Likewise.
15672 (RESTORE_FP_SUFFIX): Likewise.
15673 (SAVE_FP_PREFIX): Likewise.
15674 (SAVE_FP_SUFFIX): Likewise.
15675 (SIZE_TYPE): Likewise.
15676 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15677 (TARGET_64BIT): Likewise.
15678 (TARGET_64BIT): Likewise.
15679 (TARGET_AIX): Likewise.
15680 (WCHAR_TYPE_SIZE): Likewise.
15681 (WCHAR_TYPE): Undefine.
15682 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15683 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15684 (CPP_OS_RTEMS_SPEC): Delete.
15685 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15686 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15688 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15690 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15692 PR tree-optimization/81578
15693 * tree-parloops.c (build_new_reduction): Bail out if
15694 reduction_code isn't one of the standard OpenMP reductions.
15695 Move the details printing after that decision.
15697 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15699 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15700 related to reload_in_progress.
15701 (splat_input_operand): Likewise.
15702 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15704 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15706 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15707 (TARGET_INSTANTIATE_DECLS): Likewise.
15708 (legitimate_indexed_address_p): Delete reload_in_progress code.
15709 (rs6000_debug_legitimate_address_p): Likewise.
15710 (rs6000_eliminate_indexed_memrefs): Likewise.
15711 (rs6000_emit_le_vsx_store): Likewise.
15712 (rs6000_emit_move_si_sf_subreg): Likewise.
15713 (rs6000_emit_move): Likewise.
15714 (register_to_reg_type): Likewise.
15715 (rs6000_pre_atomic_barrier): Likewise.
15716 (rs6000_machopic_legitimize_pic_address): Likewise.
15717 (rs6000_allocate_stack_temp): Likewise.
15718 (rs6000_address_for_fpconvert): Likewise.
15719 (rs6000_address_for_altivec): Likewise.
15720 (rs6000_secondary_memory_needed_rtx): Delete function.
15721 (rs6000_check_sdmode): Likewise.
15722 (rs6000_alloc_sdmode_stack_slot): Likewise.
15723 (rs6000_instantiate_decls): Likewise.
15724 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15725 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15726 Delete reload_in_progress.
15727 (*vec_reload_and_plus_<mptrsize>): Likewise.
15728 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15729 (vsx_div_v2di): Likewise.
15730 (vsx_udiv_v2di): Likewise.
15732 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15734 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15735 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15736 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15737 (rs6000_debug_reg_global): Delete print of LRA status.
15738 (rs6000_option_override_internal): Delete dead LRA related code.
15739 (rs6000_lra_p): Delete function.
15740 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15742 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15744 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15745 * config/riscv/rtems.h: New file.
15747 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15748 Sudakshina Das <sudi.das@arm.com>
15750 * config/aarch64/aarch64.md
15751 (define_split for and<mode>3nr_compare): Move
15752 non aarch64_logical_operand to a register.
15753 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15754 register immediate operand to a register.
15755 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15757 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15759 PR middle-end/81564
15760 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15762 2017-07-27 Richard Biener <rguenther@suse.de>
15764 PR tree-optimization/81573
15765 PR tree-optimization/81494
15766 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15767 multi defuse cycle case.
15769 2017-07-27 Richard Biener <rguenther@suse.de>
15771 PR tree-optimization/81571
15772 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15775 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15777 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15778 earlier and only if MASK_FPU is set. Adjust formatting.
15780 2017-07-27 Martin Liska <mliska@suse.cz>
15782 * opt-functions.awk: Add validation of value of Init.
15783 * optc-gen.awk: Pass new argument.
15785 2017-07-27 Martin Liska <mliska@suse.cz>
15787 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15788 Fix wrong condition.
15790 2017-07-27 Martin Liska <mliska@suse.cz>
15792 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15793 BBs and edges seen by autoFDO.
15795 2017-07-27 Richard Biener <rguenther@suse.de>
15797 PR tree-optimization/81502
15798 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15799 with incompatible but same sized type.
15800 (execute_update_addresses_taken): Likewise.
15802 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15804 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15805 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15807 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15810 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15811 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15812 Change s_operand to memory_operand.
15814 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15816 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15817 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15818 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15819 Emit instructions rather than returning an expression. Handle TFmode
15820 and KFmode by casting to TImode.
15821 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15822 (rs6000_emit_le_vsx_store): Likewise.
15823 * config/rs6000/vsx.md (VSX_TI): New iterator.
15824 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15825 (*vsx_le_undo_permute_<mode>): Likewise.
15826 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15827 emit the split sequence.
15828 (*vsx_le_perm_store_<mode>): Likewise.
15830 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15832 PR tree-optimization/81555
15833 PR tree-optimization/81556
15834 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15835 if true, force CHANGED for the recursive invocation.
15836 (reassociate_bb): Remember original length of ops array, pass
15837 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15839 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15840 attributes for noipa attribute. For naked attribute use
15841 lookup_attribute first before lookup_attribute_spec.
15842 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15844 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15845 for functions with noipa attribute.
15846 (cgraph_externally_visible_p): Return true for functions with noipa
15848 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15849 for functions with noipa attribute.
15850 * doc/extend.texi: Document noipa function attribute.
15851 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15852 also for functions with noipa attribute.
15853 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15855 2017-07-26 Andrew Pinski <apinski@cavium.com>
15857 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15858 vec_unalign_load_cost and vec_unalign_store_cost.
15860 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15862 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15863 -mvsx-small-integer option.
15864 (ISA_3_0_MASKS_IEEE): Likewise.
15865 (OTHER_VSX_VECTOR_MASKS): Likewise.
15866 (POWERPC_MASKS): Likewise.
15867 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15868 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15869 code, only testing for DImode being allowed in non-VSX floating
15871 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15872 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15874 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15875 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15876 TARGET_P8_VECTOR test.
15877 (rs6000_secondary_reload_simple_move): Likewise.
15878 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15879 since TARGET_P9_VECTOR was already tested.
15880 (rs6000_opt_masks): Remove -mvsx-small-integer.
15881 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15882 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15884 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15885 test for TARGET_VEXTRACTUB was used, and that uses
15887 (p9 extract splitter): Likewise.
15888 (vsx_extract_<mode>_di_p9): Likewise.
15889 (vsx_extract_<mode>_store_p9): Likewise.
15890 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15891 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15892 the elimination of TARGET_VSX_SMALL_INTEGER.
15893 (vsx_extract_<mode>_p8): Likewise.
15894 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15895 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15896 (vsx_set_<mode>_p9): Likewise.
15897 (vsx_set_v4sf_p9): Likewise.
15898 (vsx_set_v4sf_p9_zero): Likewise.
15899 (vsx_insert_extract_v4sf_p9): Likewise.
15900 (vsx_insert_extract_v4sf_p9_2): Likewise.
15901 * config/rs6000/rs6000.md (sign extend splitter): Change
15902 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15903 (floatsi<mode>2_lfiwax_mem): Likewise.
15904 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15905 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15906 since a test for TARGET_P9_VECTOR was used.
15907 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15908 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15909 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15910 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15911 TARGET_P8_VECTOR test.
15912 (fix_trunc<mode>si2_stfiwx): Likewise.
15913 (fix_trunc<mode>si2_internal): Likewise.
15914 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15915 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15917 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15918 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15919 TARGET_P8_VECTOR test.
15920 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15921 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15922 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15924 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15925 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15926 since a test for TARGET_P9_VECTOR was used.
15927 (splitter for loading small constants): Likewise.
15929 2017-07-26 Andrew Pinski <apinski@cavium.com>
15931 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15934 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15937 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15938 (fp_valid_at): Likewise.
15940 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15942 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15943 (qdf24xx_addrcost_table): Likewise.
15944 (cortexa57_tunings): Update to use generic_branch_cost.
15945 (cortexa72_tunings): Likewise.
15946 (cortexa73_tunings): Likewise.
15947 (qdf24xx_tunings): Likewise.
15949 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15951 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15952 (thunderx2t99_branch_cost): Likewise.
15953 (cortexa35_tunings): Update to use generic_branch_cost.
15954 (cortexa53_tunings): Likewise.
15955 (cortexa57_tunings): Likewise.
15956 (cortexa72_tunings): Likewise.
15957 (cortexa73_tunings): Likewise.
15958 (thunderx2t99_tunings): Likewise.
15960 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15962 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15963 (sparc_option_override): Honour MASK_FSMULD.
15964 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15965 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15966 * config/sparc/sparc.opt (mfsmuld): New option.
15967 * doc/invoke.texi (mfsmuld): Document option.
15969 2017-07-26 Marek Polacek <polacek@redhat.com>
15971 PR middle-end/70992
15972 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15974 2017-07-26 Richard Biener <rguenther@suse.de>
15976 * gimple-match-head.c (do_valueize): Return OP if valueize
15978 (get_def): New helper to get at the def stmt of a SSA name
15979 if valueize allows.
15980 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15981 do_valueize to get at the def stmt.
15982 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15984 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15986 PR middle-end/46932
15987 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15989 2017-07-26 Martin Liska <mliska@suse.cz>
15992 * function.c (expand_function_start): Make expansion of
15993 nonlocal_goto_save_area after parm_birth_insn.
15995 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15997 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15998 from all CPU target flags enable members.
16000 2017-07-26 Richard Biener <rguenther@suse.de>
16002 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16003 (decision_tree::gen): Make 'type' const.
16004 (write_predicate): Likewise.
16006 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16008 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16009 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16010 (rs6000_option_override_internal): Likewise.
16011 (rs6000_expand_vector_set): Likewise.
16012 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16013 (TARGET_UPPER_REGS_SF): Likewise.
16014 (TARGET_UPPER_REGS_DI): Likewise.
16015 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16016 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16017 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16018 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16019 (Splitters for DI constants in Altivec registers): Likewise.
16020 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16021 (vsx_set_v4sf_p9): Likewise.
16022 (vsx_set_v4sf_p9_zero): Likewise.
16023 (vsx_insert_extract_v4sf_p9): Likewise.
16024 (vsx_insert_extract_v4sf_p9_2): Likewise.
16026 2017-07-25 Carl Love <cel@us.ibm.com>
16028 * doc/extend.texi: Update the built-in documentation file for the
16029 existing built-in functions
16030 vector signed char vec_cnttz (vector signed char);
16031 vector unsigned char vec_cnttz (vector unsigned char);
16032 vector signed short vec_cnttz (vector signed short);
16033 vector unsigned short vec_cnttz (vector unsigned short);
16034 vector signed int vec_cnttz (vector signed int);
16035 vector unsigned int vec_cnttz (vector unsigned int);
16036 vector signed long long vec_cnttz (vector signed long long);
16037 vector unsigned long long vec_cnttz (vector unsigned long long);
16039 2017-07-25 Andrew Pinski <apinski@cavium.com>
16041 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16042 accesses where the use is for the first operand of a BIT_INSERT.
16044 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16047 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16048 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16050 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16052 * config/i386/gstabs.h: Delete.
16053 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16055 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16057 * config/i386/i386.c (ix86_decompose_address): Do not check for
16058 register RTX when looking at index_reg or base_reg.
16059 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16061 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16063 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16064 to update EH info here.
16066 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16068 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16070 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16072 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16074 2017-07-25 Torsten Duwe <duwe@suse.de>
16076 * common.opt: Introduce -fpatchable-function-entry
16077 command line option, and its variables function_entry_patch_area_size
16078 and function_entry_patch_area_start.
16079 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16080 including a two-value parser.
16081 * target.def (print_patchable_function_entry): New target hook.
16082 * targhooks.h (default_print_patchable_function_entry): New function.
16083 * targhooks.c (default_print_patchable_function_entry): Likewise.
16084 * toplev.c (process_options): Switch off IPA-RA if
16085 patchable function entries are being generated.
16086 * varasm.c (assemble_start_function): Look at the
16087 patchable-function-entry command line switch and current
16088 function attributes and maybe generate NOP instructions by
16089 calling the print_patchable_function_entry hook.
16090 * doc/extend.texi: Document patchable_function_entry attribute.
16091 * doc/invoke.texi: Document -fpatchable_function_entry
16092 command line option.
16093 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16095 * doc/tm.texi: Re-generate.
16097 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16100 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16101 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16103 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16105 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16107 2017-07-25 Richard Biener <rguenther@suse.de>
16109 PR tree-optimization/81455
16110 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16111 not walk in cycles when looking for guards.
16113 2017-07-25 Richard Biener <rguenther@suse.de>
16115 PR tree-optimization/81529
16116 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16117 when optimizing backedge uses.
16119 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16121 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16123 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16124 to dl_section_ref. On AIX, append an expression to subtract
16125 the size of the section length to dl_section_ref.
16127 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16129 * configure.ac: If any of the config.* scripts fail, exit 1.
16130 * configure: Regenerate.
16132 2017-07-25 Richard Biener <rguenther@suse.de>
16134 PR middle-end/81546
16135 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16136 of immediate uses, be more verbose on errors.
16138 2017-07-25 Richard Biener <rguenther@suse.de>
16140 PR tree-optimization/81510
16141 * tree-vect-loop.c (vect_is_simple_reduction): When the
16142 reduction stmt is not inside the loop bail out.
16144 2017-07-25 Richard Biener <rguenther@suse.de>
16146 PR tree-optimization/81303
16147 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16148 profitability check against LOOP_VINFO_NITERSM1.
16150 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16152 * domwalk.c (cmp_bb_postorder): Simplify.
16153 (sort_bbs_postorder): New function. Use it...
16154 (dom_walker::walk): ...here to optimize common cases.
16156 2017-07-25 Martin Liska <mliska@suse.cz>
16159 * ipa-visibility.c (function_and_variable_visibility): Make the
16160 redirection just on target that supports aliasing.
16161 Fix GNU coding style.
16163 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16166 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16168 * config/arm/rtems.h: Likewise.
16169 * config/bfin/rtems.h: Likewise.
16170 * config/i386/rtemself.h: Likewise.
16171 * config/lm32/rtems.h: Likewise.
16172 * config/m32c/rtems.h: Likewise.
16173 * config/m68k/rtemself.h: Likewise.
16174 * config/microblaze/rtems.h: Likewise.
16175 * config/mips/rtems.h: Likewise.
16176 * config/moxie/rtems.h: Likewise.
16177 * config/nios2/rtems.h: Likewise.
16178 * config/powerpcspe/rtems.h: Likewise.
16179 * config/rs6000/rtems.h: Likewise.
16180 * config/rtems.h: Likewise.
16181 * config/sh/rtems.h: Likewise.
16182 * config/sh/rtemself.h: Likewise.
16183 * config/sparc/rtemself.h: Likewise.
16185 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16188 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16189 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16190 * tree-ssa-structalias.c (alias_get_name): Same.
16192 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16195 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16196 instructions if no du chain is found.
16198 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16200 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16202 2017-07-25 Richard Biener <rguenther@suse.de>
16204 PR middle-end/81505
16205 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16208 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16210 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16211 upper-regs options.
16212 (ISA_2_7_MASKS_SERVER): Likewise.
16213 (ISA_3_0_MASKS_IEEE): Likewise.
16214 (OTHER_P8_VECTOR_MASKS): Likewise.
16215 (OTHER_VSX_VECTOR_MASKS): Likewise.
16216 (POWERPC_MASKS): Likewise.
16217 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16218 duplicate list of options.
16219 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16220 explicit -mupper-regs options.
16221 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16222 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16223 alias for -mupper-regs-df.
16224 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16225 (rs6000_init_hard_regno_mode_ok): Likewise.
16226 (rs6000_option_override_internal): Likewise.
16227 (rs6000_opt_masks): Likewise.
16228 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16229 options in terms of whether -mvsx or -mpower8-vector was used.
16230 (TARGET_UPPER_REGS_DI): Likewise.
16231 (TARGET_UPPER_REGS_SF): Likewise.
16232 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16233 -mupper-regs-* options.
16235 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16237 * passes.c (emergency_dump_function): Print some empty lines and a
16238 header before the RTL dump.
16240 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16242 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16244 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16247 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16248 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16250 2017-07-24 Carl Love <cel@us.ibm.com>
16252 * config/rs6000/rs6000-c.c: Add support for built-in functions
16253 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16254 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16255 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16256 vec_extract_fp_from_shortl): Add defines for the two builtins.
16257 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16258 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16260 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16261 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16262 * doc/extend.texi: Update the built-in documentation file for the
16263 new built-in function.
16265 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16268 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16270 * doc/generic.texi: Likewise.
16271 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16272 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16274 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16276 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16277 (aarch64_mls_elt_merge<mode>): Likewise.
16279 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16281 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16282 having __cxa_atexit.
16284 2017-07-23 Michael Collison <michael.collison@arm.com>
16286 * config/arm/arm.c (arm_option_override): Deprecate
16287 use of -mstructure-size-boundary.
16288 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16289 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16291 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16294 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16295 Reduce cost estimate for direct moves.
16297 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16300 * config/i386/i386.c (ix86_option_override_internal): Disable
16301 BMI, BMI2 and TBM instructions for -m16.
16303 2017-07-21 Carl Love <cel@us.ibm.com>
16305 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16306 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16307 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16308 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16309 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16310 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16311 VMULOSW): New enum "unspec" values.
16312 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16313 altivec_vmulosw): New patterns.
16314 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16315 VMULOSW): Add definitions.
16317 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16319 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16320 (qdf24xx): Likewise.
16321 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16322 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16323 (AARCH64_FL_V8_1): Renumber.
16324 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16325 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16326 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16327 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16328 rdma to feature modifiers list.
16330 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16332 PR middle-end/56727
16333 * ipa-visibility (function_and_variable_visibility): Convert
16334 recursive PLT call to direct call if appropriate.
16336 2017-07-21 Andrew Pinski <apinski@cavium.com>
16338 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16339 operand 1 to see if the types precision matches.
16340 * fold-const.c (operand_equal_p): Likewise.
16342 2017-07-21 Richard Biener <rguenther@suse.de>
16344 PR tree-optimization/81303
16345 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16346 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16347 (vect_peeling_hash_get_lowest_cost): Adjust.
16348 (vect_enhance_data_refs_alignment): Likewise. Use
16349 vect_get_peeling_costs_all_drs to compute the penalty for no
16350 peeling to match up costs.
16352 2017-07-21 Richard Biener <rguenther@suse.de>
16354 PR tree-optimization/81500
16355 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16356 we didn't identify a reduction path.
16358 2017-07-21 Tom de Vries <tom@codesourcery.com>
16359 Cesar Philippidis <cesar@codesourcery.com>
16361 PR gcov-profile/81442
16362 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16365 2017-07-21 Tom de Vries <tom@codesourcery.com>
16368 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16370 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16371 nvptx_override_options_after_change.
16373 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16375 * dwarf2out.c (output_file_names): Avoid double testing for
16376 dwarf_version >= 5.
16378 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16380 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16382 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16384 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16386 (try_crossjump_to_edge): Do not punt on partitioned functions.
16388 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16390 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16391 Put all BBs reachable only via paths crossing cold region to cold
16393 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16395 2016-07-21 Richard Biener <rguenther@suse.de>
16397 PR tree-optimization/81303
16398 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16399 into account prologue and epilogue iterations when raising
16400 min_profitable_iters to sth at least covering one vector iteration.
16402 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16404 * config/arm/arm.c (arm_test_cpu_arch_dat):
16407 2017-07-20 Nathan Sidwell <nathan@acm.org>
16409 Remove TYPE_METHODS.
16410 * tree.h (TYPE_METHODS): Delete.
16411 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16412 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16413 (dbxout_type_methods): Scan TYPE_FIELDS.
16414 (dbxout_type): Don't check TYPE_METHODS here.
16415 * function.c (use_register_for_decl): Always ignore register for
16416 class types when not optimizing.
16417 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16418 * tree.c (free_lang_data_in_type): Stitch out member functions and
16419 templates from TYPE_FIELDS.
16420 (build_distinct_type_copy, verify_type_variant,
16421 verify_type): Member fns are on TYPE_FIELDS.
16422 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16423 * tree-pretty-print.c (dump_generic_node): Likewise.
16425 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16428 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16429 V2TImode and V4TImode.
16430 (ix86_expand_vector_extract): Likewise.
16431 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16432 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16433 (ssescalarmode): Handle V4TImode and V2TImode.
16434 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16435 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16436 (VEXTRACTI128_MODE): New mode iterator.
16437 (splitter for *vec_extractv?ti first element): New.
16438 (VEC_INIT_MODE): New mode iterator.
16439 (vec_init<mode>): Consolidate 3 expanders into one using
16440 VEC_INIT_MODE mode iterator.
16442 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16444 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16445 non_spilled_static_chain_regno_p.
16447 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16449 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16451 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16453 * bb-reorder.c (connect_traces): Allow copying of blocks within
16456 2017-07-20 Richard Biener <rguenther@suse.de>
16458 * gimple.h (gimple_phi_result): Add gphi * overload.
16459 (gimple_phi_result_ptr): Likewise.
16460 (gimple_phi_arg): Likewise. Adjust index assert to only
16461 allow actual argument accesses rather than all slots available
16463 (gimple_phi_arg_def): Add gphi * overload.
16464 * tree-phinodes.c (make_phi_node): Initialize only actual
16466 (resize_phi_node): Clear memory not covered by old node,
16467 do not initialize excess argument slots.
16468 (reserve_phi_args_for_new_edge): Initialize new argument slot
16471 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16473 PR tree-optimization/81388
16475 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16477 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16478 by removing computation of may_be_zero.
16480 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16481 Tom de Vries <tom@codesourcery.com>
16483 PR middle-end/81030
16484 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16485 when gimple level profile disagrees with what RTL expander did.
16487 2017-07-20 Richard Biener <rguenther@suse.de>
16489 PR tree-optimization/61171
16490 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16491 (vect_analyze_stmt): Add slp instance parameter.
16492 (vectorizable_reduction): Likewise.
16493 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16494 (vect_is_simple_reduction): Deal with chains not detected
16495 as SLP reduction chain, specifically not properly associated
16496 chains containing a mix of plus/minus.
16497 (get_reduction_op): Remove.
16498 (get_initial_defs_for_reduction): Simplify, pass in whether
16499 this is a reduction chain, pass in the SLP node for the PHIs.
16500 (vect_create_epilog_for_reduction): Get the SLP instance as
16502 (vectorizable_reduction): Get the SLP instance as arg.
16503 During analysis remember the SLP node with the PHIs in the
16504 instance. Simplify getting at the vectorized reduction PHIs.
16505 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16506 through SLP instance.
16507 (vect_slp_analyze_operations): Likewise.
16508 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16509 (vect_transform_stmt): Likewise.
16511 2017-07-20 Tom de Vries <tom@codesourcery.com>
16513 PR tree-optimization/81489
16514 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16515 read of phi arg location to before loop that modifies phi.
16517 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16519 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16522 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16524 PR middle-end/81331
16525 * except.c (execute): Fix ordering issue.
16527 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16529 PR rtl-optimization/81423
16530 * combine.c (make_compound_operation_int): Don't try to optimize
16531 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16533 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16535 PR rtl-optimization/81423
16536 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16537 with a constant that is -1 in the truncated to mode.
16539 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16541 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16542 (determine_unlikely_bbs): ... here.
16543 * predict.h (propagate_unlikely_bbs_forward): Declare.
16544 * cfgexpand.c (pass_expand::execute): Use it.
16545 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16547 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16548 propagate_unlikely_bbs_forward.
16550 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16552 PR middle-end/81331
16553 * except.c (maybe_add_nop_after_section_switch): New function.
16556 2017-07-19 Tom de Vries <tom@codesourcery.com>
16558 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16560 2017-07-19 Tom de Vries <tom@codesourcery.com>
16562 * gimple.h (gimple_phi_arg): Make assert more strict.
16564 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16566 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16567 * config/rs6000/mmintrin.h: New file.
16568 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16570 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16572 PR tree-optimization/81346
16573 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16575 2017-07-19 Tom de Vries <tom@codesourcery.com>
16577 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16579 2017-07-19 Tom de Vries <tom@codesourcery.com>
16581 * config/nvptx/nvptx-modes.def: Add V2DImode.
16582 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16583 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16584 (nvptx_output_mov_insn): Handle lack of mov.b128.
16585 (nvptx_print_operand): Handle 'H' and 'L' codes.
16586 (nvptx_vector_mode_supported): Allow V2DImode.
16587 (nvptx_preferred_simd_mode): New function.
16588 (nvptx_data_alignment): New function.
16589 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16590 nvptx_preferred_simd_mode.
16591 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16593 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16595 2017-07-19 Tom de Vries <tom@codesourcery.com>
16597 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16598 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16599 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16600 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16601 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16602 (mov<VECIM>_insn): New define_insn.
16603 (define_expand "mov<VECIM>): New define_expand.
16605 2017-07-19 Tom de Vries <tom@codesourcery.com>
16607 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16609 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16611 PR tree-optimization/81346
16612 * fold-const.h (fold_div_compare, range_check_type): Declare.
16613 * fold-const.c (range_check_type): New function.
16614 (build_range_check): Use range_check_type.
16615 (fold_div_compare): No longer static, rewritten into
16616 a match.pd helper function.
16617 (fold_comparison): Don't call fold_div_compare here.
16618 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16619 as helper function.
16621 2017-07-19 Nathan Sidwell <nathan@acm.org>
16623 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16624 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16625 * tree.c (find_decls_types_r, verify_type): Use
16626 TYPE_{MIN,MAX}_VALUE_RAW.
16627 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16628 (hash_tree): Likewise.
16629 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16631 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16634 2017-07-18 Tom de Vries <tom@codesourcery.com>
16636 PR middle-end/81464
16637 * omp-expand.c (expand_omp_for_static_chunk): Handle
16638 equal-argument loop exit phi.
16640 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16643 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16644 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16645 operand 2 predicate.
16646 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16647 operand 2 predicate.
16648 (ror,rol -> rorx splitters): Use const_int_operand as
16649 operand 2 predicate.
16651 2017-06-18 Richard Biener <rguenther@suse.de>
16653 PR tree-optimization/81410
16654 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16655 the gap in the ! slp_perm SLP case after each group.
16657 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16659 PR middle-end/81463
16660 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16663 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16665 PR middle-end/81462
16666 * predict.c (set_even_probabilities): Cleanup; do not affect
16667 probabilities that are already known.
16668 (combine_predictions_for_bb): Call even when count is set.
16670 2017-07-18 Nathan Sidwell <nathan@acm.org>
16672 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16675 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16678 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16679 optimization for loop niter analysis.
16681 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16684 * config/avr/avr.c (avr_optimize_casesi): Don't use
16685 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16687 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16689 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16690 body_cost_vec from _vect_peel_extended_info.
16691 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16692 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16695 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16697 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16699 2017-07-18 Richard Biener <rguenther@suse.de>
16701 PR tree-optimization/80620
16702 PR tree-optimization/81403
16703 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16704 info when re-using a VN table entry.
16706 2017-07-18 Richard Biener <rguenther@suse.de>
16708 PR tree-optimization/81418
16709 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16710 vectype_in. Verify that with lane-reducing reduction operations
16711 we have a single def-use cycle.
16713 2017-07-17 Carl Love <cel@us.ibm.com>
16715 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16717 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16718 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16719 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16721 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16722 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16723 VMULOSW): New enum "unspec" values.
16724 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16725 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16726 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16727 altivec_vmulosw): New patterns.
16728 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16729 VMULOSW): Add definitions.
16730 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16732 * config/alpha/alpha.c: Include predict.h.
16734 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16736 * tree-vrp.c (compare_assert_loc): Fix comparison function
16737 to return predictable results.
16739 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16741 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16743 (subdi3): Likewise.
16744 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16745 * doc/invoke.texi (mexpand-adddi): Update text.
16747 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16749 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16750 that also clobbers the CC register. The old expand code is moved
16752 (*arc_clzsi2): ... here.
16753 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16754 the CC register. The old expand code is moved to ...
16755 (arc_ctzsi2): ... here.
16757 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16759 * config/arc/arc.opt (mindexed-loads): Use initial value
16760 TARGET_INDEXED_LOADS_DEFAULT.
16761 (mauto-modify-reg): Use initial value
16762 TARGET_AUTO_MODIFY_REG_DEFAULT.
16763 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16764 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16765 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16766 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16768 2017-07-17 Martin Liska <mliska@suse.cz>
16771 * opts.c (finish_options): Do not allow -fgnu-tm
16772 w/ -fsanitize={kernel-,}address. Say sorry.
16774 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16777 * tree-loop-distribution.c (classify_partition): Only assert on
16778 numer of iterations.
16779 (merge_dep_scc_partitions): Delete prameter. Update function call.
16780 (distribute_loop): Remove code handling loop with unknown niters.
16781 (pass_loop_distribution::execute): Skip loop with unknown niters.
16783 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16786 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16787 function sort_partitions_by_post_order.
16789 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16791 PR tree-optimization/81374
16792 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16793 the max index of basic blocks, rather than number of basic blocks.
16795 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16797 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16799 (arc_legitimate_pic_operand_p): Likewise.
16800 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16802 (arc_needs_pcl_p): Likewise.
16803 (arc_legitimate_pc_offset_p): Likewise.
16804 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16805 function is also used in constrains.md.
16806 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16807 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16808 PLUS. Only return true/false in known cases, otherwise assert.
16809 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16810 is already called in arc_legitimate_constant_p.
16811 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16813 (LEGITIMATE_PIC_OPERAND_P): Use
16814 arc_raw_symbolic_reference_mentioned_p function.
16815 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16820 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16821 Andrew Burgess <andrew.burgess@embecosm.com>
16823 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16824 (arc_return_address_register): New function.
16825 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16826 (arc_handle_fndecl_attribute): Add naked attribute.
16827 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16828 (TARGET_WARN_FUNC_RETURN): Likewise.
16829 (arc_allocate_stack_slots_for_args): New function.
16830 (arc_warn_func_return): Likewise.
16831 (machine_function): Change type fn_type.
16832 (arc_compute_function_type): Consider new naked function type,
16833 change function return type.
16834 (arc_must_save_register): Adapt to handle new
16835 arc_compute_function_type's return type.
16836 (arc_expand_prologue): Likewise.
16837 (arc_expand_epilogue): Likewise.
16838 (arc_return_address_regs): Delete.
16839 (arc_return_address_register): New function.
16840 (arc_epilogue_uses): Use above function.
16841 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16842 (arc_function_type): Change encoding, add naked type.
16843 (ARC_INTERRUPT_P): Change to handle the new encoding.
16844 (ARC_FAST_INTERRUPT_P): Likewise.
16845 (ARC_NORMAL_P): Define.
16846 (ARC_NAKED_P): Likewise.
16847 (arc_compute_function_type): Delete prototype.
16848 * config/arc/arc.md (in_ret_delay_slot): Use
16849 arc_return_address_register function.
16850 (simple_return): Likewise.
16851 (p_return_i): Likewise.
16853 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16855 PR tree-optimization/81428
16856 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16857 can't be built for those types.
16859 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16861 Remove stuff dead since r239246.
16863 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16864 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16865 (avr_inform_devices): Remove dead stuff.
16867 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16869 * config/arm/arm_neon.h: Fix softp typo.
16871 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16873 PR tree-optimization/81365
16874 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16875 aggregate moves onto bb predecessor edges, make sure there are no
16876 loads that could alias the lhs in between the start of bb and the
16879 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16882 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16883 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16884 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16886 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16888 PR tree-optimization/81396
16889 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16890 (init_symbolic_number): Initialize it to 1.
16891 (perform_symbolic_merge): Add n_ops from both operands into the new
16893 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16894 without base_addr as useless if they need more than one operation.
16895 (bswap_replace): Handle !bswap case for NULL base_addr.
16897 2017-07-17 Tom de Vries <tom@codesourcery.com>
16900 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16903 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16905 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16906 conditional builtin define __FIX_LEON3FT_B2BST.
16908 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16910 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16911 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16914 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16916 PR rtl-optimization/81424
16917 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16918 to remove potential trapping from operands if -fnon-call-exceptions.
16920 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16922 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16923 profile_proability for scalling.
16924 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16926 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16928 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16930 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16932 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16933 fixpoint arithmetics.
16935 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16937 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16938 fixpoint arithmetics.
16940 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16942 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16943 fixpoint arithmetics.
16945 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16947 * profile-count.h (profile_probability::from_reg_br_prob_note,
16948 profile_probability::to_reg_br_prob_note): New functions.
16949 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16950 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16951 * predict.c (probability_reliable_p): Update.
16952 (edge_probability_reliable_p): Update.
16953 (br_prob_note_reliable_p): Update.
16954 (invert_br_probabilities): Update.
16955 (add_reg_br_prob_note): New function.
16956 (combine_predictions_for_insn): Update.
16957 * asan.c (asan_clear_shadow): Update.
16958 * cfgbuild.c (compute_outgoing_frequencies): Update.
16959 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16960 (update_br_prob_note): Update.
16961 (rtl_verify_edges): Update.
16962 (purge_dead_edges): Update.
16963 (fixup_reorder_chain): Update.
16964 * emit-rtl.c (try_split): Update.
16965 * ifcvt.c (cond_exec_process_insns): Update.
16966 (cond_exec_process_if_block): Update.
16967 (dead_or_predicable): Update.
16968 * internal-fn.c (expand_addsub_overflow): Update.
16969 (expand_neg_overflow): Update.
16970 (expand_mul_overflow): Update.
16971 * loop-doloop.c (doloop_modify): Update.
16972 * loop-unroll.c (compare_and_jump_seq): Update.
16973 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16974 * predict.h: Update.
16975 * reorg.c (mostly_true_jump): Update.
16977 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16978 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16979 * config/arc/arc.c: (emit_unlikely_jump): Update.
16980 * config/arm/arm.c: (emit_unlikely_jump): Update.
16981 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16982 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16983 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16984 (ix86_print_operand): Update.
16985 (ix86_split_fp_branch): Update.
16986 (predict_jump): Update.
16987 * config/ia64/ia64.c (ia64_print_operand): Update.
16988 * config/mmix/mmix.c (mmix_print_operand): Update.
16989 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16990 (rs6000_expand_split_stack_prologue): Update.
16991 * config/rs6000/rs6000.c: Update.
16992 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16993 (s390_expand_vec_movstr): Update.
16994 (s390_expand_cs_tdsi): Update.
16995 (s390_expand_split_stack_prologue): Update.
16996 * config/sh/sh.c (sh_print_operand): Update.
16997 (expand_cbranchsi4): Update.
16998 (expand_cbranchdi4): Update.
16999 * config/sparc/sparc.c (output_v9branch): Update.
17000 * config/spu/spu.c (get_branch_target): Update.
17001 (ea_load_store_inline): Update.
17002 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17003 * config/tilepro/tilepro.c: Update.
17005 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17007 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17008 (gimplify_save_expr): Likewise.
17010 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17012 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17014 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17016 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17018 * timevar.def (TV_IPA_FNSUMMARY): Define.
17020 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17022 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17023 to back store errata sensitive sequence from being generated.
17024 (sqrtdf2_fix): Likewise.
17026 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17028 * tree-ssa-threadupdate.c (compute_path_counts,
17029 update_joiner_offpath_counts): Use profile_probability.
17031 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17034 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17036 * config/arm/arm-c.c (arm_cpu_builtins): Define
17037 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17039 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17041 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17042 array entries to represent __ieee128 versions of the
17043 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17044 scalar_extract_sig, and scalar_insert_exp built-in functions.
17045 (altivec_resolve_overloaded_builtin): Add special case handling
17046 for the __builtin_scalar_insert_exp function, as represented by
17047 the P9V_BUILTIN_VEC_VSIEDP constant.
17048 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17049 exponent support for __ieee128 argument.
17050 (VSESQP): Add scalar extract signature support for __ieee128
17052 (VSTDCNQP): Add scalar test negative support for __ieee128
17054 (VSIEQP): Add scalar insert exponent support for __int128 argument
17055 with __ieee128 result.
17056 (VSIEQPF): Add scalar insert exponent support for __ieee128
17057 argument with __ieee128 result.
17058 (VSTDCQP): Add scalar test data class support for __ieee128
17060 (VSTDCNQP): Add overload support for scalar test negative with
17061 __ieee128 argument.
17062 (VSTDCQP): Add overload support for scalar test data class
17063 __ieee128 argument.
17064 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17065 UNSPEC_VSX_SXSIGDP.
17066 (UNSPEC_VSX_SIEXPQP): New constant.
17067 (xsxexpqp): New insn for VSX scalar extract exponent quad
17069 (xsxsigqp): New insn for VSX scalar extract significand quad
17071 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17072 precision with floating point argument.
17073 (xststdcqp): New expand for VSX scalar test data class quad
17075 (xststdcnegqp): New expand for VSX scalar test negative quad
17077 (xststdcqp): New insn to match expansions for VSX scalar test data
17078 class quad precision and VSX scalar test negative quad precision.
17079 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17080 special case operand checking to enforce that second operand of
17081 VSX scalar test data class with quad precision argument is a 7-bit
17083 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17084 prototypes and descriptions of __ieee128 versions of
17085 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17086 scalar_test_data_class, and scalar_test_neg built-in functions.
17088 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17090 PR tree-optimization/81162
17091 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17092 replace a negate with an add.
17094 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17096 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17098 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17100 * config/arm/arm-c.c (arm_cpu_builtins): Define
17101 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17103 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17105 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17106 (armv8-r): Set ARM Cortex-R52 as default CPU.
17107 * config/arm/arm-tables.opt: Regenerate.
17108 * config/arm/arm-tune.md: Regenerate.
17109 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17111 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17112 extension for -mcpu=cortex-r52.
17114 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17116 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17117 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17118 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17119 (fp-armv8): Define it as FP_ARMv8 only.
17120 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17121 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17123 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17124 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17125 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17126 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17127 than TARGET_FPU_ARMV8.
17128 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17129 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17130 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17132 * config/arm/neon.md (neon_vrint): Likewise.
17133 (neon_vcvt): Likewise.
17134 (neon_<fmaxmin_op><mode>): Likewise.
17135 (<fmaxmin><mode>3): Likewise.
17136 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17137 * config/arm/predicates.md (arm_cond_move_operator): Check against
17138 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17140 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17142 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17143 to top of function.
17145 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17147 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17148 loop in comment with memset.
17150 2017-07-14 Martin Liska <mliska@suse.cz>
17152 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17153 * dwarf2out.c (is_java): Remove the function.
17154 (output_pubname): Remove usage of the function.
17155 (lower_bound_default): Remove usage of DW_LANG_Java.
17156 (gen_compile_unit_die): Likewise.
17157 * gcc.c: Remove compiler defaults for .java and .zip files.
17158 * gimple-expr.c (remove_suffix): Change as there's no longer
17159 extension than 4-letter one.
17160 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17161 (gimplify_save_expr): Likewise.
17162 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17163 as it's possible even for other languages than Java.
17164 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17165 * lto-opts.c (lto_write_options): Remove reference to Java.
17166 * opts.c (strip_off_ending): Update file extension handling.
17167 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17168 * tree-eh.c (lower_resx): Likewise.
17169 * tree.c (free_lang_data_in_type): Remove dead code.
17170 (find_decls_types_r): Likewise.
17171 (build_common_builtin_nodes): Remove Java from a comment.
17172 (verify_type): Remove dead code.
17173 * varasm.c (assemble_external): Remove Java from a comment.
17175 2017-07-14 Martin Liska <mliska@suse.cz>
17177 * opts.c (finish_options): Add quotes.
17178 (common_handle_option): Likewise.
17180 2017-07-14 Martin Liska <mliska@suse.cz>
17182 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17183 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17184 Remove N_SO_PASCAL.
17185 * dwarf2out.c (lower_bound_default): Do not handle
17187 (gen_compile_unit_die): Likewise.
17188 * gcc.c: Remove default extension binding for GNU Pascal.
17189 * stmt.c: Remove Pascal language from a comment.
17190 * xcoffout.c: Likewise.
17192 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17195 * diagnostic-show-locus.c (fixit_cmp): New function.
17196 (layout::layout): Sort m_fixit_hints.
17197 (column_range::column_range): Assert that the values are valid.
17198 (struct char_span): New struct.
17199 (correction::overwrite): New method.
17200 (struct source_line): New struct.
17201 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17202 calls in terms of classes source_line and char_span, and
17203 correction::overwrite.
17204 (selftest::test_overlapped_fixit_printing_2): New function.
17205 (selftest::diagnostic_show_locus_c_tests): Call it.
17207 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17209 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17210 early if there is no lhs.
17212 2017-07-13 Martin Liska <mliska@suse.cz>
17214 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17215 (gen_reference_type_die): Likewise.
17216 * stor-layout.c: Remove Pascal-related comment.
17218 2017-07-13 Martin Liska <mliska@suse.cz>
17220 * opts.c (finish_options): Add quotes to error messages.
17221 (parse_sanitizer_options): Likewise.
17223 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17225 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17227 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17229 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17231 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17233 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17235 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17237 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17240 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17241 provides the hardware capability bits, define the macro
17242 __BUILTIN_CPU_SUPPORTS__.
17243 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17244 if GLIBC does not provide the hardware capability bits. Add a
17245 gcc_unreachable call if the built-in cpu function is neither
17246 __builtin_cpu_is nor __builtin_cpu_supports.
17247 (rs6000_get_function_versions_dispatcher): Change the warning
17248 that an old GLIBC is used which does not export the capability
17249 bits to be an error.
17250 * doc/extend.texi (target_clones attribute): Document the
17251 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17252 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17253 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17254 the macros defined by GCC if the newer GLIBC is available.
17256 2017-07-12 Jeff Law <law@redhat.com>
17258 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17259 remaining includes slightly.
17260 * config/riscv/riscv-builtins.c: Include profile-count.h.
17262 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17265 * config/avr/avr.c (avr_set_current_function): In diagnostic
17266 messages: Quote keywords and (parts of) identifiers.
17267 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17270 2017-07-12 Carl Love <cel@us.ibm.com>
17272 * config/rs6000/rs6000-c.c: Add support for built-in functions
17273 vector bool char vec_revb (vector bool char);
17274 vector bool short vec_revb (vector short char);
17275 vector bool int vec_revb (vector bool int);
17276 vector bool long long vec_revb (vector bool long long);
17277 * doc/extend.texi: Update the built-in documentation file for the
17278 new built-in functions.
17280 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17282 * config/s390/s390.md: Remove movcc splitter.
17284 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17286 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17287 load/store on condition.
17289 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17292 * config/avr/avr.c (avr_encode_section_info)
17293 [progmem && !TREE_READONLY]: Error if progmem object needs
17296 2017-07-11 Michael Collison <michael.collison@arm.com>
17298 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17301 2017-07-11 Carl Love <cel@us.ibm.com>
17303 * config/rs6000/rs6000-c.c: Add support for builtins
17304 vector unsigned int vec_parity_lsbb (vector signed int);
17305 vector unsigned int vec_parity_lsbb (vector unsigned int);
17306 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17307 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17308 vector unsigned long long vec_parity_lsbb (vector signed long long);
17309 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17310 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17311 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17312 * doc/extend.texi: Update the built-in documentation file for the
17313 new built-in functions.
17315 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17317 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17318 (layout::m_primary_loc): New field.
17319 (layout::layout): Initialize new field. Move location filtering
17320 logic from here to...
17321 (layout::maybe_add_location_range): ...this new method. Add
17322 support for filtering to just the lines already specified by other
17324 (layout::will_show_line_p): New method.
17325 (gcc_rich_location::add_location_if_nearby): New method.
17326 (selftest::test_add_location_if_nearby): New test function.
17327 (selftest::diagnostic_show_locus_c_tests): Call it.
17328 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17331 2017-07-11 Tom de Vries <tom@codesourcery.com>
17333 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17334 (bb_first_real_insn): New function.
17335 (nvptx_single): Add extra initialization of broadcasted condition
17338 2017-07-11 Nathan Sidwell <nathan@acm.org>
17340 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17342 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17344 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17345 Binutils doc as TEXI will mess them up.
17346 * doc/invoke.texi (AVR Options): Same here.
17348 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17350 * config/sparc/sparc.opt (mfix-ut700): New option.
17351 (mfix-gr712rc): Likewise.
17352 (sparc_fix_b2bst): New variable.
17353 * doc/invoke.texi (SPARC options): Document them.
17354 (ARM options): Fix warnings.
17355 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17356 instructions to prevent sequences that can trigger the store-store
17357 errata for certain LEON3FT processors.
17358 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17359 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17360 * config/sparc/sparc.md (fix_b2bst): New attribute.
17361 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17363 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17366 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17368 (*rsqrtsf2_sse): Ditto.
17370 (div<mode>3): Macroize insn from divdf3 and divsf3
17371 using MODEF mode iterator.
17373 2017-07-10 Martin Sebor <msebor@redhat.com>
17375 PR tree-optimization/80397
17376 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17377 instead of testing for equality to INTEGER_TYPE.
17379 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17381 * config.gcc: Remove uclibc from arc target spec.
17383 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17385 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17387 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17390 * lto-wrapper.c (remove_option): New function.
17391 (merge_and_complain): Merge PIC/PIE options more realistically.
17393 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17395 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17399 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17400 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17401 * config.in: Regenerate.
17402 * configure: Regenerate.
17403 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17404 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17405 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17406 (TARGET_GASISR_PROLOGUES): ...target mask.
17407 * common/config/avr/avr-common.c
17408 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17409 Set -mgas-isr-prologues.
17410 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17411 INSERT_PASS_BEFORE for it.
17412 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17413 * config/avr/avr.c (avr_option_override)
17414 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17415 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17416 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17417 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17418 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17420 (make_avr_pass_pre_proep): New function.
17421 (emit_push_sfr) <treg>: Add argument to function and use it
17422 instead of TMP_REG.
17423 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17424 and set machine->gasisr.yes.
17425 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17426 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17427 __gcc_isr.n_pushed to .L__stack_usage.
17428 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17429 (avr_asm_final_postscan_insn): ...this new static function.
17430 * config/avr/avr.h (machine_function)
17431 <is_no_gccisr, use_L__stack_usage>: New fields.
17432 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17433 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17434 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17435 (gasisr, *gasisr): New expander and insn.
17436 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17437 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17438 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17440 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17442 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17445 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17447 Move jump-tables out of .text again.
17450 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17451 (ASM_OUTPUT_ADDR_VEC): New function.
17452 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17453 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17454 INSN_ADDRESSes as asm comment.
17455 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17456 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17457 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17458 * config/avr/avr.md (*tablejump): Adjust comment.
17459 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17460 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17462 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17463 (avr_output_addr_vec): New proto.
17464 (avr_log_t) <insn_addresses>: New field.
17466 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17469 * config/i386/i386.c (ix86_function_arg_advance): Set
17470 outgoing_args_on_stack to true if there are outgoing arguments
17472 (ix86_function_arg): Likewise.
17473 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17474 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17475 * config/i386/i386.h (machine_function): Add
17476 outgoing_args_on_stack.
17478 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17480 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17481 supporting pthreds.
17482 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17484 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17486 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17487 (REAL_H): Remove $(MACHMODE_H).
17488 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17490 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17491 $(MACHMODE_H) and double-int.h.
17492 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17494 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17497 2017-07-07 Andrew Pinski <apinski@cavium.com>
17499 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17500 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17502 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17504 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17505 Add warning if GCC was not configured to link against a GLIBC that
17506 exports the hardware capability bits.
17507 (make_resolver_func): Make resolver function private and not a
17508 COMDAT function. Create the name with clone_function_name instead
17509 of make_unique_name.
17512 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17513 correct operand in doing the split.
17515 2017-07-07 Carl Love <cel@us.ibm.com>
17517 * config/rs6000/rs6000-c: Add support for built-in function
17518 vector unsigned short vec_pack_to_short_fp32 (vector float,
17520 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17521 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17522 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17523 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17524 (convert_4f32_8i16): Add define_expand.
17525 * doc/extend.texi: Update the built-in documentation file for the
17526 new built-in function.
17528 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17530 * config/sparc/m8.md: New file.
17531 * config/sparc/sparc.md: Include m8.md.
17533 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17535 * config/sparc/sparc.opt: New option -mvis4b.
17536 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17537 (sparc_option_override): Handle VIS4B.
17538 (enum sparc_builtins): Define
17539 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17540 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17541 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17542 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17543 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17544 (check_constant_argument): New function.
17545 (sparc_vis_init_builtins): Define builtins
17546 __builtin_vis_dictunpack{8,16,32},
17547 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17548 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17549 __builtin_vis_fpcmpde{8,16,32}shl and
17550 __builtin_vis_fpcmpur{8,16,32}shl.
17551 (sparc_expand_builtin): Check that the constant operands to
17552 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17553 constant and in range.
17554 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17556 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17557 (SPARC_IMM5_P): Likewise.
17558 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17559 (enabled): Handle vis4b.
17560 (UNSPEC_DICTUNPACK): New unspec.
17561 (UNSPEC_FPCMPSHL): Likewise.
17562 (UNSPEC_FPUCMPSHL): Likewise.
17563 (UNSPEC_FPCMPDESHL): Likewise.
17564 (UNSPEC_FPCMPURSHL): Likewise.
17565 (cpu_feature): New CPU feature `vis4b'.
17566 (dictunpack{8,16,32}): New insns.
17567 (FPCSMODE): New mode iterator.
17568 (fpcscond): New code iterator.
17569 (fpcsucond): Likewise.
17570 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17571 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17572 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17573 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17574 * config/sparc/constraints.md: Define constraints `q' for unsigned
17575 2-bit integer constants and `t' for unsigned 5-bit integer
17577 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17579 (imm5_operand_dictunpack16): Likewise.
17580 (imm5_operand_dictunpack32): Likewise.
17581 (imm2_operand): Likewise.
17582 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17583 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17584 ditunpack* and fpcmp*shl builtins.
17586 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17588 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17589 * config.in: Add HAVE_AS_SPARC6 define.
17590 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17592 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17594 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17595 (CPP_CPU_SPEC): Handle m8.
17596 (ASM_CPU_SPEC): Likewise.
17597 * config/sparc/sparc-opts.h (enum processor_type): Add
17599 * config/sparc/sparc.c (m8_costs): New struct.
17600 (sparc_option_override): Handle TARGET_CPU_m8.
17601 (sparc32_initialize_trampoline): Likewise.
17602 (sparc64_initialize_trampoline): Likewise.
17603 (sparc_issue_rate): Likewise.
17604 (sparc_register_move_cost): Likewise.
17605 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17606 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17607 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17608 (CPP_CPU_SPEC): Handle M8.
17609 (ASM_CPU_SPEC): Likewise.
17610 (AS_M8_FLAG): Define.
17611 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17612 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17613 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17615 * configure: Regenerate.
17616 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17619 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17621 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17623 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17624 ("*movdi_insn_sp32"): Do not set v3pipe.
17625 ("*movsi_insn"): Likewise.
17626 ("*movdi_insn_sp64"): Likewise.
17627 ("*movsf_insn"): Likewise.
17628 ("*movdf_insn_sp32"): Likewise.
17629 ("*movdf_insn_sp64"): Likewise.
17630 ("*zero_extendsidi2_insn_sp64"): Likewise.
17631 ("*sign_extendsidi2_insn"): Likewise.
17632 ("*mov<VM32:mode>_insn"): Likewise.
17633 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17634 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17635 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17636 ("<vlop:code><VL:mode>3"): Likewise.
17637 ("*not_<vlop:code><VL:mode>3"): Likewise.
17638 ("*nand<VL:mode>_vis"): Likewise.
17639 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17640 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17641 ("one_cmpl<VL:mode>2"): Likewise.
17642 ("faligndata<VM64:mode>_vis"): Likewise.
17643 ("alignaddrsi_vis"): Likewise.
17644 ("alignaddrdi_vis"): Likweise.
17645 ("alignaddrlsi_vis"): Likewise.
17646 ("alignaddrldi_vis"): Likewise.
17647 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17648 ("bmaskdi_vis"): Likewise.
17649 ("bmasksi_vis"): Likewise.
17650 ("bshuffle<VM64:mode>_vis"): Likewise.
17651 ("cmask8<P:mode>_vis"): Likewise.
17652 ("cmask16<P:mode>_vis"): Likewise.
17653 ("cmask32<P:mode>_vis"): Likewise.
17654 ("pdistn<P:mode>_vis"): Likewise.
17655 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17657 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17659 * config/sparc/sparc.md ("subtype"): New insn attribute.
17660 ("*wrgsr_sp64"): Set insn subtype.
17661 ("*rdgsr_sp64"): Likewise.
17662 ("alignaddrsi_vis"): Likewise.
17663 ("alignaddrdi_vis"): Likewise.
17664 ("alignaddrlsi_vis"): Likewise.
17665 ("alignaddrldi_vis"): Likewise.
17666 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17667 ("fexpand_vis"): Likewise.
17668 ("fpmerge_vis"): Likewise.
17669 ("faligndata<VM64:mode>_vis"): Likewise.
17670 ("bshuffle<VM64:mode>_vis"): Likewise.
17671 ("cmask8<P:mode>_vis"): Likewise.
17672 ("cmask16<P:mode>_vis"): Likewise.
17673 ("cmask32<P:mode>_vis"): Likewise.
17674 ("fchksm16_vis"): Likewise.
17675 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17676 ("fmean16_vis"): Likewise.
17677 ("fp<plusminus_insn>64_vis"): Likewise.
17678 ("<plusminus_insn>v8qi3"): Likewise.
17679 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17680 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17681 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17682 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17683 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17684 ("*movqi_insn"): Likewise.
17685 ("*movhi_insn"): Likewise.
17686 ("*movsi_insn"): Likewise.
17687 ("movsi_pic_gotdata_op"): Likewise.
17688 ("*movdi_insn_sp32"): Likewise.
17689 ("*movdi_insn_sp64"): Likewise.
17690 ("movdi_pic_gotdata_op"): Likewise.
17691 ("*movsf_insn"): Likewise.
17692 ("*movdf_insn_sp32"): Likewise.
17693 ("*movdf_insn_sp64"): Likewise.
17694 ("*zero_extendhisi2_insn"): Likewise.
17695 ("*zero_extendqihi2_insn"): Likewise.
17696 ("*zero_extendqisi2_insn"): Likewise.
17697 ("*zero_extendqidi2_insn"): Likewise.
17698 ("*zero_extendhidi2_insn"): Likewise.
17699 ("*zero_extendsidi2_insn_sp64"): Likewise.
17700 ("ldfsr"): Likewise.
17701 ("prefetch_64"): Likewise.
17702 ("prefetch_32"): Likewise.
17703 ("tie_ld32"): Likewise.
17704 ("tie_ld64"): Likewise.
17705 ("*tldo_ldub_sp32"): Likewise.
17706 ("*tldo_ldub1_sp32"): Likewise.
17707 ("*tldo_ldub2_sp32"): Likewise.
17708 ("*tldo_ldub_sp64"): Likewise.
17709 ("*tldo_ldub1_sp64"): Likewise.
17710 ("*tldo_ldub2_sp64"): Likewise.
17711 ("*tldo_ldub3_sp64"): Likewise.
17712 ("*tldo_lduh_sp32"): Likewise.
17713 ("*tldo_lduh1_sp32"): Likewise.
17714 ("*tldo_lduh_sp64"): Likewise.
17715 ("*tldo_lduh1_sp64"): Likewise.
17716 ("*tldo_lduh2_sp64"): Likewise.
17717 ("*tldo_lduw_sp32"): Likewise.
17718 ("*tldo_lduw_sp64"): Likewise.
17719 ("*tldo_lduw1_sp64"): Likewise.
17720 ("*tldo_ldx_sp64"): Likewise.
17721 ("*mov<VM32:mode>_insn"): Likewise.
17722 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17723 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17725 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17727 * config/sparc/sparc.md ("type"): New insn type viscmp.
17728 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17730 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17731 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17732 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17733 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17735 ("n7_vis_logical_11cycle"): Likewise.
17736 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17737 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17738 * config/sparc/niagara.md ("niag_vis"): Likewise.
17739 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17740 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17742 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17744 * config/sparc/sparc.md: New instruction type `bmask'.
17745 (bmaskdi_vis): Use the `bmask' type.
17746 (bmasksi_vis): Likewise.
17747 * config/sparc/ultra3.md (us3_array): Likewise.
17748 * config/sparc/niagara7.md (n7_array): Likewise.
17749 * config/sparc/niagara4.md (n4_array): Likewise.
17750 * config/sparc/niagara2.md (niag2_vis): Likewise.
17751 (niag3_vis): Likewise.
17752 * config/sparc/niagara.md (niag_vis): Likewise.
17754 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17756 * ipa-comdats.c: Remove optimize check from gate.
17757 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17758 for functions not optimized.
17759 (ipa_fn_summary_read): Skip optimize check.
17760 (ipa_fn_summary_write): Likewise.
17761 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17763 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17765 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17766 (check_callers): Give up on caller which is not optimized.
17767 (inline_small_functions): Likewise.
17768 (ipa_inline): Do not give up when not optimizing.
17769 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17770 away unoptimizes cdtors.
17771 (whole_program_function_and_variable_visibility): Do
17772 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17773 * ipa.c (process_references): Do not check optimize.
17774 (symbol_table::remove_unreachable_nodes): Update optimize check.
17775 (set_writeonly_bit): Update optimize check.
17776 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17777 (pass_ipa_single_use::gate): Remove.
17779 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17781 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17782 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17783 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17784 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17785 permute_load, permute_store, adjust_extract, adjust_splat,
17786 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17787 replace_swap_with_copy, dump_swap_insn_table,
17788 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17789 recombine_lvx_pattern, recombine_stvx_pattern,
17790 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17791 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17792 to file rs6000-p8swap.c.
17793 * config/rs6000/rs6000-p8swap.c: New file.
17794 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17795 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17796 and rs6000*-*-* targets.
17798 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17800 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17802 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17804 * lto-wrapper.c (merge_and_complain): Do not merge
17805 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17806 fsigned_zeros, ftrapping_math, fwrapv.
17807 (append_compiler_options): Do not track these options.
17808 (append_linker_options): Likewie
17810 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17812 * cgraphunit.c (cgraph_node::finalize_function): When
17813 !flag_toplevel_reorde set no_reorder flag.
17814 (varpool_node::finalize_decl): Likewise.
17815 (symbol_table::compile): Drop no toplevel reorder path.
17817 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17819 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17820 edges; zero probability is not better than uninitialized.
17822 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17824 * asan.h (asan_sanitize_allocas_p): Declare.
17825 * asan.c (asan_sanitize_allocas_p): New function.
17826 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17827 (handle_builtin_alloca): Likewise.
17828 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17829 if !asan_sanitize_allocas_p.
17830 * params.def (asan-instrument-allocas): Add new option.
17831 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17832 * opts.c (common_handle_option): Disable allocas sanitization for
17835 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17837 * asan.c: Include gimple-fold.h.
17838 (get_last_alloca_addr): New function.
17839 (handle_builtin_stackrestore): Likewise.
17840 (handle_builtin_alloca): Likewise.
17841 (asan_emit_allocas_unpoison): Likewise.
17842 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17843 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17844 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17845 (instrument_builtin_call): Pass gimple iterator to
17846 get_mem_refs_of_builtin_call.
17847 (last_alloca_addr): New global.
17848 * asan.h (asan_emit_allocas_unpoison): Declare.
17849 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17850 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17851 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17852 if function calls alloca.
17853 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17854 * gimple-fold.h (replace_call_with_value): Declare.
17855 * internal-fn.c: Include asan.h.
17856 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17857 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17859 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17861 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17862 (C_SELFTEST_FLAGS): New.
17863 (CPP_SELFTEST_FLAGS): New.
17864 (SELFTEST_DEPS): New, from deps of s-selftest.
17865 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17866 (CPP_SELFTEST_DEPS): New.
17867 (selftest): Add dependency on s-selftest-c++.
17868 (s-selftest): Rename to...
17869 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17870 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17871 than SELFTEST_FLAGS.
17872 (selftest-gdb): Rename to...
17873 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17875 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17876 (selftest-valgrind): Rename to...
17877 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17879 (selftest-valgrind): Reintroduce as an alias for
17880 selftest-c-valgrind.
17881 (s-selftest-c++): New.
17882 (selftest-c++-gdb): New.
17883 (selftest-c++-valgrind): New.
17885 2017-07-06 Olivier Hainque <hainque@adacore.com>
17887 * gcc.c (process_command): When deciding if undefined variables
17888 should be ignored when processing specs, accept "gcc -v" as well.
17890 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17892 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17893 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17895 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17897 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17898 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17899 * config/arm/arm-tables.opt: Regenerate.
17900 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17902 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17904 2017-07-06 Carl Love <cel@us.ibm.com>
17906 * ChangeLog: Clean up from mid air collision
17908 2017-07-06 Carl Love <cel@us.ibm.com>
17910 * config/rs6000/rs6000-c.c: Add support for built-in functions
17911 vector signed int vec_subc (vector signed int, vector signed int);
17912 vector signed __int128 vec_subc (vector signed __int128,
17913 vector signed __int128);
17914 vector unsigned __int128 vec_subc (vector unsigned __int128,
17915 vector unsigned __int128);
17916 vector signed int vec_sube (vector signed int, vector signed int,
17917 vector signed int);
17918 vector unsigned int vec_sube (vector unsigned int,
17919 vector unsigned int,
17920 vector unsigned int);
17921 vector signed __int128 vec_sube (vector signed __int128,
17922 vector signed __int128,
17923 vector signed__int128);
17924 vector unsigned __int128 vec_sube (vector unsigned __int128,
17925 vector unsigned __int128,
17926 vector unsigned __int128);
17927 vector signed int vec_subec (vector signed int, vector signed int,
17928 vector signed int);
17929 vector unsigned int vec_subec (vector unsigned int,
17930 vector unsigned int,
17931 vector unsigned int);
17932 vector signed __int128 vec_subec (vector signed __int128,
17933 vector signed __int128,
17934 vector signed__int128);
17935 vector unsigned __int128 vec_subec (vector unsigned __int128,
17936 vector unsigned __int128,
17937 vector unsigned __int128);
17938 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17939 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17940 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17941 BU_ALTIVEC_OVERLOAD_X definitions.
17942 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17943 * doc/extend.texi: Update the built-in documentation file for the new
17944 built-in functions.
17946 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17949 * diagnostic-show-locus.c (layout::layout): Use start and finish
17950 spelling location for the start and finish of each range.
17951 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17952 Add unused aspect param.
17953 * input.c (expand_location_1): Add "aspect" param, and use it
17954 to access the correct part of the location.
17955 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17957 (expand_location_to_spelling_point): Likewise.
17958 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17959 param, and pass it to expand_location_1.
17961 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17963 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17964 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17965 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17966 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17967 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17968 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17969 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17970 _mm_maskz_getmant_ss): New intrinsics.
17971 (__builtin_ia32_getexpss128_mask): Changed to ...
17972 __builtin_ia32_getexpss128_round ... this.
17973 (__builtin_ia32_getexpsd128_mask): Changed to ...
17974 __builtin_ia32_getexpsd128_round ... this.
17975 * config/i386/i386-builtin-types.def
17976 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17977 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17978 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17979 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17980 __builtin_ia32_getmantss_mask_round): New builtins.
17981 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17982 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17983 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17984 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17985 * config/i386/sse.md
17986 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17987 avx512f_sgetexp<mode><mask_scalar_name>
17988 <round_saeonly_scalar_name> ... this.
17989 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17990 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17991 vgetexp<ssescalarmodesuffix>
17992 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17993 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17994 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17995 avx512f_vgetmant<mode><mask_scalar_name>
17996 <round_saeonly_scalar_name> ... this.
17997 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17998 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17999 vgetmant<ssescalarmodesuffix>
18000 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18001 %0<mask_scalar_operand4>, %1, %2
18002 <round_saeonly_scalar_mask_op4>, %3} ... this.
18003 * config/i386/subst.md (mask_scalar_operand4,
18004 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18005 round_saeonly_scalar_nimm_predicate): New subst attributes.
18007 2017-07-06 Julia Koval <julia.koval@intel.com>
18009 * config/i386/i386.c (ix86_erase_embedded_rounding):
18010 Remove code for old rounding pattern.
18012 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18014 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18016 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18018 * doc/sourcebuild.texi (Test Directives, Variants of
18019 dg-require-support): Add documentation for dg-require-stack-check.
18021 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18023 * config/i386/subst.md (mask_scalar, round_scalar,
18024 round_saeonly_scalar): New meta-templates.
18025 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18026 round_scalar_mask_operand3, round_scalar_mask_op3,
18027 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18028 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18029 round_saeonly_scalar_constraint,
18030 round_saeonly_scalar_prefix): New subst attribute.
18031 * config/i386/sse.md
18032 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18033 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18034 <round_scalar_name> ... this.
18035 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18036 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18037 <round_scalar_name> ... this.
18038 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18039 <sse>_vm<code><mode>3<mask_scalar_name>
18040 <round_saeonly_scalar_name> ... this.
18041 (v<plusminus_mnemonic><ssescalarmodesuffix>
18042 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18043 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18044 v<plusminus_mnemonic><ssescalarmodesuffix>
18045 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18046 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18047 (v<multdiv_mnemonic><ssescalarmodesuffix>
18048 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18049 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18050 v<multdiv_mnemonic><ssescalarmodesuffix>
18051 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18052 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18053 (v<maxmin_float><ssescalarmodesuffix>
18054 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18055 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18056 v<maxmin_float><ssescalarmodesuffix>
18057 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18058 %0<mask_scalar_operand3>, %1, %<iptr>2
18059 <round_saeonly_scalar_mask_op3>} ... this.
18061 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18063 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18064 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18066 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18067 Alan Hayward <alan.hayward@arm.com>
18068 David Sherwood <david.sherwood@arm.com>
18070 * combine.c (simplify_if_then_else): Remove "enum" before
18072 * compare-elim.c (can_eliminate_compare): Likewise.
18073 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18075 (aarch64_lookup_simd_builtin_type): Likewise.
18076 (aarch64_simd_builtin_type): Likewise.
18077 (aarch64_init_simd_builtin_types): Likewise.
18078 (aarch64_simd_expand_args): Likewise.
18079 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18081 (aarch64_reverse_mask): Likewise.
18082 (aarch64_simd_emit_reg_reg_move): Likewise.
18083 (aarch64_gen_adjusted_ldpstp): Likewise.
18084 (aarch64_ccmp_mode_to_code): Likewise.
18085 (aarch64_operands_ok_for_ldpstp): Likewise.
18086 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18087 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18089 (aarch64_min_divisions_for_recip_mul): Likewise.
18090 (aarch64_reassociation_width): Likewise.
18091 (aarch64_get_condition_code_1): Likewise.
18092 (aarch64_simd_emit_reg_reg_move): Likewise.
18093 (aarch64_simd_attr_length_rglist): Likewise.
18094 (aarch64_reverse_mask): Likewise.
18095 (aarch64_operands_ok_for_ldpstp): Likewise.
18096 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18097 (aarch64_gen_adjusted_ldpstp): Likewise.
18098 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18100 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18101 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18102 (arm_lookup_simd_builtin_type): Likewise.
18103 (arm_simd_builtin_type): Likewise.
18104 (arm_init_simd_builtin_types): Likewise.
18105 (arm_expand_builtin_args): Likewise.
18106 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18107 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18108 (ft32_setup_incoming_varargs): Likewise.
18109 (ft32_function_arg): Likewise.
18110 (ft32_function_arg_advance): Likewise.
18111 (ft32_pass_by_reference): Likewise.
18112 (ft32_arg_partial_bytes): Likewise.
18113 (ft32_valid_pointer_mode): Likewise.
18114 (ft32_addr_space_pointer_mode): Likewise.
18115 (ft32_addr_space_legitimate_address_p): Likewise.
18116 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18118 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18119 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18120 (iamcu_alignment): Likewise.
18121 (canonicalize_vector_int_perm): Likewise.
18122 (ix86_noce_conversion_profitable_p): Likewise.
18123 (ix86_mpx_bound_mode): Likewise.
18124 (ix86_operands_ok_for_move_multiple): Likewise.
18125 * config/microblaze/microblaze-protos.h
18126 (microblaze_expand_conditional_branch_reg): Likewise.
18127 * config/microblaze/microblaze.c
18128 (microblaze_expand_conditional_branch_reg): Likewise.
18129 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18131 (rs6000_reassociation_width): Likewise.
18132 (rs6000_invalid_binary_op): Likewise.
18133 (fusion_p9_p): Likewise.
18134 (emit_fusion_p9_load): Likewise.
18135 (emit_fusion_p9_store): Likewise.
18136 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18138 (riscv_hard_regno_mode_ok_p): Likewise.
18139 (riscv_address_insns): Likewise.
18140 (riscv_split_symbol): Likewise.
18141 (riscv_legitimize_move): Likewise.
18142 (riscv_function_value): Likewise.
18143 (riscv_hard_regno_nregs): Likewise.
18144 (riscv_expand_builtin): Likewise.
18145 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18146 (riscv_build_integer): Likewise.
18147 (riscv_split_integer): Likewise.
18148 (riscv_legitimate_constant_p): Likewise.
18149 (riscv_cannot_force_const_mem): Likewise.
18150 (riscv_regno_mode_ok_for_base_p): Likewise.
18151 (riscv_valid_base_register_p): Likewise.
18152 (riscv_valid_offset_p): Likewise.
18153 (riscv_valid_lo_sum_p): Likewise.
18154 (riscv_classify_address): Likewise.
18155 (riscv_legitimate_address_p): Likewise.
18156 (riscv_address_insns): Likewise.
18157 (riscv_load_store_insns): Likewise.
18158 (riscv_force_binary): Likewise.
18159 (riscv_split_symbol): Likewise.
18160 (riscv_force_address): Likewise.
18161 (riscv_legitimize_address): Likewise.
18162 (riscv_move_integer): Likewise.
18163 (riscv_legitimize_const_move): Likewise.
18164 (riscv_legitimize_move): Likewise.
18165 (riscv_address_cost): Likewise.
18166 (riscv_subword): Likewise.
18167 (riscv_output_move): Likewise.
18168 (riscv_canonicalize_int_order_test): Likewise.
18169 (riscv_emit_int_order_test): Likewise.
18170 (riscv_function_arg_boundary): Likewise.
18171 (riscv_pass_mode_in_fpr_p): Likewise.
18172 (riscv_pass_fpr_single): Likewise.
18173 (riscv_pass_fpr_pair): Likewise.
18174 (riscv_get_arg_info): Likewise.
18175 (riscv_function_arg): Likewise.
18176 (riscv_function_arg_advance): Likewise.
18177 (riscv_arg_partial_bytes): Likewise.
18178 (riscv_function_value): Likewise.
18179 (riscv_pass_by_reference): Likewise.
18180 (riscv_setup_incoming_varargs): Likewise.
18181 (riscv_print_operand): Likewise.
18182 (riscv_elf_select_rtx_section): Likewise.
18183 (riscv_save_restore_reg): Likewise.
18184 (riscv_for_each_saved_reg): Likewise.
18185 (riscv_register_move_cost): Likewise.
18186 (riscv_hard_regno_mode_ok_p): Likewise.
18187 (riscv_hard_regno_nregs): Likewise.
18188 (riscv_class_max_nregs): Likewise.
18189 (riscv_memory_move_cost): Likewise.
18190 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18191 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18192 (rl78_addr_space_address_mode): Likewise.
18193 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18195 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18196 (rs6000_reassociation_width): Likewise.
18197 (rs6000_invalid_binary_op): Likewise.
18198 (fusion_p9_p): Likewise.
18199 (emit_fusion_p9_load): Likewise.
18200 (emit_fusion_p9_store): Likewise.
18201 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18202 (ok_for_simple_move_operands): Likewise.
18203 (ok_for_simple_move_strict_operands): Likewise.
18204 (ok_for_simple_arith_logic_operands): Likewise.
18205 (visium_legitimize_reload_address): Likewise.
18206 (visium_select_cc_mode): Likewise.
18207 (output_cbranch): Likewise.
18208 (visium_split_double_move): Likewise.
18209 (visium_expand_copysign): Likewise.
18210 (visium_expand_int_cstore): Likewise.
18211 (visium_expand_fp_cstore): Likewise.
18212 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18213 (visium_function_arg): Likewise.
18214 (visium_function_arg_advance): Likewise.
18215 (visium_libcall_value): Likewise.
18216 (visium_setup_incoming_varargs): Likewise.
18217 (visium_legitimate_constant_p): Likewise.
18218 (visium_legitimate_address_p): Likewise.
18219 (visium_legitimize_address): Likewise.
18220 (visium_secondary_reload): Likewise.
18221 (visium_register_move_cost): Likewise.
18222 (visium_memory_move_cost): Likewise.
18223 (prepare_move_operands): Likewise.
18224 (ok_for_simple_move_operands): Likewise.
18225 (ok_for_simple_move_strict_operands): Likewise.
18226 (ok_for_simple_arith_logic_operands): Likewise.
18227 (visium_function_value_1): Likewise.
18228 (rtx_ok_for_offset_p): Likewise.
18229 (visium_legitimize_reload_address): Likewise.
18230 (visium_split_double_move): Likewise.
18231 (visium_expand_copysign): Likewise.
18232 (visium_expand_int_cstore): Likewise.
18233 (visium_expand_fp_cstore): Likewise.
18234 (visium_split_cstore): Likewise.
18235 (visium_select_cc_mode): Likewise.
18236 (visium_split_cbranch): Likewise.
18237 (output_cbranch): Likewise.
18238 (visium_print_operand_address): Likewise.
18239 * expmed.c (flip_storage_order): Likewise.
18240 * expmed.h (emit_cstore): Likewise.
18241 (flip_storage_order): Likewise.
18242 * genrecog.c (validate_pattern): Likewise.
18243 * hsa-gen.c (gen_hsa_addr): Likewise.
18244 * internal-fn.c (expand_arith_overflow): Likewise.
18245 * ira-color.c (allocno_copy_cost_saving): Likewise.
18246 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18247 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18248 (process_invariant_for_inheritance): Likewise.
18249 * lra-eliminations.c (move_plus_up): Likewise.
18250 * omp-low.c (lower_oacc_reductions): Likewise.
18251 * simplify-rtx.c (simplify_subreg): Likewise.
18252 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18253 (TARGET_CHKP_BOUND_MODE): Likewise..
18254 * targhooks.c (default_chkp_bound_mode): Likewise.
18255 (default_setup_incoming_vararg_bounds): Likewise.
18256 * targhooks.h (default_chkp_bound_mode): Likewise.
18257 (default_setup_incoming_vararg_bounds): Likewise.
18258 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18259 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18260 (have_whole_vector_shift): Likewise.
18261 * tree-vect-stmts.c (vectorizable_load): Likewise.
18262 * doc/tm.texi: Regenerate.
18264 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18266 Graceful degrade if Binutils PR21472 is not available.
18269 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18270 .rodata in flash test fails.
18271 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18272 * confgure: Regenerate.
18273 * config.in: Regenerate.
18274 * config/avr/avr.c (avr_asm_named_section)
18275 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18276 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18277 (avr_asm_init_sections): Same.
18279 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18281 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18282 (fma<VH:mode>4_intrinsic): Likewise.
18283 (*fmsub<VCVTF:mode>4): Likewise.
18284 (*fmsub<VH:mode>4_intrinsic): Likewise.
18286 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18289 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18290 Don't depend on "optimize > 0".
18291 (out_movhi_r_mr, out_movqi_mr_r): Same.
18292 (out_movhi_mr_r, out_movqi_r_mr): Same.
18293 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18294 io_address_operand on "optimize > 0".
18296 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18298 * tree-loop-distribution.c: Add general explanantion on the pass.
18299 (generate_loops_for_partition): Mark distributed loop.
18300 (pg_add_dependence_edges): New parameter. Handle alias data
18301 dependence specially and record it in the parameter if asked.
18302 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18303 (init_partition_graph_vertices, add_partition_graph_edge): New.
18304 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18305 (free_partition_graph_vdata, build_partition_graph): New.
18306 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18307 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18308 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18309 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18310 (version_for_distribution_p, finalize_partitions): New.
18311 (distribute_loop): Handle alias data dependence specially. Factor
18312 out loop fusion code as functions and call these functions.
18314 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18316 * tree-loop-distribution.c (classify_partition): New parameter and
18317 better handle reduction statement.
18318 (rdg_build_partitions): Revise comment.
18319 (distribute_loop): Compute statements in all partitions and pass it
18320 to classify_partition.
18322 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18324 * tree-loop-distribution.c (enum partition_type): New.
18325 (struct partition): New field type.
18326 (partition_merge_into): Add parameter. Update partition type.
18327 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18328 (build_rdg_partition_for_vertex): Compute partition type.
18329 (rdg_build_partitions): Dump partition type.
18330 (distribute_loop): Update calls to partition_merge_into.
18332 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18334 * tree-loop-distribution.c (struct ddr_hasher): New.
18335 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18337 (classify_partition): Call get_data_dependence.
18338 (pg_add_dependence_edges): Ditto.
18339 (distribute_loop): Release data dependence hash table.
18341 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18343 * tree-loop-distribution.c (ref_base_address): Delete.
18344 (similar_memory_accesses): Rename ...
18345 (share_memory_accesses): ... to this. Check if partitions access
18346 the same memory reference.
18347 (distribute_loop): Call share_memory_accesses.
18349 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18351 * tree-loop-distribution.c (struct partition): New field recording
18352 its data reference.
18353 (partition_alloc, partition_free): Init and release data refs.
18354 (partition_merge_into): Merge data refs.
18355 (build_rdg_partition_for_vertex): Collect data refs for partition.
18356 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18358 (distribute_loop): Remve data refs from vertice data of partition
18361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18363 * tree-loop-distribution.c (params.h): Include header file.
18364 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18365 (datarefs_vec): New global var.
18366 (create_rdg_vertices): Use datarefs_vec directly.
18367 (free_rdg): Don't free data references.
18368 (build_rdg): Update use. Don't free data references.
18369 (distribute_loop): Compute global variable for data references.
18370 Bail out if there are too many data references.
18372 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18374 * tree-loop-distribution.c (loop_nest): New global var.
18375 (build_rdg): Use loop directly, rather than loop nest.
18376 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18378 (distribute_loop): Compute global variable loop nest. Update use.
18380 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18382 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18383 (partition_merge_into): New parameter. Dump reason for fusion.
18384 (distribute_loop): Update use of partition_merge_into.
18386 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18388 * tree-loop-distribution.c (bb_top_order_index): New.
18389 (bb_top_order_index_size, bb_top_order_cmp): New.
18390 (stmts_from_loop): Use topological order.
18391 (pass_loop_distribution::execute): Compute and release topological
18392 order for basic blocks.
18394 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18396 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18399 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18401 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18402 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18403 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18404 * internal-fn.def (LOOP_DIST_ALIAS): New.
18405 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18406 (fold_loop_internal_call): ... this.
18407 (vect_loop_dist_alias_call): New function.
18408 (set_uid_loop_bbs): Call fold_loop_internal_call.
18409 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18412 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18415 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18416 Require dead FLAGS_REG at the beginning of a peephole.
18418 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18421 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18422 arguments in the call to __builtin_ia32_sbb_u32.
18423 (_subborrow_u64): Swap _X and _Y arguments in the call to
18424 __builtin_ia32_sbb_u64.
18426 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18429 * tree-vrp.c (compare_assert_loc): Turn into a function template
18430 with stable template parameter. Only test if a->e is NULL,
18431 !a->e == !b->e has been verified already. Use e == NULL or
18432 e != NULL instead of e or ! e tests. If stable is true, don't use
18433 iterative_hash_expr, on the other side allow a or b or both NULL
18434 and sort the NULLs last.
18435 (process_assert_insertions): Sort using compare_assert_loc<false>
18436 instead of compare_assert_loc, later sort using
18437 compare_assert_loc<true> before calling process_assert_insertions_for
18438 in a loop. Use break instead of continue once seen NULL pointer.
18440 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18442 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18443 Cortex-R7 and Cortex-R8 processors.
18445 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18447 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18448 uninitialized while src is not.
18450 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18452 * common/config/arm/arm-common.c: Adjust include path for
18454 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18455 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18456 (arm-cpu-data.h): Likewise.
18457 (arm-cpu-cdata.h): Likewise.
18458 * config/arm/arm-cpu.h: Delete.
18459 * config/arm/arm-cpu-cdata.h: Delete.
18460 * config/arm/arm-cpu-data.h: Delete.
18462 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18464 * config/arm/arm-cpus.in (cortex-a55): New.
18465 (cortex-a75): Likewise.
18466 (cortex-a75.cortex-a55): Likewise.
18467 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18469 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18470 * config/arm/arm-cpu-cdata.h: Regenerate.
18471 * config/arm/arm-cpu-data.h: Regenerate.
18472 * config/arm/arm-cpu.h: Regenerate.
18473 * config/arm/arm-tables.opt: Regenerate.
18474 * config/arm/arm-tune.md: Regenerate.
18476 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18478 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18480 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18482 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18485 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18487 PR tree-optimization/81292
18488 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18489 full_string_p, also call adjust_related_strinfos if the adjustment
18490 is simple, otherwise invalidate related strinfos.
18492 2017-07-04 Martin Liska <mliska@suse.cz>
18495 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18496 newly created variable as DECL_IGNORED_P.
18498 2017-07-04 Martin Liska <mliska@suse.cz>
18501 * ipa-inline.c (inline_small_functions):
18502 Use xstrdup_for_dump.
18504 2017-07-04 Tom de Vries <tom@codesourcery.com>
18506 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18508 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18511 * config/darwin.c (darwin_function_switched_text_sections):
18514 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18516 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18518 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18520 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18522 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18524 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18525 min_profitable_iters, and th as inclusive lower bounds.
18526 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18527 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18528 for min_profitable_iters and min_profitable_estimate.
18529 (vect_transform_loop): Treat th as an inclusive lower bound.
18530 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18532 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18535 * config/darwin.c (darwin_function_switched_text_sections):
18536 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18537 in two pieces, and suppress the use of buf.
18539 2017-07-03 Nathan Sidwell <nathan@acm.org>
18541 * hash-table.h (hash_table_mod1): Fix indentation.
18543 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18545 PR middle-end/81290
18546 * predict.c (force_edge_cold): Be more careful about propagation
18548 * profile-count.h (profile_probability::guessed,
18549 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18551 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18553 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18555 * doc/invoke.texi (rcpc architecture extension): Document it.
18557 2017-07-03 Richard Biener <rguenther@suse.de>
18559 PR tree-optimization/60510
18560 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18561 the scalar reduction PHI and use it.
18562 (vectorizable_reduction): Properly guard the single_defuse_cycle
18563 path for non-SLP reduction chains where we cannot use it.
18564 Rework reduc_def/index and vector type deduction. Rework
18565 vector operand gathering during reduction op code-gen.
18566 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18567 chains dissolve the chain and leave it to non-SLP reduction
18570 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18572 * tree-data-ref.h (dr_alignment): Declare.
18573 * tree-data-ref.c (dr_alignment): New function.
18574 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18575 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18577 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18579 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18581 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18582 and base_misalignment fields.
18583 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18584 * tree-data-ref.c: Include builtins.h.
18585 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18586 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18587 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18588 * tree-vect-data-refs.c: Include tree-cfg.h.
18589 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18590 fields instead of calculating an alignment here.
18591 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18592 innermost_loop_behavior fields.
18594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18596 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18598 (DR_STEP_ALIGNMENT): New macro.
18599 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18600 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18601 (create_data_ref): Print it.
18602 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18603 to tell whether the step preserves vector (mis)alignment.
18604 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18605 Move the check for an integer step and generalise to all INTEGER_CST.
18606 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18607 Print the outer step alignment.
18609 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18611 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18612 with offset_alignment.
18613 (DR_ALIGNED_TO): Delete.
18614 (DR_OFFSET_ALIGNMENT): New macro.
18615 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18616 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18617 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18618 (create_data_ref): Likewise.
18619 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18620 (vect_analyze_data_refs): Likewise.
18621 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18622 creating dummy innermost behavior.
18624 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18626 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18627 with a "innermost_loop_behavior *" and refeence tree.
18628 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18629 (create_data_ref): Update call accordingly.
18630 * tree-predcom.c (find_looparound_phi): Likewise.
18632 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18634 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18635 fields with dr_wrt_vec_loop.
18636 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18637 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18638 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18639 (vect_dr_behavior): New function.
18640 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18641 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18642 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18643 track whether the step preserves the misalignment.
18644 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18645 Use vect_dr_behavior.
18646 (vect_setup_realignment): Update call accordingly.
18647 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18648 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18649 call to vect_create_addr_base_for_vector_ref.
18650 (vect_create_cond_for_align_checks): Likewise.
18651 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18652 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18653 (vect_recog_mask_conversion_pattern): Likewise.
18654 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18655 (new_stmt_vec_info): Remove redundant zeroing.
18657 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18659 * common/config/arm/arm-common.c (arm_be8_option): New function.
18660 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18661 (ISA_ARMv6): Add isa_bit_be8.
18662 * config/arm/arm.h (arm_be8_option): Add prototype.
18663 (BE8_SPEC_FUNCTION): New define.
18664 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18665 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18666 (mlittle-endian): Similarly.
18667 (mbe8, mbe32): New options.
18668 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18669 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18671 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18673 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18675 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18677 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18678 (cleanup_tree_cfg_bb): Use it.
18679 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18681 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18683 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18686 * loop-doloop.c (add_test): Update profile.
18688 2017-07-03 Martin Liska <mliska@suse.cz>
18691 * sanopt.c (rewrite_usage_of_param): New function.
18692 (sanitize_rewrite_addressable_params): Likewise.
18693 (pass_sanopt::execute): Call rewrite_usage_of_param.
18695 2017-07-03 Richard Biener <rguenther@suse.de>
18697 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18698 back to using VIEW_CONVERT_EXPR.
18700 2017-07-03 Martin Liska <mliska@suse.cz>
18703 * doc/extend.texi: Document when a resolver function is
18704 generated for target_clones.
18706 2017-07-03 Martin Liska <mliska@suse.cz>
18708 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18709 and shadow memory of auto variables which are subject of
18710 use-after-scope sanitization.
18711 (asan_expand_mark_ifn): Add do set only when is_poison.
18713 2016-07-03 Richard Biener <rguenther@suse.de>
18715 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18717 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18718 (vectorizable_reduction): Perform reduction PHI creation when
18719 visiting a reduction PHI and adjust and simplify code generation
18720 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18721 (vect_transform_loop): Visit reduction PHIs.
18722 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18723 defs into the SLP tree.
18724 (vect_build_slp_tree): Reduction defs terminate the recursion.
18725 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18727 (vect_get_vec_defs_for_stmt_copy): Export.
18728 (vect_get_vec_defs): Likewise.
18729 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18731 (vect_get_vec_defs_for_stmt_copy): Declare.
18732 (vect_get_vec_defs): Likewise.
18734 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18736 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18737 parameter with a "loop" parameter and use it instead of the
18738 loop containing DR_STMT. Don't check simple_iv when doing
18739 BB analysis. Describe the two analysis modes in the comment.
18741 2017-07-03 Tom de Vries <tom@codesourcery.com>
18743 PR tree-optimization/69468
18744 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18745 (find_same_succ_bb): Handle ignore_edge_flags.
18747 2017-07-03 Tom de Vries <tom@codesourcery.com>
18749 PR tree-optimization/81192
18750 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18752 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18754 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18756 2017-07-03 Tom de Vries <tom@codesourcery.com>
18758 PR tree-optimization/81192
18759 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18760 BB_SAME_SUCC (bb) == NULL.
18762 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18764 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18767 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18769 * dumpfile.c: Include profile-count.h
18770 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18772 (insert_cond_bb): Update profile.
18773 * tree-cfg.h (insert_cond_bb): Update prototype.
18774 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18775 * tree-dump.c: Do not include tree-cfg.
18777 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18779 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18781 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18783 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18788 * tree-complex.c (expand_complex_div_wide): update profile.
18790 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18791 Alan Hayward <alan.hayward@arm.com>
18792 David Sherwood <david.sherwood@arm.com>
18794 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18795 (CORETYPES_H): New define.
18796 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18797 (insn-modes-inline.h, s-modes-inline-h): New rules.
18798 (generated_files): Add insn-modes-inline.h.
18799 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18800 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18801 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18802 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18803 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18804 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18805 (build/gencodes.o, build/genconditions.o): Likewise.
18806 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18807 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18808 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18809 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18810 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18811 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18812 * coretypes.h: Include everything up to real.h for generators.
18813 Include insn-modes.h first. Include wide-int-print.h after
18814 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18815 * machmode.h: Don't include insn-modes.h here.
18816 * function-tests.c: Remove includes of signop.h, machmode.h,
18817 double-int.h and wide-int.h.
18819 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18821 * optc-save-gen.awk: Likewise.
18822 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18823 * godump.c: Remove include of wide-int-print.h.
18824 * pretty-print.h: Likewise.
18825 * wide-int-print.cc: Likewise.
18826 * wide-int.cc: Likewise.
18827 * hash-map-tests.c: Remove include of signop.h.
18828 * hash-set-tests.c: Likewise.
18829 * rtl-tests.c: Likewise.
18830 * mkconfig.sh: Remove include of machmode.h.
18831 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18833 (emit_insn_modes_inline_h): ...this new function. Emit the code
18834 into an insn-modes-inline.h header file, adding appropriate
18835 include guards and end comments.
18836 (emit_insn_modes_c_header): Remove include of machmode.h.
18837 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18839 (main): Handle -i flag and call emit_insn_modes_inline_h when
18842 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18844 * tree-ssa-strlen.c (strinfo): Rename the length field to
18845 nonzero_chars. Add a full_string_p field.
18846 (compare_nonzero_chars, zero_length_string_p): New functions.
18847 (get_addr_stridx): Add an offset_out parameter.
18848 Use compare_nonzero_chars.
18849 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18850 (new_strinfo): Update after above changes to strinfo.
18851 (set_endptr_and_length): Set full_string_p.
18852 (get_string_length): Update after above changes to strinfo.
18853 (unshare_strinfo): Update call to new_strinfo.
18854 (maybe_invalidate): Likewise.
18855 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18856 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18857 as a uhwi instead of an shwi. Update after above changes to
18858 strinfo and new_strinfo.
18859 (zero_length_string): Assert that chainsi contains full strings.
18860 Use zero_length_string_p. Update call to new_strinfo.
18861 (adjust_related_strinfos): Update after above changes to strinfo.
18862 Copy full_string_p from origsi.
18863 (adjust_last_stmt): Use zero_length_string_p.
18864 (handle_builtin_strlen): Update after above changes to strinfo and
18865 new_strinfo. Install the lhs as the string length if the previous
18866 entry didn't describe a full string.
18867 (handle_builtin_strchr): Update after above changes to strinfo
18869 (handle_builtin_strcpy): Likewise.
18870 (handle_builtin_strcat): Likewise.
18871 (handle_builtin_malloc): Likewise.
18872 (handle_pointer_plus): Likewise.
18873 (handle_builtin_memcpy): Likewise. Track nonzero characters
18874 that aren't necessarily followed by a nul terminator.
18875 (handle_char_store): Likewise.
18877 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18879 PR tree-optimization/80769
18880 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18881 for malloc and calloc. Document the new invariant that all related
18882 strinfos have delayed lengths or none do.
18883 (verify_related_strinfos): Move earlier in file.
18884 (set_endptr_and_length): New function, split out from...
18885 (get_string_length): ...here. Also set the lengths of related
18887 (zero_length_string): Assert that chainsi has known (rather than
18889 (adjust_related_strinfos): Likewise.
18891 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18893 PR tree-optimization/81136
18894 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18895 assert that two references with the same misalignment have the same
18896 compile-time misalignment if those compile-time misalignments
18899 2017-07-01 Andi Kleen <ak@linux.intel.com>
18901 * print-tree.c (print_node): Print all attributes.
18903 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18905 * cfg.c (scale_bbs_frequencies): New function.
18906 * cfg.h (scale_bbs_frequencies): Declare it.
18907 * cfgloopanal.c (single_likely_exit): Cleanup.
18908 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18910 (scale_loop_profile): Likewise.
18911 (loop_version): Likewise.
18912 (create_empty_loop_on_edge): Update.
18913 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18914 scale_loop_frequencies, scale_loop_profile, loopify,
18915 loop_version): Update prototypes.
18916 * modulo-sched.c (sms_schedule): Update.
18917 * predict.c (unlikely_executed_edge_p): Also check probability.
18918 (probably_never_executed_edge_p): Fix typo.
18919 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18920 * tree-parloops.c (gen_parallel_loop): Update.
18921 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18922 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18923 * tree-ssa-loop-split.c (split_loop): Update.
18924 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18925 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18926 (vect_loop_versioning): Update.
18927 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18929 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18931 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18933 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18935 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18936 to keep profile consistent.
18938 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18940 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18941 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18942 * profile-count.h (max_safe_multiplier): Make unsigned.
18943 (profile_count::guessed_zero): New.
18945 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18947 * bb-reorder.c (fix_up_crossing_landing_pad,
18948 fix_crossing_conditional_branches): Use make_single_succ_edge
18949 to keep profile consistent.
18951 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18953 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18956 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18959 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18960 the right scopes, make sure cond_jump isn't preserved between multiple
18961 iterations. Search for fallthru edge whenever there are 3+ edges and
18962 use find_fallthru_edge for it.
18964 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18966 Patch by Alexander Monakov <amonakov@ispras.ru>
18967 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18968 probabilities consistently.
18970 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18972 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18973 emit_cmp_and_jump_insns.
18975 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18978 * tree-inline.c (expand_call_inline): Combine profile statuses.
18980 2017-06-30 Andrew Pinski <apinski@cavium.com>
18982 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18983 fold_stmt returned true.
18985 2017-06-30 Nathan Sidwell <nathan@acm.org>
18987 * ggc.h (empty_string): Delete.
18988 * cfgexpand.c (expand_asm_stmt): Use plain "".
18989 * optabs.c (expand_asm_memory_barrier): Likewise.
18990 * stringpool.c (empty_string): Delete.
18991 (digit_vector, digit_string): Delete.
18992 (ggc_alloc_string): Use plain "", don't optimize single digit
18993 strings. Use ggc_alloc_atomic.
18995 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18997 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18998 comparison set and one other set, use the cost of the non-comparison
19001 2017-06-30 Nathan Sidwell <nathan@acm.org>
19003 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19006 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19008 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19009 loops. Remove now unneeded calls to gimple_switch_set_label() that
19010 just set removed labels to NULL_TREE.
19012 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19014 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19015 (set_range_info): ...here. Only call set_range_info_raw if domain
19017 (set_nonzero_bits): Call set_range_info_raw.
19018 * tree-ssanames.h (set_range_info_raw): New.
19020 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19023 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19024 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19025 of nonimmediate_operand and <store_mask_constraint> instead of m
19026 for the input operand. For V8FI iterator, always split if input
19027 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19028 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19029 <store_mask_predicate> instead of register_operand and
19030 <store_mask_constraint> instead of v for the input operand. Make
19031 sure both operands aren't MEMs for if not <mask_applied>.
19033 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19035 * lto-wrapper.c (copy_file) Close both file descriptors before
19038 2017-06-30 Martin Liska <mliska@suse.cz>
19041 * multiple_target.c (create_dispatcher_calls): Make ifunc
19042 also for function that don't have calls or are not referenced.
19044 2017-06-30 Richard Biener <rguenther@suse.de>
19046 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19047 analyze the first scalar stmt. Move vector type computation
19048 for the BB case here from ...
19049 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19050 live operation processing in the SLP case properly.
19052 2017-06-30 Richard Biener <rguenther@suse.de>
19054 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19056 2017-06-30 Martin Liska <mliska@suse.cz>
19059 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19060 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19062 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19064 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19067 2017-06-30 Richard Biener <rguenther@suse.de>
19069 PR tree-optimization/81249
19070 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19071 condition reduction result to original scalar type.
19073 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19075 * profile-count.h (enum profile_quality): Fix typos and whitespace
19078 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19080 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19081 type for branch probabilities.
19083 2017-06-29 Julian Brown <julian@codesourcery.com>
19084 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19086 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19087 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19088 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19089 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19091 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19093 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19094 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19095 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19096 CC usage from generic code to here.
19097 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19098 CC usage into the target macros.
19100 2017-06-29 Maya Rashish <coypu@sdf.org>
19102 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19105 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19107 * arm/arm-builtins.c: Include profile-count.h
19108 * except.c (sjlj_emit_function_enter): Use
19109 profile_probability::unlikely.
19111 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19113 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19114 and tocrel_offset be pointer args rather than implicitly using
19116 (legitimate_constant_pool_address_p, rs6000_emit_move,
19117 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19118 tocrel_offset and use in toc_relative_expr_p call.
19119 (print_operand, print_operand_address): Use static tocrel_base_oac
19120 and tocrel_offset_oac.
19121 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19124 2017-06-29 Maya Rashish <coypu@sdf.org>
19126 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19128 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19130 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19131 objects, take into account only the alignment of 'op0' and 'mode1' if
19134 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19136 * ccmp.c (ccmp_tree_comparison_p): New function.
19137 (ccmp_candidate_p): Update to use above function.
19138 (get_compare_parts): New function.
19139 (expand_ccmp_next): Update to use new functions.
19140 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19142 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19143 take mode as argument.
19144 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19145 * expr.c (expand_expr_real_1): Pass mode as argument.
19147 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19149 * combine.c (combine_instructions): Print insns to dump_file, together
19152 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19154 * asan.c (asan_emit_stack_protection): Update.
19155 (create_cond_insert_point): Update.
19156 * auto-profile.c (afdo_propagate_circuit): Update.
19157 * basic-block.h (struct edge_def): Turn probability to
19158 profile_probability.
19159 (EDGE_FREQUENCY): Update.
19160 * bb-reorder.c (find_traces_1_round): Update.
19161 (better_edge_p): Update.
19162 (sanitize_hot_paths): Update.
19163 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19164 (make_single_succ_edge): Update.
19165 (check_bb_profile): Update.
19166 (dump_edge_info): Update.
19167 (update_bb_profile_for_threading): Update.
19168 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19169 probabilitycount to 0.
19170 * cfgbuild.c (compute_outgoing_frequencies): Update.
19171 * cfgcleanup.c (try_forward_edges): Update.
19172 (outgoing_edges_match): Update.
19173 (try_crossjump_to_edge): Update.
19174 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19175 (expand_gimple_tailcall): Update.
19176 (construct_init_block): Use make_single_succ_edge.
19177 (construct_exit_block): Use make_single_succ_edge.
19178 * cfghooks.c (verify_flow_info): Update.
19179 (redirect_edge_succ_nodup): Update.
19180 (split_edge): Update.
19181 (account_profile_record): Update.
19182 * cfgloopanal.c (single_likely_exit): Update.
19183 * cfgloopmanip.c (scale_loop_profile): Update.
19184 (set_zero_probability): Remove.
19185 (duplicate_loop_to_header_edge): Update.
19186 * cfgloopmanip.h (loop_version): Update prototype.
19187 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19188 (force_nonfallthru_and_redirect): Update.
19189 (update_br_prob_note): Update.
19190 (rtl_verify_edges): Update.
19191 (purge_dead_edges): Update.
19192 (rtl_lv_add_condition_to_bb): Update.
19193 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19194 * cgraphunit.c (init_lowered_empty_function): Update.
19195 (cgraph_node::expand_thunk): Update.
19196 * cilk-common.c: Include profile-count.h
19197 * dojump.c (inv): Remove.
19198 (jumpifnot): Update.
19199 (jumpifnot_1): Update.
19200 (do_jump_1): Update.
19202 (do_jump_by_parts_greater_rtx): Update.
19203 (do_compare_rtx_and_jump): Update.
19204 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19205 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19206 * dwarf2cfi.c: Include profile-count.h
19207 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19208 (sjlj_emit_dispatch_table): Likewise.
19209 * explow.c: Include profile-count.h
19210 * expmed.c (emit_store_flag_force): Update.
19211 (do_cmp_and_jump): Update.
19212 * expr.c (compare_by_pieces_d::generate): Update.
19213 (compare_by_pieces_d::finish_mode): Update.
19214 (emit_block_move_via_loop): Update.
19215 (store_expr_with_bounds): Update.
19216 (store_constructor): Update.
19217 (expand_expr_real_2): Update.
19218 (expand_expr_real_1): Update.
19219 * expr.h (try_casesi, try_tablejump): Update prototypes.
19220 * gimple-pretty-print.c (dump_probability): Update.
19221 (dump_profile): New.
19222 (dump_gimple_label): Update.
19223 (dump_gimple_bb_header): Update.
19224 * graph.c (draw_cfg_node_succ_edges): Update.
19225 * hsa-gen.c (convert_switch_statements): Update.
19226 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19227 (find_if_case_1): Update.
19228 (find_if_case_2): Update.
19229 * internal-fn.c (expand_arith_overflow_result_store): Update.
19230 (expand_addsub_overflow): Update.
19231 (expand_neg_overflow): Update.
19232 (expand_mul_overflow): Update.
19233 (expand_vector_ubsan_overflow): Update.
19234 * ipa-cp.c (good_cloning_opportunity_p): Update.
19235 * ipa-split.c (split_function): Use make_single_succ_edge.
19236 * ipa-utils.c (ipa_merge_profiles): Update.
19237 * loop-doloop.c (add_test): Update.
19238 (doloop_modify): Update.
19239 * loop-unroll.c (compare_and_jump_seq): Update.
19240 (unroll_loop_runtime_iterations): Update.
19241 * lra-constraints.c (lra_inheritance): Update.
19242 * lto-streamer-in.c (input_cfg): Update.
19243 * lto-streamer-out.c (output_cfg): Update.
19244 * mcf.c (adjust_cfg_counts): Update.
19245 * modulo-sched.c (sms_schedule): Update.
19246 * omp-expand.c (expand_omp_for_init_counts): Update.
19247 (extract_omp_for_update_vars): Update.
19248 (expand_omp_ordered_sink): Update.
19249 (expand_omp_for_ordered_loops): Update.
19250 (expand_omp_for_generic): Update.
19251 (expand_omp_for_static_nochunk): Update.
19252 (expand_omp_for_static_chunk): Update.
19253 (expand_cilk_for): Update.
19254 (expand_omp_simd): Update.
19255 (expand_omp_taskloop_for_outer): Update.
19256 (expand_omp_taskloop_for_inner): Update.
19257 * omp-simd-clone.c (simd_clone_adjust): Update.
19258 * optabs.c (expand_doubleword_shift): Update.
19259 (expand_abs): Update.
19260 (emit_cmp_and_jump_insn_1): Update.
19261 (expand_compare_and_swap_loop): Update.
19262 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19263 * predict.c (predictable_edge_p): Update.
19264 (edge_probability_reliable_p): Update.
19265 (set_even_probabilities): Update.
19266 (combine_predictions_for_insn): Update.
19267 (combine_predictions_for_bb): Update.
19268 (propagate_freq): Update.
19269 (estimate_bb_frequencies): Update.
19270 (force_edge_cold): Update.
19271 * profile-count.c (profile_count::dump): Add missing space into dump.
19272 (profile_count::debug): Add newline.
19273 (profile_count::differs_from_p): Explicitly convert to unsigned.
19274 (profile_count::stream_in): Update.
19275 (profile_probability::dump): New member function.
19276 (profile_probability::debug): New member function.
19277 (profile_probability::differs_from_p): New member function.
19278 (profile_probability::differs_lot_from_p): New member function.
19279 (profile_probability::stream_in): New member function.
19280 (profile_probability::stream_out): New member function.
19281 * profile-count.h (profile_count_quality): Rename to ...
19282 (profile_quality): ... this one.
19283 (profile_probability): New.
19284 (profile_count): Update.
19285 * profile.c (compute_branch_probabilities): Update.
19286 * recog.c (peep2_attempt): Update.
19287 * sched-ebb.c (schedule_ebbs): Update.
19288 * sched-rgn.c (find_single_block_region): Update.
19289 (compute_dom_prob_ps): Update.
19290 (schedule_region): Update.
19291 * sel-sched-ir.c (compute_succs_info): Update.
19292 * stmt.c (struct case_node): Update.
19293 (do_jump_if_equal): Update.
19294 (get_outgoing_edge_probs): Update.
19295 (conditional_probability): Update.
19296 (emit_case_dispatch_table): Update.
19297 (expand_case): Update.
19298 (expand_sjlj_dispatch_table): Update.
19299 (emit_case_nodes): Update.
19300 * targhooks.c: Update.
19301 * tracer.c (better_p): Update.
19302 (find_best_successor): Update.
19303 * trans-mem.c (expand_transaction): Update.
19304 * tree-call-cdce.c: Update.
19305 * tree-cfg.c (gimple_split_edge): Upate.
19306 (move_sese_region_to_fn): Upate.
19307 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19308 * tree-eh.c (lower_resx): Upate.
19309 (cleanup_empty_eh_move_lp): Upate.
19310 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19311 * tree-inline.c (copy_edges_for_bb): Update.
19312 (copy_cfg_body): Update.
19313 * tree-parloops.c (gen_parallel_loop): Update.
19314 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19315 (gimple_gen_time_profiler): Update.
19316 * tree-ssa-dce.c (remove_dead_stmt): Update.
19317 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19318 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19319 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19320 (unloop_loops): Update.
19321 (try_peel_loop): Update.
19322 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19323 * tree-ssa-loop-split.c (connect_loops): Update.
19324 (split_loop): Update.
19325 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19326 (hoist_guard): Update.
19327 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19328 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19329 (value_replacement): Update.
19330 * tree-ssa-reassoc.c (branch_fixup): Update.
19331 * tree-ssa-tail-merge.c (replace_block_by): Update.
19332 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19333 (create_edge_and_update_destination_phis): Update.
19334 (compute_path_counts): Update.
19335 (recompute_probabilities): Update.
19336 (update_joiner_offpath_counts): Update.
19337 (freqs_to_counts_path): Update.
19338 (duplicate_thread_path): Update.
19339 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19340 (struct switch_conv_info): Update.
19341 (gen_inbound_check): Update.
19342 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19343 (vect_do_peeling): Update.
19344 (vect_loop_versioning): Update.
19345 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19346 (optimize_mask_stores): Update.
19347 * ubsan.c (ubsan_expand_null_ifn): Update.
19348 * value-prof.c (gimple_divmod_fixed_value): Update.
19349 (gimple_divmod_fixed_value_transform): Update.
19350 (gimple_mod_pow2): Update.
19351 (gimple_mod_pow2_value_transform): Update.
19352 (gimple_mod_subtract): Update.
19353 (gimple_mod_subtract_transform): Update.
19354 (gimple_ic): Update.
19355 (gimple_stringop_fixed_value): Update.
19356 (gimple_stringops_transform): Update.
19357 * value-prof.h: Update.
19359 2017-06-29 Carl Love <cel@us.ibm.com>
19361 * config/rs6000/rs6000-c.c: Add support for built-in functions
19362 vector signed int vec_signed (vector float);
19363 vector signed long long vec_signed (vector double);
19364 vector signed int vec_signed2 (vector double, vector double);
19365 vector signed int vec_signede (vector double);
19366 vector signed int vec_signedo (vector double);
19367 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19368 instruction generator.
19369 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19370 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19371 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19373 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19374 vunsignede_v2df): Add define_expands.
19375 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19376 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19377 VEC_UNSIGNEDO): Add definitions.
19378 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19379 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19380 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19381 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19382 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19383 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19384 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19385 vec_unsignede, vec_unsignedo): Add builtin defines.
19386 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19388 * doc/extend.texi: Update the built-in documentation file for the
19389 new built-in functions.
19391 2017-06-29 Richard Biener <rguenther@suse.de>
19393 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19394 reduction chains to LOOP_VINFO_REDUCTIONS.
19395 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19396 SLP reductions after processing reduction chains.
19398 2017-06-29 Nathan Sidwell <nathan@acm.org>
19400 * builtins.c (fold_builtin_FUNCTION): Use
19401 lang_hooks.decl_printable_name.
19403 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19405 PR middle-end/81194
19406 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19407 with only one label.
19408 * stmt.c (expand_case): Assert NCASES is greater than one.
19410 2017-06-29 Richard Biener <rguenther@suse.de>
19412 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19414 (group_case_labels): Likewise.
19415 (find_taken_edge): Push sanity checking on val to workers...
19416 (find_taken_edge_cond_expr): ... here
19417 (find_taken_edge_switch_expr): ... and here, handle cases
19418 with just a default label.
19419 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19420 (group_case_labels): Likewise.
19421 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19422 group_case_labels does anything cleanup the CFG again.
19424 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19426 PR tree-optimization/81196
19427 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19428 exit condition comparing two IVs.
19430 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19432 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19433 profile to the dummy entry at the end of the list of architectures.
19434 * config/arm/arm-cpu-cdata.h: Regenerated.
19436 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19437 Michael Collison <michael.collison@arm.com>
19440 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19442 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19443 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19444 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19445 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19446 of shift when the shift amount is masked with constant equal to
19447 the size of the mode.
19448 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19451 2017-06-29 Martin Liska <mliska@suse.cz>
19453 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19455 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19458 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19461 (edge_growth_cache_entry::edge_growth_cache_entry): New
19463 (reset_edge_growth_cache): Update to use constructor.
19465 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19467 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19468 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19469 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19471 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19473 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19474 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19476 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19478 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19479 (*-linux-uclibc*): Add t-uclibc tmake_file.
19480 * config/t-musl: New.
19481 * config/t-uclibc: New.
19483 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19485 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19487 (gen_comm_data): Emit architectural setting of arch_prof.
19488 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19490 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19491 (armv8-m.base, armv8-m.main): Likewise.
19492 * arm-protos.h (arm_build_target): Add profile field.
19493 (arch_option): Likewise.
19494 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19496 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19497 arm_active_target.profile.
19499 2017-06-28 Richard Biener <rguenther@suse.de>
19501 PR middle-end/81227
19502 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19503 TYPE_OVERFLOW_WRAPS.
19504 * match.pd (negate_expr_p): Likewise.
19505 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19506 fold_build2, not fold_binary.
19508 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19510 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19511 Convert memory address to Pmode.
19512 (aarch64_print_operand): Assert MEM operands are always Pmode.
19514 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19517 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19518 Remove redundant if.
19519 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19520 * config/arm/aarch-common-protos.h
19521 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19522 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19524 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19527 * multiple_target.c (create_dispatcher_calls): Set the default
19528 clone to be static, not public.
19530 2017-06-28 Richard Biener <rguenther@suse.de>
19532 * tree-vect-loop.c (vectorizable_reduction): Move special
19533 cond reduction IV var creation ...
19534 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19535 parameter. Use STMT_VINFO_VECTYPE.
19536 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19539 2017-06-28 Martin Liska <mliska@suse.cz>
19542 * ipa-visibility.c (non_local_p): Handle visibility.
19544 2017-06-28 Martin Liska <mliska@suse.cz>
19547 * common.opt: Add IntegerRange to various options.
19548 * opt-functions.awk (integer_range_info): New function.
19549 * optc-gen.awk: Add integer_range_info to cl_options struct.
19550 * opts-common.c (decode_cmdline_option): Handle
19551 CL_ERR_INT_RANGE_ARG.
19552 (cmdline_handle_error): Likewise.
19553 * opts.c (print_filtered_help): Show valid interval in
19554 when --help is provided.
19555 * opts.h (struct cl_option): Add range_min and range_max fields.
19556 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19558 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19560 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19561 (x * C EQ/NE y * C): New transformation.
19563 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19565 * genmultilib (combination_space): Accept '+' in option names.
19567 2017-06-28 Martin Liska <mliska@suse.cz>
19570 * asan.c (instrument_derefs): Bail out inner references
19571 that are hard register variables.
19573 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19576 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19577 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19579 2017-06-28 Richard Biener <rguenther@suse.de>
19581 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19582 (vect_get_slp_defs): Adjust.
19583 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19585 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19587 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19588 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19589 (vectorizable_reduction): Adjust.
19590 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19592 (vect_get_slp_defs): Likewise.
19593 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19594 (vectorizable_bswap): Adjust.
19595 (vectorizable_call): Likewise.
19596 (vectorizable_conversion): Likewise.
19597 (vectorizable_assignment): Likewise.
19598 (vectorizable_shift): Likewise.
19599 (vectorizable_operation): Likewise.
19600 (vectorizable_store): Likewise.
19601 (vectorizable_condition): Likewise.
19602 (vectorizable_comparison): Likewise.
19604 2017-06-28 Michael Collison <michael.collison@arm.com>
19607 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19609 (arm_gen_movmemqi): Removed unused variable 'i'.
19610 Convert 'for' loop into 'while' loop.
19611 (arm_expand_prologue): Remove last unnecessary set of insn.
19612 (thumb_pop): Remove unused variable 'pushed_words'.
19613 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19615 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19617 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19618 * config/s390/s390-protos.h: Add prototype of
19619 s390_rel_address_ok_p.
19620 * config/s390/s390.c (s390_got_symbol): New function.
19621 (s390_rel_address_ok_p): New function.
19622 (legitimize_pic_address): Use s390_rel_address_ok_p.
19623 (s390_load_got): Use s390_got_symbol.
19624 (s390_option_override): Issue error if
19625 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19626 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19628 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19630 2017-06-27 Andrew Pinski <apinski@cavium.com>
19632 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19633 (X * copysign (1.0, X)): New pattern.
19634 (X * copysign (1.0, -X)): New pattern.
19635 (copysign (-1.0, CST)): New pattern.
19637 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19639 * genmultilib (combination_space): Remove variable.
19640 Validate reuse rules against regular expression for any sequence
19641 of multilib options in any order.
19643 2017-06-27 Michael Collison <michael.collison@arm.com>
19645 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19646 call aarch64_split_simd_combine.
19647 * (aarch64_combine_internal<mode>): Delete pattern.
19648 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19649 Allow register and subreg operands.
19651 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19653 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19654 specific need, just fallback on defaults.
19655 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19657 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19658 Olivier Hainque <hainque@adacore.com>
19660 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19662 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19663 targets. Pick a default if no particular attempt applied.
19664 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19667 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19669 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19670 (x86_64-wrs-vxworks7): Likewise.
19672 2017-06-27 Marek Polacek <polacek@redhat.com>
19675 * ubsan.c (instrument_null): Check get_base_address's result for null.
19677 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19679 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19681 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19683 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19684 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19685 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19686 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19687 New function types.
19688 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19689 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19690 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19691 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19692 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19693 BUILT_IN_FEUPDATEENV): New builtins.
19694 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19695 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19696 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19697 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19699 (builtin_structptr_types): Adjust size.
19700 * tree.c (builtin_structptr_types): Add four entries.
19702 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19703 Olivier Hainque <hainque@adacore.com>
19705 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19706 (TLS_SYM): New local macro, forcing reference to __tls__ on
19707 link command lines for VxWorks 7 RTPs, triggering initialization
19709 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19710 OS features TLS support, true for RTPs on VxWorks 7.
19711 * config/vxworks.c (vxworks_override_options): Setup emutls
19714 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19716 * predict.c (test_prediction_value_range): Use -1U instead of -1
19717 to avoid narrowing conversion warning.
19718 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19719 to avoid narrowing conversion warning.
19720 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19722 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19724 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19726 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19727 64bit configurations.
19728 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19729 (SIZE_TYPE): Likewise.
19730 * config/vxworks.c (vxworks_emutls_var_fields): Use
19731 long_unsigned_type_node instead of unsigned_type_node as the offset
19732 field type, which is "pointer" mode in emutls.c.
19734 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19737 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19739 PR middle-end/81207
19740 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19741 gimple_vuse copying separately from gimple_vdef copying.
19743 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19745 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19748 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19749 Olivier Hainque <hainque@adacore.com>
19751 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19752 for all vxworks7 targets.
19753 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19754 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19755 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19756 variations for VX6/VX7 and 32/64bits later on in ...
19757 (VXWORKS_LIB_SPEC): Leverage new macros.
19758 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19759 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19761 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19763 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19764 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19766 2017-06-26 Carl Love <cel@us.ibm.com>
19768 * config/rs6000/rs6000-c.c: Add support for built-in functions
19769 vector bool char vec_reve (vector bool char);
19770 vector signed char vec_reve (vector signed char);
19771 vector unsigned char vec_reve (vector unsigned char);
19772 vector bool int vec_reve (vector bool int);
19773 vector signed int vec_reve (vector signed int);
19774 vector unsigned int vec_reve (vector unsigned int);
19775 vector bool long long vec_reve (vector bool long long);
19776 vector signed long long vec_reve (vector signed long long);
19777 vector unsigned long long vec_reve (vector unsigned long long);
19778 vector bool short vec_reve (vector bool short);
19779 vector signed short vec_reve (vector signed short);
19780 vector double vec_reve (vector double);
19781 vector float vec_reve (vector float);
19782 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19783 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19784 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19785 (altivec_vreve): New pattern.
19786 * config/rs6000/altivec.h (vec_reve): New define.
19787 * doc/extend.texi (vec_rev): Update the built-in documentation file
19788 for the new built-in functions.
19790 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19792 PR tree-optimization/71815
19793 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19795 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19797 (slsr_process_phi): Likewise.
19798 (replace_uncond_cands_and_profitable_phis): Don't replace a
19799 multiply candidate with a stride of 1 (copy or cast).
19800 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19802 (lowest_cost_path): Likewise.
19803 (total_savings): Likewise.
19805 2017-06-26 Richard Biener <rguenther@suse.de>
19808 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19809 Use def_builtin_pure for all gather builtins.
19811 2017-06-26 Richard Biener <rguenther@suse.de>
19813 PR tree-optimization/81203
19814 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19815 non-dominating BBs.
19817 2017-06-26 Marek Polacek <polacek@redhat.com>
19820 * doc/invoke.texi: Document -Wmultistatement-macros.
19822 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19824 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19825 arm_neon_ok_no_float_abi effective target.
19827 2017-06-26 Richard Biener <rguenther@suse.de>
19829 PR tree-optimization/80928
19830 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19831 (copy_bbs): Set BB_DUPLICATED flag early.
19832 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19834 (execute_on_shrinking_pred): Likewise.
19835 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19836 BB_DUPLICATED blocks.
19837 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19838 iterate over all PHIs considering removal of *gsi.
19840 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19842 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19845 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19847 * config/rs6000/rs6000-string.c: (expand_block_clear,
19848 do_load_for_compare, select_block_compare_mode,
19849 compute_current_alignment, expand_block_compare,
19850 expand_strncmp_align_check, expand_strn_compare,
19851 expand_block_move, rs6000_output_load_multiple)
19852 Move functions related to string/block move/compare
19853 to a separate file.
19854 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19855 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19856 for this function which is now used in two files.
19857 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19858 * config.gcc: Add rs6000-string.o to extra_objs for
19859 targets powerpc*-*-* and rs6000*-*-*.
19861 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19864 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19865 32-bit, since indexed is not valid for DImode.
19866 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19867 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19868 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19869 (define_peephole2 for Altivec d-form store): Likewise.
19872 * multiple_target.c (create_dispatcher_calls): Only create the
19873 dispatcher call if the function is the default clone of a
19874 versioned function.
19876 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19878 PR middle-end/80902
19879 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19880 a call, force the call to not be a tail call.
19882 2017-06-23 Jeff Law <law@redhat.com>
19884 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19887 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19889 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19890 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19891 handling for early expansion of vector shifts (sl,sr,sra,rl).
19892 (builtin_function_type): Add vector shift right instructions
19893 to the unsigned argument list.
19895 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19897 rtl-optimizatoin/79286
19898 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19899 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19900 trap. PIC register plus a const unspec without offset can never trap.
19902 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19904 * tree.h (builtin_structptr_type): New type.
19905 (builtin_structptr_types): Declare new array.
19906 * tree.c (builtin_structptr_types): New array.
19907 (free_lang_data, build_common_tree_nodes): Use it.
19909 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19912 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19915 2017-06-22 Matt Turner <mattst88@gmail.com>
19917 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19918 Lake models to skylake case. Assume skylake for unknown
19919 models with clflushopt.
19921 2017-06-22 Jeff Law <law@redhat.com>
19923 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19924 frame sizes that do not satisfy aarch64_uimm12_shift.
19926 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19928 * profile-count.h (apply_probability,
19929 apply_scale, probability_in): Fix checks for zero.
19931 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19933 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19934 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19936 2016-06-22 Richard Biener <rguenther@suse.de>
19938 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19939 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19940 REDUC_MAX_EXPR support.
19941 (vectorizable_reduction): Likewise.
19942 (vect_create_epilog_for_reduction): Likewise.
19944 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19946 * match.pd (A / (1 << B) -> A >> B): New.
19947 * generic-match-head.c: Include optabs-tree.h.
19948 * gimple-match-head.c: Likewise.
19949 * optabs-tree.h (target_supports_op_p): New.
19950 * optabs-tree.c (target_supports_op_p): New.
19952 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19954 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19955 $gcc_cv_ld --help output.
19956 (gcc_cv_ld_demangle): Likewise.
19957 (gcc_cv_ld_eh_frame_hdr): Likewise.
19958 (gcc_cv_ld_pie): Likewise.
19959 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19960 (gcc_cv_ld_buildid): Likewise.
19961 (gcc_cv_ld_sysroot): Likewise.
19962 (ld_bndplt_support): Likewise.
19963 (ld_pushpopstate_support): Likewise.
19964 * configure: Regenerate.
19965 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19967 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19970 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19971 operands indexes to avoid gap between operands and match_dups.
19973 2017-06-21 Andrew Pinski <apinski@cavium.com>
19975 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19976 Increment Arith_shift and Arith_shift_reg by 1.
19977 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19979 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19980 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19981 (aarch64_strip_extend): Add new argument and test for it.
19982 (aarch64_cheap_mult_shift_p): New function.
19983 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19984 add a cost if it is true.
19985 Update calls to aarch64_strip_extend.
19986 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19988 2017-06-21 Andrew Pinski <apinski@cavium.com>
19990 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19992 (thunderxt88): Likewise.
19993 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19994 (thunderx_prefetch_tune): New variable.
19995 (thunderx2t99_prefetch_tune): Update for the correct values.
19996 (thunderxt88_tunings): New variable.
19997 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19998 generic_prefetch_tune.
19999 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20001 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20003 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20004 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20005 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20006 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20007 (aarch64_atomic_cas<mode>, GPI): Likewise.
20009 2017-06-21 Martin Liska <mliska@suse.cz>
20011 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20012 statements on cold and hot labels.
20013 * predict.c (tree_estimate_probability_bb): Remove the
20014 prediction from this place.
20016 2017-06-21 Martin Liska <mliska@suse.cz>
20018 PR tree-optimization/79489
20019 * gimplify.c (maybe_add_early_return_predict_stmt): New
20021 (gimplify_return_expr): Call the function.
20022 * predict.c (tree_estimate_probability_bb): Remove handling
20024 * predict.def: Update comment about early return predictor.
20025 * gimple-predict.h (is_gimple_predict): New function.
20026 * predict.def: Change default value of early return to 66.
20027 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20029 * passes.def: Put pass_strip_predict_hints to the beginning of
20032 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20034 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20035 FUNCTION_DECL declarations.
20036 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20038 (dwaf2out_decl): Likewise.
20039 * godump.c (go_early_global_decl): Skip call to the real debug hook
20040 for FUNCTION_DECL declarations.
20041 * passes.c (rest_of_decl_compilation): Skip call to the
20042 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20043 -fdump-go-spec is passed.
20045 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20047 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20048 Reorder for compactness.
20049 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20050 (def_builtin_pure, def_builtin_pure2): New functions.
20051 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20053 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20055 * match.pd (nop_convert): New predicate.
20056 ((A +- CST1) +- CST2): Allow some NOP conversions.
20058 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20061 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20062 with ctors/dtors if GOVD_SHARED is set.
20064 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20066 * config/aarch64/aarch64.md (movti_aarch64):
20067 Emit mov rather than orr.
20068 (movtf_aarch64): Likewise.
20069 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20070 Emit mov rather than orr.
20072 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20074 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20075 Swap alternatives, make integer dup more expensive.
20077 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20079 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20080 Return true for non-tls symbols.
20082 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20084 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20085 (cortex-a75): Likewise.
20086 (cortex-a75.cortex-a55): Likewise.
20087 * config/aarch64/aarch64-tune.md: Regenerate.
20088 * doc/invoke.texi (-mtune): Document new values for -mtune.
20090 2017-06-21 Tom de Vries <tom@codesourcery.com>
20092 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20093 stack_size feature.
20094 (Effective-Target Keywords, Other attributes): Suggest using
20095 dg-add-options stack_size feature to get stack limit in stack_size
20096 effective target documentation.
20098 2017-06-21 Julian Brown <julian@codesourcery.com>
20099 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20101 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20102 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20103 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20105 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20106 attribute type list for neon_multiply.
20107 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20108 attribute type list for neon_multiply.
20109 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20110 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20111 attribute type list for neon_multiply.
20112 * config/arm/types.md (crypto_pmull): Add.
20113 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20114 attribute type list.
20116 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20118 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20121 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20123 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20124 to make sure not to dereference a NULL cost_classes_ptr pointer.
20126 2017-06-20 Carl Love <cel@us.ibm.com>
20128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20129 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20130 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20132 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20133 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20134 VMULOSW): New enum "unspec" values.
20135 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20136 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20137 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20138 altivec_vmulosw): New patterns.
20139 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20140 VMULOSW): Add definitions.
20142 2017-06-20 Julia Koval <julia.koval@intel.com>
20144 * config/i386/i386.c: Fix rounding expand for new pattern.
20145 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20147 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20149 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20150 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20152 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20154 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20157 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20159 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20160 architecture, then by implementer ID.
20161 * config/aarch64/aarch64-tune.md: Regenerate.
20163 2017-06-20 Richard Biener <rguenther@suse.de>
20165 PR middle-end/81097
20166 * fold-const.c (split_tree): Fold to type before negating.
20168 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20170 * diagnostic-show-locus.c
20171 (selftest::test_fixit_deletion_affecting_newline): New function.
20172 (selftest::diagnostic_show_locus_c_tests): Call it.
20174 2017-06-20 Andreas Schwab <schwab@suse.de>
20177 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20180 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20182 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20183 __ARM_FEATURE_COPROC according to support.
20185 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20187 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20188 Rewritten to avoid overflow for > 32-bit pointers.
20191 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20192 by removing enum keyword.
20193 (ubsan_type_descriptor): Likewise. Formatting fix.
20196 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20197 splitter): Require TARGET_SSE2 in the condition.
20199 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20202 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20203 for doing vector set of SFmode on ISA 3.0.
20204 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20205 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20207 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20208 SFmode value into a V4SF variable that was extracted from another
20209 V4SF variable without converting the element to double precision
20210 and back to single precision vector format.
20211 (vsx_insert_extract_v4sf_p9_2): Likewise.
20213 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20215 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20216 in UWHI to avoid undefined overflow.
20219 * ubsan.h (enum ubsan_encode_value_phase): New.
20220 (ubsan_encode_value): Change second argument to
20221 enum ubsan_encode_value_phase with default value of
20222 UBSAN_ENCODE_VALUE_GENERIC.
20223 * ubsan.c (ubsan_encode_value): Change second argument to
20224 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20225 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20226 create_tmp_var_raw instead of create_tmp_var and use a
20228 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20229 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20230 ubsan_encode_value callers.
20233 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20234 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20235 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20237 2017-06-19 Richard Biener <rguenther@suse.de>
20239 PR middle-end/81118
20240 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20241 estimates if we changed anything.
20243 2017-06-19 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/80887
20246 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20247 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20248 simplified lookups, then reset mprts_hook.
20249 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20251 (try_to_simplify): Likewise.
20253 2017-06-19 Martin Liska <mliska@suse.cz>
20256 * gimplify.c (gimplify_switch_expr):
20257 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20259 2017-06-19 Martin Liska <mliska@suse.cz>
20261 * doc/install.texi: Document that PGO runs in 4 stages.
20263 2017-06-19 Martin Liska <mliska@suse.cz>
20266 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20267 to dispatcher function name.
20268 * multiple_target.c (replace_function_decl): New function.
20269 (create_dispatcher_calls): Redirect both edges and references.
20271 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20273 * profile-count.c (profile_count::dump): Dump quality.
20274 (profile_count::differs_from_p): Update for unsigned val.
20275 * profile-count.h (profile_count_quality): New enum.
20276 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20278 2017-06-19 Richard Biener <rguenther@suse.de>
20280 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20281 struct function as arg.
20282 (estimate_numbers_of_iterations): Export overload with loop arg.
20283 (free_numbers_of_iterations_estimates_loop): Use an overload of
20284 free_numbers_of_iterations_estimates instead.
20285 * tree-cfg.c (remove_bb): Adjust.
20286 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20287 * tree-parloops.c (gen_parallel_loop): Likewise.
20288 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20290 (tree_unroll_loops_completely): Likewise.
20291 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20292 Use an overload instead and export.
20293 (estimated_loop_iterations): Adjust.
20294 (max_loop_iterations): Likewise.
20295 (likely_max_loop_iterations): Likewise.
20296 (estimate_numbers_of_iterations): Take struct function as arg
20298 (loop_exits_before_overflow): Adjust.
20299 (free_numbers_of_iterations_estimates_loop): Use an overload.
20300 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20301 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20303 2017-06-19 Richard Biener <rguenther@suse.de>
20306 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20307 RANGE_EXPR conservatively.
20309 2017-06-16 Carl Love <cel@us.ibm.com>
20311 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20312 definitions for vec_float, vec_float2, vec_floato,
20313 vec_floate built-ins.
20314 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20315 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20317 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20318 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20319 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20320 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20321 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20322 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20323 vec_floato): Add builtin defines.
20324 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20325 Update the built-in documentation file for the new built-in
20328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20330 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20331 (mthumb): Mark as the negative of -marm.
20333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20335 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20337 (ARM Options, -mtune): Document that this accepts the same
20338 extension options as -mcpu.
20339 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20343 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20344 permitted extensions.
20346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20348 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20349 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20350 (armv8-m.main): Add option +nodsp.
20351 * config/arm/arm-cpu-cdata.h: Regenerated.
20353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20355 * config/arm/t-fuchsia: New file.
20356 * config.gcc (arm*-*-fuchsia*): Use it.
20358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20360 * config/arm/t-symbian: Rewrite for new option infrastructure.
20362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20364 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20365 (MULTILIB_REQUIRED): Likewise.
20367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20369 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20370 (MULTILIB_RESUE): Likewise.
20371 (MULTILIB_MATCHES): Likewise.
20372 (MULTLIB_REQUIRED): Likewise.
20374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20376 * config/arm/t-rtems: Rewrite for new option framework.
20378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20380 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20381 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20382 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20383 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20384 * config/arm/t-multilib: ... here.
20385 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20386 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20387 armv7-a and armv8*-a when A-profile libraries have not been built.
20388 * config/arm/t-rmprofile: Rewrite.
20390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20392 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20393 with a backslash. Remove the backslash after substituting unescaped
20395 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20399 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20400 the driver to pass through the default setting of -mfloat-abi.
20401 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20403 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20404 (all_feat_combs): New rule.
20405 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20407 * config/arm/t-aprofile: Rewrite.
20409 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20411 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20412 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20413 fpu is not specified by the user/command-line.
20414 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20415 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20416 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20417 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20418 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20419 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20423 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20424 * config/arm/t-arm-elf: Rewritten.
20426 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20428 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20429 have some floating-point instructions.
20430 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20431 (TARGET_MAYBE_HARD_FLOAT): New macro.
20432 * config/arm/arm-builtins.c (arm_init_builtins): Use
20433 TARGET_MAYBE_HARD_FLOAT.
20434 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20438 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20439 (configargs.h): Include it.
20440 (arm_print_hint_for_fpu_option): New function.
20441 (arm_parse_fpu_option): New function.
20442 (candidate_extension): New class.
20443 (arm_canon_for_multilib): New function.
20444 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20445 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20446 (ARCH_CANONICAL_SPECS): New macro.
20447 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20451 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20452 are set after handling multilib fragments. Set target_cpu_default2
20455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20457 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20459 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20460 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20462 (check_arch): Likewise.
20463 * config/arm/arm.c (arm_configure_build_target): Handle
20464 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20465 options in the default.
20467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20469 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20470 when an option is an alias of another.
20471 * config/arm/parsecpu.awk (optalias): New parser token.
20472 (gen_comm_data): Mark non-alias options as such. Emit entries
20473 for extension aliases.
20474 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20475 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20476 (armv6kz, armv6zk, armv6t2): Likewise.
20477 (armv7): Make vfpv3-d16 an alias.
20478 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20480 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20481 Sort in canonical order.
20482 (armv8-a): Sort in canonical order.
20483 (armv8.1-a, armv8.2-a): Likewise.
20484 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20486 (cortex-a9): Sort in canonical order.
20487 * config/arm/arm.c (selftests.h): Include it.
20488 (arm_test_cpu_arch_data): New function.
20489 (arm_run_self_tests): New function.
20490 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20491 (targetm): Move declaration to the end of the file.
20492 * arm-cpu-cdata.h: Regenerated.
20494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20496 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20497 call to target_mode_check describing the type of option passed.
20498 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20499 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20500 arm_parse_cpu_option_name to match parameters against list of
20502 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20503 arm_arch_core_flags data structure.
20504 * config/arm/arm-cpu_cdata.h: Regenerated.
20506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20508 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20510 (arm_print_hint_for_cpu_option): Likewise.
20511 (arm_print_hint_for_arch_option): Likewise.
20512 (arm_parse_cpu_option_name): Likewise.
20513 (arm_parse_arch_option_name): Likewise.
20514 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20515 of entries in the all_fpus list.
20516 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20517 (arm_parse_cpu_option_name): Declare.
20518 (arm_parse_arch_option_name): Declare.
20519 (arm_parse_option_features): Declare.
20520 (arm_intialize_isa): Declare.
20521 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20523 (gen_comm_data): ... here. Make definitions non-static.
20524 * config/arm/arm-cpu-data.h: Regenerated.
20525 * config/arm/arm-cpu-cdata.h: Regenerated.
20527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20529 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20530 (cpu_arch_extension): New structure.
20531 (cpu_arch_option, arch_option, cpu_option): New structures.
20532 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20533 architecture types.
20534 (gen_data): Generate new format data tables.
20535 * config/arm/arm.c (cpu_tune): New structure.
20536 (cpu_option, processors): Delete.
20537 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20538 (arm_print_hint_for_cpu_option): ... this and ...
20539 (arm_print_hint_for_arch_option): ... this.
20540 (arm_parse_arch_cpu_name): Delete. Replace with ...
20541 (arm_parse_cpu_option_name): ... this and ...
20542 (arm_parse_arch_option_name): ... this.
20543 (arm_unrecognized_feature): Change type of target parameter to
20545 (arm_parse_arch_cpu_features): Delete. Replace with ...
20546 (arm_parse_option_features): ... this.
20547 (arm_configure_build_target): Rework to use new configuration data
20549 (arm_print_tune_info): Rework for new configuration data tables.
20550 * config/arm/arm-cpu-data.h: Regenerated.
20551 * config/arm/arm-cpu.h: Regenerated.
20553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20555 * Makefile.in (OBJS): Move sbitmap.o from here ...
20556 (OBJS-libcommon): ... to here.
20558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20560 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20561 (ISA_ALL_CRYPTO): New macro.
20562 (ISA_ALL_SIMD): New macro
20563 (ISA_ALL_FP): New macro.
20564 * config/arm/arm.c (fpu_bitlist): Update initializer.
20565 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20567 (arm9e): Add fpu. Add option for nofp
20568 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20569 (arm926ej-s, arm1026ej-s): Likewise.
20570 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20571 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20572 neon-fp16, neon-vfpv4, nofp and nosimd.
20573 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20574 (cortex-a8): Add fpu. Add option for nofp.
20575 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20576 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20577 (cortex-r4f): Add fpu.
20578 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20579 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20581 (cortex-r8): Likewise.
20582 (cortex-m4): Add fpu. Add option for nofp.
20583 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20584 (cortex-a17.cortex-a7): Likewise.
20585 (cortex-a32): Add fpu. Add options for crypto and nofp.
20586 (cortex-a35, cortex-a53): Likewise.
20587 (cortex-a57): Add fpu. Add option for crypto.
20588 (cortex-a72, cortex-a73): Likewise.
20589 (exynos-m1): Likewise.
20590 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20591 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20592 (cortex-m33): Add fpu. Add option for nofp.
20593 * config/arm/arm-cpu-cdata.h: Regenerated
20594 * config/arm/arm-cpu-data.h: Regenerated.
20596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20598 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20599 (armv5te, armv5tej): Likewise.
20600 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20601 (armv7): Add options fp and vfpv3-d16.
20602 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20603 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20605 (armv7ve): Likewise.
20606 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20607 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20608 (armv8-a): Add nocrypto option.
20609 (armv8.1-a, armv8.2-a): Likewise.
20610 (armv8-m.main): add options fp, fp.dp and nofp.
20612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20614 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20616 (armv8-a+crc): Delete.
20617 (armv8.1-a): Add options simd, crypto and nofp.
20618 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20619 (armv8.2-a+fp16): Delete.
20620 (armv8-m.main): Add option dsp.
20621 (armv8-m.main+dsp): Delete.
20622 (cortex-a8): Add fpu. Add nofp option.
20623 (cortex-a9): Add fpu. Add nofp and nosimd options.
20624 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20625 link to main cpu and architecture data structures.
20626 (gen_comm_data): Only put isa attributes from the main architecture
20628 (option): New statement for architecture and CPU entries.
20629 * arm.c (struct cpu_option): New structure.
20630 (struct processors): Add entry for options.
20631 (arm_unrecognized_feature): New function.
20632 (arm_parse_arch_cpu_name): Ignore any characters after the first
20634 (arm_parse_arch_cpu_feature): New function.
20635 (arm_configure_build_target): Separate out any CPU and architecture
20636 features and parse separately. Don't error out if -mfpu=auto is
20637 used with only an architecture string.
20638 (arm_print_asm_arch_directives): New function.
20639 (arm_file_start): Call it.
20640 * config/arm/arm-cpu-cdata.h: Regenerated.
20641 * config/arm/arm-cpu-data.h: Likewise.
20642 * config/arm/arm-tables.opt: Likewise.
20644 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20646 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20647 assembler when it is not -mfpu=auto.
20649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20651 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20652 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20653 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20654 (ASM_CPU_SPEC): Rewrite.
20655 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20656 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20657 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20658 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20659 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20660 copied string is NUL-terminated. Also strip any characters prefixed
20662 (arm_rewrite_selected_arch): New function.
20663 (arm_rewrite_march): New function.
20665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20667 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20668 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20669 (march, mcpu, mtune): Convert to string-based options.
20670 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20671 (arm_parse_arch_cpu_name): New function.
20672 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20673 identify selected architecture or CPU.
20674 (arm_option_save): New function.
20675 (TARGET_OPTION_SAVE): Redefine.
20676 (arm_option_restore): Restore string options.
20677 (arm_option_print): Print string options.
20679 2017-06-16 Martin Sebor <msebor@redhat.com>
20681 PR tree-optimization/80933
20682 PR tree-optimization/80934
20683 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20684 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20685 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20686 (gimple_fold_builtin): Call them.
20688 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20690 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20691 as unlikely; update profile.
20693 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20695 * predict.c (force_edge_cold): Handle declaring edges impossible
20698 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20700 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20702 (try_unroll_loop_completely): Fix reporting.
20704 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20706 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20708 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20711 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20712 if given a non-constant argument for an intrinsic which requires a
20715 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20717 * profile.c (compare_freqs): New function.
20718 (branch_prob): Sort edge list.
20719 (find_spanning_tree): Assume that the list is priority sorted.
20721 2017-06-16 Richard Biener <rguenther@suse.de>
20723 PR tree-optimization/81090
20724 * passes.def (pass_record_bounds): Remove.
20725 * tree-pass.h (make_pass_record_bounds): Likewise.
20726 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20727 make_pass_record_bounds): Likewise.
20728 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20729 not free niter estimates at the beginning but at the end.
20730 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20732 2017-06-16 Richard Biener <rguenther@suse.de>
20734 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20735 initializer to workaround ICE in host GCC 4.8.
20737 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20739 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20741 (clone_inlined_nodes): Update.
20743 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20745 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20746 prefetch settings, and enable prefetching by default at -O3.
20748 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20750 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20751 Set flag_prefetch_loop_arrays according to tuning data.
20753 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20755 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20756 New tune structure.
20757 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20758 [Unrelated to main purpose of the patch] Place the pointer field last
20759 to enable type checking errors when tune structure are wrongly merged.
20760 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20761 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20762 (thunderx2t99_prefetch_tune): New tune constants.
20763 (tune_params *_tunings): Update all tunings (no functional change).
20764 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20765 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20766 from tunings structures.
20768 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20771 * ubsan.c (instrument_null): Add T argument, use it instead
20772 of computing it based on IS_LHS.
20773 (instrument_object_size): Likewise.
20774 (pass_ubsan::execute): Adjust instrument_null and
20775 instrument_object_size callers to pass gimple_get_lhs or
20776 gimple_assign_rhs1 result to it. Use instrument_null instead of
20777 calling get_base_address and instrument_mem_ref. Handle
20778 aggregate call arguments for object-size sanitization.
20780 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20782 PR tree-optimization/81089
20783 * tree-vrp.c (is_masked_range_test): Validate operands of
20786 2017-06-15 Martin Sebor <msebor@redhat.com>
20789 * dumpfile.c (dump_register): Avoid calling memset to initialize
20790 a class with a default ctor.
20791 * gcc.c (struct compiler): Remove const qualification.
20792 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20793 * hash-table.h: Ditto.
20794 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20796 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20797 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20799 * params.h (struct param_info): Make struct members non-const.
20800 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20801 with default initialization.
20802 * vec.h (vec_copy_construct, vec_default_construct): New helper
20804 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20805 with vec_copy_construct.
20806 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20807 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20808 * doc/invoke.texi (-Wclass-memaccess): Document.
20810 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20812 * emit-rtl.h (is_leaf): Update comment about local
20813 register allocator.
20815 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20818 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20819 for a variable to have a section before checking if the section has a
20821 Set section to.persistent if persistent attribute is set.
20822 Warn if .persistent attribute is used on an automatic variable.
20824 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20826 PR rtl-optimization/80474
20827 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20829 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20831 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20834 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20836 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20837 (casesi): Emit bounds checking as RTL.
20838 (casesi_internal_mips16_<mode>): Remove bounds checking.
20840 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20842 * config/xtensa/xtensa.c (xtensa_option_override): Append
20843 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20844 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20845 xtensa_doloop_hooks): Define unconditionally.
20846 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20848 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20849 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20850 for it in xtensa_option_override.
20851 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20852 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20854 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20856 * doc/cppopts.texi: Document '-' special value to -MF.
20858 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20860 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20861 (cortex_a53_fconst): Likewise.
20862 (cortex_a53_fpmul): Likewise.
20863 (cortex_a53_f_load_64): Likewise.
20864 (cortex_a53_f_load_many): Likewise.
20865 (cortex_a53_advsimd_alu): Likewise.
20866 (cortex_a53_advsimd_alu_q): Likewise.
20867 (cortex_a53_advsimd_mul): Likewise.
20868 (cortex_a53_advsimd_mul_q): Likewise.
20869 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20870 Add missing fmul, r2f_cvt and fconst cases.
20872 2017-06-14 Richard Biener <rguenther@suse.de>
20874 PR middle-end/81088
20875 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20877 (fold_binary_loc): When associating do not treat pre-existing
20878 TREE_OVERFLOW on literal constants as a reason to allow
20879 TREE_OVERFLOW on associated literal constants.
20881 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20883 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20884 (MASK_FEATURES): New macro.
20885 * config/sparc/sparc.c (sparc_option_override): Remove the special
20886 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20888 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20890 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20891 a division of 0 if non-call exceptions are enabled.
20893 2017-06-14 Andrew Pinski <apinski@cavium.com>
20894 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20897 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20898 Improve vector initialization code gen for only variable case.
20900 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20902 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20904 2017-06-14 Richard Biener <rguenther@suse.de>
20906 PR tree-optimization/81083
20907 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20910 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20912 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20913 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20914 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20915 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20916 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20917 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20919 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20921 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20922 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20924 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20926 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20928 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20930 * config/rs6000/t-rtems: Don't handle SPE.
20932 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20934 * config/rs6000/t-linux: Don't handle SPE.
20936 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20938 * config/rs6000/eabispe.h: Delete file.
20940 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20942 * config/rs6000/t-spe: Delete file.
20944 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20946 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20947 (rs6000_legitimate_offset_address_p): Return false for anything in
20948 V2SImode or V2SFmode.
20950 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20952 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20953 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20954 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20956 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20957 (rs6000_legitimate_offset_address_p): Ditto.
20958 (rs6000_emit_move): Ditto.
20959 (rs6000_init_builtins): Remove V4HI_type_node.
20961 2017-06-13 Martin Liska <mliska@suse.cz>
20964 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20965 (gate_asan): Likewise.
20966 * asan.h (asan_no_sanitize_address_p): Remove the function.
20967 (sanitize_flags_p): New function.
20968 * builtins.def: Fix coding style.
20969 * common.opt: Use renamed enum value.
20970 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20971 * doc/extend.texi: Document no_sanitize attribute.
20972 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20973 to SANITIZE_UNDEFINED_NONDEFAULT.
20974 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20975 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20976 Use sanitize_flags_p.
20977 * gimplify.c (gimplify_function_tree): Likewise.
20978 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20979 * opts.c (parse_no_sanitize_attribute): New function.
20980 (common_handle_option): Use renamed enum value.
20981 * opts.h (parse_no_sanitize_attribute): Declare.
20982 * tree.c (sanitize_flags_p): New function.
20983 * tree.h: Declared here.
20984 * tsan.c: Use sanitize_flags_p.
20985 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20986 (instrument_mem_ref): Likewise.
20987 (instrument_bool_enum_load): Likewise.
20988 (do_ubsan_in_current_function): Remove the function.
20989 (pass_ubsan::execute): Use sanitize_flags_p.
20990 * ubsan.h: Remove do_ubsan_in_current_function
20991 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20992 (dump_function_to_file): Use it here.
20994 2017-06-13 Martin Jambor <mjambor@suse.cz>
20996 PR tree-optimization/80803
20997 PR tree-optimization/81063
20998 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20999 (propagate_subaccesses_across_link): Enqueue subtree whenever
21000 necessary instead of relying on the caller.
21002 2017-06-13 Martin Jambor <mjambor@suse.cz>
21004 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21005 that have a first_link.
21006 (sort_and_splice_var_accesses): Do not check first_link before
21008 (subtree_mark_written_and_enqueue): Likewise.
21009 (propagate_all_subaccesses): Likewise and do not stop at first
21010 parent with a first_link.
21012 2017-06-13 Martin Jambor <mjambor@suse.cz>
21014 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21017 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21019 * match.pd: New pattern.
21021 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21023 * tree-vrp.c (is_masked_range_test): New function.
21024 (register_edge_assert_for): Determine ranges for
21027 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21029 PR tree-optimization/67328
21030 * fold-const.c (maskable_range_p): New function.
21031 (build_range_check): Generate bittests if possible.
21033 2017-06-13 Martin Liska <mliska@suse.cz>
21035 * gimple-pretty-print.c (dump_probability): Add new argument.
21036 (dump_edge_probability): Dump both probability and count.
21037 (dump_gimple_label): Likewise.
21038 (dump_gimple_bb_header): Likewise.
21040 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21043 * config/avr/avr-devices.c: Fix indentation.
21044 * config/avr/gen-avr-mmcu-specs.c: Dito.
21046 2017-06-13 Richard Biener <rguenther@suse.de>
21048 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21049 instead get vector type from stmt_info.
21050 (vectorizable_reduction): Adjust. Remove dead code.
21052 2017-06-13 Richard Biener <rguenther@suse.de>
21054 PR middle-end/81065
21055 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21056 case of C * (x * C2 + C3).
21057 (fold_addr_of_array_ref_difference): Properly fold index difference.
21059 2017-06-12 David S. Miller <davem@davemloft.net>
21062 * config/sparc/sparc.md (return expander): Emit frame blockage if
21063 function uses alloca.
21065 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21067 * combine.c (make_field_assignment): Check len rather than the mode
21068 precision when calling force_to_mode.
21070 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21072 Support multilibs and devices that see flash in RAM address range.
21075 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21076 (avr_mcu_t) <flash_pm_offset>: New field.
21077 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21078 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21079 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21080 (AVR_TINY_PM_OFFSET): Remove macro.
21081 * config/avr/avr.opt (-mshort-calls): New option.
21082 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21083 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21084 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21085 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21086 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21087 instead of avr_arch->have_jmp_call.
21088 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21089 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21090 avr_arch->flash_pm_offset to define.
21091 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21092 new field flash_pm_offset. Add entry for avrxmega3.
21093 (avr_texinfo): Add entry for avrxmega3.
21094 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21095 attiny212, attiny214,
21096 attiny412, attiny414, attiny416, attiny417,
21097 attiny814, attiny816, attiny817,
21098 attiny1614, attiny1616, attiny1617,
21099 attiny3214, attiny3216, attiny3217.
21100 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21101 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21102 (avr_print_operand_address) [AVR_TINY]: Same.
21103 (avr_asm_init_sections) <readonly_data_section>: Only patch
21104 callback if avr_arch->flash_pm_offset = 0.
21105 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21106 for rodata if avr_arch->flash_pm_offset != 0.
21107 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21108 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21109 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21110 (m_options): Append opt_rcall.
21111 (m_dirnames): Append dir_rcall.
21112 * config/avr/t-multilib: Regenerate.
21114 * configure.ac [target=avr]: Check whether avrxmega3 default
21115 linker description file works as needed.
21116 * configure: Regenerate.
21117 * doc/avr-mmcu.texi: Regenerate.
21118 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21119 <__AVR_ARCH__>: Document avrxmega3 and 103.
21120 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21121 <__AVR_SHORT_CALLS__>: Document it.
21122 <__AVR_PM_BASE_ADDRESS__>: Document it.
21123 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21124 (AVR Variable Attributes) <progmem>: Document this is
21125 not needed for avrxmega3.
21126 (AVR Named Address Spaces) <__flash>: Dito.
21128 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21130 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21132 2017-06-12 Doug Rupp <rupp@adacore.com>
21134 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21135 Append vxworks-stdint.h to the tm_file list.
21136 * config/vxworks-stdint.h: New file.
21138 2017-06-12 Martin Liska <mliska@suse.cz>
21140 PR tree-optimization/81041
21141 * tree-profile.c (gimple_gen_ic_func_profiler):
21142 Create an extra BB in profile-generate
21143 (gimple_gen_time_profiler): Likewise.
21145 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21147 PR tree-optimization/81003
21148 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21149 (update_range_test): Use it instead of force_gimple_operand_gsi.
21151 2017-06-12 Richard Biener <rguenther@suse.de>
21153 PR tree-optimization/81053
21154 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21155 with backedge value not defined in loop. Simplify def stmt
21158 2017-06-11 Tom de Vries <tom@codesourcery.com>
21161 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21163 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21164 nvptx_cannot_force_const_mem.
21166 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21168 * opts.c (finish_options): Move test for flag_split_stack after
21169 it has been initialized.
21171 2017-06-11 Jason Merrill <jason@redhat.com>
21173 * tree.h (id_equal): New.
21174 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21175 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21176 instead of strcmp of IDENTIFIER_POINTER.
21178 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21180 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21181 (mark_all_inlined_calls_cdtor): Fix formating.
21182 (inline_transform): Rescale profile before inlining.
21184 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21186 * cgraph.h (cgraph_edge::clone): Update prototype.
21187 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21188 (cgraph_node::create_clone): Update.
21189 (cgraph_node::create_version_clone): Update.
21190 * tree-inline.c (copy_bb): Update.
21191 (expand_call_inline): Update.
21193 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21195 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21196 factored out from ...
21197 (rs6000_emit_prologue): ... here.
21199 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21201 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21202 factored out from ...
21203 (rs6000_emit_prologue): ... here.
21205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21207 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21209 (handle_missing_profiles): Fix computation of tp_first_run.
21210 (counts_to_freqs): Do not touch freqs when count is 0.
21212 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21214 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21217 2017-06-10 Tom de Vries <tom@codesourcery.com>
21219 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21220 attributes): Document signal effective target.
21222 2017-06-10 Tom de Vries <tom@codesourcery.com>
21224 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21225 Document effective target stack_size.
21227 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21229 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21230 to the edit_context if they can be auto-applied.
21232 2017-06-9 Ian Lance Taylor <iant@golang.org>
21234 * opts.c (finish_options): If -fsplit-stack, disable implicit
21235 -forder-blocks-and-partition.
21236 * doc/invoke.texi (Optimize Options): Document that when using
21237 -fsplit-stack -forder-blocks-and-partition is not implicitly
21240 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21242 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21243 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21244 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21245 * builtins.def (abort, trap, unreachable): Declare cold.
21246 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21247 * tree-core.h (ECF_COLD): New.
21248 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21249 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21251 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21253 * predict.c (unlikely_executed_stmt_p): Cleanup.
21255 2017-06-09 Richard Biener <rguenther@suse.de>
21257 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21258 model if the ref is always written to.
21260 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21262 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21264 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21266 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21269 2017-06-09 Tom de Vries <tom@codesourcery.com>
21272 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21273 "target cannot support label values" when encountering LABEL_REF.
21275 2017-06-09 Martin Liska <mliska@suse.cz>
21277 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21278 (gimple_gen_ic_func_profiler): Emit direct comparison
21279 of __gcov_indirect_call_callee with NULL.
21280 (gimple_gen_time_profiler): Change probability from
21281 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21283 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21285 * profile.c (edge_gcov_counts): Turn to pointer.
21286 (compute_branch_probabilities, compute_branch_probabilities): Update.
21287 (branch_prob): Do not clear edge_gcov_count.
21288 * profile.h (edge_gcov_counts): Turn to pointer.
21289 (edge_gcov_count): Update.
21291 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21293 * gimple.h (gimple_check_failed): Mark cold.
21295 2017-06-09 Richard Biener <rguenther@suse.de>
21297 PR tree-optimization/66623
21298 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21299 refactor check_reduction into two parts, properly computing
21300 whether we have to check reduction validity for outer loop
21303 2017-06-09 Richard Biener <rguenther@suse.de>
21305 PR tree-optimization/79483
21306 * graphite-scop-detection.c (order): New global.
21307 (get_order): Compute bb to order mapping that satisfies code
21308 generation constraints.
21309 (cmp_pbbs): New helper.
21310 (build_scops): Start domwalk at entry block, sort generated
21313 2017-06-09 Richard Biener <rguenther@suse.de>
21315 PR middle-end/81007
21316 * ipa-polymorphic-call.c
21317 (ipa_polymorphic_call_context::restrict_to_inner_class):
21318 Skip FIELD_DECLs with error_mark_node type.
21319 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21322 2017-06-09 Martin Liska <mliska@suse.cz>
21324 * predict.c (struct branch_predictor): New struct.
21325 (test_prediction_value_range): New test.
21326 (predict_c_tests): New function.
21327 * selftest-run-tests.c (selftest::run_tests): Run the function.
21328 * selftest.h: Declare new tests.
21330 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21333 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21334 gen_add3_insn did not fail.
21335 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21336 r0, construct that number in a temporary reg and add that reg to r0.
21337 If asked to put the result in r0 as well, fail.
21339 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21342 for early expansion of vec_eqv.
21344 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21346 PR middle-end/81005
21347 * ubsan.c (instrument_null): Avoid pointless code temporary.
21348 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21350 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21354 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21357 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21360 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21362 * predict.c (unlikely_executed_edge_p): Move ahead.
21363 (probably_never_executed_edge_p): Use it.
21365 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21367 PR middle-end/79988
21368 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21369 gimple_call_builtin_p call.
21371 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21373 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21374 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21375 rtl_check_failed_type2, rtl_check_failed_code1,
21376 rtl_check_failed_code2, rtl_check_failed_code_mode,
21377 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21378 rtvec_check_failed_bounds, rtl_check_failed_flag,
21379 _fatal_insn_not_found, _fatal_insn): Likewise.
21380 * tree.h (tree_contains_struct_check_failed,
21381 tree_check_failed, tree_not_check_failed,
21382 tree_class_check_failed, tree_range_check_failed,
21383 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21384 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21385 tree_operand_check_failed, omp_clause_check_failed,
21386 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21389 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21391 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21392 flag_branch_probabilities.
21393 * ipa-inline.c (edge_badness): Likewise.
21394 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21395 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21396 * predict.c (maybe_hot_frequency_p): Likewise.
21397 (probably_never_executed): Likewise.
21398 * sched-ebb.c (schedule_ebbs): Likewise.
21399 * sched-rgn.c (find_single_block_region): Likewise.
21400 * tracer.c (tail_duplicate): Likewise.
21402 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21404 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21405 longer requires x_flag_profile_use.
21407 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21409 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21410 instead of flag_reorder_blocks_and_partition.
21411 * dbxout.c (dbxout_function_end): Likewise.
21412 * dwarf2out.c (gen_subprogram_die): Likewise.
21413 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21414 * hw-doloop.c (reorg_loops): Likewise.
21415 * varasm.c (assemble_start_function,
21416 assemble_end_function): Likewise.
21417 (decide_function_section): Do not check for
21418 flag_reorder_blocks_and_partition.
21420 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21422 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21424 (chkp_get_hard_register_fake_addr_expr): Ditto.
21425 (chkp_build_addr_expr): Add check for hard reg case.
21426 (chkp_parse_array_and_component_ref): Ditto.
21427 (chkp_find_bounds_1): Ditto.
21428 (chkp_process_stmt): Don't generate bounds store for
21431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21433 * predict.c (maybe_hot_bb_p): Do not check profile status.
21434 (maybe_hot_edge_p): Likewise.
21435 (probably_never_executed): Check for zero counts even if profile
21437 (unlikely_executed_edge_p): New function.
21438 (unlikely_executed_stmt_p): New function.
21439 (unlikely_executed_bb_p): New function.
21440 (set_even_probabilities): Use unlikely predicates.
21441 (combine_predictions_for_bb): Likewise.
21442 (predict_paths_for_bb): Likewise.
21443 (predict_paths_leading_to_edge): Likewise.
21444 (determine_unlikely_bbs): New function.
21445 (estimate_bb_frequencies): Use it.
21446 (compute_function_frequency): Use zero counts even if profile is
21448 * profile-count.h: Fix typo.
21450 2017-08-08 Julia Koval <julia.koval@intel.com>
21452 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21453 _mm512_mask_cvtsepi16_storeu_epi8,
21454 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21455 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21456 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21457 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21458 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21459 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21460 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21461 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21462 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21463 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21464 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21465 __builtin_ia32_pmovuswb256mem_mask,
21466 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21467 __builtin_ia32_pmovwb512mem_mask): New builtins.
21469 2017-08-08 Julia Koval <julia.koval@intel.com>
21471 PR target/73350,80862
21472 * config/i386/subst.md (round): Fix round pattern.
21473 * config/i386/i386.c (ix86_erase_embedded_rounding):
21474 Fix erasing rounding for the fixed pattern.
21476 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21478 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21480 2017-06-08 Martin Liska <mliska@suse.cz>
21482 PR gcov-profile/80911
21483 * gcov.c (block_info::block_info): New constructor.
21485 2017-06-07 Carl Love <cel@us.ibm.com>
21487 * config/rs6000/rs6000-c: The return type of the following
21488 built-in functions was implemented as int not long long. Fix sign
21489 of return value for the unsigned version of vec_mulo and vec_mule.
21490 vector unsigned long long vec_bperm (vector unsigned long long,
21491 vector unsigned char)
21492 vector signed long long vec_mule (vector signed int,
21494 vector unsigned long long vec_mule (vector unsigned int,
21495 vector unsigned int)
21496 vector signed long long vec_mulo (vector signed int,
21498 vector unsigned long long vec_mulo (vector unsigned int,
21499 vector unsigned int)
21500 * doc/extend.texi: Fix the documentation for the built-in
21503 2017-06-07 Carl Love <cel@us.ibm.com>
21506 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21509 2017-06-07 Carl Love <cel@us.ibm.com>
21511 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21512 support, Generate doublehv for signed int/float for BE case only.
21514 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21516 * doc/invoke.texi (mcx16): Rewrite.
21518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21520 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21521 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21522 *mov<mode>_softfloat, and an anonymous splitter): Use
21523 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21525 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21527 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21529 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21530 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21531 (rs6000_debug_reg_global): Adjust.
21532 (rs6000_init_hard_regno_mode_ok): Adjust.
21533 (rs6000_dbx_register_number): Adjust.
21534 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21535 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21536 Remove SPE_ACC and SPEFSCR.
21537 (REG_ALLOC_ORDER): Ditto.
21538 (FRAME_POINTER_REGNUM): Change to 111.
21539 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21540 (REG_CLASS_NAMES): Ditto.
21541 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21542 (REGISTER_NAMES): Ditto.
21543 (ADDITIONAL_REG_NAMES): Ditto.
21544 (rs6000_reg_names): Ditto.
21545 * config/rs6000/rs6000.md: Renumber some register number
21548 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21550 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21552 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21553 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21555 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21556 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21557 Delete the SPE high registers.
21558 (REG_ALLOC_ORDER): Ditto.
21559 (enum reg_class): Remove SPE_HIGH_REGS.
21560 (REG_CLASS_NAMES): Ditto.
21561 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21562 (REGISTER_NAMES): Ditto.
21563 (rs6000_reg_names): Ditto.
21564 * doc/tm.texi.in: Remove SPE as example.
21565 * doc/tm.texi: Regenerate.
21567 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21569 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21570 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21571 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21572 * config/rs6000/rs6000.md (type): Remove "brinc".
21574 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21576 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21577 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21578 * config/rs6000/linuxspe.h: Delete file.
21579 * config/rs6000/rs6000.md: Don't include spe.md.
21580 * config/rs6000/spe.h: Delete file.
21581 * config/rs6000/spe.md: Delete file.
21582 * config/rs6000/t-rs6000: Remove spe.md.
21584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21586 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21587 (reg_or_none500mem_operand): Delete.
21588 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21589 instead of reg_or_none500mem_operand.
21591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21594 handling of SPE flags.
21595 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21597 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21599 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21601 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21602 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21603 paired_divv2sf3): Similar.
21604 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21605 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21606 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21608 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21609 Rename the paired_* instruction patterns.
21610 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21612 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21613 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21614 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21615 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21616 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21617 PAIRED_VECTOR_MODE.
21618 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21619 (spe_func_has_64bit_regs_p): Delete.
21620 (spe_expand_predicate_builtin): Delete.
21621 (spe_expand_evsel_builtin): Delete.
21622 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21623 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21624 (invalid_e500_subreg): Delete.
21625 (rs6000_legitimize_address): Always force_reg op2 as well, for
21626 paired single memory accesses.
21627 (rs6000_member_type_forces_blk): Delete.
21628 (rs6000_spe_function_arg): Delete.
21629 (rs6000_expand_unop_builtin): Delete SPE handling.
21630 (rs6000_expand_binop_builtin): Ditto.
21631 (spe_expand_stv_builtin): Delete.
21632 (bdesc_2arg_spe): Delete.
21633 (spe_expand_builtin): Delete.
21634 (spe_expand_predicate_builtin): Delete.
21635 (spe_expand_evsel_builtin): Delete.
21636 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21637 (spe_init_builtins): Delete.
21638 (spe_func_has_64bit_regs_p): Delete.
21639 (savres_routine_name): Delete "info" parameter. Adjust callers.
21640 (rs6000_emit_stack_reset): Ditto.
21641 (rs6000_dwarf_register_span): Delete.
21642 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21643 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21644 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21646 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21648 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21649 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21650 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21651 mulv2sf3, divv2sf3): Delete expanders.
21653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21655 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21657 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21659 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21660 * config/rs6000/rs6000.c: Ditto.
21662 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21664 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21665 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21666 comparison_operator.
21668 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21670 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21671 * config/rs6000/rs6000.opt: Ditto.
21672 * config/rs6000/t-rtems: Ditto.
21674 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21676 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21677 TARGET_E500_SINGLE by 0, simplify.
21678 * config/rs6000/rs6000.c: Ditto.
21679 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21680 (spe_build_register_parallel): Delete.
21681 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21682 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21683 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21684 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21685 (E500_CONVERT): Delete.
21686 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21690 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21691 * config/rs6000/dfp.md: Ditto.
21692 (negdd2, *negdd2_fpr): Merge.
21693 (absdd2, *absdd2_fpr): Merge.
21694 (negtd2, *negtd2_fpr): Merge.
21695 (abstd2, *abstd2_fpr): Merge.
21696 * config/rs6000/e500.h: Delete file.
21697 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21698 TARGET_FPRS by 1 and simplify.
21699 * config/rs6000/rs6000-c.c: Ditto.
21700 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21701 TARGET_DF_SPE by 0.
21702 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21704 * config/rs6000/rs6000.md: Ditto.
21705 (floatdidf2, *floatdidf2_fpr): Merge.
21706 (move_from_CR_gt_bit): Delete.
21707 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21708 (E500_CR_IOR_COMPARE): Delete.
21709 (All patterns that require !TARGET_FPRS): Delete.
21710 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21712 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21714 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21716 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21718 * graphds.c (add_edge): Intitialize edge's attached data.
21719 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21720 pointer parameter. Call pointed function on each edge during
21721 graph traversing. Skip traversing the edge when the function
21723 (graphds_dfs, graphds_scc): Ditto.
21724 (for_each_edge): New parameter. Pass the new parameter to callback
21726 * graphds.h (skip_edge_callback): New function pointer type.
21727 (graphds_dfs, graphds_scc): New function pointer parameter.
21728 (graphds_edge_callback, for_each_edge): New parameter.
21730 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21732 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21733 out code checking if runtime alias check is possible to below ...
21734 Call the new function.
21735 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21736 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21738 2017-06-07 Marek Polacek <polacek@redhat.com>
21741 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21742 TYPE_OVERFLOW_WRAPS check.
21744 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21746 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21747 if versioning is required.
21748 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21749 peeling with the check for versioning.
21751 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21753 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21754 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21755 Set true to new parameter if new ssa variable is defined.
21756 (vect_gen_vector_loop_niters): Refactor. Set range information
21757 for the new vector loop bound variable.
21758 (vect_do_peeling): Ditto.
21760 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21762 * tree-affine.c (ssa.h): Include header file.
21763 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21764 has wrapping overflow behavior.
21766 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21768 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21770 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21772 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21773 (tree_to_aff_combination): ... here.
21775 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21777 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21778 reg_pressure model function.
21779 (ivopts_global_cost_for_size): Delete.
21780 (determine_set_costs, iv_ca_recount_cost): Call new model function
21781 ivopts_estimate_reg_pressure.
21783 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21785 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21786 expensive than udiv. Remove floating point cases from mod.
21788 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21790 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21791 Increase idiv cost.
21793 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21795 * config/aarch64/aarch64.md
21796 (copysignsf3): Fix mask generation.
21798 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21800 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21802 (class dump_manager): Add register_dumps method.
21803 * dumpfile.c: Include langhooks.h.
21804 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21805 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21806 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21807 (dump_manager::dump_register): Start with 512 entries instead of 32.
21808 (dump_manager::register_dumps): New method.
21809 * toplev.c (general_init): Instead of invoking register_dumps
21810 langhook, invoke register_dumps method on the dump manager.
21811 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21814 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21816 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21817 Say that # requires an associated define_split to exist, and that
21818 the define_split must be suitable for use after register allocation.
21820 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21822 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21823 (compute_outgoing_frequencies): Also initialize zero counts.
21824 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21825 around loops; preserve more of profile when nothing changes.
21827 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21829 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21831 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21832 * config/arm/arm-cpu-cdata.h: Regenerate.
21833 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21834 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21835 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21836 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21837 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21839 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21841 * config/arm/t-rmprofile: Likewise.
21842 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21844 2017-06-06 David S. Miller <davem@davemloft.net>
21847 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21848 blockage if function uses alloca.
21850 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21852 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21853 New "uid" fields to hold pretty-print IDs of group and ref.
21854 Memory references are now identified as <group_id>:<ref_id>
21855 instead of using [random] addresses.
21856 (dump_mem_details): Simplify, no functional change.
21857 (dump_mem_ref): Simplify and make output more concise.
21858 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21859 (find_or_create_group): Initialize group uid.
21860 (record_ref): Initialize ref uid. Improve debug output.
21861 (prune_group_by_reuse, should_issue_prefetch_p,)
21862 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21863 (mark_nontemporal_store, determine_loop_nest_reuse):
21864 Improve debug output.
21866 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21868 * dbgcnt.def (prefetch): New debug counter.
21869 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21870 (schedule_prefetches): Stop issueing prefetches if debug counter
21873 2017-06-06 Tom de Vries <tom@codesourcery.com>
21875 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21876 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21878 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21880 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21881 Use aarch64_reg_or_zero predicate for operand 4.
21882 (aarch64_compare_and_swap<mode> define_insn_and_split):
21883 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21884 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21886 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21888 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21889 (arm_compute_save_core_reg_mask): This.
21890 (thumb1_compute_save_reg_mask): Rename into ...
21891 (thumb1_compute_save_core_reg_mask): This.
21892 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21893 (arm_compute_frame_layout): Likewise.
21895 2017-06-06 Richard Biener <rguenther@suse.de>
21897 PR tree-optimization/80974
21898 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21899 keep or clear leaders SSA info.
21901 2017-06-06 Tom de Vries <tom@codesourcery.com>
21903 * config/nvptx/nvptx.c (split_mode_p): New function.
21904 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21906 2017-06-06 Tom de Vries <tom@codesourcery.com>
21908 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21910 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21913 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21916 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21918 * shrink-wrap.c (handle_simple_exit): Update profile.
21919 (try_shrink_wrapping): Upate profile.
21921 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21923 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21924 (tree_guess_outgoing_edge_probabilities): New.
21925 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21926 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21928 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21930 * ipa-split.c (split_function): Initialize return bb profile.
21932 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21934 * profile.c (compute_branch_probabilities): Also initialize
21935 EXIT_BLOCK profile.
21937 2017-06-06 Richard Biener <rguenther@suse.de>
21939 PR tree-optimization/80928
21940 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21941 (vect_analyze_loop_operations): Properly guard analysis for
21943 (vect_transform_loop): Likewise.
21944 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21945 (vect_model_induction_cost): Do not cost for pure SLP.
21946 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21947 of induction in inner loop vectorization.
21948 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21949 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21950 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21952 (vect_analyze_slp_cost_1): Cost induction.
21953 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21954 (vect_get_slp_vect_defs): Likewise.
21955 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21956 (vect_transform_stmt): Handle SLP reductions.
21957 * tree-vectorizer.h (vectorizable_induction): Adjust.
21959 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21961 * config/rs6000/rs6000.c (make_resolver_func): Update
21962 init_lowered_empty_function call.
21964 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21966 * doc/invoke.texi: Document the -fprofile-abs-path option.
21967 * common.opt (fprofile-abs-path): New option.
21968 * gcov-io.h (gcov_write_filename): Declare.
21969 * gcov-io.c (gcov_write_filename): New function.
21970 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21971 * profile.c (output_location): Likewise.
21973 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21975 * shring-wrap.c: Revert accidental commit.
21977 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21979 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21981 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21983 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21985 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21986 profile in callgraph edge.
21987 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21988 (apply_scale): Likewise.
21989 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21990 Also scale profile when inlining function with zero profile.
21991 (initialize_cfun): Update exit block profile even when it is zero.
21992 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21993 when profile is read.
21995 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21997 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21998 (CLONE_*): New constants to define the processors we can generate
21999 code for with the target_clone attribute.
22000 (rs6000_clone_map): New array to identify which clone processors
22001 the current program is running on.
22002 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22003 target_clone attribute.
22004 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22005 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22006 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22007 (cpu_expand_builtin): Add support for target_clone attribute.
22008 (rs6000_valid_attribute_p): Allow "default" attribute.
22009 (get_decl_name): New debug function to simplify printing the
22010 current function name in debugging statements.
22011 (rs6000_clone_priority): New functions to support the target_clone
22012 attribute, and be able to generate code to switch between ISA 2.05
22013 through ISA 3.0 (power6 through power9).
22014 (rs6000_compare_version_priority): Likewise.
22015 (rs6000_get_function_versions_dispatcher): Likewise.
22016 (make_resolver_func): Likewise.
22017 (add_condition_to_bb): Likewise.
22018 (dispatch_function_versions): Likewise.
22019 (rs6000_generate_version_dispatcher_body): Likewise.
22020 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22021 (fusion_gpr_load_p): Fix a spacing issue.
22022 * doc/extend.texi (Common Function Attributes): Document that the
22023 PowerPC supports the target_clone attribute.
22025 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22027 * config/arm/arm.h: explain F symbol found in description of ARM
22028 register allocation in its legend.
22030 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22032 * config/mips/frame-header-opt.c: Include profile-count.h.
22033 * config/riscv/riscv.c: Include profile-count.h
22035 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22037 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22039 (sm_set_flag_if_changed): Add bbs field.
22040 (execute_sm_if_changed_flag_set): Pass BBS.
22041 (execute_sm): Update.
22043 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22045 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22048 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22050 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22051 (peephole2): New peephole2 to emit the above.
22052 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22054 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22056 * config/aarch64/aarch64.c (define_peephole2 above
22057 *sub_<shift>_<mode>): New peephole.
22059 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22061 * config/i386/i386.c (make_resolver_func): Update.
22062 * Makefile.in: Add profile-count.h and profile-count.o
22063 * auto-profile.c (afdo_indirect_call): Update to new API.
22064 (afdo_set_bb_count): Update.
22065 (afdo_propagate_edge): Update.
22066 (afdo_propagate_circuit): Update.
22067 (afdo_calculate_branch_prob): Update.
22068 (afdo_annotate_cfg): Update.
22069 * basic-block.h: Include profile-count.h
22070 (struct edge_def): Turn count to profile_count.
22071 (struct basic_block_def): Likewie.
22072 (REG_BR_PROB_BASE): Move to profile-count.h
22073 (RDIV): Move to profile-count.h
22074 * bb-reorder.c (max_entry_count): Turn to profile_count.
22075 (find_traces): Update.
22076 (rotate_loop):Update.
22077 (connect_traces):Update.
22078 (sanitize_hot_paths):Update.
22079 * bt-load.c (migrate_btr_defs): Update.
22080 * cfg.c (RDIV): Remove.
22081 (init_flow): Use alloc_block.
22082 (alloc_block): Uninitialize count.
22083 (unchecked_make_edge): Uninitialize count.
22084 (check_bb_profile): Update.
22085 (dump_edge_info): Update.
22086 (dump_bb_info): Update.
22087 (update_bb_profile_for_threading): Update.
22088 (scale_bbs_frequencies_int): Update.
22089 (scale_bbs_frequencies_gcov_type): Update.
22090 (scale_bbs_frequencies_profile_count): New.
22091 * cfg.h (update_bb_profile_for_threading): Update.
22092 (scale_bbs_frequencies_profile_count): Declare.
22093 * cfgbuild.c (compute_outgoing_frequencies): Update.
22094 (find_many_sub_basic_blocks): Update.
22095 * cfgcleanup.c (try_forward_edges): Update.
22096 (try_crossjump_to_edge): Update.
22097 * cfgexpand.c (expand_gimple_tailcall): Update.
22098 (construct_exit_block): Update.
22099 * cfghooks.c (verify_flow_info): Update.
22100 (dump_bb_for_graph): Update.
22101 (split_edge): Update.
22102 (make_forwarder_block): Update.
22103 (duplicate_block): Update.
22104 (account_profile_record): Update.
22105 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22106 (get_estimated_loop_iterations): Update.
22107 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22108 (single_likely_exit): Update.
22109 * cfgloopmanip.c (scale_loop_profile): Update.
22111 (set_zero_probability): Update.
22112 (lv_adjust_loop_entry_edge): Update.
22113 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22114 (purge_dead_edges): Update.
22115 (rtl_account_profile_record): Update.
22116 * cgraph.c (cgraph_node::create): Uninitialize count.
22117 (symbol_table::create_edge): Uninitialize count.
22118 (cgraph_update_edges_for_call_stmt_node): Update.
22119 (cgraph_edge::dump_edge_flags): Update.
22120 (cgraph_node::dump): Update.
22121 (cgraph_edge::maybe_hot_p): Update.
22122 * cgraph.h: Include profile-count.h
22123 (create_clone), create_edge, create_indirect_edge): Update.
22124 (cgraph_node): Turn count to profile_count.
22125 (cgraph_edge0: Likewise.
22126 (make_speculative, clone): Update.
22127 (create_edge): Update.
22128 (init_lowered_empty_function): Update.
22129 * cgraphclones.c (cgraph_edge::clone): Update.
22130 (duplicate_thunk_for_node): Update.
22131 (cgraph_node::create_clone): Update.
22132 * cgraphunit.c (cgraph_node::analyze): Update.
22133 (cgraph_node::expand_thunk): Update.
22134 * final.c (dump_basic_block_info): Update.
22135 * gimple-streamer-in.c (input_bb): Update.
22136 * gimple-streamer-out.c (output_bb): Update.
22137 * graphite.c (print_global_statistics): Update.
22138 (print_graphite_scop_statistics): Update.
22139 * hsa-brig.c: Include basic-block.h.
22140 * hsa-dump.c: Include basic-block.h.
22141 * hsa-gen.c (T sum_slice): Update.
22142 (convert_switch_statements):Update.
22143 * hsa-regalloc.c: Include basic-block.h.
22144 * ipa-chkp.c (chkp_produce_thunks): Update.
22145 * ipa-cp.c (struct caller_statistics): Update.
22146 (init_caller_stats): Update.
22147 (gather_caller_stats): Update.
22148 (ipcp_cloning_candidate_p): Update.
22149 (good_cloning_opportunity_p): Update.
22150 (get_info_about_necessary_edges): Update.
22151 (dump_profile_updates): Update.
22152 (update_profiling_info): Update.
22153 (update_specialized_profile): Update.
22154 (perhaps_add_new_callers): Update.
22155 (decide_about_value): Update.
22156 (ipa_cp_c_finalize): Update.
22157 * ipa-devirt.c (struct odr_type_warn_count): Update.
22158 (struct decl_warn_count): Update.
22159 (struct final_warning_record): Update.
22160 (possible_polymorphic_call_targets): Update.
22161 (ipa_devirt): Update.
22162 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22163 * ipa-icf.c (sem_function::merge): Update.
22164 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22165 * ipa-inline.c (compute_uninlined_call_time): Update.
22166 (compute_inlined_call_time): Update.
22167 (want_inline_small_function_p): Update.
22168 (want_inline_self_recursive_call_p): Update.
22169 (edge_badness): Update.
22170 (lookup_recursive_calls): Update.
22171 (recursive_inlining): Update.
22172 (inline_small_functions): Update.
22173 (dump_overall_stats): Update.
22174 (dump_inline_stats): Update.
22175 * ipa-profile.c (ipa_profile_generate_summary): Update.
22176 (ipa_propagate_frequency): Update.
22177 (ipa_profile): Update.
22178 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22179 * ipa-utils.c (ipa_merge_profiles): Update.
22180 * loop-doloop.c (doloop_modify): Update.
22181 * loop-unroll.c (report_unroll): Update.
22182 (unroll_loop_runtime_iterations): Update.
22183 * lto-cgraph.c (lto_output_edge): Update.
22184 (lto_output_node): Update.
22185 (input_node): Update.
22186 (input_edge): Update.
22187 (merge_profile_summaries): Update.
22188 * lto-streamer-in.c (input_cfg): Update.
22189 * lto-streamer-out.c (output_cfg): Update.
22190 * mcf.c (create_fixup_graph): Update.
22191 (adjust_cfg_counts): Update.
22192 (sum_edge_counts): Update.
22193 * modulo-sched.c (sms_schedule): Update.
22194 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22195 * predict.c (maybe_hot_count_p): Update.
22196 (probably_never_executed): Update.
22197 (dump_prediction): Update.
22198 (combine_predictions_for_bb): Update.
22199 (propagate_freq): Update.
22200 (handle_missing_profiles): Update.
22201 (counts_to_freqs): Update.
22202 (rebuild_frequencies): Update.
22203 (force_edge_cold): Update.
22204 * predict.h: Include profile-count.h
22205 (maybe_hot_count_p, counts_to_freqs): UPdate.
22206 * print-rtl-function.c: Do not include cfg.h
22207 * print-rtl.c: Include basic-block.h
22208 * profile-count.c: New file.
22209 * profile-count.h: New file.
22210 * profile.c (is_edge_inconsistent): Update.
22211 (correct_negative_edge_counts): Update.
22212 (is_inconsistent): Update.
22213 (set_bb_counts): Update.
22214 (read_profile_edge_counts): Update.
22215 (compute_frequency_overlap): Update.
22216 (compute_branch_probabilities): Update; Initialize and deinitialize
22218 (branch_prob): Update.
22219 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22220 (edge_gcov_count): New.
22221 (bb_gcov_count): New.
22222 * shrink-wrap.c (try_shrink_wrapping): Update.
22223 * tracer.c (better_p): Update.
22224 * trans-mem.c (expand_transaction): Update.
22225 (ipa_tm_insert_irr_call): Update.
22226 (ipa_tm_insert_gettmclone_call): Update.
22227 * tree-call-cdce.c: Update.
22228 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22229 (gimple_duplicate_sese_tail): Update.
22230 (gimple_account_profile_record): Update.
22231 (execute_fixup_cfg): Update.
22232 * tree-inline.c (copy_bb): Update.
22233 (copy_edges_for_bb): Update.
22234 (initialize_cfun): Update.
22235 (freqs_to_counts): Update.
22236 (copy_cfg_body): Update.
22237 (expand_call_inline): Update.
22238 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22239 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22240 (try_unroll_loop_completely): Update.
22241 (try_peel_loop): Update.
22242 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22243 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22244 * tree-ssa-loop-split.c (connect_loops): Update.
22245 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22246 * tree-ssa-reassoc.c (branch_fixup): Update.
22247 * tree-ssa-tail-merge.c (replace_block_by): Update.
22248 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22249 (compute_path_counts): Update.
22250 (update_profile): Update.
22251 (recompute_probabilities): Update.
22252 (update_joiner_offpath_counts): Update.
22253 (estimated_freqs_path): Update.
22254 (freqs_to_counts_path): Update.
22255 (clear_counts_path): Update.
22256 (ssa_fix_duplicate_block_edges): Update.
22257 (duplicate_thread_path): Update.
22258 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22259 (struct switch_conv_info): Update.
22260 * tree-tailcall.c (decrease_profile): Update.
22261 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22262 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22263 * value-prof.c (check_counter): Update.
22264 (gimple_divmod_fixed_value): Update.
22265 (gimple_mod_pow2): Update.
22266 (gimple_mod_subtract): Update.
22267 (gimple_ic_transform): Update.
22268 (gimple_stringop_fixed_value): Update.
22269 * value-prof.h (gimple_ic): Update.
22271 2017-06-02 Carl Love <cel@us.ibm.com>
22273 * config/rs6000/rs6000-c: Add support for built-in functions
22274 vector double vec_doublee (vector signed int);
22275 vector double vec_doublee (vector unsigned int);
22276 vector double vec_doublee (vector float);
22277 vector double vec_doubleh (vector signed int);
22278 vector double vec_doubleh (vector unsigned int);
22279 vector double vec_doubleh (vector float);
22280 vector double vec_doublel (vector signed int);
22281 vector double vec_doublel (vector unsigned int);
22282 vector double vec_doublel (vector float);
22283 vector double vec_doubleo (vector signed int);
22284 vector double vec_doubleo (vector unsigned int);
22285 vector double vec_doubleo (vector float);.
22286 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22287 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22289 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22290 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22291 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22293 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22294 vec_doublel, vec_doubleh.
22295 * doc/extend.texi: Update the built-in documentation file for the
22296 new built-in functions.
22298 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22301 * ipa-inline-analysis.c (free_growth_caches): Set
22302 edge_removal_hook_holder to NULL after removing it.
22304 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22306 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22307 comparision with zero.
22309 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22310 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22311 for early expansion of vec_min and vec_max builtins.
22312 (builtin_function_type): Add min/max unsigned variants to those
22313 identified as having unsigned arguments.
22315 2017-06-02 Olivier Hainque <hainque@adacore.com>
22317 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22319 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22321 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22322 Use VALL_F16 iterator rather than VALL.
22324 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22326 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22327 Emit CBNZ inside loop when doing a strong exchange and comparing
22328 against zero. Generate the CC flags after the loop.
22330 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22332 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22333 (dl_section_ref): New.
22334 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22335 On AIX, append an expression to subtract the size of the
22336 section length to dl_section_ref.
22338 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22340 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22341 for early expansion of vector absolute builtins.
22343 2017-06-02 Richard Biener <rguenther@suse.de>
22345 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22346 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22348 2017-06-02 Richard Biener <rguenther@suse.de>
22350 PR tree-optimization/80948
22351 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22352 stmt order as well.
22354 2017-06-02 Richard Biener <rguenther@suse.de>
22356 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22358 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22359 for inductions as relevant.
22361 2017-06-02 Richard Biener <rguenther@suse.de>
22363 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22364 (vectorizable_induction): ... this. Remove dead code.
22366 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22368 * builtins. (expand_builtin_alloca): Remove second parameter and
22369 infer its value from the first parameter instead.
22370 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22372 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22374 PR rtl-optimization/80903
22375 * loop-doloop.c (add_test): Unshare sequence.
22377 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22379 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22381 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22383 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22385 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22386 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22387 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22388 (xlogue_layout::get_stub_rtx): Make static.
22389 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22390 (xlogue_layout::compute_stub_managed_regs): Rename to...
22391 (xlogue_layout::count_stub_managed_regs): ...this.
22392 (xlogue_layout::is_stub_managed_reg): New function.
22393 (xlogue_layout::m_stub_names): Rename to...
22394 (xlogue_layout::s_stub_names): ...this, make static.
22395 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22396 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22397 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22398 xlogue_layout::s_stub_names): Instantiate statics.
22399 (stub_managed_regs): Remove.
22400 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22401 (disable_call_ms2sysv_xlogues): Rename to...
22402 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22403 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22405 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22406 change after reload_completed.
22407 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22409 (ix86_expand_prologue): Likewise.
22410 (ix86_expand_epilogue): Likewise.
22411 (ix86_expand_split_stack_prologue): Likewise.
22412 (ix86_compute_frame_layout): Remove frame parameter ...
22413 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22414 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22416 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22417 (ix86_frame): Move from here ...
22418 * config/i386/i386.h (ix86_frame): ... to here.
22419 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22420 complete ix86_frame data structure instead. Remove some_ld_name.
22422 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22424 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22425 symbols that hold a DECL_VALUE_EXPR.
22427 2017-06-01 Martin Jambor <mjambor@suse.cz>
22429 PR tree-optimization/80898
22430 * tree-sra.c (process_subtree_disqualification): Removed.
22431 (disqualify_candidate): Do not acll
22432 process_subtree_disqualification.
22433 (subtree_mark_written_and_enqueue): New function.
22434 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22435 RHS has been disqualified and re-queue LHS if necessary. Apart
22436 from that, ignore disqualified RHS.
22438 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22440 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22441 address fetch for z10 or later.
22443 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22445 * config/arc/arc.md (tst_movb): Add guard when splitting.
22447 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22449 * config/arc/arc.c (arc_can_eliminate): Test against
22450 arc_frame_pointer_needed.
22452 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22454 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22455 to prevent store reordering.
22456 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22457 (type): Add block type.
22458 (stack_tie): Define special instruction to be used in
22461 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22463 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22464 constraint. It is not valid for the pattern.
22465 (noncommutative_binary_comparison): Likewise.
22467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22469 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22474 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22475 be used by the reg-alloc.
22477 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22479 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22480 reg-alloc when having mul64 or mul32x16 instructions.
22481 (mulsidi3): Likewise.
22482 (umulsidi3): Likewise.
22483 (mulsi32x16): New pattern.
22484 (mulsi64): Likewise.
22485 (mulsidi64): Likewise.
22486 (umulsidi64): Likewise.
22487 (MUL32x16_REG): Define.
22488 (mul64_600): Use MUL32x16_REG.
22489 (mac64_600): Likewise.
22490 (umul64_600): Likewise.
22491 (umac64_600): Likewise.
22493 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22495 * config/arc/arc.md (mulsi3_700): Make it commutative.
22497 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22499 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22501 (*sign_extendsidi2_insn): Likewise for movstosw.
22503 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22505 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22506 the type of the input discriminant value. Convert the
22507 discriminant value of signedness vary.
22509 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22511 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22512 Add to -Wall section.
22514 2017-06-01 Richard Biener <rguenther@suse.de>
22516 PR middle-end/66313
22517 * fold-const.c (fold_plusminus_mult_expr): If the factored
22518 factor may be zero use a wrapping type for the inner operation.
22519 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22520 and handle moved defs.
22521 (process_assignment): Properly guard the unary op case. Return a
22522 tri-state indicating that moving the stmt before the call may allow
22523 to continue. Pass through to_move.
22524 (find_tail_calls): Handle moving unrelated defs before
22527 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22530 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22531 splitter result in the canonical way.
22533 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22535 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22536 also for 32bit target. Update insn attributes.
22537 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22539 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22541 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22542 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22543 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22544 (_mm_maskz_min_ss): New intrinsics.
22546 2017-05-31 Martin Liska <mliska@suse.cz>
22548 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22549 Change comment style to one we normally use.
22550 (vectorizable_reduction): Likewise.
22551 (vectorizable_induction): Likewise.
22552 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22553 (vectorizable_call): Likewise.
22554 (vectorizable_simd_clone_call): Likewise.
22555 (vectorizable_conversion): Likewise.
22556 (vectorizable_assignment): Likewise.
22557 (vectorizable_shift): Likewise.
22558 (vectorizable_operation): Likewise.
22559 (vectorizable_store): Likewise.
22560 (vectorizable_load): Likewise.
22561 * tree-vectorizer.h: Likewise.
22563 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22565 * passes.c (emergency_dump_function): New.
22566 * tree-pass.h (emergency_dump_function): Declare.
22567 * plugin.c (plugins_internal_error_function): Remove.
22568 * plugin.h (plugins_internal_error_function): Remove declaration.
22569 * toplev.c (internal_error_function): New static function. Use it...
22570 (general_init): ...here.
22572 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22574 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22575 (arc_rtx_costs): Add costs for new patterns.
22576 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22577 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22579 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22581 * tree-ssa-strlen.c (get_next_strinfo): New function.
22582 (get_stridx_plus_constant): Use it.
22583 (zero_length_string): Likewise.
22584 (adjust_related_strinfos): Likewise.
22585 (adjust_last_stmt): Likewise.
22587 2017-05-31 Richard Biener <rguenther@suse.de>
22590 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22591 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22593 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22595 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22596 loop_vinfo argument and use of dependence distance vectors.
22597 Check instead whether the two references differ only in their
22598 initial value and assume that they have the same alignment if the
22599 difference is a multiple of the vector alignment.
22600 (vect_analyze_data_refs_alignment): Update call accordingly.
22602 2017-05-31 Martin Liska <mliska@suse.cz>
22605 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22607 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22609 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22610 (create_intersect_range_checks): Move from ...
22611 * tree-data-ref.c (create_intersect_range_checks_index)
22612 (create_intersect_range_checks): ... to here.
22613 (create_runtime_alias_checks): New function factored from ...
22614 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22615 here. Call above function.
22616 * tree-data-ref.h (create_runtime_alias_checks): New function.
22618 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22620 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22621 segment length for dr_b and compute it in wide_int.
22623 2017-05-31 Richard Biener <rguenther@suse.de>
22625 PR tree-optimization/80906
22626 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22627 and pass through iv_map.
22628 (copy_bb_and_scalar_dependences): Adjust.
22629 (translate_pending_phi_nodes): Likewise.
22630 (copy_loop_close_phi_args): Handle code-generating IVs instead
22633 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22635 * diagnostic-color.c (color_dict): Add "type-diff".
22636 (parse_gcc_colors): Update comment.
22637 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22638 -fdiagnostics-show-template-tree and -fno-elide-type.
22639 (GCC_COLORS): Add type-diff to example.
22641 (-fdiagnostics-show-template-tree): New.
22642 (-fno-elide-type): New.
22643 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22644 the pp_format_decoder callback. Call any m_format_postprocessor's
22646 (pretty_printer::pretty_printer): Initialize
22647 m_format_postprocessor.
22648 (pretty_printer::~pretty_printer): Delete any
22649 m_format_postprocessor.
22650 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22651 (class format_postprocessor): New class.
22652 (struct pretty_printer::format_decoder): Document the new parameters.
22653 (struct pretty_printer::m_format_postprocessor): New field.
22654 * tree-diagnostic.c (default_tree_printer): Update for new
22655 bool and const char ** params.
22656 * tree-diagnostic.h (default_tree_printer): Likewise.
22658 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22660 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22661 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22662 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22663 rs6000_gen_cell_microcode code.
22664 (rs6000_final_prescan_insn): Delete.
22665 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22666 "warn-cell-microcode" entries.
22667 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22668 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22669 throughout. Change cc_reg_not_micro_cr0_operand to
22670 cc_reg_not_cr0_operand throughout.
22671 (*extendhi<mode>2_noload): Delete.
22672 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22673 (mwarn-cell-microcode): Delete.
22674 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22675 -mgen-cell-microcode and -mwarn-cell-microcode.
22677 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22680 * config/i386/constraints.md (Yd): New constraint.
22682 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22683 and (?Yd, r) alternatives. Update insn attributes.
22684 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22685 and (?*Yd, r) alternatives. Update insn attributes.
22686 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22688 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22690 * gimplify.c (gimplify_modify_expr): Don't create a
22691 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22694 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22696 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22698 2017-05-30 Richard Biener <rguenther@suse.de>
22700 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22701 and reduc_def fields.
22702 (STMT_VINFO_REDUC_TYPE): New define.
22703 (STMT_VINFO_REDUC_DEF): Likewise.
22704 (vect_force_simple_reduction): Adjust prototype.
22705 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22706 (vect_is_simple_reduction): Remove check_reduction argument.
22707 (vect_force_simple_reduction): Adjust and set
22708 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22709 (vectorizable_reduction): Do not re-do reduction analysis
22710 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22711 * tree-parloops.c (gather_scalar_reductions): Adjust.
22713 2017-05-30 Richard Biener <rguenther@suse.de>
22715 PR middle-end/80901
22716 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22719 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22721 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22722 Introduce unknown_misalignment parameter and remove vf.
22723 (vect_peeling_hash_get_lowest_cost):
22724 Pass unknown_misalignment parameter.
22725 (vect_enhance_data_refs_alignment):
22726 Fix unsupportable data ref treatment.
22728 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22730 * tree-vect-data-refs.c (vect_get_data_access_cost):
22731 Workaround for SLP handling.
22732 (vect_enhance_data_refs_alignment):
22733 Compute costs for doing no peeling at all, compare to the best
22734 peeling costs so far and avoid peeling if cheaper.
22736 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22738 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22739 Return peeling info and set costs to zero for unlimited cost
22741 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22742 with unknown misalignment. Compute and costs for unknown
22743 misalignment, compare them to the costs for known misalignment
22744 and choose the cheapest for peeling.
22746 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22748 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22749 (vect_get_peeling_costs_all_drs): Create function.
22750 (vect_peeling_hash_get_lowest_cost):
22751 Use vect_get_peeling_costs_all_drs.
22752 (vect_peeling_supportable): Create function.
22753 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22755 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22757 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22758 DR_HAS_NEGATIVE_STEP.
22759 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22760 (vect_enhance_data_refs_alignment): Use.
22761 (vect_duplicate_ssa_name_ptr_info): Use.
22762 * tree-vectorizer.h (dr_misalignment): Use.
22763 (known_alignment_for_access_p): Use.
22765 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22768 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22770 (has_section_name): New function.
22772 2017-05-30 Martin Liska <mliska@suse.cz>
22775 * auto-profile.c (get_function_decl_from_block): Fix
22778 2017-05-30 Richard Biener <rguenther@suse.de>
22780 PR middle-end/80876
22781 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22783 2017-05-30 Martin Liska <mliska@suse.cz>
22785 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22786 * dumpfile.h (struct dump_file_info): Remove ctors.
22788 2017-05-30 Martin Liska <mliska@suse.cz>
22790 * predict.def: Fix GNU coding style.
22792 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22794 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22795 Mark 'to' argument with ATTRIBUTE_UNUSED.
22797 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22799 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22800 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22801 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22804 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22806 * doc/install.texi (Options specification): Restore entry of
22807 --enable-sjlj-exceptions.
22809 2017-05-27 Michael Eager <eager@eagercon.com>
22812 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22814 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22816 * config/microblaze/microblaze.h
22817 (FIXED_REGISTERS): Update in macro.
22818 (CALL_USED_REGISTERS): Update in macro.
22820 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22822 * doc/install.texi: Add links to macOS binary distributions.
22824 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22828 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22830 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22832 2017-05-26 Martin Liska <mliska@suse.cz>
22834 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22836 2017-05-26 Martin Liska <mliska@suse.cz>
22838 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22839 always leading ';; '.
22840 (dump_bb_info): Likewise.
22841 (brief_dump_cfg): Likewise.
22842 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22843 * dumpfile.c: Remove usage of TDF_VERBOSE.
22844 * dumpfile.h (enum dump_kind): Likewise.
22845 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22846 * print-tree.c (debug_verbose): Remove.
22847 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22848 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22849 * tree-diagnostic.c (default_tree_printer): Replace
22850 TDF_DIAGNOSTIC with TDF_SLIM.
22852 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22854 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22855 in parameter loop, rather than loop_vinfo.
22856 (create_intersect_range_checks): Ditto.
22857 (vect_create_cond_for_alias_checks): Update call to above functions.
22859 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22861 PR tree-optimization/80815
22862 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22863 for merging runtime alias checks. Handle negative DR_STEPs.
22865 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22867 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22869 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22870 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22871 out code pruning runtime alias checks.
22872 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22873 factored out from above.
22874 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22876 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22878 (prune_runtime_alias_test_list): New decalaration.
22880 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22882 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22883 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22884 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22885 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22886 (operator==, comp_dr_with_seg_len_pair): Ditto.
22887 (vect_prune_runtime_alias_test_list): Ditto.
22889 2017-05-26 Martin Liska <mliska@suse.cz>
22892 * params.def: Bound partial-inlining-entry-probability param.
22894 2017-05-26 Marek Polacek <polacek@redhat.com>
22897 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22900 2017-05-26 Richard Biener <rguenther@suse.de>
22902 PR tree-optimization/80842
22903 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22906 2017-05-26 Richard Biener <rguenther@suse.de>
22908 PR tree-optimization/80844
22909 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22911 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22913 * doc/md.texi (Machine Constraints): Update x86 family
22914 machine constraints section to match 'config/i386/constraints.md'.
22916 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22918 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22920 2017-05-25 Nathan Sidwell <nathan@acm.org>
22922 * doc/invoke.texi (--enable-languages): Update documentation.
22924 2017-05-25 Martin Liska <mliska@suse.cz>
22926 * dumpfile.c: Add TDF_FOLDING.
22927 * dumpfile.h (enum dump_kind): Likewise.
22928 * genmatch.c (dt_simplify::gen_1): Use it.
22930 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22932 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22934 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22936 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22937 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22939 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22941 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22942 * match.pd (X == C): Rewrite it here.
22943 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22944 with_certain_nonzero_bits2): New predicates.
22945 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22947 2017-05-24 Nathan Sidwell <nathan@acm.org>
22949 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22952 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22955 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22957 * config/powerpcspe: New port. Files are copied from the rs6000
22958 port, with "rs6000" in filenames replaced by "powerpcspe".
22960 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22962 PR rtl-optimization/80754
22963 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22965 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22967 * config/sparc/sparc.md (length): Return the correct value for -mflat
22968 sibcalls to match output_sibcall.
22970 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22974 * config/rs6000/rs6000.c (struct machine_function): Add new field
22976 (rs6000_get_separate_components): Init that field, use it.
22977 (rs6000_components_for_bb): Use the field.
22979 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22981 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22983 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22985 PR middle-end/80823
22986 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22988 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22991 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22992 address against address_operand predicate.
22993 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22995 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22997 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22998 parameters passed indirectly.
23000 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23002 * config/i386/i386.md (*movdi_internal): Remove SSE4
23003 alternative 18 (?r, *v). Update insn attributes.
23004 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23005 Update insn attributes.
23006 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23007 Update insn attributes.
23008 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23009 alternative 1 (r, v). Remove isa attribute.
23010 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23011 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23012 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23014 2017-05-23 Tom de Vries <tom@codesourcery.com>
23016 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23019 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23021 * cgraphunit.c (symbol_table::process_new_functions): Update.
23022 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23023 (inline_generate_summary): Rename to ...
23024 (ipa_fn_summary_generate): ... this one.
23025 (inline_read_summary): Rename to ...
23026 (ipa_fn_summary_read): ... this one.
23027 (inline_write_summary): Rename to ...
23028 (ipa_fn_summary_write): ... this one.
23029 (inline_free_summary): Rename to ...
23030 (ipa_free_fn_summary): ... this one.
23031 (pass_data_local_fn_summary, pass_local_fn_summary,
23032 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23033 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23034 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23035 make_pass_ipa_fn_summary): New.
23036 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23037 inline_write_summary, inline_free_summary): Remove.
23038 (ipa_free_fn_summary) : New.
23039 * ipa-inline.c (ipa_inline): Update.
23040 (pass_ipa_inline): Do not generate summaries.
23041 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23043 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23044 and add pass_ipa_fn_summary.
23045 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23047 (make_pass_inline_parameters): Remove.
23049 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23051 * omp-low.c (struct omp_context): Remove "default_kind" member.
23054 * omp-offload.c (execute_oacc_device_lower): Remove the
23055 parallelism dimensions function attributes for unparallelized
23056 OpenACC kernels constructs.
23058 2017-05-23 Martin Liska <mliska@suse.cz>
23060 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23062 (cgraph_edge::make_speculative): Likewise.
23063 (cgraph_edge::resolve_speculation): Likewise.
23064 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23065 (cgraph_node::dump): Likewise.
23066 * cgraph.h: Likewise.
23067 * cgraphunit.c (analyze_functions): Likewise.
23068 (symbol_table::compile): Likewise.
23069 * ipa-cp.c (print_all_lattices): Likewise.
23070 (determine_versionability): Likewise.
23071 (initialize_node_lattices): Likewise.
23072 (ipcp_verify_propagated_values): Likewise.
23073 (estimate_local_effects): Likewise.
23074 (update_profiling_info): Likewise.
23075 (create_specialized_node): Likewise.
23076 (perhaps_add_new_callers): Likewise.
23077 (decide_about_value): Likewise.
23078 (decide_whether_version_node): Likewise.
23079 (identify_dead_nodes): Likewise.
23080 (ipcp_store_bits_results): Likewise.
23081 * ipa-devirt.c (dump_targets): Likewise.
23082 (ipa_devirt): Likewise.
23083 * ipa-icf.c (sem_item::dump): Likewise.
23084 (sem_function::equals): Likewise.
23085 (sem_variable::equals): Likewise.
23086 (sem_item_optimizer::read_section): Likewise.
23087 (sem_item_optimizer::execute): Likewise.
23088 (congruence_class::dump): Likewise.
23089 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23090 (dump_inline_summary): Likewise.
23091 (estimate_node_size_and_time): Likewise.
23092 (inline_analyze_function): Likewise.
23093 * ipa-inline-transform.c (inline_call): Likewise.
23094 * ipa-inline.c (report_inline_failed_reason): Likewise.
23095 (want_early_inline_function_p): Likewise.
23096 (edge_badness): Likewise.
23097 (update_edge_key): Likewise.
23098 (inline_small_functions): Likewise.
23099 * ipa-profile.c (ipa_profile): Likewise.
23100 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23101 (ipa_make_edge_direct_to_target): Likewise.
23102 (remove_described_reference): Likewise.
23103 (ipa_impossible_devirt_target): Likewise.
23104 (propagate_controlled_uses): Likewise.
23105 (ipa_print_node_params): Likewise.
23106 (ipcp_transform_function): Likewise.
23107 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23108 (propagate_pure_const): Likewise.
23109 * ipa-reference.c (generate_summary): Likewise.
23110 (read_write_all_from_decl): Likewise.
23111 (propagate): Likewise.
23112 (ipa_reference_read_optimization_summary): Likewise.
23113 * ipa-utils.c (ipa_merge_profiles): Likewise.
23114 * ipa.c (walk_polymorphic_call_targets): Likewise.
23115 (symbol_table::remove_unreachable_nodes): Likewise.
23116 (ipa_single_use): Likewise.
23117 * passes.c (execute_todo): Likewise.
23118 * predict.c (drop_profile): Likewise.
23119 * symtab.c (symtab_node::get_dump_name): New function.
23120 (symtab_node::dump_name): Likewise.
23121 (symtab_node::dump_asm_name): Likewise.
23122 (symtab_node::dump_references): Likewise.
23123 (symtab_node::dump_referring): Likewise.
23124 (symtab_node::dump_base): Likewise.
23125 (symtab_node::debug_symtab): Likewise.
23126 * tree-sra.c (convert_callers_for_node): Likewise.
23127 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23128 * value-prof.c (init_node_map): Likewise.
23130 2017-05-23 Martin Liska <mliska@suse.cz>
23132 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23133 and symtab_node::debug_symtab to symbol_table::debug.
23134 * cgraphunit.c (analyze_functions): Use the renamed function.
23135 (symbol_table::compile): Likewise.
23136 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23137 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23138 * passes.c (execute_todo): Likewise.
23139 * symtab.c (symbol_table::dump): New function.
23140 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23142 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23144 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23145 that nonconst implies exec.
23147 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23149 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23150 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23151 (inline_edge_summary_vec): Turn into ...
23152 (ipa_call_summaries): ... this one.
23153 (redirect_to_unreachable, edge_set_predicate,
23154 evaluate_properties_for_edge, inline_summary_alloc,
23155 reset_ipa_call_summary, reset_inline_summary,
23156 inline_summary_t::duplicate): Update.
23157 (inline_edge_duplication_hook): Turn to ...
23158 (ipa_call_summary_t::duplicate): ... this one.
23159 (inline_edge_removal_hook): Turn to ...
23160 (ipa_call_summary_t::remove): ... this one.
23161 (dump_inline_edge_summary): Turn to ...
23162 (dump_ipa_call_summary): ... this one.
23163 (estimate_function_body_sizes): Update.
23164 (inline_update_callee_summaries): Update.
23165 (remap_edge_change_prob): Update.
23166 (remap_edge_summaries): Update.
23167 (inline_merge_summary): Update.
23168 (do_estimate_edge_time): Update.
23169 (inline_generate_summary): Update.
23170 (inline_read_section): Update.
23171 (inline_read_summary): Update.
23172 (inline_free_summary): Update.
23173 * ipa-inline.c (can_inline_edge_p): Update.
23174 (compute_inlined_call_time): Update.
23175 (want_inline_small_function_p): Update.
23176 (edge_badness): Update.
23177 (early_inliner): Update.
23178 * ipa-inline.h (inline_edge_summary): Turn to ...
23179 (ipa_call_summary): ... this one.
23180 (ipa_call_summary_t): New class.
23181 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23182 (ipa_call_summaries): New.
23183 (inline_edge_summary): Remove.
23184 (estimate_edge_growth): Update.
23185 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23186 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23187 * ipa-split.c (execute_split_functions): Update.
23188 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23190 2017-05-23 Tom de Vries <tom@codesourcery.com>
23192 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23193 attributes): Document rdrand effective target.
23195 2017-05-23 Tom de Vries <tom@codesourcery.com>
23197 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23198 attributes): Sort alphabetically.
23200 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23202 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23204 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23207 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23208 V2DF/V2DI splat into two separate patterns, one that handles
23209 registers, and the other that only handles memory. Drop support
23210 for splatting from a GPR on ISA 2.07 and then splitting the
23211 splat into direct move and splat.
23212 (vsx_splat_<mode>_reg): Likewise.
23213 (vsx_splat_<mode>_mem): Likewise.
23215 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23217 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23219 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23221 PR middle-end/80809
23222 * omp-low.c (finish_taskreg_remap): New function.
23223 (finish_taskreg_scan): If unit size of ctx->record_type
23224 is non-constant, unshare the size expression and replace
23225 decls in it with possible outer var refs.
23227 PR middle-end/80809
23228 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23229 GOVD_SHARED rather than GOVD_PRIVATE with it.
23230 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23231 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23233 PR middle-end/80853
23234 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23235 as last argument to build_outer_var_ref for pointer bases of array
23236 section reductions.
23238 2017-05-19 Martin Sebor <msebor@redhat.com>
23240 * print-tree.c (print_node): Print DECL_READ_P flag.
23242 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23244 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23245 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23246 * cgraph.c: Likewise.
23247 * cgraphunit.c: Likewise.
23248 * gengtype.c: Likewise.
23249 * ipa-cp.c: Likewise.
23250 * ipa-devirt.c: Likewise.
23251 * ipa-icf.c: Likewise.
23252 * ipa-predicate.c: Likewise.
23253 * ipa-profile.c: Likewise.
23254 * ipa-prop.c: Likewise.
23255 * ipa-split.c: Likewise.
23257 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23258 edge_predicate_pool, dump_inline_hints,
23259 inline_summary::account_size_time, redirect_to_unreachable,
23260 edge_set_predicate, set_hint_predicate,
23261 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23262 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23263 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23264 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23265 ipa_call_summary_t::remove, initialize_growth_caches,
23266 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23267 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23268 mark_modified, unmodified_parm_1, unmodified_parm,
23269 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23270 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23271 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23272 will_be_nonconstant_predicate, record_modified_bb_info,
23273 get_minimal_bb, record_modified, param_change_prob,
23274 phi_result_unknown_predicate, predicate_for_phi_result,
23275 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23276 estimate_function_body_sizes, compute_inline_parameters,
23277 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23278 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23279 inline_update_callee_summaries, remap_edge_change_prob,
23280 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23281 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23282 inline_analyze_function, inline_summary_t::insert,
23283 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23284 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23285 inline_free_summary): Move to ipa-fnsummary.h
23286 (predicate_t): Remove.
23287 * ipa-fnsummary.c: New file.
23288 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23289 (enum inline_hints_vals, inline_hints, agg_position_info,
23290 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23291 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23292 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23293 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23294 inline_read_summary, inline_write_summary, inline_free_summary,
23295 inline_analyze_function, initialize_inline_failed,
23296 inline_merge_summary, inline_update_overall_summary,
23297 compute_inline_parameters): Move to ipa-fnsummary.h
23298 * ipa-fnsummary.h: New file.
23299 * ipa-inline-transform.h: Include ipa-inline.h.
23300 * ipa-inline.c: LIkewise.
23302 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23304 * ipa-inline.c (edge_badness): Use inlined_time instead of
23305 inline_summaries->get.
23307 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23309 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23311 2017-05-22 Nathan Sidwell <nathan@acm.org>
23313 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23314 (fdump-lang): Document 'raw' option.
23315 * dumpfile.h (TDI_tu): Delete.
23316 * dumpfile.c (dump_files): Remove translation-unit.
23317 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23319 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23321 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23322 command option from $(AWK) call.
23323 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23324 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23325 [FORMAT]: Remove handling of variable.
23326 * config/avr/t-multilib: Regenerate.
23328 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23330 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23332 (dump_inline_summary): Do not print self_time.
23333 (estimate_function_body_sizes): Do not set self_time.
23334 (compute_inline_parameters): Likewise.
23335 (inline_read_section, inline_write_summary): Do not stream self_time.
23336 * ipa-inline.h (inline_summary): Drop self_time.
23338 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23340 * ipa-inline-analysis.c (account_size_time): Rename to ...
23341 (inline_summary::account_size_time): ... this one.
23342 (reset_ipa_call_summary): Turn to ...
23343 (ipa_call_summary::reset): ... this one.
23344 (reset_inline_summary): Turn to ...
23345 (inline_summary::reset): ... this one.
23346 (inline_summary_t::remove): Update.
23347 (inline_summary_t::duplicate): Update.
23348 (ipa_call_summary_t::remove): Update.
23349 (dump_inline_summary): Update.
23350 (estimate_function_body_sizes): Update.
23351 (compute_inline_parameters): Update.
23352 (estimate_node_size_and_time): Update.
23353 (inline_merge_summary): Update.
23354 (inline_update_overall_summary): Update.
23355 (inline_read_section): Update.
23356 (inline_write_summary): Update.
23357 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23358 add account_size_time and reset member functions.
23359 (ipa_call_summary): Add reset function.
23360 * ipa-predicate.h (predicate::operator &): Constify.
23362 2017-05-22 Richard Biener <rguenther@suse.de>
23364 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23366 2017-05-19 Jason Merrill <jason@redhat.com>
23368 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23370 2017-05-19 Marek Polacek <polacek@redhat.com>
23373 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23374 TYPE_OVERFLOW_WRAPS checks.
23376 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23378 * tree-core.h (enum omp_clause_default_kind): Add
23379 "OMP_CLAUSE_DEFAULT_PRESENT".
23380 * tree-pretty-print.c (dump_omp_clause): Handle it.
23381 * gimplify.c (enum gimplify_omp_var_data): Add
23382 "GOVD_MAP_FORCE_PRESENT".
23383 (gimplify_adjust_omp_clauses_1): Map it to
23384 "GOMP_MAP_FORCE_PRESENT".
23385 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23387 * gimplify.c (oacc_default_clause): Clarify.
23389 2017-05-19 Nathan Sidwell <nathan@acm.org>
23391 LANG_HOOK_REGISTER_DUMPS
23392 * toplev.c (general_init): Call register dump lang hook.
23393 * doc/invoke.texi: Document -fdump-lang option family.
23394 * dumpfile.c (dump_files): Remove class dump here.
23395 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23396 * dumpfile.h (tree_dump_index): Remove TDI_class.
23397 * langhooks-def.h (lhd_register_dumps): Declare.
23398 (LANG_HOOKS_REGISTER_DUMPS): Define.
23399 (LANG_HOOKS_INITIALIZER): Add it.
23400 * langhooks.c (lhd_register_dumps): Define.
23401 * langhooks.h (struct lang_hooks): Add register_dumps.
23403 2017-05-19 Nathan Sidwell <nathan@acm.org>
23405 * context.h (context::set_passes): New.
23406 * context.c (context::context): Do not create pass manager.
23407 * toplev.c (general_init): Create pass manager here.
23409 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23411 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23412 use this splitter if two add or or instructions would also work for
23413 the constant we want to generate.
23415 2017-05-19 Richard Biener <rguenther@suse.de>
23418 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23419 predicate evaluation.
23421 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23423 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23425 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23428 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23430 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23431 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23432 (inline_edge_summary_vec): Turn into ...
23433 (ipa_call_summaries): ... this one.
23434 (redirect_to_unreachable, edge_set_predicate,
23435 evaluate_properties_for_edge, inline_summary_alloc,
23436 reset_ipa_call_summary, reset_inline_summary,
23437 inline_summary_t::duplicate): Update.
23438 (inline_edge_duplication_hook): Turn to ...
23439 (ipa_call_summary_t::duplicate): ... this one.
23440 (inline_edge_removal_hook): Turn to ...
23441 (ipa_call_summary_t::remove): ... this one.
23442 (dump_inline_edge_summary): Turn to ...
23443 (dump_ipa_call_summary): ... this one.
23444 (estimate_function_body_sizes): Update.
23445 (inline_update_callee_summaries): Update.
23446 (remap_edge_change_prob): Update.
23447 (remap_edge_summaries): Update.
23448 (inline_merge_summary): Update.
23449 (do_estimate_edge_time): Update.
23450 (inline_generate_summary): Update.
23451 (inline_read_section): Update.
23452 (inline_read_summary): Update.
23453 (inline_free_summary): Update.
23454 * ipa-inline.c (can_inline_edge_p): Update.
23455 (compute_inlined_call_time): Update.
23456 (want_inline_small_function_p): Update.
23457 (edge_badness): Update.
23458 (early_inliner): Update.
23459 * ipa-inline.h (inline_edge_summary): Turn to ...
23460 (ipa_call_summary): ... this one.
23461 (ipa_call_summary_t): New class.
23462 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23463 (ipa_call_summaries): New.
23464 (inline_edge_summary): Remove.
23465 (estimate_edge_growth): Update.
23466 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23467 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23468 * ipa-split.c (execute_split_functions): Update.
23469 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23471 2017-05-19 Richard Biener <rguenther@suse.de>
23473 PR middle-end/80764
23474 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23476 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23478 * config/rs6000/rs6000.c (struct machine_function): Add field
23479 fpr_is_wrapped_separately.
23480 (rs6000_get_separate_components): Use 64 components. Handle the
23481 new FPR components.
23482 (rs6000_components_for_bb): Handle the FPR components.
23483 (rs6000_emit_prologue_components): Handle the FPR components.
23484 (rs6000_emit_epilogue_components): Handle the FPR components.
23485 (rs6000_set_handled_components): Handle the FPR components.
23486 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23487 that are already separately shrink-wrapped.
23488 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23489 that are already separately shrink-wrapped.
23491 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23494 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23497 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23498 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23499 cases where the register allocator uses a move and an offsettable
23500 memory operation to/from a FPR register on ISA 2.06/2.07.
23501 (define_peephole2 for Altivec d-form store): Likewise.
23503 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23506 * config/i386/mmx.md (*mov<mode>_internal): Enable
23507 alternatives 11, 12, 13 and 14 also for 32bit targets.
23508 Remove alternatives 15, 16, 17 and 18.
23509 * config/i386/sse.md (vec_concatv2di): Change
23510 alternative (!x, *y) to (x, ?!*Yn).
23512 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23514 * dumpfile.h (enum dump_kind): Remove stray comma.
23516 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23518 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23519 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23520 predicate::num_conditions
23521 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23522 (CHANGED): turn into predicate::changed.
23523 (agg_position_info): Move to ipa-predicate.h
23524 (add_condition, predicate::add_clause, predicate::operator &=,
23525 predicate::or_with, predicate::evaluate, predicate::probability,
23526 dump_condition, dump_clause, predicate::dump,
23527 predicate::remap_after_duplication, predicate::remap_after_inlining,
23528 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23529 (evaluate_conditions_for_known_args): Update.
23530 (set_cond_stmt_execution_predicate): Update.
23531 * ipa-inline.h: Include ipa-predicate.h
23532 (condition, inline_param_summary, conditions, agg_position_info,
23533 predicate): Move to ipa-predicate.h
23534 * ipa-predicate.c: New file.
23535 * ipa-predicate.h: New file.
23537 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23539 * final.c (leaf_function_p): Check we are not in a sequence.
23541 2017-05-18 Martin Liska <mliska@suse.cz>
23543 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23544 * dumpfile.c (dump_register): Use new enum dump_kind.
23545 (get_dump_file_name): Likewise.
23546 (dump_enable_all): Likewise.
23547 (dump_switch_p_1): Likewise.
23548 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23549 * dumpfile.h (enum dump_kind): New enum type.
23550 (struct dump_file_info): Create constructor and
23551 format fields and comments.
23552 * passes.c (pass_manager::register_one_dump_file):
23554 * statistics.c (statistics_early_init): Likewise.
23555 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23556 TDF_TREE with TDF_SLIM.
23557 (gather_memory_references_ref): Likewise.
23559 2017-05-18 Martin Liska <mliska@suse.cz>
23561 * vec.h (struct vnull): Use it.
23563 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23565 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23566 (true_predicate, false_predicate, true_predicate_p,
23567 false_predicate_p): Remove.
23568 (single_cond_predicate, not_inlined_predicate): Turn to member function
23570 (add_condition): Update.
23571 (add_clause): Turn to...
23572 (predicate::add_clause): ... this one; update; allow passing NULL
23574 (and_predicates): Turn to ...
23575 (predicate::operator &=): ... this one.
23576 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23577 (or_predicates): Turn to ...
23578 (predicate::or_with): ... this one.
23579 (evaluate_predicate): Turn to ...
23580 (predicate::evaluate): ... this one.
23581 (predicate_probability): Turn to ...
23582 (predicate::probability): ... this one.
23583 (dump_condition): Update.
23584 (dump_predicate): Turn to ...
23585 (predicate::dump): ... this one.
23586 (account_size_time): Update.
23587 (edge_set_predicate): Update.
23588 (set_hint_predicate): UPdate.
23589 (evaluate_conditions_for_known_args): Update.
23590 (evaluate_properties_for_edge): Update.
23591 (remap_predicate_after_duplication): Turn to...
23592 (predicate::remap_after_duplication): ... this one.
23593 (remap_hint_predicate_after_duplication): Update.
23594 (inline_summary_t::duplicate): UPdate.
23595 (dump_inline_edge_summary): Update.
23596 (dump_inline_summary): Update.
23597 (set_cond_stmt_execution_predicate): Update.
23598 (set_switch_stmt_execution_predicate): Update.
23599 (compute_bb_predicates): Update.
23600 (will_be_nonconstant_expr_predicate): Update.
23601 (will_be_nonconstant_predicate): Update.
23602 (phi_result_unknown_predicate): Update.
23603 (predicate_for_phi_result): Update.
23604 (array_index_predicate): Update.
23605 (estimate_function_body_sizes): Update.
23606 (estimate_node_size_and_time): Update.
23607 (estimate_ipcp_clone_size_and_time): Update.
23608 (remap_predicate): Rename to ...
23609 (predicate::remap_after_inlining): ... this one.
23610 (remap_hint_predicate): Update.
23611 (inline_merge_summary): Update.
23612 (inline_update_overall_summary): Update.
23613 (estimate_size_after_inlining): Update.
23614 (read_predicate): Rename to ...
23615 (predicate::stream_in): ... this one.
23616 (read_inline_edge_summary): Update.
23617 (write_predicate): Rename to ...
23618 (predicate::stream_out): ... this one.
23619 (write_inline_edge_summary): Update.
23620 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23621 (clause_t): Turn to uint32_t
23622 (predicate): Turn to class; implement constructor and operators
23624 (size_time_entry): Update.
23625 (inline_summary): Update.
23626 (inline_edge_summary): Update.
23628 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23630 * fold-const.c (fold_binary_loc): Move transformation...
23631 * match.pd (C - X CMP X): ... here.
23633 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23635 * config/sparc/sparc.c (sparc_option_override): Set function
23636 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23637 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23639 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23641 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23643 2017-05-18 Marek Polacek <polacek@redhat.com>
23646 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23647 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23648 gimple_assign_load_p.
23650 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23652 PR middle-end/80692
23653 * real.c (do_compare): Give decimal_do_compare preference over
23654 comparing just the signs.
23656 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23658 * doc/md.texi (Canonicalization of Instructions): Describe the
23659 canonical form of instructions that inherently set a condition
23662 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23664 PR middle-end/80775
23665 * tree-cfg.c: Move deletion of unreachable case statements to after
23666 the merging of consecutive case labels.
23668 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23670 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23671 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23672 restoring of callee-saved registers.
23674 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23676 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23677 * config/visium/visium.c (single_set_and_flags): Likewise.
23678 * config/visium/visium.md (Substitutions): Likewise.
23680 2017-05-17 Martin Liska <mliska@suse.cz>
23682 * cfg.c: Introduce dump_flags_t type and
23683 use it instead of int type.
23685 * cfghooks.c: Likewise.
23686 * cfghooks.h (struct cfg_hooks): Likewise.
23687 * cfgrtl.c: Likewise.
23688 * cfgrtl.h: Likewise.
23689 * cgraph.c (cgraph_node::get_body): Likewise.
23690 * coretypes.h: Likewise.
23691 * domwalk.c: Likewise.
23692 * domwalk.h: Likewise.
23693 * dumpfile.c (struct dump_option_value_info): Likewise.
23694 (dump_enable_all): Likewise.
23695 (dump_switch_p_1): Likewise.
23696 (opt_info_switch_p): Likewise.
23697 * dumpfile.h (enum tree_dump_index): Likewise.
23698 (struct dump_file_info): Likewise.
23699 * genemit.c: Likewise.
23700 * generic-match-head.c: Likewise.
23701 * gengtype.c (open_base_files): Likewise.
23702 * gimple-pretty-print.c: Likewise.
23703 * gimple-pretty-print.h: Likewise.
23704 * graph.c (print_graph_cfg): Likewise.
23705 * graphite-scop-detection.c (dot_all_sese): Likewise.
23706 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23707 * loop-unroll.c (report_unroll): Likewise.
23708 * passes.c (pass_manager::register_one_dump_file): Likewise.
23709 * print-tree.c: Likewise.
23710 * statistics.c: Likewise.
23711 * tree-cfg.c: Likewise.
23712 * tree-cfg.h: Likewise.
23713 * tree-dfa.c: Likewise.
23714 * tree-dfa.h: Likewise.
23715 * tree-dump.c (dump_function): Likewise.
23716 * tree-dump.h (struct dump_info): Likewise.
23717 * tree-pretty-print.c: Likewise.
23718 * tree-pretty-print.h: Likewise.
23719 * tree-ssa-live.c: Likewise.
23720 * tree-ssa-live.h: Likewise.
23721 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23722 * tree-vect-loop.c: Likewise.
23723 * tree-vect-slp.c: Likewise.
23725 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23726 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23728 PR tree-optimization/80457
23729 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23730 of all arguments to a statement as scalar_to_vec operations.
23731 (vectorizable_call): Adjust call to vect_model_simple_cost for
23733 (vectorizable_conversion): Likewise.
23734 (vectorizable_assignment): Likewise.
23735 (vectorizable_shift): Likewise.
23736 (vectorizable_operation): Likewise.
23737 (vectorizable_comparison): Likewise.
23738 (vect_is_simple_cond): Record the def types for operands.
23739 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23740 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23741 for statement argument count.
23743 2017-05-16 Carl Love <cel@us.ibm.com>
23745 * config/rs6000/rs6000-c: Add support for built-in functions
23746 vector unsigned long long vec_bperm (vector unsigned long long,
23747 vector unsigned char)
23748 vector signed long long vec_mule (vector signed int,
23750 vector unsigned long long vec_mule (vector unsigned int,
23751 vector unsigned int)
23752 vector signed long long vec_mulo (vector signed int,
23754 vector unsigned long long vec_mulo (vector unsigned int,
23755 vector unsigned int)
23756 vector signed char vec_sldw (vector signed char,
23757 vector signed char,
23759 vector unsigned char vec_sldw (vector unsigned char,
23760 vector unsigned char,
23762 vector signed short vec_sldw (vector signed short,
23763 vector signed short,
23765 vector unsigned short vec_sldw (vector unsigned short,
23766 vector unsigned short,
23768 vector signed int vec_sldw (vector signed int,
23771 vector unsigned int vec_sldw (vector unsigned int,
23772 vector unsigned int,
23774 vector signed long long vec_sldw (vector signed long long,
23775 vector signed long long,
23777 vector unsigned long long vec_sldw (vector unsigned long long,
23778 vector unsigned long long,
23780 * config/rs6000/rs6000-c: Add support for built-in functions
23781 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23782 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23783 * doc/extend.texi: Update the built-in documentation for the
23784 new built-in functions.
23786 2017-05-16 Marek Polacek <polacek@redhat.com>
23790 * tree.c (save_expr): Don't fold the expression.
23792 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23794 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23795 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23796 and (?*y,m). Update insn attributes.
23798 2017-05-16 Martin Liska <mliska@suse.cz>
23800 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23801 flags argument of print_gimple_stmt, print_gimple_expr,
23802 print_generic_stmt and print_generic_expr.
23803 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23804 * coretypes.h: Likewise.
23805 * except.c (dump_eh_tree): Likewise.
23806 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23807 * gimple-pretty-print.h: Likewise.
23808 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23809 (backprop::push_to_worklist): Likewise.
23810 (backprop::pop_from_worklist): Likewise.
23811 (backprop::process_use): Likewise.
23812 (backprop::intersect_uses): Likewise.
23813 (note_replacement): Likewise.
23814 * gimple-ssa-store-merging.c
23815 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23816 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23817 (pass_store_merging::execute): Likewise.
23818 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23819 (ssa_base_cand_dump_callback): Likewise.
23820 (dump_incr_vec): Likewise.
23821 (replace_refs): Likewise.
23822 (replace_mult_candidate): Likewise.
23823 (create_add_on_incoming_edge): Likewise.
23824 (create_phi_basis): Likewise.
23825 (insert_initializers): Likewise.
23826 (all_phi_incrs_profitable): Likewise.
23827 (introduce_cast_before_cand): Likewise.
23828 (replace_one_candidate): Likewise.
23829 * gimplify.c (gimplify_expr): Likewise.
23830 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23831 (set_rename): Likewise.
23832 (rename_uses): Likewise.
23833 (copy_loop_phi_nodes): Likewise.
23834 (add_close_phis_to_merge_points): Likewise.
23835 (copy_loop_close_phi_args): Likewise.
23836 (copy_cond_phi_args): Likewise.
23837 (graphite_copy_stmts_from_block): Likewise.
23838 (translate_pending_phi_nodes): Likewise.
23839 * graphite-poly.c (print_pdr): Likewise.
23840 (dump_gbb_cases): Likewise.
23841 (dump_gbb_conditions): Likewise.
23842 (print_scop_params): Likewise.
23843 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23844 (build_cross_bb_scalars_use): Likewise.
23845 (gather_bbs::before_dom_children): Likewise.
23846 * hsa-dump.c (dump_hsa_immed): Likewise.
23847 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23848 (get_replacement_map): Likewise.
23849 * ipa-inline-analysis.c (dump_condition): Likewise.
23850 (estimate_function_body_sizes): Likewise.
23851 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23852 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23853 * ipa-prop.c (ipa_dump_param): Likewise.
23854 (ipa_print_node_jump_functions_for_edge): Likewise.
23855 (ipa_modify_call_arguments): Likewise.
23856 (ipa_modify_expr): Likewise.
23857 (ipa_dump_param_adjustments): Likewise.
23858 (ipa_dump_agg_replacement_values): Likewise.
23859 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23860 * ipa-pure-const.c (check_stmt): Likewise.
23861 (pass_nothrow::execute): Likewise.
23862 * ipa-split.c (execute_split_functions): Likewise.
23863 * omp-offload.c (dump_oacc_loop_part): Likewise.
23864 (dump_oacc_loop): Likewise.
23865 * trans-mem.c (tm_log_emit): Likewise.
23866 (tm_memopt_accumulate_memops): Likewise.
23867 (dump_tm_memopt_set): Likewise.
23868 (dump_tm_memopt_transform): Likewise.
23869 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23870 (print_loop): Likewise.
23871 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23872 (chkp_gather_checks_info): Likewise.
23873 (chkp_get_check_result): Likewise.
23874 (chkp_remove_check_if_pass): Likewise.
23875 (chkp_use_outer_bounds_if_possible): Likewise.
23876 (chkp_reduce_bounds_lifetime): Likewise.
23877 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23878 (chkp_mark_completed_bounds): Likewise.
23879 (chkp_register_incomplete_bounds): Likewise.
23880 (chkp_mark_invalid_bounds): Likewise.
23881 (chkp_maybe_copy_and_register_bounds): Likewise.
23882 (chkp_build_returned_bound): Likewise.
23883 (chkp_get_bound_for_parm): Likewise.
23884 (chkp_build_bndldx): Likewise.
23885 (chkp_get_bounds_by_definition): Likewise.
23886 (chkp_generate_extern_var_bounds): Likewise.
23887 (chkp_get_bounds_for_decl_addr): Likewise.
23888 * tree-chrec.c (chrec_apply): Likewise.
23889 * tree-data-ref.c (dump_data_reference): Likewise.
23890 (dump_subscript): Likewise.
23891 (dump_data_dependence_relation): Likewise.
23892 (analyze_overlapping_iterations): Likewise.
23893 * tree-inline.c (expand_call_inline): Likewise.
23894 (tree_function_versioning): Likewise.
23895 * tree-into-ssa.c (dump_defs_stack): Likewise.
23896 (dump_currdefs): Likewise.
23897 (dump_names_replaced_by): Likewise.
23898 (dump_update_ssa): Likewise.
23899 (update_ssa): Likewise.
23900 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23901 * tree-parloops.c (build_new_reduction): Likewise.
23902 (try_create_reduction_list): Likewise.
23903 (ref_conflicts_with_region): Likewise.
23904 (oacc_entry_exit_ok_1): Likewise.
23905 (oacc_entry_exit_single_gang): Likewise.
23906 * tree-pretty-print.h: Likewise.
23907 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23908 (get_scalar_evolution): Likewise.
23909 (add_to_evolution): Likewise.
23910 (get_loop_exit_condition): Likewise.
23911 (analyze_evolution_in_loop): Likewise.
23912 (analyze_initial_condition): Likewise.
23913 (analyze_scalar_evolution): Likewise.
23914 (instantiate_scev): Likewise.
23915 (number_of_latch_executions): Likewise.
23916 (gather_chrec_stats): Likewise.
23917 (final_value_replacement_loop): Likewise.
23918 (scev_const_prop): Likewise.
23919 * tree-sra.c (dump_access): Likewise.
23920 (disqualify_candidate): Likewise.
23921 (create_access): Likewise.
23922 (reject): Likewise.
23923 (maybe_add_sra_candidate): Likewise.
23924 (create_access_replacement): Likewise.
23925 (analyze_access_subtree): Likewise.
23926 (analyze_all_variable_accesses): Likewise.
23927 (sra_modify_assign): Likewise.
23928 (initialize_constant_pool_replacements): Likewise.
23929 (find_param_candidates): Likewise.
23930 (decide_one_param_reduction): Likewise.
23931 (replace_removed_params_ssa_names): Likewise.
23932 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23933 * tree-ssa-copy.c (dump_copy_of): Likewise.
23934 (copy_prop_visit_cond_stmt): Likewise.
23935 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23936 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23937 (record_equivalences_from_stmt): Likewise.
23938 * tree-ssa-dse.c (compute_trims): Likewise.
23939 (delete_dead_call): Likewise.
23940 (delete_dead_assignment): Likewise.
23941 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23942 (forward_propagate_into_cond): Likewise.
23943 (pass_forwprop::execute): Likewise.
23944 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23945 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23947 (move_computations_worker): Likewise.
23948 (execute_sm): Likewise.
23949 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23950 (remove_exits_and_undefined_stmts): Likewise.
23951 (remove_redundant_iv_tests): Likewise.
23952 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23953 (adjust_iv_update_pos): Likewise.
23954 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23955 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23956 (value_replacement): Likewise.
23957 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23958 * tree-ssa-pre.c (print_pre_expr): Likewise.
23959 (get_representative_for): Likewise.
23960 (create_expression_by_pieces): Likewise.
23961 (insert_into_preds_of_block): Likewise.
23962 (eliminate_insert): Likewise.
23963 (eliminate_dom_walker::before_dom_children): Likewise.
23964 (eliminate): Likewise.
23965 (remove_dead_inserted_code): Likewise.
23966 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23967 * tree-ssa-reassoc.c (get_rank): Likewise.
23968 (eliminate_duplicate_pair): Likewise.
23969 (eliminate_plus_minus_pair): Likewise.
23970 (eliminate_not_pairs): Likewise.
23971 (undistribute_ops_list): Likewise.
23972 (eliminate_redundant_comparison): Likewise.
23973 (update_range_test): Likewise.
23974 (optimize_range_tests_var_bound): Likewise.
23975 (optimize_vec_cond_expr): Likewise.
23976 (rewrite_expr_tree): Likewise.
23977 (rewrite_expr_tree_parallel): Likewise.
23978 (linearize_expr): Likewise.
23979 (break_up_subtract): Likewise.
23980 (linearize_expr_tree): Likewise.
23981 (attempt_builtin_powi): Likewise.
23982 (attempt_builtin_copysign): Likewise.
23983 (transform_stmt_to_copy): Likewise.
23984 (transform_stmt_to_multiply): Likewise.
23985 (dump_ops_vector): Likewise.
23986 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23987 (print_scc): Likewise.
23988 (set_ssa_val_to): Likewise.
23989 (visit_reference_op_store): Likewise.
23990 (visit_use): Likewise.
23991 (sccvn_dom_walker::before_dom_children): Likewise.
23992 (run_scc_vn): Likewise.
23993 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23995 (expr_hash_elt::print): Likewise.
23996 (const_and_copies::pop_to_marker): Likewise.
23997 (const_and_copies::record_const_or_copy_raw): Likewise.
23998 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23999 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24000 (dump_predicates): Likewise.
24001 (find_uninit_use): Likewise.
24002 (warn_uninitialized_phi): Likewise.
24003 (pass_late_warn_uninitialized::execute): Likewise.
24004 * tree-ssa.c (verify_vssa): Likewise.
24005 (verify_ssa): Likewise.
24006 (maybe_optimize_var): Likewise.
24007 * tree-vrp.c (dump_value_range): Likewise.
24008 (dump_all_value_ranges): Likewise.
24009 (dump_asserts_for): Likewise.
24010 (register_edge_assert_for_2): Likewise.
24011 (vrp_visit_cond_stmt): Likewise.
24012 (vrp_visit_switch_stmt): Likewise.
24013 (vrp_visit_stmt): Likewise.
24014 (vrp_visit_phi_node): Likewise.
24015 (simplify_cond_using_ranges_1): Likewise.
24016 (fold_predicate_in): Likewise.
24017 (evrp_dom_walker::before_dom_children): Likewise.
24018 (evrp_dom_walker::push_value_range): Likewise.
24019 (evrp_dom_walker::pop_value_range): Likewise.
24020 (execute_early_vrp): Likewise.
24022 2017-05-16 Richard Biener <rguenther@suse.de>
24024 * dwarf2out.c (loc_list_from_tree_1): Do not create
24025 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24027 2017-05-16 Richard Biener <rguenther@suse.de>
24029 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24031 (note_variable_value_in_expr): If we resolved the decl ref
24032 do not push to the stack.
24034 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24036 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24037 operations in fast-math mode.
24038 (vaddq_f16): Likewise.
24039 (vmul_f16): Likewise.
24040 (vmulq_f16): Likewise.
24041 (vsub_f16): Likewise.
24042 (vsubq_f16): Likewise.
24043 * config/arm/neon.md (add<mode>3): New.
24045 (fma:<VH:mode>3): New. Also remove outdated comment.
24048 2017-05-16 Martin Liska <mliska@suse.cz>
24052 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24053 (odr_types_equivalent_p): Likewise.
24055 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24057 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24059 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24062 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24063 non-interunit SSE move alternatives with '?'.
24064 (zero-extendsidi peephole2): New peephole to skip intermediate
24065 general register in SSE zero-extend sequence.
24067 2017-05-15 Jeff Law <law@redhat.com>
24069 * reorg.c (relax_delay_slots): Create a new variable to hold
24070 the temporary target rather than clobbering TARGET_LABEL.
24072 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24073 missing argument to extract_bit_field call.
24074 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24076 2017-05-15 Martin Liska <mliska@suse.cz>
24079 * gcc.c (process_command): Do not allow empty argument of -o option.
24081 2017-05-15 Renlin Li <renlin.li@arm.com>
24083 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24084 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24085 * config/aarch64/constraints.md (Usf): Add long call check.
24086 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24087 (call_value): Likewise.
24088 (sibcall): Likewise.
24089 (sibcall_value): Likewise.
24091 (call_value_insn): New.
24092 (sibcall_insn): Update rtx pattern.
24093 (sibcall_value_insn): Likewise.
24094 (call_internal): Remove.
24095 (call_value_internal): Likewise.
24096 (sibcall_internal): Likewise.
24097 (sibcall_value_internal): Likewise.
24098 (call_reg): Likewise.
24099 (call_symbol): Likewise.
24100 (call_value_reg): Likewise.
24101 (call_value_symbol): Likewise.
24103 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24106 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24108 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24110 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24111 compatible with CCGOCmode and with CCZmode.
24113 2017-05-14 Martin Sebor <msebor@redhat.com>
24115 PR middle-end/77671
24116 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24117 (gimple_fold_builtin_snprintf): Same.
24118 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24119 (gimple_fold_builtin_snprintf): Same.
24120 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24121 of character types.
24122 (is_call_safe): New function.
24123 (try_substitute_return_value): Call it.
24124 (try_simplify_call): New function.
24125 (pass_sprintf_length::handle_gimple_call): Call it.
24127 2017-05-14 Martin Sebor <msebor@redhat.com>
24129 PR middle-end/80669
24130 * builtins.c (expand_builtin_stpncpy): Simplify.
24132 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24134 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24135 * config/i386/i386.h
24136 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24137 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24138 (struct machine_function): Add new members call_ms2sysv,
24139 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24140 (struct machine_frame_state): New fields sp_realigned and
24141 sp_realigned_offset.
24142 * config/i386/i386.c
24143 (enum xlogue_stub): New enum.
24144 (enum xlogue_stub_sets): New enum.
24145 (class xlogue_layout): New class.
24146 (struct ix86_frame): New fields stack_realign_allocate_offset,
24147 stack_realign_offset and outlined_save_offset. Modify comments to
24148 detail stack layout when using out-of-line stubs.
24149 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24150 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24151 -mcall-ms2sysv-xlogues.
24152 (stub_managed_regs): New static variable.
24153 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24154 registers managed by out-of-line stub.
24155 (disable_call_ms2sysv_xlogues): New function.
24156 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24157 m->call_ms2sysv when appropriate and compute frame layout for
24159 (sp_valid_at, fp_valid_at): New inline functions.
24160 (choose_basereg): New function.
24161 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24163 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24164 Use align parameter of choose_baseaddr to generated aligned SSE movs
24166 (pro_epilogue_adjust_stack): Modify to track
24167 machine_frame_state::sp_realigned.
24168 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24169 (ix86_nsaved_sseregs): Likewise.
24170 (ix86_emit_save_regs): Likewise.
24171 (ix86_emit_save_regs_using_mov): Likewise.
24172 (ix86_emit_save_sse_regs_using_mov): Likewise.
24173 (get_scratch_register_on_entry): Likewise.
24174 (gen_frame_set): New function.
24175 (gen_frame_load): Likewise.
24176 (gen_frame_store): Likewise.
24177 (emit_outlined_ms2sysv_save): Likewise.
24178 (emit_outlined_ms2sysv_restore): Likewise.
24179 (ix86_expand_prologue): Modify stack re-alignment code and call
24180 emit_outlined_ms2sysv_save when appropriate.
24181 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24182 parameter rtx_insn *insn, which allows the function to be used to only
24183 generate the notes.
24184 (ix86_expand_epilogue): Modify validity checks of frame and stack
24185 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24186 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24187 * config/i386/predicates.md
24188 (save_multiple): New predicate.
24189 (restore_multiple): Likewise.
24190 * config/i386/sse.md
24191 (save_multiple<mode>): New pattern.
24192 (save_multiple_realign<mode>): Likewise.
24193 (restore_multiple<mode>): Likewise.
24194 (restore_multiple_and_return<mode>): Likewise.
24195 (restore_multiple_leave_return<mode>): Likewise.
24196 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24198 2017-05-14 Julia Koval <julia.koval@intel.com>
24200 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24201 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24202 (__builtin_ia32_xsetbv): New builtins.
24203 * config/i386/i386.c (ix86_expand_special_args_builtin):
24205 (ix86_expand_builtin): Special expand for new intrinsics.
24206 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24207 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24208 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24210 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24212 * cfganal.c (inverted_post_order_compute): Change argument type
24214 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24215 * df-core.c (rest_of_handle_df_initialize): Adjust.
24216 (rest_of_handle_df_finish): Likewise.
24217 (df_analyze_1): Likewise.
24218 (df_analyze): Likewise.
24219 (loop_inverted_post_order_compute): Change argument to be a vec *.
24220 (df_analyze_loop): Adjust.
24221 (df_get_n_blocks): Likewise.
24222 (df_get_postorder): Likewise.
24223 * df.h (struct df_d): Change field to be a vec.
24224 * lcm.c (compute_laterin): Adjust.
24225 (compute_available): Likewise.
24226 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24227 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24228 * tree-ssa-pre.c (compute_antic): Likewise.
24230 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24232 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24233 (depth_first_search::depth_first_search): Change structure init
24234 function to this constructor.
24235 (depth_first_search::add_bb): Rename function to this member.
24236 (depth_first_search::execute): Likewise.
24237 (flow_dfs_compute_reverse_finish): Adjust.
24239 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24241 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24242 (longest_simple_path): Likewise.
24243 * shrink-wrap.c (spread_components): Likewise.
24244 (disqualify_problematic_components): Likewise.
24245 (emit_common_heads_for_components): Likewise.
24246 (emit_common_tails_for_components): Likewise.
24247 (insert_prologue_epilogue_for_components): Likewise.
24249 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24251 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24256 * df-core.c (df_set_blocks): Start using auto_bitmap.
24257 (df_compact_blocks): Likewise.
24258 * df-problems.c (df_rd_confluence_n): Likewise.
24259 * df-scan.c (df_insn_rescan_all): Likewise.
24260 (df_process_deferred_rescans): Likewise.
24261 (df_update_entry_block_defs): Likewise.
24262 (df_update_exit_block_uses): Likewise.
24263 (df_entry_block_bitmap_verify): Likewise.
24264 (df_exit_block_bitmap_verify): Likewise.
24265 (df_scan_verify): Likewise.
24266 * lra-constraints.c (lra_constraints): Likewise.
24267 (undo_optional_reloads): Likewise.
24268 (lra_undo_inheritance): Likewise.
24269 * lra-remat.c (calculate_gen_cands): Likewise.
24270 (do_remat): Likewise.
24271 * lra-spills.c (assign_spill_hard_regs): Likewise.
24272 (spill_pseudos): Likewise.
24273 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24274 (bitmap_set_subtract_values): Likewise.
24276 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24278 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24279 management with auto_bitmap.
24280 (fix_inter_tick): Likewise.
24281 (fix_recovery_deps): Likewise.
24282 * ira.c (add_store_equivs): Likewise.
24283 (find_moveable_pseudos): Likewise.
24284 (split_live_ranges_for_shrink_wrap): Likewise.
24285 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24286 (rtx_reuse_manager::seen_def_p): Likewise.
24287 (rtx_reuse_manager::set_seen_def): Likewise.
24288 * print-rtl.h (class rtx_reuse_manager): Likewise.
24290 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24292 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24294 (migrate_btr_def): Likewise.
24295 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24296 * df-core.c (loop_post_order_compute): Likewise.
24297 (loop_inverted_post_order_compute): Likewise.
24298 * hsa-common.h: Likewise.
24299 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24300 * init-regs.c (initialize_uninitialized_regs): Likewise.
24301 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24302 (inline_small_functions): Likewise.
24303 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24304 * ira.c (combine_and_move_insns): Likewise.
24305 (build_insn_chain): Likewise.
24306 * loop-invariant.c (find_invariants): Likewise.
24307 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24308 * predict.c (tree_predict_by_opcode): Likewise.
24309 (predict_paths_leading_to): Likewise.
24310 (predict_paths_leading_to_edge): Likewise.
24311 (estimate_loops_at_level): Likewise.
24312 (estimate_loops): Likewise.
24313 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24314 (spread_components): Likewise.
24315 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24316 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24317 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24318 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24319 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24320 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24321 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24322 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24323 (mark_threaded_blocks): Likewise.
24324 (thread_through_all_blocks): Likewise.
24325 * tree-ssa.c (verify_ssa): Likewise.
24326 (execute_update_addresses_taken): Likewise.
24327 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24329 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24331 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24333 (post_order_compute): Likewise.
24334 (inverted_post_order_compute): Likewise.
24335 (pre_and_rev_post_order_compute_fn): Likewise.
24337 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24339 * genrecog.c (int_set::int_set): Explicitly construct our
24340 auto_vec base class.
24341 * vec.h (auto_vec::auto_vec): New constructor.
24343 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24345 * bitmap.h (class auto_bitmap): New constructor taking
24346 bitmap_obstack * argument.
24348 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24350 * bitmap.h (class auto_bitmap): Change type of m_bits to
24351 bitmap_head, and adjust ctor / dtor and member operators.
24353 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24355 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24356 when returned register mode doesn't match original mode.
24358 2017-05-12 Jeff Law <law@redhat.com>
24359 Jakub Jelinek <jakub@redhat.com>
24361 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24362 we look for cc setter after the compare-elim changes.
24363 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24364 within the vector to match what compare-elim now expects.
24365 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24366 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24368 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24369 after the compare-elim changes.
24370 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24371 the vector to match what compare-elim now expects.
24372 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24373 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24374 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24375 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24376 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24378 * config/visium/visium.c (single_set_and_flags): Fix where
24379 we look for cc setter after the compare-elim changes.
24380 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24381 with the vector to match what compare-elim now expects.
24382 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24383 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24384 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24385 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24386 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24387 (neg<mode>2_insn_set_overflow): Likewise.
24389 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24391 PR middle-end/79794
24392 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24393 maybe_expand_insn call, set ops[0].target. If still set after call,
24394 set alt_rtl. Add extra arg to recursive calls.
24395 (extract_bit_field): Add alt_rtl argument. Pass to
24397 * expmed.h (extract_bit_field): Fix prototype.
24398 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24399 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24400 to extract_bit_field_calls.
24401 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24402 Pass alt_rtl to extract_bit_field calls.
24403 * calls.c (store_unaligned_arguments_into_psuedos)
24404 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24405 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24407 * optabs.h (struct expand_operand): Add target bitfield.
24409 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24411 * compare-elim.c (try_eliminate_compare): Canonicalize
24412 operation with embedded compare to
24413 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24414 (set (reg) (operation)].
24416 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24418 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24421 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24422 cost of adding a carry flag for ADC instruction.
24423 [case MINUS]: Ignore the cost of subtracting a carry flag
24424 for SBB instruction.
24426 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24428 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24430 * config/rs6000/bmiintrin.h: New file.
24431 * config/rs6000/bmi2intrin.h: New file.
24432 * config/rs6000/x86intrin.h: New file.
24434 2017-05-12 Jeff Law <law@redhat.com>
24436 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24439 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24441 PR middle-end/80707
24442 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24444 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24446 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24447 early expansion of vector divide builtins.
24448 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24449 builtins identified as having unsigned arguments.
24451 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24453 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24454 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24455 expansion of vector logical operations (and, andc, or, xor,
24458 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24460 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24461 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24463 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24465 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24466 early GIMPLE expansion of vector multiplies.
24468 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24470 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24471 TARGET_HAVE_MOVT conditional.
24472 (movt splitter): Likewise.
24474 2017-05-12 Richard Biener <rguenther@suse.de>
24476 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24477 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24478 Fold all stmts not inplace.
24480 2017-05-12 Richard Biener <rguenther@suse.de>
24482 PR tree-optimization/80713
24483 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24484 inserted_exprs bit for not removed stmts.
24486 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24488 PR middle-end/69921
24489 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24490 parallelized" attribute for parallelized OpenACC kernels.
24491 * omp-offload.c (execute_oacc_device_lower): Use it.
24493 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24494 Set "oacc kernels" attribute.
24495 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24496 parameter. Adjust all users.
24497 (oacc_fn_attrib_kernels_p): Remove function.
24498 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24499 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24500 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24501 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24502 assert "oacc kernels" attribute is set.
24504 2017-05-11 Carl Love <cel@us.ibm.com>
24506 * config/rs6000/rs6000-c: Add support for built-in functions
24507 vector unsigned char vec_popcnt (vector signed char)
24508 vector unsigned char vec_popcnt (vector unsigned char)
24509 vector unsigned short vec_popcnt (vector signed short)
24510 vector unsigned short vec_popcnt (vector unsigned short)
24511 vector unsigned int vec_popcnt (vector signed int)
24512 vector unsigned int vec_popcnt (vector unsigned int)
24513 vector unsigned long long vec_popcnt (vector signed long long)
24514 vector unsigned long long vec_popcnt (vector unsigned long long)
24515 vector signed long long vec_slo (vector signed long long,
24516 vector signed char)
24517 vector signed long long vec_slo (vector signed long long,
24518 vector unsigned char)
24519 vector unsigned long long vec_slo (vector unsigned long long,
24520 vector signed char)
24521 vector unsigned long long vec_slo (vector unsigned long long,
24522 vector unsigned char)
24523 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24524 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24525 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24526 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24527 * doc/extend.texi: Update the built-in documentation file for the
24528 new built-in functions.
24530 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24532 * attribs.h (sorted_attr_string): Move machine independent
24533 functions for target clone support from the i386 port to common
24534 code. Rename ix86_function_versions to common_function_versions.
24535 Rename make_name to make_unique_name.
24536 (common_function_versions): Likewise.
24537 (make_unique_name): Likewise.
24538 (make_dispatcher_decl): Likewise.
24539 (is_function_default_version): Likewise.
24540 * attribs.c (attr_strcmp): Likewise.
24541 (sorted_attr_string): Likewise.
24542 (common_function_versions): Likewise.
24543 (make_unique_name): Likewise.
24544 (make_dispatcher_decl): Likewise.
24545 (is_function_default_version): Likewise.
24546 * config/i386/i386.c (attr_strcmp): Likewise.
24547 (sorted_attr_string): Likewise.
24548 (ix86_function_versions): Likewise.
24549 (make_name): Likewise.
24550 (make_dispatcher_decl): Likewise.
24551 (is_function_default_version): Likewise.
24552 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24554 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24557 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24558 Account for direct move costs for vec_construct of integer
24561 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24564 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24565 (UNSPEC_STX_ATOMIC): Ditto.
24566 (loaddi_via_sse): New insn.
24567 (storedi_via_sse): Ditto.
24568 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24569 Update corresponding peephole2 patterns.
24570 (atomic_storedi_fpu): Ditto.
24572 2017-05-11 Julia Koval <julia.koval@intel.com>
24574 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24575 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24577 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24578 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24579 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24581 2017-05-11 Nathan Sidwell <nathan@acm.org>
24583 * graphite-poly.c: Include dumpfile.h.
24585 * dumpfle.h (dump_function): Declare here ...
24586 * tree-dump.h (dump_function): ... not here.
24587 * dumpfile.c: #include tree-cfg.h.
24588 (dump_function): Move here from ...
24589 * tree-dump.c (dump_function): ... here.
24590 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24591 * graphite-poly.c: Don't include tree-dump.h.
24592 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24593 * print-tree.c: Likewise.
24594 * stor-layout.c: Likewise.
24595 * tree-nested.c: Likewise.
24597 * dumpfile.c (dump_start): Use TDF_FLAGS.
24598 (dump_enable_all): Fix TDF_KIND check thinko.
24600 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24603 array entries to represent two legal parameterizations of the
24604 overloaded __builtin_cmpb function, as represented by the
24605 P6_OV_BUILTIN_CMPB constant.
24606 (altivec_resolve_overloaded_builtin): Add special case handling
24607 for the __builtin_cmpb function, as represented by the
24608 P6_OV_BUILTIN_CMPB constant.
24609 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24610 (BU_P6_64BIT_2): New macro.
24611 (BU_P6_OVERLOAD_2): New macro
24612 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24613 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24614 (CMPB): Add overload support to represent both 32-bit and 64-bit
24615 compare-bytes function.
24616 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24617 support for TARGET_CMPB.
24618 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24619 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24620 documentation of the __builtin_cmpb overloaded built-in function.
24622 2017-05-11 Richard Biener <rguenther@suse.de>
24624 PR tree-optimization/80705
24625 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24626 bases are not vectorizable.
24628 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24630 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24631 when counting register pressure.
24633 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24635 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24636 (iv_ca_more_deps): Renamed to ...
24637 (iv_ca_compare_deps): ... this.
24638 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24642 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24644 (determine_group_iv_costs): ... here.
24645 (find_inv_vars_cb): Record inv var if it's not recorded before.
24647 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24649 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24650 (get_shiftadd_cost): Ditto.
24652 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24654 * tree-ssa-address.c: Include header file.
24655 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24657 (add_to_parts): Refactor.
24658 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24659 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24664 PR tree-optimization/53090
24665 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24667 (extract_cond_operands): Detect condition with IV on both sides
24668 and return COMP_IV_EXPR_2.
24669 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24670 (rewrite_use_compare): Simplify by removing call to function
24671 extract_cond_operands.
24673 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24675 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24676 (extract_cond_operands): Detect condition comparing against non-
24677 invariant bound and return appropriate enum value.
24678 (find_interesting_uses_cond): Update use of extract_cond_operands.
24679 Handle its return value accordingly.
24680 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24684 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24685 nonlinear iv_use computation in loop invariant sensitive way.
24687 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24689 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24690 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24692 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24694 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24695 (dump_cand): Support iv_cand.inv_exprs.
24696 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24698 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24703 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24705 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24706 as local function. Include necessary header files.
24707 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24711 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24713 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24715 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24716 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24717 RSHIFT_EXPR and BIT_NOT_EXPR.
24719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24721 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24722 (adjust_setup_cost): New parameter supporting round up adjustment.
24723 (struct address_cost_data): Delete.
24724 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24725 (split_address_cost, ptr_difference_cost): Delete.
24726 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24727 (struct ainc_cost_data): New struct.
24728 (get_address_cost_ainc): New function.
24729 (get_address_cost, get_computation_cost): Reimplement.
24730 (determine_group_iv_cost_address): Record inv_expr for all uses of
24732 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24733 (iv_ca_has_deps): Reimplemented to ...
24734 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24736 (iv_ca_extend): Call iv_ca_more_deps.
24738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24740 * tree-ssa-address.c (struct mem_address): Move to header file.
24741 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24742 * tree-ssa-address.h (struct mem_address): Move from C file.
24743 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24745 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24747 * tree-affine.h (aff_combination_type): New interface.
24748 (aff_combination_zero_p): Remove static.
24749 (aff_combination_const_p): New interface.
24750 (aff_combination_singleton_var_p): New interfaces.
24752 2017-05-11 Richard Biener <rguenther@suse.de>
24754 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24755 Skip unreachable blocks and destinations.
24756 (eliminate): Move stmt removal and fixup ...
24757 (fini_eliminate): ... here. Skip inserted exprs.
24758 (pass_pre::execute): Move fini_pre after fini_eliminate.
24759 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24760 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24761 PRE to get rid of dead code that has invalid SSA form and
24762 split critical edges again.
24764 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24766 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24768 2017-05-11 Richard Biener <rguenther@suse.de>
24770 * passes.c (execute_function_todo): Verify loops if they are
24771 said to be up-to-date.
24772 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24773 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24775 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24778 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24779 handle calling assemble_external ourself.
24782 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24783 modes with zero size. Enhance comment.
24785 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24787 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24788 built-ins for vec_xl and vec_xst with short and char pointer
24791 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24793 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24794 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24795 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24796 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24797 (_mm_maskz_min_round_ss): New intrinsics.
24798 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24799 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24800 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24801 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24802 (__builtin_ia32_minss_mask_round): New builtins.
24803 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24804 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24805 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24807 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24808 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24810 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24813 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24815 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24816 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24817 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24818 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24819 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24820 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24821 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24822 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24823 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24824 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24825 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24826 (__builtin_ia32_mulss_mask_round): New builtins.
24827 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24828 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24829 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24831 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24832 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24834 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24837 2017-05-10 Julia Koval <julia.koval@intel.com>
24839 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24840 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24841 (_mm256_setr_m128i): New intrinsics.
24843 2017-05-10 Julia Koval <julia.koval@intel.com>
24845 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24846 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24847 (_mm_maskz_rcp14_ss): New intrinsics.
24848 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24849 (__builtin_ia32_rcp14ss_mask): New builtins.
24850 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24852 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24854 PR tree-optimization/51513
24855 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24856 (assert_unreachable_fallthru_edge_p): Use it.
24857 (group_case_labels_stmt): Likewise.
24858 * tree-cfg.h: Prototype it.
24859 * stmt.c: Include cfghooks.h and tree-cfg.h.
24860 (emit_case_dispatch_table) <gap_label>: New local variable.
24861 Use it to fill dispatch table gaps.
24862 Test for default_label before updating probabilities.
24863 (expand_case) <default_label>: Remove unneeded initialization.
24864 Test for unreachable default case statement and remove its edge.
24865 Set default_label accordingly.
24866 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24868 2017-05-10 Carl Love <cel@us.ibm.com>
24870 * config/rs6000/rs6000-c: Add support for built-in functions
24871 vector signed char vec_neg (vector signed char)
24872 vector signed short int vec_neg (vector short int)
24873 vector signed int vec_neg (vector signed int)
24874 vector signed long long vec_neg (vector signed long long)
24875 vector float vec_neg (vector float)
24876 vector double vec_neg (vector double)
24877 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24879 * config/rs6000/altivec.h: Add define for vec_neg
24880 * doc/extend.texi: Update the built-in documentation for the
24881 new built-in functions.
24883 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24885 PR tree-optimization/77644
24886 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24888 2017-05-10 Nathan Sidwell <nathan@acm.org>
24890 * dumpfile.h (TDI_lang_all): New.
24891 (TDF_KIND): New. Renumber others
24892 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24894 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24896 (get_dump_file_name): Adjust suffix generation.
24897 (dump_enable_all): Use TDF_KIND.
24898 * doc/invoke.texi (-fdump-lang-all): Document.
24900 * dumpfile.h: Tabify.
24902 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24905 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24906 Move member access before delete.
24908 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24910 * tree-inline.c (expand_call_inline): Split block at stmt
24913 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24916 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24917 are now unused after splitting mov{sf,sd}_hardfloat.
24918 (f32_lr2): Likewise.
24919 (f32_lm): Likewise.
24920 (f32_lm2): Likewise.
24921 (f32_li): Likewise.
24922 (f32_li2): Likewise.
24923 (f32_lv): Likewise.
24924 (f32_sr): Likewise.
24925 (f32_sr2): Likewise.
24926 (f32_sm): Likewise.
24927 (f32_sm2): Likewise.
24928 (f32_si): Likewise.
24929 (f32_si2): Likewise.
24930 (f32_sv): Likewise.
24931 (f32_dm): Likewise.
24932 (f32_vsx): Likewise.
24933 (f32_av): Likewise.
24934 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24935 For movsf, order stores so the VSX stores occur before the GPR
24936 store which encourages the register allocator to use a traditional
24937 FPR instead of a GPR. For movsd, order the stores so that the GPR
24938 store comes before the VSX stores to allow the power6 to work.
24939 This is due to the power6 not having a 32-bit integer store
24940 instruction from a FPR.
24941 (movsf_hardfloat): Likewise.
24942 (movsd_hardfloat): Likewise.
24944 2017-05-09 Martin Sebor <msebor@redhat.com>
24946 PR translation/80280
24947 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24950 PR translation/80280
24951 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24952 data member added in r247778.
24953 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24955 2017-05-09 Nathan Sidwell <nathan@acm.org>
24957 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24959 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24962 2017-05-09 Marek Polacek <polacek@redhat.com>
24964 * doc/invoke.texi: Fix typo.
24966 2017-05-09 Richard Biener <rguenther@suse.de>
24968 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24969 (vrp_val_is_min): Likewise.
24970 (set_value_range_to_value): Likewise.
24971 (set_value_range_to_nonnegative): Likewise.
24972 (gimple_assign_nonzero_p): Likewise.
24973 (gimple_stmt_nonzero_p): Likewise.
24974 (vrp_int_const_binop): Likewise. Remove unreachable case.
24975 (adjust_range_with_scev): Adjust comments.
24976 (compare_range_with_value): Likewise.
24977 (extract_range_from_phi_node): Likewise.
24978 (test_for_singularity): Likewise.
24980 2017-05-09 Richard Biener <rguenther@suse.de>
24982 * tree-vrp.c (get_single_symbol): Add assert that we don't
24983 get overflowed constants as invariant part.
24984 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24985 checks. Use wi::cmp instead of recursing for integer constants.
24986 (compare_values): Just ignore whether we assumed undefined
24987 overflow instead of failing the compare.
24988 (extract_range_for_var_from_comparison_expr): Add comment before the
24989 TREE_NO_WARNING sets.
24990 (test_for_singularity): Likewise.
24991 (extract_range_from_comparison): Do not disable optimization
24992 when we assumed undefined overflow.
24993 (extract_range_basic): Remove init of unused var.
24995 2017-05-09 Richard Biener <rguenther@suse.de>
24997 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24998 (extract_range_from_multiplicative_op_1): Adjust.
24999 (extract_range_from_binary_expr_1): Use int_const_binop.
25001 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25004 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25005 rs6000_store_data_bypass_p in seven define_bypass directives and
25006 in several comments.
25007 * config/rs6000/rs6000-protos.h: Add prototype for
25008 rs6000_store_data_bypass_p function.
25009 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25010 function implements slightly different (rs6000-specific) semantics
25011 than store_data_bypass_p, returning false rather than aborting
25012 with assertion error when arguments do not satisfy the
25013 requirements of store data bypass.
25014 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25015 rs6000_store_data_bypass_p.
25017 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25019 * config/xtensa/xtensa-protos.h
25020 (xtensa_initial_elimination_offset): New declaration.
25021 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25022 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25023 macro definition, add case for FRAME_POINTER_REGNUM when
25024 FRAME_GROWS_DOWNWARD.
25025 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25026 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25027 xtensa_initial_elimination_offset.
25029 2017-05-08 Nathan Sidwell <nathan@acm.org>
25031 * doc/invoke.texi: Alphabetize -fdump options.
25033 2017-05-08 Martin Sebor <msebor@redhat.com>
25035 PR translation/80280
25036 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25038 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25040 * target.def (compute_frame_layout): New optional target hook.
25041 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25042 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25043 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25045 * reload1.c (verify_initial_elim_offsets): Likewise.
25046 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25047 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25048 (arm_get_frame_offsets): Split up into this ...
25049 (arm_compute_frame_layout): ... and this function.
25051 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25053 * config/aarch64/constraints.md (Usa): New constraint.
25054 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25056 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25058 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25059 with_multilib_list after it has been checked.
25061 2017-05-08 Richard Biener <rguenther@suse.de>
25063 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25064 (bitmap_set_subtract_values): Likewise.
25066 2017-05-08 Richard Biener <rguenther@suse.de>
25068 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25069 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25071 (gimple_stmt_nonzero_warnv_p): Rename to ...
25072 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25074 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25075 (extract_range_basic): Adjust, do not disable propagation on
25076 strict overflow sensitive simplification.
25077 (vrp_visit_cond_stmt): Likewise.
25079 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25081 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25082 body size unconditionally.
25084 2017-05-07 Jeff Law <law@redhat.com>
25087 2017-05-06 Jeff Law <law@redhat.com>
25088 PR tree-optimization/78496
25089 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25092 PR tree-optimization/78496
25093 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25094 (simplify_stmt_using_ranges): Call it.
25095 (vrp_dom_walker::before_dom_children): Extract equivalences
25096 from an ASSERT_EXPR with an equality comparison against a
25099 2017-05-06 Jeff Law <law@redhat.com>
25101 PR tree-optimization/78496
25102 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25105 PR tree-optimization/78496
25106 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25107 (simplify_stmt_using_ranges): Call it.
25108 (vrp_dom_walker::before_dom_children): Extract equivalences
25109 from an ASSERT_EXPR with an equality comparison against a
25112 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25114 * lra-constraints.c (lra_copy_reg_equiv): New function.
25115 (split_reg): Use it to copy equivalence information from the
25116 original register to the spill register.
25118 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25120 PR rtl-optimization/75964
25121 * simplify-rtx.c (simplify_const_relational_operation): Remove
25122 invalid handling of comparisons of integer ABS.
25124 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25126 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25127 initialize to zero.
25128 (init_regs): Remove declaration.
25129 (function_arg_advance_32): Initialize error_p as boolean variable.
25131 2017-05-05 Nathan Sidwell <nathan@acm.org>
25133 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25134 lines. Use for (;;).
25136 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25138 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25139 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25140 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25141 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25142 VF=2 that require versioning.
25144 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25146 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25149 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25151 * diagnostic.h (diagnostic_override_option_index): Convert from
25152 macro to inline function.
25154 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25156 * diagnostic.c (last_module_changed_p): New function.
25157 (set_last_module): New function.
25158 (diagnostic_report_current_module): Convert macro usage to
25159 the above functions.
25160 * diagnostic.h (diagnostic_context::last_module): Strengthen
25161 from const line_map * to const line_map_ordinary *.
25162 (diagnostic_last_module_changed): Delete macro.
25163 (diagnostic_set_last_module): Delete macro.
25165 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25167 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25168 with diagnostic_report_diagnostic.
25169 (diagnostic_n_impl_richloc): Likewise.
25170 * diagnostic.h (report_diagnostic): Delete macro.
25171 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25172 with diagnostic_report_diagnostic.
25173 * substring-locations.c (format_warning_va): Likewise.
25175 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25177 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25178 save/restor of format_spec. Move option-printing code to...
25179 (print_option_information): ...this new function, and
25180 reimplement by simply printing to the pretty_printer,
25181 rather than appending to the format string.
25183 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25185 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25186 handling logic into...
25187 (update_effective_level_from_pragmas): ...this new function.
25189 2017-05-04 Andrew Waterman <andrew@sifive.com>
25191 * config/riscv/riscv.opt (mstrict-align): New option.
25192 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25193 (SLOW_UNALIGNED_ACCESS): Define.
25194 (riscv_slow_unaligned_access): Declare.
25195 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25197 (riscv_slow_unaligned_access): New variable.
25198 (rocket_tune_info): Set slow_unaligned_access to true.
25199 (optimize_size_tune_info): Set slow_unaligned_access to false.
25200 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25201 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25202 (riscv_option_override): Set riscv_slow_unaligned_access.
25203 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25205 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25207 * config/riscv/riscv.md: Unify indentation.
25209 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25214 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25216 (extendsi<mode>2): Add support for doing sign extension via
25217 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25218 don't have ISA 3.0 instructions.
25219 (extendsi<mode>2 splitter): Likewise.
25220 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25221 generate the normal insns since SImode can now go in vector
25222 registers. Disallow the special UNSPECs needed for previous
25223 machines to hide SImode being used. Add new insns
25224 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25225 (fix_trunc<mode>si2_stfiwx): Likewise.
25226 (fix_trunc<mode>si2_internal): Likewise.
25227 (fixuns_trunc<mode>si2): Likewise.
25228 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25229 (fctiw<u>z_<mode>_smallint): Likewise.
25230 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25231 of floating point to 32-bit integer from doing a direct move to
25232 the GPR registers to do a store.
25233 (fctiwz_<mode>): Break long line.
25235 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25237 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25238 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25239 (addr_list, addr_offset_valid_p): New.
25240 (split_address_groups): Check offset validity with above function.
25241 (gt-tree-ssa-loop-ivopts.h): Include header file.
25243 2017-05-05 Nathan Sidwell <nathan@acm.org>
25245 * config.gcc (arm*-*-*): Add missing 'fi'.
25247 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25249 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25250 included in -fopt-info does not matter.
25251 * doc/optinfo.texi (-fopt-info): Fix description of default
25252 behavour. Explicitly say order of options included in -fopt-info
25255 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25257 * config.gcc: Allow combinations of aprofile and rmprofile values for
25258 --with-multilib-list.
25259 * config/arm/t-multilib: New file.
25260 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25261 variables. Remove setting of ISA and floating-point ABI in
25262 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25263 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25264 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25265 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25267 * config/arm/t-rmprofile: Likewise except for the matches changes.
25268 * doc/install.texi (--with-multilib-list): Document the combination of
25269 aprofile and rmprofile values and warn about pitfalls in doing that.
25271 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25273 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25274 (movdi_aarch64): Likewise.
25276 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25278 PR tree-optimization/80632
25279 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25281 (build_arrays): Initialize it for virtual phis.
25282 (fix_phi_nodes): Use it for virtual phis.
25284 PR tree-optimization/80558
25285 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25286 [x, y] op z into [x op, y op z] for op & or | if conditions
25289 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25290 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25293 * config/arm/arm.md (use_literal_pool): Remove.
25294 (64-bit immediate split): No longer takes cost into consideration
25295 if arm_disable_literal_pool is enabled.
25296 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25297 used when arm_disable_literal_pool is enabled.
25298 (arm_max_const_double_inline_cost): Remove use of
25299 arm_disable_literal_pool.
25300 (push_minipool_fix): Add assert.
25301 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25302 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25303 (no_literal_pool_sf_immediate): New.
25305 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25307 PR tree-optimization/80613
25308 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25309 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25311 2017-05-05 Richard Biener <rguenther@suse.de>
25313 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25315 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25317 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25318 of this flag from insn conditions due to removal from r247495.
25320 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25322 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25324 (arm_early_store_addr_dep_ptr): Likewise.
25325 * config/arm/aarch-common-protos.h
25326 (arm_early_load_addr_dep_ptr): Add prototype.
25327 (arm_early_store_addr_dep_ptr): Likewise.
25328 * config/arm/cortex-a53.md: Add new bypasses.
25330 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25332 * tree.c (next_type_uid): Change type to unsigned.
25333 (type_hash_canon): Decrement back next_type_uid if
25334 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25335 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25338 2017-05-04 Martin Sebor <msebor@redhat.com>
25340 * builtins.c: Fix a trivial typo in a comment.
25342 PR middle-end/79234
25343 * builtins.c (check_sizes): Adjust to handle reading past the end.
25344 Avoid printing excessive upper bound of ranges. Use %E to print
25345 tree nodes instead of converting them to %wu.
25346 (expand_builtin_memchr): New function.
25347 (compute_dest_size): Rename...
25348 (compute_objsize): ...to this.
25349 (expand_builtin_memcpy): Adjust.
25350 (expand_builtin_mempcpy): Adjust.
25351 (expand_builtin_strcat): Adjust.
25352 (expand_builtin_strcpy): Adjust.
25353 (check_strncat_sizes): Adjust.
25354 (expand_builtin_strncat): Adjust.
25355 (expand_builtin_strncpy): Adjust and simplify.
25356 (expand_builtin_memset): Adjust.
25357 (expand_builtin_bzero): Adjust.
25358 (expand_builtin_memcmp): Adjust.
25359 (expand_builtin): Handle memcmp.
25360 (maybe_emit_chk_warning): Check strncat just once.
25362 2017-05-04 Martin Sebor <msebor@redhat.com>
25364 PR preprocessor/79214
25365 PR middle-end/79222
25366 PR middle-end/79223
25367 * builtins.c (check_sizes): Add inlining context and issue
25368 warnings even when -Wno-system-headers is set.
25369 (check_strncat_sizes): Same.
25370 (expand_builtin_strncat): Same.
25371 (expand_builtin_memmove): New function.
25372 (expand_builtin_stpncpy): Same.
25373 (expand_builtin): Handle memmove and stpncpy.
25375 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25377 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25378 which is not used any more.
25380 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25382 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25384 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25386 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25387 (cortexa53_tunings): Likewise.
25388 (cortexa57_tunings): Likewise.
25389 (cortexa72_tunings): Likewise.
25390 (cortexa73_tunings): Likewise.
25392 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25394 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25395 Set loop alignment to 8.
25397 2017-05-04 Martin Sebor <msebor@redhat.com>
25399 PR translation/80280
25400 * builtins.c (expand_builtin_object_size): Add missing quoting to
25401 %D and like directives.
25402 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25403 (hsa_type_for_tree_type): Same.
25404 (verify_function_arguments): Same.
25405 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25406 * varasm.c (get_section): Same.
25409 2017-05-04 Martin Sebor <msebor@redhat.com>
25411 PR translation/80280
25412 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25414 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25416 * config/aarch64/aarch64.c (generic_addrcost_table):
25417 Change HI/TI mode setting.
25419 2017-05-04 Martin Jambor <mjambor@suse.cz>
25421 PR tree-optimization/80622
25422 * tree-sra.c (comes_initialized_p): New function.
25423 (build_accesses_from_assign): Only set write lazily when
25424 comes_initialized_p is false.
25425 (analyze_access_subtree): Use comes_initialized_p.
25426 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25427 instead of testing for PARM_DECL.
25429 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25431 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25432 constraint on operand 0 to allow more general addressing modes.
25433 Adjust output template.
25434 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25436 * config/aarch64/aarch64-protos.h
25437 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25438 * config/aarch64/constraints.md (Dp): New address constraint.
25439 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25442 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25444 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25445 update use of estimate_ipcp_clone_size_and_time.
25446 (estimate_local_effects): Update use of
25447 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25448 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25449 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25450 Return nonspecialized time.
25452 2017-05-04 Richard Biener <rguenther@suse.de>
25454 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25455 for the last VUSE which def dominates the PHI. Directly call
25457 (get_continuation_for_phi_1): Remove.
25459 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25461 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25462 to explain the use of truncating division. Cap the number of
25463 iterations to the maximum given by nb_iterations_upper_bound,
25466 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25468 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25469 * configure: Regenerate.
25470 * config.in: Regenerate.
25471 * config/i386/driver-mingw32.c: new file.
25472 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25473 * config.host: Link driver-mingw32.o on MinGW host.
25474 * doc/install.texi: Document new --enable-mingw-wildcard configure
25477 2017-05-04 Marek Polacek <polacek@redhat.com>
25479 PR tree-optimization/80612
25480 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25482 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25483 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25485 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25486 (movt splitter): Likewise.
25487 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25488 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25489 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25490 block for Thumb-1 with MOVT.
25491 (thumb2_legitimate_address_p): Move code block ...
25492 (can_avoid_literal_pool_for_label_p): ... into this new function.
25493 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25495 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25496 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25497 "M-profile targets with the MOVT instruction".
25499 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25501 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25502 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25503 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25505 2017-05-04 Martin Liska <mliska@suse.cz>
25507 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25508 variable cond_code.
25510 2017-05-04 Richard Biener <rguenther@suse.de>
25512 * tree.c (array_at_struct_end_p): Handle arrays at struct
25513 end with flexarrays more conservatively. Refactor and treat
25514 arrays of arrays or aggregates more strict. Fix
25515 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25516 * tree.c (array_at_struct_end_p): Adjust prototype.
25517 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25518 * gimple-fold.c (get_range_strlen): Likewise.
25519 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25521 2017-05-04 Richard Biener <rguenther@suse.de>
25523 PR tree-optimization/31130
25524 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25526 (supports_overflow_infinity): Likewise.
25527 (is_negative_overflow_infinity): Likewise.
25528 (is_positive_overflow_infinity): Likewise.
25529 (is_overflow_infinity): Likewise.
25530 (stmt_overflow_infinity): Likewise.
25531 (overflow_infinity_range_p): Likewise.
25532 (usable_range_p): Remove as always returning true.
25533 (make_overflow_infinity): Remove.
25534 (negative_overflow_infinity): Likewise.
25535 (positive_overflow_infinity): Likewise.
25536 (avoid_overflow_infinity): Likewise.
25537 (set_value_range): Adjust accordingly.
25538 (set_value_range_to_nonnegative): Likewise, remove now unused
25539 overflow_infinity arg.
25540 (vrp_operand_equal_p): Adjust.
25541 (update_value_range): Likewise.
25542 (range_int_cst_singleton_p): Likewise.
25543 (operand_less_p): Likewise.
25544 (compare_values_warnv): Likewise.
25545 (extract_range_for_var_from_comparison_expr): Likewise.
25546 (vrp_int_const_binop): Likewise.
25547 (zero_nonzero_bits_from_vr): Likewise.
25548 (extract_range_from_multiplicative_op_1): Likewise.
25549 (extract_range_from_binary_expr_1): Likewise.
25550 (extract_range_from_unary_expr): Likewise.
25551 (extract_range_from_comparison): Likewise.
25552 (extract_range_basic): Likewise.
25553 (adjust_range_with_scev): Likewise.
25554 (compare_ranges): Likewise.
25555 (compare_range_with_value): Likewise.
25556 (dump_value_range): Likewise.
25557 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25559 (simplify_cond_using_ranges): Adjust.
25561 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25563 * brig-builtins.def: Added a builtin for class_f64.
25564 * builtin-types.def: Added a builtin type needed by class_f64.
25566 2017-05-03 Jason Merrill <jason@redhat.com>
25568 * timevar.def: Add TV_CONSTEXPR.
25570 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25572 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25574 2017-05-03 Martin Jambor <mjambor@suse.cz>
25576 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25577 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25578 * ipa-cp.c (ipcp_propagate_stage): Do not call
25579 ipa_update_after_lto_read.
25580 * ipa-inline.c (ipa_inline): Likewise.
25582 2017-05-03 Martin Jambor <mjambor@suse.cz>
25584 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25585 tag. Added a default constructor and a destructor.
25586 (ipa_edge_args_sum_t): New class;
25587 (ipa_edge_args_sum): Declare.
25588 (ipa_edge_args_vector): Remove declaration.
25589 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25590 (ipa_free_edge_args_substructures): Remove declaration.
25591 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25592 (ipa_edge_args_info_available_for_edge_p): Likewise.
25593 * ipa-prop.c (ipa_edge_args_vector): Removed.
25594 (edge_removal_hook_holder): Likewise.
25595 (edge_duplication_hook_holder): Likewise.
25596 (ipa_edge_args_sum): New variable.
25597 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25598 ipa_edge_args_vector.
25599 (ipa_free_edge_args_substructures): Likewise.
25600 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25601 ipa_edge_args_vector.
25602 (ipa_edge_removal_hook): Turned into method
25603 ipa_edge_args_sum_t::remove.
25604 (ipa_edge_duplication_hook): Turned into method
25605 ipa_edge_args_sum_t::duplicate.
25606 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25607 registering edge hooks.
25608 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25609 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25610 ipa_edge_args_sum instead of ipa_edge_args_vector.
25611 * ipa-profile.c (ipa_profile): Likewise.
25613 2017-05-03 Martin Jambor <mjambor@suse.cz>
25615 * symbol-summary.h (function_summary): New method exists.
25616 (function_summary::symtab_removal): Deallocate through release.
25617 (call_summary): New class.
25618 (gt_ggc_mx): New overload.
25619 (gt_pch_nx): Likewise.
25620 (gt_pch_nx): Likewise.
25622 2017-05-03 Jeff Law <law@redhat.com>
25624 PR tree-optimization/78496
25625 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25626 from simplify_cond_using_ranges. Split off code to walk
25627 backwards through casts into ...
25628 (simplify_cond_using_ranges_2): New function.
25629 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25630 (execute_vrp): After identifying jump threads, call
25631 simplify_cond_using_ranges_2.
25633 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25636 * ipa-inline.h (inline_summary): Add ctor.
25637 (create_ggc): Do not use ggc_cleared_alloc.
25639 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25640 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25642 * gcc.c (handle_braces): Support escaping in switch matching
25644 * doc/invoke.texi (Spec Files): Document it.
25645 Remove superfluous @code markup in items.
25647 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25649 * diagnostic-show-locus.c (struct column_range): New struct.
25650 (get_affected_columns): New function.
25651 (get_printed_columns): New function.
25652 (struct correction): New struct.
25653 (correction::ensure_capacity): New function.
25654 (correction::ensure_terminated): New function.
25655 (struct line_corrections): New struct.
25656 (line_corrections::~line_corrections): New dtor.
25657 (line_corrections::add_hint): New function.
25658 (layout::print_trailing_fixits): Reimplement in terms of the new
25660 (selftest::test_overlapped_fixit_printing): New function.
25661 (selftest::diagnostic_show_locus_c_tests): Call it.
25663 2017-05-03 Nathan Sidwell <nathan@acm.org>
25665 Canonicalize canonical type hashing
25666 * tree.h (type_hash_canon_hash): Declare.
25667 * tree.c (type_hash_list, attribute_hash_list): Move into
25668 type_hash_canon_hash.
25669 (build_type_attribute_qual_variant): Break out hash code calc into
25670 type_hash_canon_hash.
25671 (type_hash_canon_hash): New. Generic type hash computation.
25672 (build_range_type_1, build_array_type_1, build_function_type,
25673 build_method_type_directly, build_offset_type, build_complex_type,
25674 make_vector_type): Call it.
25676 2017-05-03 Richard Biener <rguenther@suse.de>
25678 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25679 When all DRs have unknown misaligned do not always peel
25680 when there is a store but apply the same costing model as if
25681 there were only loads.
25683 2017-05-03 Richard Biener <rguenther@suse.de>
25686 PR tree-optimization/80492
25687 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25688 compare_base_decls returning dont-know properly.
25690 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25692 * config/arm/iterators.md (CCSI): New mode iterator.
25693 (arch): New mode attribute.
25694 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25695 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25696 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25697 code iterator for success result mode.
25698 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25699 the corresponding new insn generators.
25701 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25704 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25705 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25707 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25709 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25710 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25711 (DDR_A): Wrap DDR argument in brackets.
25712 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25713 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25714 (DDR_REVERSED_P): Likewise.
25716 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25718 PR tree-optimization/79472
25719 * tree-switch-conversion.c (struct switch_conv_info): Add
25720 contiguous_range and default_case_nonstandard fields.
25721 (collect_switch_conv_info): Compute contiguous_range and
25722 default_case_nonstandard fields, don't clear final_bb if
25723 contiguous_range and only the default case doesn't have the required
25725 (check_all_empty_except_final): Set default_case_nonstandard instead
25726 of failing if contiguous_range and the default case doesn't have empty
25728 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25729 and only the default case doesn't have the required constants. Skip
25731 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25732 if default_case_nonstandard.
25733 (build_constructors): Build constant 1 just once. Assert that default
25734 values aren't inserted in between cases if contiguous_range. Skip
25736 (build_arrays): Skip virtual phis.
25737 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25738 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25739 Handle virtual phis.
25740 (gen_inbound_check): Handle default_case_nonstandard case.
25741 (process_switch): Adjust check_final_bb caller. Call
25742 gather_default_values with the first non-default case instead of
25743 default case if default_case_nonstandard.
25745 2017-05-02 Nathan Sidwell <nathan@acm.org>
25747 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25748 check. Fix formatting.
25750 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25752 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25753 errors when comparing specialized and unspecialized times.
25755 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25757 * diagnostic-show-locus.c
25758 (layout::should_print_annotation_line_p): Make private.
25759 (layout::print_annotation_line): Make private.
25760 (layout::annotation_line_showed_range_p): Make private.
25761 (layout::show_ruler): Make private.
25762 (layout::print_source_line): Make private. Pass in line and
25763 line_width, rather than calling location_get_source_line. Drop
25765 (layout::print_leading_fixits): New method.
25766 (layout::print_any_fixits): Rename to...
25767 (layout::print_trailing_fixits): ...this, and make private.
25768 Don't print newline fixits.
25769 (diagnostic_show_locus): Move logic for printing one row into...
25770 (layout::print_line): ...this new function. Move the
25771 location_get_source_line call and error-handling from
25772 print_source_line to here. Call print_leading_fixits, and rename
25773 print_any_fixits to print_trailing_fixits.
25774 (selftest::test_fixit_insert_containing_newline): Update now that
25775 newlines are partially supported.
25776 (selftest::test_fixit_insert_containing_newline_2): New test.
25777 (selftest::test_fixit_replace_containing_newline): Update comments.
25778 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25779 * edit-context.c (class added_line): New class.
25780 (class edited_line): Describe newline handling in comment.
25781 (edited_line::actually_edited_p): New method.
25782 (edited_line::print_content): Delete redundant decl.
25783 (edited_line::m_predecessors): New field.
25784 (edited_file::print_content): Call edited_line::print_content.
25785 (edited_file::print_diff): Update to support newlines.
25786 (edited_file::print_diff_hunk): Likewise.
25787 (edited_file::print_run_of_changed_lines): New function.
25788 (edited_file::print_diff_line): Convert to...
25789 (print_diff_line): ...this.
25790 (edited_file::get_effective_line_count): New function.
25791 (edited_line::edited_line): Initialize new field m_predecessors.
25792 (edited_line::~edited_line): Clean up m_predecessors.
25793 (edited_line::apply_fixit): Handle newlines.
25794 (edited_line::get_effective_line_count): New function.
25795 (edited_line::print_content): New function.
25796 (edited_line::print_diff_lines): New function.
25797 (selftest::test_applying_fixits_insert_containing_newline): New
25799 (selftest::test_applying_fixits_replace_containing_newline): New
25801 (selftest::insert_line): New function.
25802 (selftest::test_applying_fixits_multiple_lines): Add example of
25804 (selftest::edit_context_c_tests): Call the new tests.
25806 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25808 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25809 parameter cand. Update dump information.
25810 (get_computation_cost): Update uses.
25812 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25814 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25815 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25816 (get_computation_at, rewrite_use_address): Update use of
25817 get_computation_aff.
25819 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25821 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25822 (get_computation): Delete.
25823 (get_computation_cost): Implement like get_computation_cost_at.
25824 Use get_computation_at.
25825 (get_computation_cost_at): Delete.
25826 (rewrite_use_nonlinear_expr): Use get_computation_at.
25827 (rewrite_use_compare, remove_unused_ivs): Ditto.
25829 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25831 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25833 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25835 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25836 (ivopts_global_cost_for_size): Rename parameter and update uses.
25837 (iv_ca_recount_cost): Update uses.
25838 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25839 candidates seperately in n_invs and n_cands.
25840 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25842 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25844 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25845 (find_inv_vars_cb): New.
25846 (find_depends): Renamed to ...
25847 (find_inv_vars): ... this.
25848 (add_candidate_1, force_var_cost): Call find_inv_vars.
25849 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25851 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25853 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25854 inv_vars. Add inv_exprs.
25855 (struct iv_cand): Rename depends_on to inv_vars.
25856 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25857 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25858 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25859 (dump_cand): Dump inv_vars.
25860 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25861 (record_invariant, find_depends, add_candidate_1): Ditto.
25862 (set_group_iv_cost, force_var_cost): Ditto.
25863 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25864 (get_computation_cost_at, get_computation_cost): Ditto.
25865 (determine_group_iv_cost_generic): Ditto.
25866 (determine_group_iv_cost_address): Ditto.
25867 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25868 (determine_group_iv_costs): Ditto.
25869 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25870 (iv_ca_set_remove_invariants): Renamed to ...
25871 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25872 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25873 (iv_ca_set_add_invariants): Renamed to ...
25874 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25875 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25876 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25877 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25878 (create_new_ivs): Remove useless dump.
25880 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25882 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25884 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25885 (iv_ca_set_no_cp, create_new_iv): Ditto.
25887 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25889 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25891 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25893 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25894 function tree_check2.
25896 2017-05-02 Martin Liska <mliska@suse.cz>
25898 * doc/gcov.texi: Add missing preposition.
25899 * gcov.c (function_info::function_info): Properly fill up
25900 all member variables.
25902 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25904 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25906 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25908 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25910 2017-05-02 Martin Liska <mliska@suse.cz>
25913 * lto-streamer-in.c (lto_read_tree_1): Remove
25914 LTO_STREAMER_DEBUG.
25915 * lto-streamer.c (struct tree_hash_entry): Likewise.
25916 (struct tree_entry_hasher): Likewise.
25917 (tree_entry_hasher::hash): Likewise.
25918 (tree_entry_hasher::equal): Likewise.
25919 (lto_streamer_init): Likewise.
25920 (lto_orig_address_map): Likewise.
25921 (lto_orig_address_get): Likewise.
25922 (lto_orig_address_remove): Likewise.
25923 * lto-streamer.h: Likewise.
25924 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25925 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25927 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25929 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25930 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25931 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25932 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25933 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25934 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25935 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25936 (mm_maskz_sub_ss): New intrinsics.
25937 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25939 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25940 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25941 (__builtin_ia32_subss_mask_round): New builtins.
25942 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25943 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25944 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25946 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25947 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25949 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25952 2017-05-02 Martin Jambor <mjambor@suse.cz>
25954 PR tree-optimization/78687
25955 * tree-sra.c (access): New field parent.
25956 (process_subtree_disqualification): New function.
25957 (disqualify_candidate): Call it.
25958 (build_accesses_from_assign): Reset write flag if creating an
25960 (build_access_subtree): Fill in parent field and also prpagate
25961 down grp_write flag.
25962 (create_artificial_child_access): New parameter set_grp_write, set
25963 grp_write to its value.
25964 (propagate_subaccesses_across_link): Also propagate grp_write flag
25966 (propagate_all_subaccesses): Push the closest parent back to work
25967 queue if add_access_to_work_queue returned true.
25969 2017-05-02 Richard Biener <rguenther@suse.de>
25971 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25972 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25973 -fstrict-overflow documentation.
25974 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25975 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25976 flag_strict_overflow.
25977 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25978 * lto-opts.c (lto_write_options): Do not stream it.
25979 * lto-wrapper.c (merge_and_complain): Do not handle it.
25980 * opts.c (default_options_table): Do not set -fstrict-overflow.
25981 (finish_options): Likewise do not clear it when sanitizing.
25982 * simplify-rtx.c (simplify_const_relational_operation): Do not
25983 test flag_strict_overflow.
25985 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25987 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25988 using enabled attribute.
25989 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25990 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25991 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25992 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25993 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25994 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25995 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25996 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25997 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25998 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26000 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26002 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26004 2017-05-02 Richard Biener <rguenther@suse.de>
26006 PR tree-optimization/80591
26008 2017-04-10 Richard Biener <rguenther@suse.de>
26010 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26013 2017-05-02 Richard Biener <rguenther@suse.de>
26015 PR tree-optimization/80549
26016 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26017 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26018 headers if they do not have a preheader.
26020 2017-05-02 Martin Liska <mliska@suse.cz>
26023 * common.opt: Fix typo.
26024 * doc/invoke.texi: Likewise.
26026 2017-05-01 Jan Beulich <jbeulich@suse.com>
26028 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26029 swapping, add (x,x,m,x,n) alternative.
26031 2017-05-01 Nathan Sidwell <nathan@acm.org>
26033 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26034 unnecessary unadjusted_alignment check.
26036 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26039 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26041 * gimplify.c (gimplify_cilk_detach): New function.
26042 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26043 * tree-core.h: Document EXPR_CILK_SPAWN.
26044 * tree.h (EXPR_CILK_SPAWN): Define.
26046 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26048 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26049 to use new fixit_hint representation, using the "replace" logic.
26050 (get_line_span_for_fixit_hint): Likewise.
26051 (layout::print_any_fixits): Likewise.
26052 (selftest::test_one_liner_many_fixits): Rename to...
26053 (selftest::test_one_liner_many_fixits_1): ...this, and update
26054 comment and expected output to reflect that the multiple fix-it
26055 hints are now consolidated into one insertion.
26056 (selftest::test_one_liner_many_fixits_2): New test.
26057 (selftest::test_diagnostic_show_locus_one_liner): Update for
26059 (selftest::test_fixit_consolidation): Update for fix-it API
26061 * diagnostic.c (print_parseable_fixits): Likewise.
26062 * edit-context.c (edited_line::m_line_events): Convert from
26063 auto_vec <line_event *> to auto_vec <line_event>.
26064 (class line_event): Convert from abstract base class to a concrete
26065 class, taking over the role of replace_event.
26066 (class insert_event): Delete.
26067 (class replace_event): Rename to class line_event. Convert to
26069 (edit_context::add_fixits): Reimplement.
26070 (edit_context::apply_insert): Delete.
26071 (edit_context::apply_replace): Rename to...
26072 (edit_context::apply_fixit): ...this. Convert to half-open range.
26073 (edited_file::apply_insert): Delete.
26074 (edited_file::apply_replace): Rename to...
26075 (edited_file::apply_fixit): ...this.
26076 (edited_line::~edited_line): Drop deletion of events.
26077 (edited_line::apply_insert): Delete.
26078 (edited_line::apply_replace): Rename to...
26079 (edited_line::apply_fixit): ...this. Convert to half-open range.
26080 Update for change to type of m_line_events.
26081 * edit-context.h (edit_context::apply_insert): Delete.
26082 (edit_context::apply_replace): Rename to...
26083 (edit_context::apply_fixit): ...this.
26085 2017-05-01 Martin Sebor <msebor@redhat.com>
26087 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26090 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26093 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26094 __get_cpuid_max returns 0.
26095 (__get_cpuid_count): Ditto.
26097 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26099 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26100 replacement expression is another instance of one of its arguments.
26102 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26105 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26106 check for stack push/pop autoinc.
26107 * config/i386/i386.c (ix86_agi_dependent): Return false
26108 if the only reason why modified_in_p returned true is that
26109 addr is SP based and set_insn is a push or pop.
26111 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26113 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26116 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26119 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26120 (account_size_time): Use two predicates - exec_pred and
26122 (evaluate_conditions_for_known_args): Compute both clause and
26124 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26125 (inline_summary_t::duplicate): Update.
26126 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26128 (compute_inline_parameters): Likewise.
26129 (estimate_edge_size_and_time): Update caluclation of time.
26130 (estimate_node_size_and_time): Compute both time and nonspecialized
26132 (estimate_ipcp_clone_size_and_time): Update.
26133 (inline_merge_summary): Update.
26134 (do_estimate_edge_time): Update.
26135 (do_estimate_edge_size): Update.
26136 (do_estimate_edge_hints): Update.
26137 (inline_read_section, inline_write_summary): Stream both new predicates.
26138 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26140 (compute_inlined_call_time): Cleanup.
26141 (big_speedup_p): Update.
26142 (edge_badness): Update.
26143 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26144 (size_time_entry): Replace predicate by exec_predicate and
26145 nonconst_predicate.
26146 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26147 (estimate_edge_time): Return also nonspec_time.
26148 (reset_edge_growth_cache): Update.
26150 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26152 PR rtl-optimization/80491
26153 * ifcvt.c (noce_process_if_block): When looking for x setter
26154 with missing else_bb, don't check only the insn right before
26155 cond_earliest, but look for the last insn that x is modified in
26156 within the same bb.
26158 PR rtl-optimization/80491
26159 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26161 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26163 PR tree-optimization/80487
26164 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26166 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26168 PR tree-optimization/79697
26169 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26170 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26171 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26173 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26174 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26176 2017-04-28 Martin Sebor <msebor@redhat.com>
26178 PR tree-optimization/80523
26179 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26180 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26182 (maybe_warn, format_directive, parse_directive): Use new functions.
26183 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26185 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26187 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26189 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26191 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26192 target_header_dir): Set correctly.
26193 * configure: Regenerated.
26194 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26195 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26196 instead of SYSTEM_HEADER_DIR.
26198 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26200 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26201 (estimate_local_effects): Likewise.
26202 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26203 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26204 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26205 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26206 do_estimate_edge_time, estimate_edge_time): Likewise.
26207 * ipa-inline-analysis.c (estimate_node_size_and_time,
26208 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26209 (estimate_time_after_inlining): Remove.
26211 2017-04-28 Martin Liska <mliska@suse.cz>
26213 * doc/gcov.texi: Enhance documentation of gcov.
26215 2017-04-28 Martin Liska <mliska@suse.cz>
26217 * doc/gcov.texi: Sort options in alphabetic order.
26218 * doc/gcov-dump.texi: Likewise.
26219 * doc/gcov-tool.texi: Likewise.
26220 * gcov.c (print_usage): Likewise.
26221 * gcov-dump.c (print_usage): Likewise.
26222 * gcov-tool.c (print_merge_usage_message): Likewise.
26223 (print_rewrite_usage_message): Likewise.
26224 (print_overlap_usage_message): Likewise.
26226 2017-04-28 Martin Liska <mliska@suse.cz>
26228 PR gcov-profile/53915
26229 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26231 2017-04-28 Martin Liska <mliska@suse.cz>
26233 PR gcov-profile/79891
26234 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26235 is marked by compiler as living on a line.
26236 (get_cycles_count): Remove usage of the union.
26237 (output_intermediate_file): Likewise.
26238 (find_source): Fix GNU coding style.
26239 (accumulate_line_counts): Remove old non-all block mode.
26240 (output_lines): Remove usage of the union.
26241 * profile.c (output_location): Include all BBs, even if
26242 belonging to a same line (and file) as a previous BB.
26244 2017-04-28 Martin Liska <mliska@suse.cz>
26246 * gcov.c (process_args): Handle new argument 'w'.
26247 (read_graph_file): Assign ID to BBs.
26248 (output_branch_count): Display BB # if verbose flag is set.
26249 (output_lines): Likewise for arcs.
26250 (print_usage): Add '--verbose' option help.
26251 * doc/gcov.texi: Document --verbose (-w) option.
26253 2017-04-28 Martin Liska <mliska@suse.cz>
26255 * gcov.c (struct block_location_info): New struct.
26256 (process_file): Fill up the new structure.
26257 (read_graph_file): Replace usage of encoding by the newly added
26259 (add_line_counts): Likewise.
26260 (accumulate_line_counts): Remove usage of the union.
26261 (function_info::function_info): New function.
26262 (function_info::~function_info): Likewise.
26263 (process_file): Call delete instead of release_function.
26264 (release_function): Release the function.
26265 (release_structures): Call delete instead of release_function.
26266 (solve_flow_graph): Replace usage of num_blocks.
26267 (find_exception_blocks): Likewise.
26268 (output_lines): Fix GNU coding style.
26270 2017-04-28 Martin Liska <mliska@suse.cz>
26273 * coverage.c (coverage_remove_note_file): New function.
26274 * coverage.h: Declare the function.
26275 * toplev.c (finalize): Clean if an error has been seen.
26277 2017-04-28 Martin Liska <mliska@suse.cz>
26279 PR gcov-profile/80031
26280 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26281 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26282 * gcov.c (read_graph_file): Read just number of blocks.
26283 * profile.c (branch_prob): Do not stream 0 flags per a basic
26286 2017-04-28 Martin Liska <mliska@suse.cz>
26288 * gcov-dump.c (tag_*): Add new argument to declarations.
26289 (dump_gcov_file): Likewise.
26290 (tag_blocks): Add and use new argument depth.
26291 (tag_arcs): Likewise.
26292 (tag_lines): Likewise.
26293 (tag_counters): Likewise.
26294 (tag_summary): Likewise.
26295 (dump_working_sets): Use depth to do a proper indentation.
26297 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26300 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26301 * symtab.c (symtab_node::debug_symtab): Move definition here.
26303 2017-04-28 Richard Biener <rguenther@suse.de>
26305 * lto-streamer.h (LTO_major_version): Bump to 7.
26307 2017-04-28 Richard Biener <rguenther@suse.de>
26309 * tree-vrp.c (assert_info): New struct.
26310 (add_assert_info): New helper.
26311 (register_edge_assert_for_2): Refactor to add asserts to a vector
26313 (register_edge_assert_for_1): Likewise.
26314 (register_edge_assert_for): Likewise.
26315 (finish_register_edge_assert_for): New helper actually registering
26316 asserts where live on edge.
26317 (find_conditional_asserts): Adjust.
26318 (find_switch_asserts): Likewise.
26319 (evrp_dom_walker::try_find_new_range): Generalize.
26320 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26322 2017-04-27 Marek Polacek <polacek@redhat.com>
26325 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26326 arg10 and arg11 to itype.
26328 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26330 * doc/extend.texi (Object Size Checking): Improve grammar.
26332 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26335 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26336 that the logic for permitting reciprocal estimates matches that
26339 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26342 * tree.c (type_cache_hasher::equal): Only compare
26343 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26344 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26345 non-aggregate element types.
26346 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26347 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26349 2017-04-27 Richard Biener <rguenther@suse.de>
26351 PR middle-end/80533
26352 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26353 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26354 keeping a reference to a trailing array.
26356 2017-04-27 Richard Biener <rguenther@suse.de>
26358 PR middle-end/80539
26359 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26360 being in loop-closed SSA form conservatively.
26361 (chrec_fold_multiply_poly_poly): Likewise.
26363 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26365 PR middle-end/79665
26366 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26367 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26369 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26372 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26373 (aarch64_function_arg_alignment): Return unsigned int again, but still
26374 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26375 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26376 Don't emit -Wpsabi note.
26377 (aarch64_function_arg_boundary): Likewise.
26378 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26381 2017-04-26 Nathan Sidwell <nathan@acm.org>
26383 * tree.h (crc32_unsigned_n): Declare.
26384 (crc32_unsigned, crc32_unsigned): Make inline.
26385 * tree.c (crc32_unsigned_bits): Replace with ...
26386 (crc32_unsigned_n): ... this.
26387 (crc32_unsigned, crc32_byte): Remove.
26388 (crc32_string): Remove unnecessary braces.
26390 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26392 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26393 * ipa-inline-analysis.c (MAX_TIME): Remove.
26394 (account_size_time): Use sreal for time.
26395 (dump_inline_summary): Update.
26396 (estimate_function_body_sizes): Update.
26397 (estimate_edge_size_and_time): Update.
26398 (estimate_calls_size_and_time): Update.
26399 (estimate_node_size_and_time): Update.
26400 (inline_merge_summary): Update.
26401 (inline_update_overall_summary): Update.
26402 (estimate_time_after_inlining): Update.
26403 (inline_read_section): Update.
26404 (inline_write_summary): Update.
26405 * ipa-inline.c (compute_uninlined_call_time): Update.
26406 (compute_inlined_call_time): Update.
26407 (recursive_inlining): Update.
26408 (inline_small_functions): Update.
26409 (dump_overall_stats): Update.
26410 * ipa-inline.h: Include sreal.h.
26411 (size_time_entry): Turn time to sreal.
26412 (inline_summary): Turn self_time nad time to sreal.
26414 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26416 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26418 (sreal::stream_out, sreal::stream_in): New.
26419 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26421 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26423 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26426 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26429 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26431 (dimode_scalar_chain::compute_convert_gain): Ditto.
26432 (dimode_scalar_chain::make_vector_copies): Ditto.
26433 (dimode_scalar_chain::convert_reg): Ditto.
26434 (dimode_scalar_chain::convert_insn): Ditto.
26435 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26436 (VI248_AVX512BW_1): New mode iterator.
26437 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26438 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26441 2017-04-25 Martin Sebor <msebor@redhat.com>
26443 PR tree-optimization/80497
26444 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26445 constants are representable in HOST_WIDE_INT.
26446 (parse_directive): Ditto.
26448 2017-04-25 Martin Sebor <msebor@redhat.com>
26451 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26452 (new_zero_array): Adjust signature.
26453 (dom_info::dom_init): Used unsigned rather that size_t.
26454 (dom_info::dom_info): Same.
26456 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26457 Jakub Jelinek <jakub@redhat.com>
26460 * config/arm/arm.c: Include gimple.h.
26461 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26462 returns negative, increment ncrn only if it returned positive.
26463 (arm_needs_doubleword_align): Return int instead of bool,
26464 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26465 members, but if there is any such non-FIELD_DECL
26466 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26467 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26468 returns negative, increment nregs only if it returned positive.
26469 (arm_setup_incoming_varargs): Likewise.
26470 (arm_function_arg_boundary): Emit -Wpsabi note if
26471 arm_needs_doubleword_align returns negative, return
26472 DOUBLEWORD_ALIGNMENT only if it returned positive.
26474 2017-04-25 Marek Polacek <polacek@redhat.com>
26477 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26478 first argument to type.
26480 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26484 type checks to test for compatibility instead of equality.
26486 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26487 Jakub Jelinek <jakub@redhat.com>
26490 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26492 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26493 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26494 the alignment computation, but return their maximum in warn_alignment.
26495 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26496 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26498 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26499 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26502 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26504 * config/arc/simdext.md (dmpyh): Fix typo.
26506 2017-04-25 Richard Biener <rguenther@suse.de>
26508 PR tree-optimization/80492
26509 * alias.c (compare_base_decls): Handle registers with asm
26510 specification conservatively.
26511 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26512 compare_base_decls returning dont-know properly.
26514 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26516 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26517 (legitimate_offset_address_p): New function.
26518 (arc_legitimate_address_p): Use above function.
26520 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26522 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26524 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26526 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26527 ACCH registers whenever they are available.
26529 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26531 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26532 double regs fix when not used.
26534 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26536 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26538 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26539 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26543 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26544 when emitting short ADD instructions.
26546 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26548 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26550 (cmpsi_cc_c_insn): Likewise.
26551 (cbranchsi4_scratch): Compute proper instruction length using
26552 compact_hreg_operand.
26553 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26555 2017-04-25 Richard Biener <rguenther@suse.de>
26557 PR middle-end/80509
26558 * passes.c (pass_manager::pass_manager): Initialize
26559 m_name_to_pass_map.
26561 2017-04-25 Richard Biener <rguenther@suse.de>
26563 PR tree-optimization/79201
26564 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26566 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26569 * config/s390/vector.md: Split MEM->GPR vector moves for
26570 non-s_operand addresses.
26572 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26575 * config/s390/predicates.md (reload_const_wide_int_operand): New
26577 * config/s390/s390.md ("movti"): Remove d/P alternative.
26578 ("movti_bigconst"): New pattern definition.
26580 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26583 * s390-protos.h (s390_expand_cs_hqi): Removed.
26584 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26585 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26586 modes as well as CCZ1mode and CCZmode.
26587 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26588 signature of s390_emit_compare_and_swap.
26589 (s390_expand_cs_hqi): Likewise, make static.
26590 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26591 compare-and-swap, in some cases.
26592 (s390_expand_cs): Wrapper function.
26593 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26595 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26596 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26597 patterns for small and large integers. Forbid symref memory operands.
26598 Move expander to s390.c. Require cc register.
26599 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26600 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26601 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26602 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26603 symref memory operands. Remove CC mode and call s390_match_ccmode
26605 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26607 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26609 * config/s390/s390.md (define_peephole2): New peephole to help
26610 combining the load-and-test pattern with volatile memory.
26612 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26614 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26615 with CCZmode for TARGET_Z196.
26617 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26619 PR rtl-optimization/80501
26620 * combine.c (make_compound_operation_int): Set subreg_code to SET
26621 even for AND with mask of the sign bit of mode.
26623 PR rtl-optimization/80500
26624 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26625 sum's initial value.
26627 2017-04-25 Julian Brown <julian@codesourcery.com>
26628 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26630 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26632 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26634 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26636 2017-04-25 Julian Brown <julian@codesourcery.com>
26637 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26639 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26640 (thunderx2t99_sha): New Reservation.
26642 2017-04-25 Julian Brown <julian@codesourcery.com>
26643 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26645 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26646 type for 1-element load.
26648 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26650 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26652 2017-04-24 Martin Jambor <mjambor@suse.cz>
26654 PR tree-optimization/80293
26655 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26656 char arrays not totally scalarizable if it is false.
26657 (analyze_all_variable_accesses): Pass correct value in the new
26658 parameter. Add a statistics counter.
26660 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26662 PR middle-end/79931
26663 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26665 2017-04-24 Richard Biener <rguenther@suse.de>
26667 PR tree-optimization/80494
26668 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26669 out for complex types.
26671 2017-04-24 Richard Biener <rguenther@suse.de>
26673 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26674 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26675 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26676 (DFS): Adjust and never fail.
26677 (sccvn_dom_walker::fail): Remove.
26678 (sccvn_dom_walker::before_dom_children): Adjust.
26679 (run_scc_vn): Likewise and never fail.
26680 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26681 (pass_fre::execute): Likewise.
26683 2017-04-24 Richard Biener <rguenther@suse.de>
26685 PR tree-optimization/79725
26686 * tree-ssa-sink.c (statement_sink_location): Return whether
26687 failure reason was zero uses. Move that check later.
26688 (sink_code_in_bb): Deal with zero uses by removing the stmt
26691 2017-04-24 Richard Biener <rguenther@suse.de>
26694 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26695 pointer-based references.
26697 2017-04-24 Richard Biener <rguenther@suse.de>
26700 * pass_manager.h (pass_manager::operator new): Remove.
26701 (pass_manager::operator delete): Likewise.
26702 * passes.c (pass_manager::operator new): Remove.
26703 (pass_manager::operator delete): Likewise.
26704 (pass_manager::pass_manager): Zero individual pass members.
26706 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26709 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26710 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26711 Check "XEXP (src, 1)" operand here.
26712 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26713 Check "XEXP (src, 1)" operand here.
26714 (dimode_scalar_chain::make_vector_copies): Detect count register
26715 of a shift instruction. Zero extend count register from QImode
26716 to DImode to satisfy vector shift pattern count operand predicate.
26717 Substitute vector shift count operand with a DImode copy.
26718 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26721 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26723 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26724 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26725 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26726 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26727 (UNSPEC_NOREX_MEM): Remove definition.
26729 2017-04-21 Richard Biener <rguenther@suse.de>
26731 PR tree-optimization/79547
26732 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26733 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26734 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26735 without any constraints.
26737 2017-04-21 Richard Biener <rguenther@suse.de>
26739 PR tree-optimization/78847
26740 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26742 2017-04-21 Richard Biener <rguenther@suse.de>
26744 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26745 (build_distinct_type_copy): Likewise.
26746 (build_variant_type_copy): Likewise.
26747 * tree.c (build_qualified_type): Pass down mem-stat info.
26748 (build_distinct_type_copy): Likewise.
26749 (build_variant_type_copy): Likewise.
26751 2017-04-21 Richard Biener <rguenther@suse.de>
26753 PR tree-optimization/80237
26754 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26756 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26757 for a simplified result.
26759 2016-04-21 Richard Biener <rguenther@suse.de>
26761 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26762 sth as strict as a simple_iv but a chrec without symbols and an
26763 operand defined in the loop we are peeling (and not some subloop).
26764 (propagate_constants_for_unrolling): Propagate all constants.
26766 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26769 * config/i386/i386.c (print_reg): Remove assert for disalowed
26770 regno values, call output_operand_lossage instead.
26772 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26775 * config/i386/constraints.md (Yc): New register constraint.
26776 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26777 Use Yc constraint for alternative 2 of operand 0. Remove
26778 preferred_for_speed attribute.
26780 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26782 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26783 lastprivate clauses in SIMT case.
26785 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26787 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26789 2017-04-20 Richard Biener <rguenther@suse.de>
26791 PR tree-optimization/57796
26792 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26793 as N scalar stores.
26794 (vect_model_load_cost): Cost gathers as N scalar loads.
26796 2017-04-20 Richard Biener <rguenther@suse.de>
26798 * ggc-page.c (ggc_allocated_p): Rename to ...
26799 (safe_lookup_page_table_entry): ... this and return the lookup
26801 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26803 2017-04-20 Richard Biener <rguenther@suse.de>
26805 PR tree-optimization/80453
26806 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26807 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26808 from the conditions.
26809 (vn_phi_eq): Pass them down.
26810 (vn_phi_lookup): Record them.
26811 (vn_phi_insert): Likewise.
26813 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26815 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26816 uninitialized variable warning to avoid buffer overrun.
26818 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26821 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26822 is suppressed for '{ 0 }' in C.
26824 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26826 * BASE-VER: Set to 8.0.0.
26828 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26830 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26831 priority .init_array and .fini_array section with SECTION_NOTYPE
26834 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26836 PR middle-end/80423
26837 * tree.h (build_array_type): Add typeless_storage default argument.
26838 * tree.c (type_cache_hasher::equal): Also compare
26839 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26840 (build_array_type): Add typeless_storage argument, set
26841 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26843 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26844 (build_array_type): Likewise. Add typeless_storage argument.
26846 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26847 Jakub Jelinek <jakub@redhat.com>
26849 PR tree-optimization/80426
26850 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26851 operation on symbolic operands, also compute the overflow for the
26852 invariant part when the operation degenerates into a negation.
26854 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26857 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26858 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26861 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26863 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26866 * config/avr/avr.c (tree.h): Include it.
26867 (cgraph.h): Include it.
26868 (avr_encode_section_info): Don't warn for uninitialized progmem
26869 variable if it's just an alias.
26871 2017-04-19 Richard Biener <rguenther@suse.de>
26874 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26875 when needed by AutoPGO.
26877 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26880 * doc/lto.texi: Remove an extra 'that'.
26882 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26884 PR rtl-optimization/80429
26885 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26886 are only used in debug insns.
26888 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26889 Vladimir Makarov <vmakarov@redhat.com>
26891 * config/sparc/predicates.md (input_operand): Add comment. Return
26892 true for any memory operand when LRA is in progress.
26893 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26895 2017-04-18 Jeff Law <law@redhat.com>
26898 * mips.md ({return,simple_return}_internal): Do not overwrite
26901 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26903 PR tree-optimization/80443
26904 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26905 instead of adding 1, subtract -1 and similarly instead of subtracting
26908 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26910 PR rtl-optimization/80357
26911 * haifa-sched.c (tmp_bitmap): New variable.
26912 (model_recompute): Handle duplicate use records.
26913 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26914 (free_global_sched_pressure_data): Free it.
26916 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26919 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26920 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26921 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26922 instead of SYSTEM_HEADER_DIR.
26924 2017-04-18 Jeff Law <law@redhat.com>
26926 PR middle-end/80422
26927 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26928 predecessors after walking up the insn chain.
26930 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26933 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26934 sizetype type into debug info.
26936 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26939 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26940 unneeded test for TARGET_UPPER_REGS_SF.
26941 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26943 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26946 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26947 instead of gsi_after_labels.
26949 2017-04-18 Jeff Law <law@redhat.com>
26951 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26955 2017-04-13 Jeff Law <law@redhat.com>
26956 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26957 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26959 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26962 * config/avr/avr.c (intl.h): Include it.
26963 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26965 2017-04-18 Martin Liska <mliska@suse.cz>
26967 PR gcov-profile/78783
26968 * gcov-tool.c (gcov_output_files): Validate that destination
26969 file is either removed by the tool or by a user.
26971 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26972 Guy Benyei <guybe@mellanox.com>
26974 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26975 block, and do not negate it, the stored id is already negative.
26977 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26979 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26981 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26984 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26985 masks of options that should be turned off if the VSX vector
26986 options are turned off.
26987 (OTHER_P8_VECTOR_MASKS): Likewise.
26988 (OTHER_VSX_VECTOR_MASKS): Likewise.
26989 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26990 rs6000_disable_incompatible_switches to validate no type switches
26992 (rs6000_incompatible_switch): New function to disallow turning on
26993 other vector options if -mno-vsx, -mno-power8-vector, or
26994 -mno-power9-vector are specified.
26996 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26998 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27000 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27002 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27003 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27004 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27005 (ARG_POINTER_CFA_OFFSET): Likewise.
27007 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27009 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27010 conditions to take advantage of various optimizations.
27012 2017-04-13 Jeff Law <law@redhat.com>
27014 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27015 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27016 (zero_extendsidi2_dext): Likewise.
27018 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27021 * fold-const.c (fold_ternary_loc): Revert
27022 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27025 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27027 PR rtl-optimization/80343
27028 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27029 new scratch pseudo.
27031 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27034 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27035 to ubsan_encode_value.
27037 2017-04-13 Jeff Law <law@redhat.com>
27039 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27040 appearing in DEBUG_INSNs.
27042 2017-04-13 Martin Liska <mliska@suse.cz>
27044 PR gcov-profile/80413
27045 * gcov-io.c (gcov_write_string): Copy to buffer just when
27046 allocated size is greater than zero.
27048 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27051 * dwarf2out.c (decls_for_scope): Ignore declarations of
27052 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27054 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27057 * ipa-visibility.c (non_local_p): Fix typos.
27058 (localize_node): When localizing symbol in same comdat group,
27059 dissolve the group only when we know external symbols are going
27061 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27063 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27065 PR tree-optimization/79390
27066 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27067 order does not result in usable sequence, retry with reversed operand
27073 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27074 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27075 op0 instead of fold_convert_loc (loc, type, arg0).
27077 2017-04-12 Jeff Law <law@redhat.com>
27079 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27080 has a delay slot in the generated code.
27082 * config/cris/cris.md (cris_preferred_reload_class): Return
27083 GENNONACR_REGS rather than GENERAL_REGS.
27085 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27088 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27089 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27090 signedness of the result type.
27092 2017-04-12 Richard Biener <rguenther@suse.de>
27093 Jeff Law <law@redhat.com>
27095 PR tree-optimization/80359
27096 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27097 trim stores to TARGET_MEM_REFs.
27099 2017-04-12 Richard Biener <rguenther@suse.de>
27101 PR tree-optimization/79390
27102 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27103 threading case even more.
27105 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27108 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27109 for quad_address_p for TImode, instead of just not indexed_address.
27111 2017-04-12 Richard Biener <rguenther@suse.de>
27112 Bernd Edlinger <bernd.edlinger@hotmail.de>
27114 PR middle-end/79671
27115 * alias.c (component_uses_parent_alias_set_from): Handle
27116 TYPE_TYPELESS_STORAGE.
27117 (get_alias_set): Likewise.
27118 * tree-core.h (tree_type_common): Add typeless_storage flag.
27119 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27120 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27121 for types containing members with TYPE_TYPELESS_STORAGE.
27122 (place_field): Likewise.
27123 (layout_type): Likewise for ARRAY_TYPE.
27124 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27125 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27126 TYPE_TYPELESS_STORAGE.
27127 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27129 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27132 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27133 first argument to type.
27135 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27139 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27140 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27141 (rs6000_expand_binop_builtin): Likewise.
27142 (rs6000_expand_ternop_builtin): Likewise; also add missing
27143 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27144 vshasigma built-ins.
27145 * doc/extend.texi: Document that vec_xxpermdi's third argument
27146 must be a constant.
27148 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27150 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27151 Use shift_const cost parameter when calculating gain of STV shifts.
27153 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27155 PR rtl-optimization/70478
27156 * lra-constraints.c (process_alt_operands): Check memory for
27157 disfavoring memory insn operand.
27159 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27161 PR middle-end/80100
27162 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27163 left shift in unsigned HOST_WIDE_INT type.
27165 PR rtl-optimization/80385
27166 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27167 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27170 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27171 if they have any depend clauses.
27173 2017-04-11 Martin Liska <mliska@suse.cz>
27176 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27177 * ipa-split.c (split_function): Create a local comdat symbol
27178 if caller is in a comdat group.
27180 2017-04-11 Martin Liska <mliska@suse.cz>
27183 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27186 2017-04-11 Martin Sebor <msebor@redhat.com>
27188 PR middle-end/80364
27189 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27190 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27192 (directive::set_width, directive::set_precision, format_character):
27194 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27197 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27200 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27201 conflict, set target->arch_name instead of target->cpu_name.
27203 2017-04-11 Richard Biener <rguenther@suse.de>
27205 PR tree-optimization/80374
27206 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27207 build_zero_cst, remove fold_convertible_p check again.
27209 2017-04-11 Martin Liska <mliska@suse.cz>
27212 * ubsan.c (instrument_object_size): Do not instrument register
27215 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27218 * config/i386/i386-builtin-types.def
27219 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27220 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27221 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27222 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27223 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27224 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27225 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27226 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27227 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27228 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27229 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27230 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27231 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27232 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27233 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27234 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27235 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27236 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27237 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27238 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27239 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27240 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27241 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27242 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27243 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27244 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27245 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27246 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27247 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27248 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27249 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27250 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27251 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27252 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27253 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27254 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27255 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27256 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27257 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27258 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27259 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27260 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27261 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27262 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27263 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27264 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27266 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27267 flag to second_arg_count, handle 4 argument function type _COUNT
27268 aliases, handle second_arg_count on second argument rather than last.
27270 2017-04-10 Jeff Law <law@redhat.com>
27272 PR tree-optimization/80374
27273 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27274 record anything if we can not convert integer_zero_node to the
27277 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27280 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27281 Enhance special handling given to the TARGET_P9_MINMAX option in
27282 relation to certain other options.
27284 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27286 PR tree-optimization/80153
27287 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27288 remove POINTER_PLUS_EXPR's base part directly, rather than through
27291 2017-04-10 Richard Biener <rguenther@suse.de>
27292 Bin Cheng <bin.cheng@arm.com>
27294 PR tree-optimization/80153
27295 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27296 the first element of pointer type aff_tree. Build result expr in
27298 (add_elt_to_tree): Convert to type unconditionally. Remove other
27299 fold_convert calls.
27300 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27301 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27303 2017-04-10 Richard Biener <rguenther@suse.de>
27305 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27308 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27310 PR rtl-optimization/70478
27311 * lra-constraints.c (curr_small_class_check): New.
27312 (update_and_check_small_class_inputs): New.
27313 (process_alt_operands): Update curr_small_class_check. Disfavor
27314 alternative insn memory operands. Check available regs for small
27317 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27320 * config/mips/mips.opt (-mvirt): Update description.
27321 * doc/invoke.texi (-mvirt): Likewise.
27323 2017-04-10 Richard Biener <rguenther@suse.de>
27325 PR middle-end/80362
27326 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27327 looking for NEGATE_EXPR in -A / -B to A / B folding.
27329 2017-04-10 Martin Liska <mliska@suse.cz>
27331 PR gcov-profile/80224
27332 * gcov.c (print_usage): Fix usage string.
27333 (get_gcov_intermediate_filename): Remove.
27334 (output_gcov_file): Use both for normal and intermediate format.
27335 (generate_results): Do not initialize special file for
27336 intermediate format.
27338 2017-04-10 Richard Biener <rguenther@suse.de>
27340 PR tree-optimization/80304
27341 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27344 2017-04-10 Nathan Sidwell <nathan@acm.org>
27347 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27348 (rs6000_init_builtins): Use it.
27350 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27352 * config/arm/arm.md (<mrc>): Add mode to SET source.
27353 (<mrrc>): Likewise.
27355 2017-04-10 Richard Biener <rguenther@suse.de>
27357 PR middle-end/80344
27358 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27360 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27363 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27364 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27365 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27366 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27367 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27368 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27369 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27370 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27371 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27372 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27373 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27374 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27375 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27376 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27377 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27378 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27379 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27380 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27381 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27382 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27383 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27384 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27385 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27387 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27389 PR rtl-optimization/70478
27390 * lra-constraints.c: Reverse the last patch.
27392 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27394 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27395 Add comment for WCHAR_T.
27397 2017-04-08 Martin Liska <mliska@suse.cz>
27400 2017-04-07 Martin Liska <mliska@suse.cz>
27403 * ipa-split.c (split_function): Add function part to a same comdat
27406 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27409 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27411 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27413 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27414 (vec_pairing): ...to this.
27415 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27416 (rs6000_sched_init): Adjust for name change.
27417 (struct rs6000_sched_context): Likewise.
27418 (rs6000_init_sched_context): Likewise.
27419 (rs6000_set_sched_context): Likewise.
27421 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27427 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27429 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27430 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27431 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27433 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27435 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27437 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27439 PR rtl-optimization/70703
27440 * ira-color.c (update_conflict_hard_regno_costs): Use
27441 int64_t instead of HOST_WIDE_INT.
27443 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27445 PR rtl-optimization/70478
27446 * lra-constraints.c (process_alt_operands): Disfavor alternative
27447 insn memory operands.
27449 2017-04-07 Jeff Law <law@redhat.com>
27451 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27452 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27454 2017-04-07 Martin Liska <mliska@suse.cz>
27457 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27458 Show error message instead of an ICE.
27460 2017-04-07 Martin Liska <mliska@suse.cz>
27463 * ipa-split.c (split_function): Add function part to a same comdat
27466 2017-04-07 Richard Biener <rguenther@suse.de>
27468 PR middle-end/80341
27469 * tree.c (get_unwidened): Also handle ! for_type case for
27471 * convert.c (do_narrow): Split out from ...
27472 (convert_to_integer_1): ... here. Do not pass final truncation
27473 type to get_unwidened for TRUNC_DIV_EXPR.
27475 2017-04-07 Richard Biener <rguenther@suse.de>
27477 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27479 (aff_combination_const): Adjust.
27480 (aff_combination_scale): Likewise.
27481 (aff_combination_add_elt): Likewise.
27482 (aff_combination_add_cst): Likewise.
27483 (aff_combination_convert): Likewise.
27484 (add_elt_to_tree): Likewise. Remove unused argument.
27485 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27487 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27489 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27491 * config/arm/arm.c (arm_default_short_enums): Use
27492 ARM_DEFAULT_SHORT_ENUMS.
27493 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27495 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27498 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27499 members with redundant out-of-class redeclaration.
27501 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27504 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27505 * config/i386/i386.md (*zero_extendsidi2):
27506 Add (?*x,*x) and (?*v,*v) alternatives.
27508 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27511 * config/i386/i386.c (ix86_expand_builtin)
27512 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27513 mode from insn data. Convert operands to insn operand mode.
27514 Copy operands that don't satisfy insn predicate to a register.
27516 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27518 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27521 2017-04-06 Richard Biener <rguenther@suse.de>
27523 PR tree-optimization/80334
27524 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27525 preserve alignment of accesses.
27527 2017-04-06 Richard Biener <rguenther@suse.de>
27529 PR tree-optimization/80262
27530 * tree-sra.c (build_ref_for_offset): Preserve address-space
27532 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27533 Drop useless address-space information on MEM_REF offsets.
27535 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27537 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27539 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27541 PR rtl-optimization/70703
27542 * ira-color.c (update_conflict_hard_regno_costs): Use
27543 HOST_WIDE_INT instead of long.
27545 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27548 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27549 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27551 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27552 for x86_64 target. Handle -m3dnowa option.
27554 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27556 PR rtl-optimization/70703
27557 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27558 (update_conflict_hard_regno_costs): Use long instead of unsigned
27559 arithmetic for cost calculation.
27561 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27562 Bernd Edlinger <bernd.edlinger@hotmail.de>
27565 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27568 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27571 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27572 ptr_mode with Pmode throughout.
27573 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27574 into probe_stack_range and use DImode.
27576 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27579 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27580 call_eh_return is true.
27582 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27584 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27585 Initialize last_match_fntype_index.
27587 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27590 * tree-nvr.c: Include internal-fn.h.
27591 (pass_return_slot::execute): Ignore internal calls without
27594 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27595 Richard Biener <rguenther@suse.de>
27598 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27599 captures used multiple times, except for the last use.
27600 * generic-match-head.c: Include gimplify.h.
27602 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27604 PR tree-optimization/79390
27605 * target.h (struct noce_if_info): Declare.
27606 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27607 * target.def (noce_conversion_profitable_p): New target hook.
27608 * ifcvt.h (struct noce_if_info): New type, moved from ...
27609 * ifcvt.c (struct noce_if_info): ... here.
27610 (noce_conversion_profitable_p): Renamed to ...
27611 (default_noce_conversion_profitable_p): ... this. No longer
27613 (noce_try_store_flag_constants, noce_try_addcc,
27614 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27615 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27616 instead of noce_conversion_profitable_p.
27617 * config/i386/i386.c: Include ifcvt.h.
27618 (ix86_option_override_internal): Don't override
27619 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27620 (ix86_noce_conversion_profitable_p): New function.
27621 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27622 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27623 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27624 * doc/tm.texi: Regenerated.
27626 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27628 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27631 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27634 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27635 instructions for small multiply cores.
27637 2017-04-04 Jeff Law <law@redhat.com>
27639 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27641 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27642 that are not set by the loop over the elements.
27644 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27647 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27648 int mode, convert_modes it to mode as unsigned, otherwise use
27649 lowpart_subreg to mode rather than SImode.
27650 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27651 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27652 Use DImode instead of SImode for the shift count operand.
27653 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27656 2017-04-04 Richard Biener <rguenther@suse.de>
27658 PR middle-end/80281
27659 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27660 arithmetic done for the negate or the plus. Simplify.
27661 (A - (-B) -> A + B): Likewise.
27662 * fold-const.c (split_tree): Make sure to not negate pointers.
27664 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27666 PR rtl-optimization/60818
27667 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27668 a compare of comparisons with the thing compared if this results
27669 in a different machine mode.
27671 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27673 * alias.c (base_alias_check): Fix typo in comment.
27674 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27675 * cgraphunit.c (symbol_table::compile): Likewise.
27676 * collect2.c (maybe_run_lto_and_relink): Likewise.
27677 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27678 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27679 * config/avr/avr.c (avr_map_op_t): Likewise.
27680 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27681 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27682 * config/epiphany/epiphany.md (movcc): Likewise.
27683 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27684 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27686 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27687 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27688 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27689 * config/sh/sh.c (sh_rtx_costs): Likewise.
27690 * fold-const.c (fold_truth_andor): Likewise.
27691 * genautomata.c (collapse_flag): Likewise.
27692 * gengtype.h (struct type::u::s): Likewise.
27693 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27694 * input.c (FORMAT_AMOUNT): Likewise.
27695 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27696 (known_aggs_to_agg_replacement_list): Likewise.
27697 * ipa-inline-analysis.c: Likewise.
27698 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27699 * ipa-polymorphic-call.c
27700 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27701 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27702 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27704 * modulo-sched.c (apply_reg_moves): Likewise.
27705 * omp-expand.c (build_omp_regions_1): Likewise.
27706 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27707 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27708 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27709 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27710 * value-prof.c: Likewise.
27711 * var-tracking.c (val_reset): Likewise.
27713 2017-04-03 Richard Biener <rguenther@suse.de>
27715 PR tree-optimization/80275
27716 * fold-const.c (split_address_to_core_and_offset): Handle
27719 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27721 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27722 descriptors is at least equal to that of functions.
27724 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27726 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27728 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27731 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27732 (mov<IMOD4:mode>): New expander.
27733 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27735 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27737 PR rtl-optimization/79405
27738 * fwprop.c (propagations_left): New variable.
27739 (forward_propagate_into): Decrement it.
27740 (fwprop_init): Initialize it.
27741 (fw_prop): If the variable has reached zero, stop propagating.
27742 (fwprop_addr): Ditto.
27744 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27747 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27748 a FUNCTION_DECL, pass it as decl instead of origin to
27751 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27753 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27756 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27759 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27760 TARGET_VSX_SMALL_INTEGER.
27762 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27764 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27765 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27767 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27769 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27770 extraction from odd-numbered MSA register.
27772 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27774 PR middle-end/80173
27775 * expmed.c (store_bit_field_1): Don't attempt to create
27776 a word subreg out of hard registers wider than word if they
27777 have HARD_REGNO_NREGS of 1 for their mode.
27779 PR middle-end/80163
27780 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27781 conversions to integer types wider than word and pointer.
27784 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27785 (rtx_equal_for_cselib_p): Pass 0 to it.
27786 * cselib.c (cselib_hasher::equal): Likewise.
27787 (rtx_equal_for_cselib_1): Add depth argument. If depth
27788 is 128, don't look up VALUE locs and punt. Increment
27789 depth in recursive calls when walking VALUE locs.
27791 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27793 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27794 (make_gcov_file_name): Use the canonical path name for generating
27796 (read_line): Fix handling of files with ascii null bytes.
27798 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27800 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27801 to initialise a vector register instead
27802 of using a const_int.
27804 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27806 PR translation/80189
27807 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27808 diagnostic messages.
27810 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27813 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27814 (dfp_diex_<mode>): Update mode of operand 1.
27815 * doc/extend.texi (dxex, dxexq): Document change to return type.
27816 (diex, diexq): Document change to argument type.
27818 2017-03-30 Martin Jambor <mjambor@suse.cz>
27821 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27822 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27823 it reflects the signature changes performed at the callee side.
27824 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27825 to cgraph_build_function_type_skip_args.
27826 (build_function_decl_skip_args): Adjust call to the above function.
27828 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27831 * config/i386/sse.md
27832 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27833 register as dest whenever it is a MEM not rtx_equal_p to the
27834 corresponding dup operand, and when forcing into reg move the
27835 reg into the memory afterwards.
27836 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27837 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27838 for the force_reg mode.
27839 (avx512vl_vextractf128<mode>): Use register as dest either
27840 always when a MEM, or when it is a MEM not rtx_equal_p to the
27841 corresponding dup operand, or even not when it is a CONST_VECTOR
27842 depending on the mode and lo vs. hi.
27843 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27845 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27846 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27847 Likewise. Require that operands[2] is even.
27848 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27849 Remove extraneous parens. Require that operands[2] is a multiple
27851 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27852 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27853 disallow memory then.
27855 2017-03-30 Richard Biener <rguenther@suse.de>
27857 PR tree-optimization/77498
27858 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27859 to non-constants over backedges.
27861 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27863 PR rtl-optimization/80233
27864 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27865 as last_combined_insn. Do not test for BARRIER_P separately.
27867 2017-03-29 Andreas Schwab <schwab@suse.de>
27870 * calls.c (prepare_call_address): Convert funexp to Pmode before
27871 copying to temp reg.
27873 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27875 PR tree-optimization/80158
27876 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27877 Handle possible future case of more than one alternate
27879 (replace_rhs_if_not_dup): Likewise.
27880 (replace_one_candidate): Likewise.
27882 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27884 PR rtl-optimization/80193
27885 * ira.c (ira): Do not check allocation for LRA.
27887 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27889 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27890 (nvptx_output_simt_exit): Declare.
27891 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27892 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27893 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27894 (nvptx_declare_function_name): ...here. Emit declaration of local
27895 memory space buffer for omp_simt_enter insn.
27896 (nvptx_output_unisimt_switch): New.
27897 (nvptx_output_softstack_switch): New.
27898 (nvptx_output_simt_enter): New.
27899 (nvptx_output_simt_exit): New.
27900 * config/nvptx/nvptx.h (struct machine_function): New fields
27901 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27902 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27903 (UNSPECV_SIMT_EXIT): Ditto.
27904 (omp_simt_enter_insn): New insn.
27905 (omp_simt_enter): New expansion.
27906 (omp_simt_exit): New insn.
27907 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27909 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27910 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27911 (expand_GOMP_SIMT_EXIT): New.
27912 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27913 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27914 (GOMP_SIMT_EXIT): Ditto.
27915 * target-insns.def (omp_simt_enter): New insn.
27916 (omp_simt_exit): Ditto.
27917 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27919 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27920 (lower_rec_input_clauses): Likewise.
27921 (lower_lastprivate_clauses): Handle SIMT privatization.
27923 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27924 (ompdevlow_adjust_simt_enter): New.
27925 (find_simtpriv_var_op): New.
27926 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27927 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27929 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27930 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27931 (copy_decl_for_dup_finish): Ditto.
27933 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27935 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27938 * config/i386/i386.c (ix86_option_override_internal): Always
27939 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27941 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27943 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27945 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27947 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27948 mark new edge's irreducible flag accordign to it.
27949 (vect_do_peeling): Check loop preheader edge's irreducible flag
27950 and pass it to function slpeel_add_loop_guard.
27952 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27954 PR tree-optimization/80218
27955 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27956 Update block frequencies and counts.
27958 2017-03-28 Richard Biener <rguenther@suse.de>
27960 PR tree-optimization/78644
27961 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27962 of a simplification result we may not use it at all.
27964 2017-03-28 Richard Biener <rguenther@suse.de>
27967 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27968 without arguments, generate default definition of a SSA name.
27970 2017-03-28 Richard Biener <rguenther@suse.de>
27972 PR middle-end/80222
27973 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27974 TYPE_REF_CAN_ALIAS_ALL references.
27975 * fold-const.c (fold_indirect_ref_1): Likewise.
27977 2017-03-28 Martin Liska <mliska@suse.cz>
27980 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27981 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27983 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27984 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27986 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27987 (EXTRA_SPECS): Define.
27988 (SUBTARGET_EXTRA_SPECS): Likewise.
27989 (SUBTARGET_CPP_SPEC): Likewise.
27990 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27991 SUBTARGET_EXTRA_SPECS.
27992 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27994 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27996 * config/arc/simdext.md (vst64_insn): Update pattern.
27997 (vld32wh_insn): Likewise.
27998 (vld32wl_insn): Likewise.
27999 (vld64_insn): Likewise.
28000 (vld32_insn): Likewise.
28002 2017-03-28 Marek Polacek <polacek@redhat.com>
28005 * fold-const.c (fold_comparison): Use protected_set_expr_location
28006 instead of SET_EXPR_LOCATION.
28008 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28010 * tree.c (add_expr): Avoid name lookup warning.
28012 2017-03-27 Jeff Law <law@redhat.com>
28014 PR tree-optimization/80216
28015 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28016 function name. Limit recursion depth.
28017 (record_temporary_equivalences): Corresponding changes.
28019 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28021 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28024 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28027 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28029 * cfgcleanup.c (reg_note_cfa_p): New array.
28030 (insns_have_identical_cfa_notes): New function.
28031 (old_insns_match_p): Don't cross-jump in between /f
28032 and non-/f instructions. If both i1 and i2 are frame related,
28033 verify all CFA notes, their order and content.
28035 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28038 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28039 HImode and SImode with zero extend to DImode to one insn.
28040 (bswap<mode>2_extenddi): Likewise.
28041 (bswapsi2_extenddi): Likewise.
28042 (bswaphi2_extendsi): Likewise.
28043 (bswaphi2): Combine bswap HImode and SImode into one insn.
28044 Separate memory insns from swapping register.
28045 (bswapsi2): Likewise.
28046 (bswap<mode>2): Likewise.
28047 (bswaphi2_internal): Delete, no longer used.
28048 (bswapsi2_internal): Likewise.
28049 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28050 store, and gpr<-gpr swap insns.
28051 (bswap<mode>2_store): Likewise.
28052 (bswaphi2_reg): Register only splitter, combine with the splitter.
28053 (bswaphi2 splitter): Likewise.
28054 (bswapsi2_reg): Likewise.
28055 (bswapsi2 splitter): Likewise.
28056 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28057 the insns into load, store, and register/register insns.
28058 (bswapdi2_ldbrx): Likewise.
28059 (bswapdi2_load): Likewise.
28060 (bswapdi2_store): Likewise.
28061 (bswapdi2_reg): Likewise.
28063 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28065 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28066 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28068 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28071 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28073 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28074 special handling for target option conflicts between dform
28075 options (-mpower9-dform, -mpower9-dform-vector,
28076 -mpower9-dform-scalar) and -mno-direct-move.
28078 2017-03-27 Richard Biener <rguenther@suse.de>
28080 PR tree-optimization/80181
28081 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28083 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28085 * config/arc/predicates.md (move_double_src_operand): Replace the
28086 call to move_double_src_operand with a call to address_operand.
28088 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28090 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28091 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28092 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28094 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28096 * config/arc/predicates.md (long_immediate_loadstore_operand):
28097 Consider scaled addresses cases.
28099 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28101 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28102 restored when in interrupt.
28103 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28104 doesn't have delay slot.
28106 2017-03-27 Richard Biener <rguenther@suse.de>
28109 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28110 inlined thunk clones.
28112 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28115 * asan.c (instrument_derefs): Copy over last operand from
28116 original COMPONENT_REF to the new COMPONENT_REF with
28117 DECL_BIT_FIELD_REPRESENTATIVE.
28118 * ubsan.c (instrument_object_size): Likewise.
28120 2017-03-27 Richard Biener <rguenther@suse.de>
28122 PR tree-optimization/80170
28123 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28124 sure DR/SCEV didnt fold in constants we do not see when looking
28125 at the reference base alignment.
28127 2017-03-27 Richard Biener <rguenther@suse.de>
28129 PR middle-end/80171
28130 * gimple-fold.c (fold_ctor_reference): Properly guard against
28131 NULL return value from canonicalize_constructor_val.
28133 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28136 * config/i386/i386.c (ix86_expand_builtin)
28137 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28138 flags reg setting and flags reg using instructions.
28139 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28140 clobbering instructions to zero extend op2.
28142 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28144 * doc/install.texi (Configuration) <--with-aix-soname>:
28145 Update link to AIX ld.
28147 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28149 PR rtl-optimization/80160
28150 PR rtl-optimization/80159
28151 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28152 reg_alternate_class into account.
28154 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28157 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28158 to consider in curr_insn_transform.
28160 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28162 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28163 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28164 and emit_mode_inner.
28166 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28168 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28169 argument to the overloaded builtin variants. Use the new flag to
28170 deprecate certain builtin variants.
28171 * config/s390/s390-builtin-types.def: Add new builtin types.
28172 * config/s390/s390-builtins.h: Support new flags field for
28173 overloaded builtins.
28174 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28175 (s390_macro_to_expand): Enable vector float data type.
28176 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28177 builtins by incrementing the __VEC__ version number.
28178 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28180 (s390_resolve_overloaded_builtin): Emit error messages depending
28181 on the builtin flags.
28182 * config/s390/s390.c (s390_expand_builtin): Support additional
28183 flags argument. Change error message to match the messages
28184 emitted in s390-c.c.
28185 * config/s390/s390.md: New UNSPEC_* constants.
28186 (op_type): Add new instruction types.
28187 * config/s390/vecintrin.h: Add new builtins and test data class
28189 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28190 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28191 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28192 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28193 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28194 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28196 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28197 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28198 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28199 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28201 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28202 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28203 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28204 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28206 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28207 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28208 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28209 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28210 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28211 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28212 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28214 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28215 ("vec_scatter_element<V_HW_4:mode>_DI")
28216 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28217 ("vec_fpint<mode>", "vflls")
28218 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28219 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28220 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28221 ("*vec_cmphe<mode>_cc"): ... these.
28223 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28224 mode constant instead of magic value.
28226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28228 * config/s390/s390.c (s390_expand_vec_compare): Support other
28229 vector floating point modes than just V2DF.
28230 (s390_expand_vcond): Likewise.
28231 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28232 (s390_cannot_change_mode_class): Prevent mode changes between TF
28233 and V1TF in vector registers.
28234 * config/s390/s390.md (DF, SF): New mode attributes.
28235 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28236 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28237 SFmode support for VRs.
28238 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28240 (VFT, VF_HW): New mode iterators.
28241 (vw, sdx): New mode attributes.
28242 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28243 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28244 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28245 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28246 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28247 also the new vector floating point modes. Renaming to ...
28249 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28250 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28251 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28252 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28253 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28254 ("vec_unordered<mode>"): ... these.
28256 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28257 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28258 ("*vec_extendv2df"): New insn definitions.
28260 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28262 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28263 ("mulditi3_2", "*muldi3_sign"): New patterns.
28264 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28265 rename the pattern definition.
28267 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28269 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28271 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28273 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28275 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28276 instruction if possible.
28277 * config/s390/vector.md (vec_halfnumelts): New mode
28279 ("*vec_vllezlf<mode>"): New pattern.
28281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28283 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28284 ("popcountv4si2", "popcountv2di2"): Rename to ...
28285 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28286 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28288 ("popcount<mode>2_vxe"): New pattern.
28290 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28292 * common/config/s390/s390-common.c (processor_flags_table): Add
28294 * config.gcc: Add arch12.
28295 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28296 Default to arch12 for unknown CPU model numbers.
28297 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28298 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28299 PROCESSOR_max sanity check.
28300 * config/s390/s390-opts.h (enum processor_type): Add
28302 * config/s390/s390.c (processor_table): Add arch12.
28303 (s390_expand_builtin): Add check for B_VXE flag.
28304 (s390_issue_rate): Add PROCESSOR_ARCH12.
28305 (s390_get_sched_attrmask): Likewise.
28306 (s390_get_unit_mask): Likewise.
28307 (s390_sched_score): Enable z13 scheduling for arch12.
28308 (s390_sched_reorder): Likewise.
28309 (s390_sched_variable_issue): Likewise.
28310 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28312 (s390_tune_attr): Use z13 scheduling also for arch12.
28313 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28314 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28315 (TARGET_VXE_P): New macros.
28316 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28317 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28318 * config/s390/s390.opt: Add arch12 as processor_type.
28320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28322 * config/s390/s390.md
28323 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28324 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28325 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28327 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28328 Rename expanders to ...
28330 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28331 ("fixuns_truncdddi2_emu"): ... these.
28333 ("fixuns_trunc<mode>si2_emu"): New expander.
28335 ("*fixuns_truncdfdi2_z13"): Rename to ...
28336 ("*fixuns_truncdfdi2_vx"): ... this.
28338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28340 * config/s390/2964.md: Remove the single element vector compare
28341 instructions which are no longer used.
28342 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28344 (s390_canonicalize_comparison): Remove handling of DFmode
28346 (s390_expand_vec_compare_scalar): Remove function.
28347 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28348 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28350 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28352 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28354 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28356 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28357 will anyway by matched by mov<mode>_64dfp.
28359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28361 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28362 vlef/vstef. Add missing operand to vleif.
28364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28366 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28367 pair for all vector types with 64 bit elements.
28368 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28369 * config/s390/vector.md (V_HW_64): ... here.
28370 (V_128_NOSINGLE): New mode iterator.
28371 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28372 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28373 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28374 ("*vec_load_pairv2di"): Change to ...
28375 ("*vec_load_pair<mode>"): ... this one.
28377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28379 * config/s390/constraints.md: Add comments.
28380 (jKK): Reject element sizes > 8 bytes.
28381 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28383 * config/s390/s390.md: Add the s_operand checks formerly in
28384 s390_split_ok_p to various splitters where they are still
28386 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28387 for 128 bit vectors. Plus two splitters.
28389 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28391 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28397 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28398 error if the boundary argument is not constant.
28400 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28402 PR rtl-optimization/80112
28403 * loop-doloop.c (doloop_condition_get): Don't check condition
28404 if cmp isn't SET with IF_THEN_ELSE src.
28406 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28408 PR tree-optimization/80158
28409 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28410 replacing a candidate statement, also replace it for the
28411 candidate's alternate interpretation.
28412 (replace_rhs_if_not_dup): Likewise.
28413 (replace_one_candidate): Likewise.
28415 2017-03-24 Richard Biener <rguenther@suse.de>
28417 PR tree-optimization/80167
28418 * graphite-isl-ast-to-gimple.c
28419 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28421 (translate_isl_ast_to_gimple::get_rename): Likewise.
28423 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28425 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28426 handling of certain combinations of target options, including the
28427 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28428 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28430 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28433 * config/arm/arm.md (*load_multiple): Add reload_completed to
28434 matching condition.
28436 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28437 Richard Biener <rguenth@suse.de>
28439 PR tree-optimization/79908
28440 PR tree-optimization/80136
28441 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28442 been cast away, gimplify_and_add suffices.
28444 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28446 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28448 2017-03-23 Richard Biener <rguenther@suse.de>
28450 PR tree-optimization/80032
28451 * gimplify.c (gimple_push_cleanup): Forced unconditional
28452 cleanups still have to go to the conditional_cleanups
28455 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28457 PR tree-optimization/80072
28458 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28460 (next_operand_entry_id): Change type to unsigned int.
28461 (sort_by_operand_rank): Make sure to return the right return value
28462 even if unsigned fields are bigger than INT_MAX.
28463 (struct oecount): Change cnt and id type to unsigned int.
28464 (oecount_hasher::equal): Formatting fix.
28465 (oecount_cmp): Make sure to return the right return value
28466 even if unsigned fields are bigger than INT_MAX.
28467 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28470 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28471 TREE_READONLY on result if writing it more than once.
28474 * doc/invoke.texi (-fsanitize=thread): Document that with
28475 -fnon-call-exceptions atomics are not able to throw
28479 * tsan.c: Include tree-eh.h.
28480 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28481 maybe_clean_or_replace_eh_stmt where needed.
28482 (instrument_memory_accesses): Add cfg_changed argument.
28483 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28484 if it returned true.
28485 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28487 PR rtl-optimization/63191
28488 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28489 wrapper function, moved the whole old content into ...
28490 (ix86_delegitimize_address_1): ... this. New inline function.
28491 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28492 true as last argument instead of ix86_delegitimize_address.
28494 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28496 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28497 cortexa57_branch_cost.
28499 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28501 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28503 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28506 * doc/md.texi (Constraints): Document wA constraint.
28507 * config/rs6000/constraints.md (wA): New.
28508 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28509 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28510 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28511 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28513 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28516 * gimplify.c (is_oacc_declared): New function.
28517 (oacc_default_clause): Use it to set default flags for acc declared
28518 variables inside parallel regions.
28519 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28520 declared variables.
28521 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28522 declare attribute to any decl as necessary.
28524 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28527 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28528 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28529 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28530 (arm_arch_lpae): This.
28531 * config/arm/arm.c (arm_arch7ve): Rename into ...
28532 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28533 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28536 2017-03-22 Martin Liska <mliska@suse.cz>
28539 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28540 error message instead of an ICE.
28542 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28544 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28546 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28548 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28553 2017-03-21 Martin Sebor <msebor@redhat.com>
28555 * doc/extend.texi: Use "cannot" instead of "can't."
28556 * doc/hostconfig.texi: Same.
28557 * doc/install.texi: Same.
28558 * doc/invoke.texi: Same.
28559 * doc/loop.texi: Same.
28560 * doc/md.texi: Same.
28561 * doc/objc.texi: Same.
28562 * doc/rtl.texi: Same.
28563 * doc/tm.texi: Same.
28564 * doc/tm.texi.in: Same.
28565 * doc/trouble.texi: Same.
28567 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28570 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28571 (collect_checksum_attributes): Set it.
28572 (die_checksum_ordered): Use it.
28574 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28576 PR tree-optimization/79908
28577 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28578 change: For a VA_ARG whose LHS has been cast away, use
28579 force_gimple_operand to construct the side effects.
28581 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28583 PR translation/80001
28584 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28585 more amenable to translation.
28586 (oacc_loop_auto_partitions): Likewise.
28588 2017-03-21 Marek Polacek <polacek@redhat.com>
28589 Martin Sebor <msebor@redhat.com>
28591 PR tree-optimization/80109
28592 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28593 on INTEGRAL_TYPE_P.
28595 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28596 Segher Boessenkool <segher@kernel.crashing.org>
28599 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28600 check reg_used_between_p between insn and one of succ or succ2
28601 depending on if succ is artificial insn not inserted into insn
28604 2017-03-21 Martin Liska <mliska@suse.cz>
28606 PR gcov-profile/80081
28607 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28608 * doc/gcc.texi: Include gcov-dump stuff.
28609 * doc/gcov-dump.texi: New file.
28611 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28613 PR rtl-optimization/79150
28614 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28615 conditional jump, if the jump is the last insn of the loop.
28617 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28618 Richard Biener <rguenth@suse.de>
28620 PR tree-optimization/79908
28621 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28622 been cast away, use force_gimple_operand to construct the side
28625 2017-03-21 Martin Liska <mliska@suse.cz>
28627 PR libfortran/79956
28628 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28631 2017-03-21 Brad Spengler <spender@grsecurity.net>
28634 * plugin.c (htab_hash_plugin): New function.
28635 (add_new_plugin): Use it and adjust.
28636 (parse_plugin_arg_opt): Adjust.
28637 (init_one_plugin): Likewise.
28639 2017-03-21 Richard Biener <rguenther@suse.de>
28641 PR tree-optimization/80032
28642 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28643 if set force the cleanup to happen unconditionally.
28644 (gimplify_target_expr): Push inserted clobbers with force_uncond
28645 to avoid them being removed by control-dependent DCE.
28647 2017-03-21 Richard Biener <rguenther@suse.de>
28649 PR tree-optimization/80122
28650 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28651 va_arg_pack_len when the inlined call stmt requires pack
28653 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28655 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28658 * tsan.c (instrument_builtin_call): If the memory model argument
28659 is not a constant, assume it is valid.
28662 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28665 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28667 PR rtl-optimization/79910
28668 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28669 if its dest is used by an insn before I2 (other than the combined
28670 insns themselves, which are properly handled already).
28672 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28675 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28677 * combine.c (record_used_regs): New static function.
28678 (try_combine): Handle situations where there is an additional
28679 instruction between I2 and I3 which needs to have a LOG_LINK
28683 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28685 * combine.c (try_combine): Delete redundant i1 test. Call
28686 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28688 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28691 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28692 alternatives 13/14.
28694 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28696 PR tree-optimization/80054
28697 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28698 the optimization if a PHI or any of its arguments is not dominated
28699 by the candidate's basis. Use gphi* rather than gimple* as
28701 (replace_profitable_candidates): Clean up a gimple* variable that
28702 should be a gphi* variable.
28704 2017-03-20 Martin Sebor <msebor@redhat.com>
28707 * doc/extend.texi (attribute constructor): Document present limitation.
28709 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28712 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28713 __POWER9_VECTOR__ #ifdef control, change template definition to
28714 use Power9-specific built-in function.
28715 (vec_any_eq): Likewise.
28716 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28717 to control outcomes from this test.
28718 (vector_ae_<mode>p): For VEC_F modes, likewise.
28720 2017-03-20 Ian Lance Taylor <iant@google.com>
28722 * config/i386/i386.c (ix86_function_regparm): Save an extra
28723 register for -fsplit-stack with DECL_STATIC_CHAIN.
28725 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28728 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28729 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28731 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28733 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28735 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28738 2017-03-20 Marek Polacek <polacek@redhat.com>
28741 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28743 2017-03-20 Richard Biener <rguenther@suse.de>
28745 PR tree-optimization/80113
28746 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28747 allocate extra SSA name for PHI def.
28748 (add_close_phis_to_outer_loops): Likewise.
28749 (add_close_phis_to_merge_points): Likewise.
28750 (copy_loop_close_phi_args): Likewise.
28751 (copy_cond_phi_nodes): Likewise.
28753 2017-03-20 Martin Liska <mliska@suse.cz>
28755 PR middle-end/79753
28756 * tree-chkp.c (chkp_build_returned_bound): Do not build
28757 returned bounds for a LHS that's not a BOUNDED_P type.
28759 2017-03-20 Martin Liska <mliska@suse.cz>
28763 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28764 COMPLEX_CST and VECTOR_CST.
28766 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28769 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28770 target operand. A new splitter adds the clobber statement in case
28771 the target operand is dead anyway.
28773 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28775 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28776 to age-old versions of binutils and glibc.
28778 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28780 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28782 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28784 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28786 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28788 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28789 requirement for binutils 2.13.
28791 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28793 * combine.c (try_combine): Delete redundant i1 test. Call
28794 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28796 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28798 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28799 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28801 <riscv64-*-elf>: Re-arrange section
28802 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28803 <riscv32-*-linux>: Likewise.
28804 <riscv64-*-elf>: Likewise
28805 <riscv64-*-linux>: Likewise.
28807 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28810 * aarch64.opt(verbose-cost-dump): Fix typo.
28812 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28815 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28816 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28818 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28820 * reload.c (find_reloads): When reloading a nonoffsettable address,
28821 use RELOAD_OTHER for it and its address reloads.
28823 PR rtl-optimization/79910
28824 * combine.c (record_used_regs): New static function.
28825 (try_combine): Handle situations where there is an additional
28826 instruction between I2 and I3 which needs to have a LOG_LINK
28829 2017-03-17 Jeff Law <law@redhat.com>
28831 PR tree-optimization/71437
28832 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28833 conditional in the hash table first.
28834 (vrp_dom_walker::before_dom_children): Extract condition from
28835 ASSERT_EXPR. Record condition, its inverion and any implied
28836 conditions as well.
28838 2017-03-17 Marek Polacek <polacek@redhat.com>
28839 Markus Trippelsdorf <markus@trippelsdorf.de>
28841 PR tree-optimization/80079
28842 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28845 2017-03-17 Richard Biener <rguenther@suse.de>
28847 PR middle-end/80075
28848 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28849 Properly verify the LHS before the RHS possibly claims to be
28851 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28854 2017-03-17 Martin Jambor <mjambor@suse.cz>
28856 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28857 (List of -O2 options): Likewise.
28858 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28861 2017-03-17 Tom de Vries <tom@codesourcery.com>
28863 * gcov-dump.c (print_usage): Print bug_report_url.
28865 2017-03-17 Richard Biener <rguenther@suse.de>
28867 PR middle-end/80050
28868 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28869 (parser::peek): Likewise.
28871 2017-03-17 Richard Biener <rguenther@suse.de>
28873 PR tree-optimization/80048
28874 * sese.c (free_sese_info): Properly release rename_map and
28875 copied_bb_map elements.
28877 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28879 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28880 Add linked-list forward and backlinks. Insert on
28881 construction, remove on destruction.
28882 (class pass_store_merging): Add m_stores_head field.
28883 (pass_store_merging::terminate_and_process_all_chains):
28884 Iterate over m_stores_head list.
28885 (pass_store_merging::terminate_all_aliasing_chains):
28887 (pass_store_merging::execute): Check for debug stmts first.
28888 Push new chains onto the m_stores_head stack.
28890 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28893 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28894 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28895 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28897 2017-03-16 Jeff Law <law@redhat.com>
28899 PR tree-optimization/71437
28900 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28901 member function. Implementation moved into after_dom_children
28902 member function and into the threader's thread_outgoing_edges
28904 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28905 some code into new thread_outgoing_edges.
28906 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28907 definition. Simplify marker handling (do it here). Assume we always
28908 have the available expression and the const/copies tables.
28909 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28911 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28912 * tree-vrp.c (equiv_stack): No longer file scoped.
28913 (vrp_dom_walker): New class.
28914 (vrp_dom_walker::before_dom_children): New member function.
28915 (vrp_dom_walker::after_dom_children): Likewise.
28916 (identify_jump_threads): Setup domwalker. Use it rather than
28917 walking edges in a random order by hand. Simplify setup/finalization.
28918 (finalize_jump_threads): Remove.
28919 (vrp_finalize): Do not call identify_jump_threads here.
28920 (execute_vrp): Do it here instead and call thread_through_all_blocks
28923 PR tree-optimization/71437
28924 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28926 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28928 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28929 (dom_opt_dom_walker::thread_across_edge): Remove
28930 handle_dominating_asserts argument. All callers changed.
28931 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28932 changes. Remove calls to lhs_of_dominating_assert. Other
28933 uses of handle_dominating_asserts turn into unconditional code
28934 (simplify_control_stmt_condition_1): Likewise.
28935 (simplify_control_stmt_condition): Likewise.
28936 (thread_through_normal_block, thread_across_edge): Likewise.
28937 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28938 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28939 object if it is not an SSA_NAME.
28940 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28941 before calling into the VRP specific simplifiers.
28942 (identify_jump_threads): Remove handle_dominating_asserts
28945 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28948 * tree-diagnostic.c (default_tree_printer): No longer static.
28949 * tree-diagnostic.h (default_tree_printer): New prototype.
28951 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28953 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28954 Change ins into fmov.
28956 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28958 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28959 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28960 Use h_con constraint for operand 1.
28961 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28962 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28964 2017-03-15 Jeff Law <law@redhat.com>
28966 PR tree-optimization/71437
28967 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28968 (record_temporary_equivalences): Use it.
28970 PR tree-optimization/71437
28971 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28972 tree-ssa-scopedtables.
28973 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28974 (record_conditions, record_cond, vuse_eq): Likewise.
28975 (record_edge_info): Adjust to API tweak of record_conditions.
28976 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28977 (record_temporary_equivalences, optimize_stmt): Likewise.
28978 (eliminate_redundant_computations): Likewise.
28979 (record_equivalences_from_stmt): Likewise.
28980 * tree-ssa-scopedtables.c: Include options.h and params.h.
28981 (vuse_eq): New function, moved from tree-ssa-dom.c
28982 (build_and_record_new_cond): Likewise.
28983 (record_conditions): Likewise. Accept vector of conditions rather
28984 than edge_equivalence structure for first argument.
28985 for the first argument.
28986 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28987 from tree-ssa-dom.c.
28988 (avail_exprs_stack::record_cond): Likewise.
28989 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28990 from tree-ssa-dom.c.
28991 (avail_exprs_stack): Add new member functions lookup_avail_expr
28993 (record_conditions): Declare.
28995 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28998 * lra-constraints.c (process_alt_operands): Increase reject for
28999 reloading an input/output operand.
29001 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29004 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29005 insns to convert from signed/unsigned char/short to IEEE 128-bit
29007 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29009 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29012 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29013 subreg of inner mode for values already in registers.
29015 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29017 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29018 iteration reg is used after the loop.
29020 2017-03-14 Martin Sebor <msebor@redhat.com>
29022 PR tree-optimization/79800
29023 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29024 precision in negative-positive range.
29025 (format_floating): Call non-const overload with adjusted precision.
29027 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29030 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29033 2017-03-14 Martin Sebor <msebor@redhat.com>
29035 PR middle-end/80020
29036 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29037 * builtins.def (aligned_alloc): Use it.
29040 * Makefile.in (GTFILES): Add calls.c.
29041 * calls.c: Include "gt-calls.h".
29043 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29045 PR rtl-optimization/79728
29046 * regs.h (struct target_regs): New field
29047 x_contains_allocatable_regs_of_mode.
29048 (contains_allocatable_regs_of_mode): New macro.
29049 * reginfo.c (init_reg_sets_1): Initialize it, and change
29050 contains_reg_of_mode so it includes global regs as well.
29051 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29052 rather than contains_regs_of_mode.
29054 2017-03-14 Martin Liska <mliska@suse.cz>
29056 * doc/invoke.texi: Document options that can't be combined with
29057 -fcheck-pointer-bounds.
29059 2017-03-14 Martin Liska <mliska@suse.cz>
29061 PR middle-end/79831
29062 * doc/invoke.texi (-Wchkp): Document the option.
29064 2017-03-14 Martin Liska <mliska@suse.cz>
29066 * Makefile.in: Install gcov-dump.
29068 2017-03-14 Martin Liska <mliska@suse.cz>
29070 * multiple_target.c (expand_target_clones): Bail out for
29071 an invalid attribute.
29073 2017-03-14 Richard Biener <rguenther@suse.de>
29075 * alias.c (struct alias_set_entry): Pack properly.
29076 * cfgloop.h (struct loop): Likewise.
29077 * cse.c (struct set): Likewise.
29078 * ipa-utils.c (struct searchc_env): Likewise.
29079 * loop-invariant.c (struct invariant): Likewise.
29080 * lra-remat.c (struct cand): Likewise.
29081 * recog.c (struct change_t): Likewise.
29082 * rtl.h (struct address_info): Likewise.
29083 * symbol-summary.h (function_summary): Likewise.
29084 * tree-loop-distribution.c (struct partition): Likewise.
29085 * tree-object-size.c (struct object_size_info): Likewise.
29086 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29087 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29088 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29089 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29090 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29091 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29092 (struct _stmt_vec_info): Likewise.
29094 2017-03-14 Martin Liska <mliska@suse.cz>
29097 * multiple_target.c (create_dispatcher_calls): Check that
29098 a target can create a function dispatcher.
29100 2017-03-14 Martin Liska <mliska@suse.cz>
29103 * multiple_target.c (expand_target_clones): Drop local.local
29104 flag for default implementation.
29106 2017-03-14 Richard Biener <rguenther@suse.de>
29108 PR tree-optimization/80030
29109 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29111 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29113 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29114 gcc_fallthrough() instead of __attribute__((fallthrough));
29116 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29118 * doc/gcc.texi: Remove "up" link to (DIR).
29119 * doc/gccint.texi: Ditto.
29121 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29123 * doc/install.texi (Specific) <avr>: Remove reference to
29126 2017-03-13 Jeff Law <law@redhat.com>
29128 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29129 attribute rather than comments.
29131 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29132 match_scratch operand is highest.
29134 2017-03-13 Martin Liska <mliska@suse.cz>
29136 PR middle-end/78339
29137 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29138 is a CHKP clone, use original declaration.
29140 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29142 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29143 (arc_conditional_register_usage): Use a different allocation order
29144 when optimizing for size.
29145 * common/config/arc/arc-common.c (arc_option_optimization_table):
29146 Section anchors default on when optimizing for size.
29148 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29150 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29152 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29154 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29155 * config/arc/arc.md (cpu_facility): Add cd variant.
29156 (*movqi_insn): Add code density variant.
29157 (*movhi_insn): Likewise.
29158 (*movqi_insn): Likewise.
29159 (*addsi3_mixed): Likewise.
29160 (subsi3_insn): Likewise.
29162 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29164 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29166 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29168 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29169 expressions with MINUS and UNARY ops.
29171 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29174 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29176 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29177 between vec_select and vector argument.
29178 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29179 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29180 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29181 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29182 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29183 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29185 2017-03-13 Richard Biener <rguenther@suse.de>
29188 * params.def (vect-max-peeling-for-alignment): Fix typo.
29190 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29192 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29193 issue that only occurred with binutils below 2.18.
29195 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29197 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29198 refer to binutils 2.11/2.12 minimum.
29200 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29202 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29203 ftp.kernel.org and simplify binutils requirement.
29205 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29207 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29209 (Optimize Options): Ditto. Also remove redundancy.
29211 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29213 PR translation/79848
29214 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29216 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29217 to G_ to avoid double translation.
29219 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29221 PR translation/79923
29222 * auto-profile.c (get_combined_location): Convert leading
29223 character of diagnostics to lower case and remove trailing period.
29224 (read_profile): Likewise for various diagnostics.
29225 * config/arm/arm.c (arm_option_override): Remove trailing period
29226 from various diagnostics.
29227 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29228 (msp430_expand_delay_cycles): Likewise.
29230 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29233 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29234 full command-line argument, rather than just "str".
29235 (aarch64_validate_march): Likewise.
29236 (aarch64_validate_mtune): Likewise.
29238 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29240 PR rtl-optimization/78911
29241 * lra-assigns.c (must_not_spill_p): New function.
29242 (spill_for): Use it.
29244 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29246 PR tree-optimization/79981
29247 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29248 ATOMIC_COMPARE_EXCHANGE ifn result.
29249 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29250 IFN_ATOMIC_COMPARE_EXCHANGE.
29252 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29255 * opts.c (parse_sanitizer_options): Add missing question mark to
29256 "did you mean" message.
29258 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29260 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29262 (VMULEUH_UNS): Likewise.
29263 (VMULOUB_UNS): Likewise.
29264 (VMULOUH_UNS): Likewise.
29265 * config/rs6000/rs6000.c (builtin_function_type): Remove
29266 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29268 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29271 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29272 x with result of extra_parsing_for_operand_code_0.
29273 (function_reader::extra_parsing_for_operand_code_0): Convert
29274 return type from void to rtx, returning x. When reading
29275 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29276 larger size containing struct block_symbol.
29278 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29280 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29281 -mfloat128-hardware without -m64.
29283 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29286 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29287 entries to the case statement that marks unsigned arguments to
29288 overloaded functions.
29290 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29292 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29293 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29295 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29299 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29301 2017-03-10 Martin Liska <mliska@suse.cz>
29305 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29306 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29309 2017-03-10 Olivier Hainque <hainque@adacore.com>
29311 * tree-switch-conversion (array_value_type): Start by resetting
29312 candidate type to it's main variant.
29314 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29316 PR rtl-optimization/79909
29317 * combine.c (try_combine): Use simplify_replace_rtx on individual
29318 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29319 of the whole CALL_INSN_FUNCTION_USAGE.
29321 PR tree-optimization/79972
29322 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29323 get_range_info on SSA_NAMEs. Formatting fixes.
29325 2017-03-10 Richard Biener <rguenther@suse.de>
29326 Jakub Jelinek <jakub@redhat.com>
29328 PR tree-optimization/77975
29329 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29330 edge to be constant.
29331 (get_val_for): For constant x return it. Formatting fix.
29332 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29333 would use the same bases as the current one.
29335 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29337 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29338 instead of vec_select for V1TImode.
29339 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29341 (VSX_LE_128): Add V1TI to this mode iterator.
29342 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29343 (*vsx_le_perm_store_<mode>): Likewise.
29344 (pre-reload splitter for VSX stores): Likewise.
29345 (post-reload splitter for VSX stores): Likewise.
29346 (*vsx_xxpermdi2_le_<mode>): Likewise.
29347 (*vsx_lxvd2x2_le_<mode>): Likewise.
29348 (*vsx_stxvd2x2_le_<mode>): Likewise.
29350 2017-03-09 Michael Eager <eager@eagercon.com>
29352 Correct failures with --enable-checking=yes,rtl.
29354 * config/microblaze/microblaze.c (microblaze_expand_shift):
29355 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29356 test for const0_rtx.
29357 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29358 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29360 2017-03-09 Richard Biener <rguenther@suse.de>
29362 PR tree-optimization/79977
29363 * graphite-scop-detection.c (scop_detection::merge_sese):
29364 Handle the case of extra exits to blocks dominating the entry.
29366 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29368 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29371 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29373 PR rtl-optimization/79949
29374 * lra-constraints.c (process_alt_operands): Check memory when
29375 trying to predict a cycle. Print about the overall increase.
29377 2017-03-09 Richard Biener <rguenther@suse.de>
29379 PR middle-end/79971
29380 * gimple-expr.c (useless_type_conversion_p): Preserve
29381 TYPE_SATURATING for fixed-point types.
29383 2017-03-09 Richard Biener <rguenther@suse.de>
29386 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29387 alignment of BLKmode params.
29389 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29392 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29393 (VALL_NO_V2Q): Likewise.
29395 * config/aarch64/aarch64-simd.md
29396 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29398 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29399 VALL_NO_V2Q mode iterator.
29400 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29402 2017-03-09 Martin Liska <mliska@suse.cz>
29404 PR tree-optimization/79631
29405 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29406 tree_int_cst_sign_bit just for INTEGER constants.
29408 2017-03-09 Martin Liska <mliska@suse.cz>
29412 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29415 2017-03-09 Marek Polacek <polacek@redhat.com>
29418 * tree.c (inchash::add_expr): Handle TREE_VEC.
29420 2017-03-09 Martin Liska <mliska@suse.cz>
29423 (chkp_narrow_size_and_offset): New function.
29424 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29425 (void chkp_parse_bit_field_ref): New function.
29426 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29427 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29429 2017-03-09 Martin Liska <mliska@suse.cz>
29432 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29433 (chkp_find_bounds_1): Remove gcc_unreachable.
29435 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29438 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29439 BUILT_IN_SYNC*, determine the access type from the size suffix and
29440 always build a MEM_REF with that type. Handle forgotten
29441 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29444 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29445 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29446 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29447 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29448 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29449 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29450 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29451 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29452 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29453 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29454 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29455 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29456 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29457 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29458 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29459 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29460 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29461 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29462 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29463 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29464 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29465 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29466 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29467 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29468 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29469 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29470 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29471 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29472 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29473 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29474 definitions outside of __OPTIMIZE__ guarded section.
29477 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29478 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29479 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29480 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29483 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29485 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29486 ("vfenez<mode>"): Add missing constraints.
29488 2017-03-08 Martin Sebor <msebor@redhat.com>
29491 * config/nds32/nds32.c (nds32_option_override):
29492 Fix misspelled diagnostic.
29494 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29497 * gimplify.c (gimplify_omp_for): Replace index var in outer
29498 taskloop statement with an artificial variable and add
29499 OMP_CLAUSE_PRIVATE clause for it.
29501 2017-03-08 Richard Biener <rguenther@suse.de>
29503 PR tree-optimization/79955
29504 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29505 for accesses that are completely outside of the variable.
29507 2017-03-08 Andrew Haley <aph@redhat.com>
29509 PR tree-optimization/79943
29510 * tree-ssa-loop-split.c (compute_new_first_bound): When
29511 calculating the new upper bound, (END-BEG) should be added, not
29514 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29516 * config/avr/avr.md (setmemhi): Make sure match_dup
29517 operand number comes before match_scratch.
29519 2017-03-08 Richard Biener <rguenther@suse.de>
29521 PR tree-optimization/79920
29522 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29523 with ncopies == 1 to ...
29524 (vect_transform_slp_perm_load): ... here. Properly compute
29525 all element loads by iterating VF times over the group. Do
29526 not handle ncopies (computed in a broken way) in
29527 vect_create_mask_and_perm.
29529 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29532 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29533 is a uniform vector, use uniform_vector_p return value instead of
29534 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29536 2017-03-07 Marek Polacek <polacek@redhat.com>
29538 PR middle-end/79809
29539 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29540 (alloca_call_type): Likewise.
29542 2017-03-07 Martin Liska <mliska@suse.cz>
29544 * gcov.c (process_args): Put comment to correct location.
29546 2017-03-07 Martin Liska <mliska@suse.cz>
29548 PR middle-end/68270
29549 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29550 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29551 (chkp_narrow_bounds_for_field): Likewise.
29552 (chkp_parse_array_and_component_ref): Pass one more argument to
29555 2017-03-07 Richard Biener <rguenther@suse.de>
29557 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29560 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29562 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29563 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29565 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29568 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29569 to end of description.
29570 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29572 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29574 PR rtl-optimization/79901
29575 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29577 (*avx512f_<code><mode>3<mask_name>): ... this.
29578 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29579 iterator instead of VI8_AVX2_AVX512BW.
29581 PR rtl-optimization/79901
29582 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29583 min/max expander, expand it using expand_vec_cond_expr.
29586 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29589 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29592 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29593 to void * for PCH reasons.
29594 * dwarf2out.c (output_loc_operands, output_die): Cast
29595 v.val_vec.array to unsigned char *.
29597 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29600 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29603 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29605 PR rtl-optimization/79571
29606 * lra-constraints.c (process_alt_operands): Calculate static
29607 reject and subtract it from overall when only addresses will be
29610 2017-03-06 Julia Koval <julia.koval@intel.com>
29613 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29614 incoming stack boundary to 128 for 64-bit targets.
29616 2017-03-06 Richard Biener <rguenther@suse.de>
29618 PR tree-optimization/79894
29619 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29620 to NULL after folding it.
29622 2017-03-06 Richard Biener <rguenther@suse.de>
29624 PR tree-optimization/79824
29625 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29626 check disabling peeling for gaps.
29628 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29630 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29631 attributes): Document gettimeofday.
29633 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29635 * config/s390/s390.c (s390_option_override_internal): Set
29636 PARAM_MIN_VECT_LOOP_BOUND
29638 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29640 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29641 * config/s390/s390.md: Likewise.
29643 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29646 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29647 (<avx2_avx512>_perm<mode>): Rename to ...
29648 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29650 (<avx512>_perm<mode>_mask): Rename to ...
29651 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29653 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29654 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29655 instead of VI8F_256_512.
29656 (avx512f_perm<mode>): New define_expand.
29657 (avx512f_perm<mode>_mask): Likewise.
29658 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29659 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29661 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29663 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29664 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29666 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29668 2017-03-06 Martin Liska <mliska@suse.cz>
29671 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29672 when having a SSA NAME w/o VAR_DECL assigned to it.
29674 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29676 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29677 msa_dpsub_<su>_d): Fix MODE for vec_select.
29679 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29681 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29683 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29685 2017-03-06 Richard Biener <rguenther@suse.de>
29687 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29688 * plugin.c (register_plugin_info): Likewise.
29689 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29691 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29693 * config/i386/sse.md (sse_storehps, sse_storelps,
29694 avx_<castmode><avxsizesuffix>_<castmode>,
29695 avx512f_<castmode><avxsizesuffix>_<castmode>,
29696 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29697 in condition that at least one operand is not a MEM.
29699 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29701 PR middle-end/79805
29702 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29703 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29705 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29706 gimple_call_nothrow_p flag based on whether original builtin can throw.
29707 If it can, emit following stmts on the fallthrough edge.
29708 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29709 don't create new bb if inserting just debug stmts on the edge, try to
29710 insert them on the fallthru bb or just reset debug stmts.
29712 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29715 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29716 restore recog_data (including the operand rtxes inside it) around
29717 the call to get_insn_template.
29719 2017-03-03 Martin Sebor <msebor@redhat.com>
29721 PR tree-optimization/79699
29722 * context.c (context::~context): Free MPFR caches to avoid
29723 a memory leak on program exit.
29725 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29727 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29728 Use wide_int::ulow () instead of .elt (0).
29730 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29732 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29733 (*pushxf): Limit oF constraint to 32bit targets and add oC
29734 constraint for 64bit targets.
29735 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29736 (*pushdf): Change rmF constraint to rmC.
29738 2017-03-03 Martin Liska <mliska@suse.cz>
29740 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29741 Remove unused variable.
29743 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29746 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29747 is a memory operand, increase num_memory.
29748 (ix86_expand_args_builtin): Likewise.
29750 2017-03-03 Jan Hubicka <jh@suse.cz>
29753 * ipa-devirt.c (maybe_record_node): Properly handle
29754 __cxa_pure_virtual visibility.
29756 2017-03-03 Martin Liska <mliska@suse.cz>
29758 PR tree-optimization/79803
29759 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29761 (pass_loop_prefetch::execute): Disabled optimization if an
29762 assumption about L1 cache size is not met.
29764 2017-03-03 Martin Liska <mliska@suse.cz>
29766 PR rtl-optimization/79574
29767 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29768 (hash_scan_set): Likewise.
29769 (dump_hash_table): Likewise.
29770 (hoist_code): Likewise.
29772 2017-03-03 Richard Biener <rguenther@suse.de>
29774 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29775 in place of uhigh ().
29776 (fixed_convert_from_real): Likewise.
29778 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29781 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29783 2017-03-03 Richard Biener <rguenther@suse.de>
29785 PR middle-end/79818
29786 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29787 TYPE_OVERFLOW_UNDEFINED check.
29789 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29791 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29793 (vector_ae_<mode>_p): Likewise.
29794 (vector_nez_<mode>_p): Likewise.
29795 (vector_ne_v2di_p): Likewise.
29796 (vector_ae_v2di_p): Likewise.
29797 (vector_ne_<mode>_p): Likewise.
29798 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29800 (vsx_tsqrt<mode>2_fe): Likewise.
29802 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29805 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29807 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29809 PR rtl-optimization/79780
29810 * cprop.c (one_cprop_pass): When second and further conditional trap
29811 in a single basic block is turned into an unconditional trap, turn it
29812 into a deleted note to avoid RTL verification failures.
29814 2017-03-02 Richard Biener <rguenther@suse.de>
29816 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29818 2017-03-02 Richard Biener <rguenther@suse.de>
29820 PR tree-optimization/79345
29822 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29823 param and abort the walk, returning -1 if it is hit.
29824 (walk_aliased_vdefs): Take a limit param and pass it on.
29825 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29826 defaulting to 0 and return a signed int.
29827 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29828 (check_defs): New helper.
29829 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29830 about uninitialized memory.
29831 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29832 bogus uninitialized warning.
29833 (fixed_convert_from_real): Likewise.
29835 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29837 PR tree-optimization/66768
29838 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29839 iv_use if base object can't be determined.
29841 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29843 PR tree-optimization/79345
29844 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29845 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29846 (get_pattern_stats): Initialize it.
29847 * genemit.c (gen_expand): Verify match_scratch numbers come after
29848 match_operand/match_dup numbers.
29849 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29850 match_scratch numbers.
29851 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29853 * config/s390/s390.md (trunctdsd2): Likewise.
29855 2017-03-02 Richard Biener <rguenther@suse.de>
29857 * wide-int.h (wide_int_storage::operator=): Implement in terms
29860 2017-03-02 Richard Biener <rguenther@suse.de>
29862 PR tree-optimization/79777
29863 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29864 the to insert expression to sth existing.
29866 2017-03-01 Martin Sebor <msebor@redhat.com>
29868 PR middle-end/79692
29869 * gimple-ssa-sprintf.c
29870 (directive::known_width_and_precision): New function.
29871 (format_integer): Use it.
29872 (get_mpfr_format_length): Consider the full range of precision
29873 when computing %g output with the # flag. Set the likely byte
29874 count to 3 rather than 1 when precision is indeterminate.
29875 (format_floating): Correct the lower bound of precision.
29877 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29879 * doc/invoke.texi: Document default code model for 64-bit Linux.
29881 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29884 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29885 udiv rather than div since input pattern is unsigned.
29887 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29889 * config/i386/i386.c (print_reg): Warn for values of
29890 unsupported size in integer register.
29892 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29895 * config/rs6000/predicates.md (current_file_function_operand): Do
29896 not allow self calls to be local if the function is replaceable.
29898 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29901 * config/rs6000/altivec.h (vec_ctz and others): Change the
29902 preprocessor macro that controls conditional compilation from
29903 _ARCH_PWR9 to __POWER9_VECTOR__.
29904 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29905 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29906 control (instead of _ARCH_PWR9 control) so that template
29907 definition uses power9-specific function.
29908 (vec_any_eq): Likewise.
29909 (vec_all_ne): Change macro definition to use a power9-specific
29910 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29911 _ARCH_PWR9 control).
29912 (vec_any_eq) Likewise.
29913 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29914 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29915 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29916 support for xvcmpnedp instruction.
29917 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29918 macro expansion so that Power9 implementation of vec_all_ne does
29919 not use the AltiVec predicate framework.
29920 (VCMPNEH_P): Likewise.
29921 (VCMPNEW_P): Likewise.
29922 (VCMPNED_P): Likewise.
29923 (VCMPNEFP_P): Likewise.
29924 (VCMPNEDP_P): Likewise.
29925 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29926 implementation of vec_any_eq to not use AltiVec predicate
29928 (VCMPAEH_P): Likewise.
29929 (VCMPAEW_P): Likewise.
29930 (VCMPAED_P): Likewise.
29931 (VCMPAEFP_P): Likewise.
29932 (VCMPAEDP_P): Likewise.
29933 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29934 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29935 not use the AltiVec predicate framework.
29936 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29937 of vec_any_eq to not use AltiVec predicate framework.
29938 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29939 support for predefined __POWER9_VECTOR__ macro to indicate that
29940 Power9 instruction selection is enabled.
29941 (altivec_overloaded_builtins): Remove extraneous
29942 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29943 function argument types RS6000_BTI_bool_V16QI and
29944 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29945 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29946 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29947 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29948 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29949 Power9 for implementations of vec_cmpne. Change the signature for
29950 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29951 (representing vec_all_ne) to remove the previously described first
29952 argument of type RS6000_BTI_INTSI, as this was an artifact of
29953 reliance on the AltiVec predicate framework, which is no longer
29954 used in the implementation of these functions. Add
29955 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29956 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29957 since, unlike the AltiVec predicate framework implementation, we
29958 do not share function descriptors between vec_alle and vec_anyeq.
29959 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29960 set of modes that receive special treatment even when
29961 TARGET_P9_VECTOR is true. The special treatment emits code that
29962 does not depend on Power9 instructions.
29963 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29964 define_expand to not rely on AltiVec predicate framework.
29965 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29967 (vector_ne_v2di_p): Change this define_expand to not rely on
29968 AltiVec predicate framework.
29969 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29971 (vector_ne_<mode>_p): Change this define_expand to not rely on
29972 AltiVec predicate framework.
29973 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29975 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29976 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29977 define_insn pattern.
29978 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29979 define_insn pattern because the xvcmpne<VSs>. instruction is not
29981 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29982 instruction is not supported.
29984 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29986 * config/nvptx/nvptx.c: Include intl.h.
29988 2017-03-01 Martin Jambor <mjambor@suse.cz>
29991 * ipa-prop.h (ipa_bits): Removed field known.
29992 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29993 to pointers. Adjusted their comments to warn about their sharing.
29994 (ipcp_transformation_summary): Change bits to a vector of pointers.
29995 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29996 (ipa_get_ipa_bits_for_value): Declare.
29997 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29998 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29999 (ipa_bits_hash_table): Likewise.
30000 (ipa_vr_ggc_hash_traits): Likewise.
30001 (ipa_vr_hash_table): Likewise.
30002 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30003 being pointers and vr_known being removed.
30004 (ipa_set_jf_unknown): Likewise.
30005 (ipa_get_ipa_bits_for_value): New function.
30006 (ipa_set_jfunc_bits): Likewise.
30007 (ipa_get_value_range): New overloaded functions.
30008 (ipa_set_jfunc_vr): Likewise.
30009 (ipa_compute_jump_functions_for_edge): Use the above functions to
30010 construct bits and vr parts of jump functions.
30011 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30012 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30014 (ipcp_grow_transformations_if_necessary): Also allocate
30015 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30017 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30018 them. Fix too long lines.
30019 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30020 vr_known being removed.
30021 (ipa_read_jump_function): Use new setter functions to construct bits
30022 and vr parts of jump functions or set them to NULL.
30023 (write_ipcp_transformation_info): Adjust for bits being pointers.
30024 (read_ipcp_transformation_info): Likewise.
30025 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30027 Include gt-ipa-prop.h.
30028 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30030 (ipcp_store_bits_results): Likewise.
30031 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30032 Do not write to existing jump functions but use a temporary instead.
30034 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30037 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30038 attempt to use its first operand as BIT_FIELD_REF base.
30040 2017-03-01 Richard Biener <rguenther@suse.de>
30042 PR middle-end/79721
30043 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30044 interpolating formula in wrapping arithmetic.
30045 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30047 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30049 PR tree-optimization/79734
30050 * tree-vect-generic.c (expand_vector_condition): Optimize
30051 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30052 Handle VEC_COND_EXPR where comparison has different inner width from
30053 type's inner width.
30055 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30057 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30058 markup, and similar issues. Remove @opindex entries for things
30059 that aren't options. Add missing -mmpy-option entries.
30061 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30063 PR tree-optimization/79737
30064 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30065 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30066 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30067 instead of byte_size. Formatting fix.
30068 (shift_bytes_in_array_right): Formatting fix.
30070 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30073 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30074 condition on optimize for the leaf function test.
30076 2017-02-28 Martin Liska <mliska@suse.cz>
30079 * read-rtl-function.c (function_reader::handle_unknown_directive):
30080 Bail out when one uses -flto.
30082 2017-02-28 Martin Liska <mliska@suse.cz>
30084 * common.opt: Replace space with tabular for options of <number>
30086 * config/i386/i386.opt: Show <number> value for
30087 -mlarge-data-threshold.
30088 * opts.c (print_filtered_help): Do not display number in hexadecimal
30091 2017-02-28 Martin Liska <mliska@suse.cz>
30093 * common.opt: Fix --help=option -Q for options which are of
30096 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30098 * config/i386/i386.c (print_reg): Error out for values
30099 of 8-bit size in invalid integer register.
30101 2017-02-28 Martin Sebor <msebor@redhat.com>
30103 PR tree-optimization/79691
30104 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30106 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30109 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30110 gcc_unreachable with output_operand_lossage.
30112 2017-02-28 Richard Biener <rguenther@suse.de>
30114 PR tree-optimization/79740
30115 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30117 (visit_nary_op): Insert the nary into the hashtable if we
30118 pattern-matched sth.
30119 * tree-ssa-pre.c (eliminate_insert): Robustify.
30121 2017-02-28 Richard Biener <rguenther@suse.de>
30123 PR middle-end/79731
30124 * fold-const.c (decode_field_reference): Reject out-of-bound
30127 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30129 * config/i386/i386.c: Include intl.h.
30130 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30131 instead of just cond ? "..." : "...".
30132 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30133 * coverage.c (read_counts_file): Likewise.
30134 * omp-offload.c: Include intl.h.
30135 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30136 of just cond ? "..." : "...".
30137 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30138 of just cond ? "..." : "...".
30140 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30143 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30145 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30146 'tune for' CPU name.
30147 * config/arm/arm-cpu-data.h: Regenerated.
30149 2017-02-28 Richard Biener <rguenther@suse.de>
30151 PR tree-optimization/79732
30152 * tree-inline.c (expand_call_inline): Do not shadow var.
30154 2017-02-28 Richard Biener <rguenther@suse.de>
30156 PR tree-optimization/79723
30157 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30158 address-space properly.
30160 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30162 * doc/optinfo.texi (Optimization groups): Fix option used for
30164 * doc/invoke.texi (-fopt-info): Document "omp".
30165 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30166 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30167 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30168 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30169 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30171 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30173 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30174 OPTGROUP_OMP with "omp".
30176 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30179 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30180 for arithmetic shift of unsigned V2DI.
30182 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30184 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30185 arc/linux.h headers.
30186 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30187 (LINK_SPEC): Likewise.
30188 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30189 (EXTRA_SPECS): Likewise.
30190 (STARTFILE_SPEC): Likewise.
30191 (ENDFILE_SPEC): Likewise.
30192 (LIB_SPEC): Likewise.
30193 (TARGET_SDATA_DEFAULT): Likewise.
30194 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30195 (MULTILIB_DEFAULTS): Likewise.
30196 (DWARF2_UNWIND_INFO): Likewise.
30197 * config/arc/big.h: New file.
30198 * config/arc/elf.h: Likewise.
30199 * config/arc/linux.h: Likewise.
30200 * config/arc/t-uClibc: Remove.
30202 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30204 PR tree-optimization/77536
30205 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30206 (tree_transform_and_unroll_loop): Use above function to compute the
30207 estimated niter of unrolled loop and use it when scaling profile.
30208 Also use count info rather than frequency if it's non-zero.
30209 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30210 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30211 (vect_transform_loop): Call above function.
30213 2017-02-27 Richard Biener <rguenther@suse.de>
30215 PR tree-optimization/45397
30216 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30217 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30218 (visit_nary_op): Add pattern matching for CSEing sign-changed
30219 or truncated operations with wider ones.
30221 2017-02-27 Richard Biener <rguenther@suse.de>
30223 PR tree-optimization/79690
30224 * tree-vect-stmts.c (vectorizable_store): Use vector type
30225 built from the DR with address-space.
30227 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30229 * doc/invoke.texi (Optimize Options): Refine the description
30230 of asan-use-after-return.
30232 2017-02-25 Alan Modra <amodra@gmail.com>
30234 PR rtl-optimization/79584
30235 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30236 base, not ad->base_term, the reg within base. Remove assertion
30237 that ad->base == ad->base_term. Replace gen_int_mode using
30238 bogus mode with const0_rtx.
30240 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30242 PR middle-end/79396
30243 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30244 FMA_EXPR like tcc_binary or tcc_unary.
30246 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30249 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30251 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30252 (output_loc_operands): Handle DW_OP_call_ref and
30253 DW_OP_GNU_variable_value.
30254 (struct variable_value_struct): New type.
30255 (struct variable_value_hasher): Likewise.
30256 (variable_value_hash): New variable.
30257 (string_types): Remove.
30258 (copy_loc_descr): New function.
30259 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30260 (prepend_loc_descr_to_each): New function.
30261 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30262 instead of add_loc_descr_to_each if the first argument is single
30263 location list and the second has multiple.
30264 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30265 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30266 when looking for variable value which doesn't have other location info.
30267 (loc_list_from_tree): Formatting fix.
30268 (gen_array_type_die): Simplify DW_AT_string_length handling.
30269 (adjust_string_types): Remove.
30270 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30271 string_types. Call resolve_variable_values.
30272 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30273 (resolve_addr_in_expr): Likewise. Add A argument.
30274 (copy_deref_exprloc): Remove deref argument. Adjust for the
30275 original expression being DW_OP_GNU_variable_value with optionally
30276 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30277 optionally after it.
30278 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30279 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30280 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30281 (variable_value_hasher::hash, variable_value_hasher::equal): New
30283 (resolve_variable_value_in_expr, resolve_variable_value,
30284 resolve_variable_values, note_variable_value_in_expr,
30285 note_variable_value): New functions.
30286 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30289 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30292 * opts.h (handle_generated_option): Add GENERATED_P argument.
30293 * opts-common.c (handle_option): Adjust function comment.
30294 (handle_generated_option): Add GENERATED_P argument, pass it to
30296 (control_warning_option): Pass false to handle_generated_option
30298 * opts.c (maybe_default_option): Pass true to handle_generated_option
30300 * optc-gen.awk: Likewise.
30302 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30304 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30305 a REG, look at the REG it is a SUBREG of.
30306 (splitter for cmpeqsi_t): Ditto.
30308 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30310 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30311 the special USEs with the pattern of the insn, not the insn itself.
30313 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30316 * doc/invoke.texi: Document -mload-store-pairs.
30318 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30319 Sandra Loosemore <sandra@codesourcery.com>
30321 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30322 argument isn't a CONST_INT.
30323 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30324 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30325 (nios2_validate_compare): Bypass alternate compare logic if *op2
30326 is not a CONST_INT.
30327 (ldstwm_operation_p): Return false if first_base is not a REG or
30328 if first_offset is not a CONST_INT.
30330 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30332 * config/cris/cris.md: Use correct operand in a define_peephole2.
30334 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30336 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30338 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30340 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30341 this_insn if it is an INSN or JUMP_INSN.
30342 (force_offsettable): Look at base, not at addr.
30343 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30344 on things that aren't necessarily CONST_INTs.
30346 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30348 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30349 -mfpmath=sse is the default also for x86-32 targets with SSE2
30350 instruction set when @option{-ffast-math} is enabled
30352 2017-02-24 Jeff Law <law@redhat.com>
30354 PR rtl-optimizatoin/79286
30355 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30358 2017-02-24 Richard Biener <rguenther@suse.de>
30360 PR tree-optimization/79389
30361 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30364 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30366 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30367 function comment to reflect reality.
30368 (loop_exits_before_overflow): Fix typo in function description.
30370 2017-02-24 Richard Biener <rguenther@suse.de>
30372 PR tree-optimization/79389
30373 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30374 properly that a threading opportunity exists. Detect conditional
30375 copy/constant propagation opportunities.
30377 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30379 * config/visium/visium.md (type): Add trap.
30380 (b): New mode attribute.
30381 (*btst): Rename into...
30382 (*btst<mode>): ...this and adjust.
30383 (*cbranchsi4_btst_insn): Rename into...
30384 (*cbranch<mode>4_btst_insn): ...this and adjust.
30385 (trap): New define_insn.
30387 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30389 PR tree-optimization/79389
30390 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30391 (noce_reversed_cond_code): New function.
30392 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30393 reversed_comparison_code. Formatting fix.
30394 (noce_try_store_flag): Test rev_cond != NULL in addition to
30395 reversed_comparison_code.
30396 (noce_try_store_flag_constants): Likewise.
30397 (noce_try_store_flag_mask): Likewise.
30398 (noce_try_addcc): Use rev_cond if non-NULL instead of
30399 reversed_comparison_code.
30400 (noce_try_cmove_arith): Likewise. Formatting fixes.
30401 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30402 (noce_find_if_block): Initialize rev_cond.
30403 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30404 instead of false as last argument never attempt to reverse it
30407 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30409 PR tree-optimization/79663
30410 * tree-predcom.c (combine_chains): Process refs in reverse order
30411 only for ZERO length chains, and add explaining comment.
30413 2017-02-23 Jeff Law <law@redhat.com>
30415 PR tree-optimization/79578
30416 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30417 in call to operand_equal_p.
30419 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30422 * config/i386/cpuid.h: Fix another undefined behavior.
30424 2017-02-23 Richard Biener <rguenther@suse.de>
30426 PR tree-optimization/79683
30427 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30428 vector types for data-refs.
30430 2017-02-23 Martin Liska <mliska@suse.cz>
30432 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30434 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30436 PR middle-end/79665
30437 * internal-fn.c (get_range_pos_neg): Moved to ...
30438 * tree.c (get_range_pos_neg): ... here. No longer static.
30439 * tree.h (get_range_pos_neg): New prototype.
30440 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30441 are known to be in between 0 and signed maximum inclusive, try to
30442 expand both unsigned and signed divmod and use the cheaper one from
30445 2017-02-22 Jeff Law <law@redhat.com>
30447 PR tree-optimization/79578
30448 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30449 to compare base operands.
30451 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30454 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30455 gpc_reg_operand instead of fpr_reg_operand.
30457 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30459 * config/mips/mips.c (mips_return_in_memory): Force FP
30460 vector types to be returned in memory for o32 ABI.
30462 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30464 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30465 instead of DW_TAG_member for static data member declarations and don't
30466 set no_linkage_name for static inline data members.
30467 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30470 2017-02-22 Martin Liska <mliska@suse.cz>
30472 * doc/invoke.texi: Replace inequality signs with square brackets
30475 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30478 * lra-constraints.c (simplify_operand_subreg): Handle
30479 WORD_REGISTER_OPERATIONS targets.
30481 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30484 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30485 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30486 elimination by swapping fld*.
30488 2017-02-22 Richard Biener <rguenther@suse.de>
30490 PR tree-optimization/79673
30491 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30492 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30493 irrelevant address-space qualifiers and avoiding a
30494 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30496 2017-02-22 Richard Biener <rguenther@suse.de>
30498 PR tree-optimization/79666
30499 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30500 to not symbolically negate if that may introduce undefined
30503 2017-02-22 Martin Liska <mliska@suse.cz>
30506 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30507 * data-streamer-out.c (streamer_write_gcov_count_stream):
30509 * value-prof.c (stream_out_histogram_value): Make assert more
30510 precise based on type of counter.
30512 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30515 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30516 Use nonimmediate_operand instead of memory_operand for operand 1.
30517 (float-extend standard_x87sse_constant_load splitter): Ditto.
30519 2017-02-21 Jeff Law <law@redhat.com>
30521 PR tree-optimization/79621
30522 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30523 blocks with edges to themselves.
30525 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30528 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30529 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30530 Use gimple_call_builtin_p.
30533 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30534 on temporarily removed DEBUG_INSNs.
30536 PR tree-optimization/79649
30537 * tree-loop-distribution.c (classify_partition): Give up on
30538 non-generic address space loads/stores.
30540 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30542 * doc/loop.texi (Loop manipulation): Remove nonexistent
30543 tree_ssa_loop_version from the documentation.
30544 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30546 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30549 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30550 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30551 * config/rs6000/rs6000.c: Include except.h.
30552 (rs6000_expand_split_stack_prologue): Call
30553 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30555 2017-02-21 Martin Jambor <mjambor@suse.cz>
30558 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30559 have been analyzed.
30561 2017-02-21 Martin Jambor <mjambor@suse.cz>
30563 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30564 for backward compatibility only.
30565 * doc/invoke.texi (Option Summary): Remove all references to
30566 -fipa-cp-alignment.
30568 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30572 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30574 * lra-constraints.c (curr_insn_transform): Handle
30575 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30577 2017-02-21 Martin Liska <mliska@suse.cz>
30579 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30581 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30583 PR translation/79638
30584 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30586 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30589 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30590 (arm_function_ok_for_sibcall): Return false for an indirect call by
30591 descriptor if all the argument registers are used.
30592 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30593 alignment of the function.
30595 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30597 PR tree-optimization/61441
30598 * simplify-rtx.c (simplify_const_unary_operation): For
30599 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30600 the sNaN unmodified.
30602 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30604 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30605 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30606 instead of SYSTEM_HEADER_DIR.
30608 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30609 Martin Liška <mliska@suse.cz>
30611 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30612 Fix typos and grammar, use active voice, and clarify.
30614 2017-02-20 Marek Polacek <polacek@redhat.com>
30616 PR middle-end/79537
30617 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30620 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30622 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30625 * config/i386/i386.c (ix86_expand_builtin): Handle
30626 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30627 ix86_builtins_isa[fcode].isa as a requirement of those
30628 flags and any other flag in the bitmask.
30629 (ix86_init_mmx_sse_builtins): Use 0 instead of
30630 ~OPTION_MASK_ISA_64BIT as mask.
30631 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30632 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30633 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30634 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30636 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30639 * lra-constraints.c (split_reg): Check requested split mode
30640 is supported by the register.
30642 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30644 * lra-constraints.c (simplify_operand_subreg): Remove early
30647 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30650 * lra-constraints.c (curr_insn_transform): Tighten condition
30651 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30656 * lra-constraints.c (curr_insn_transform): Handle
30657 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30659 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30662 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30664 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30666 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30669 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30672 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30674 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30677 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30679 * final.c (last_columnnum, override_columnnum): New variables.
30680 (final_start_function): Set last_columnnum, pass it to begin_prologue
30681 hook and pass 0 to dwarf2out_begin_prologue.
30682 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30683 to source_line debug hook.
30684 (notice_source_line): Compute last_columnnum and for debug_column_info
30685 return true on column changes.
30686 * debug.h (struct gcc_debug_hooks): Add column argument to
30687 source_line and begin_prologue hooks.
30688 (debug_nothing_int_charstar_int_bool): Remove prototype.
30689 (debug_nothing_int_int_charstar,
30690 debug_nothing_int_int_charstar_int_bool): New prototypes.
30691 (dwarf2out_begin_prologue): Add column argument.
30692 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30693 begin_prologue hooks.
30694 (debug_nothing_int_charstar_int_bool): Remove.
30695 (debug_nothing_int_int_charstar,
30696 debug_nothing_int_int_charstar_int_bool): New functions.
30697 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30698 through to dwarf2out_source_line.
30699 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30700 (dwarf2out_source_line): Add column argument, emit it if requested.
30701 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30703 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30704 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30705 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30706 through to dwarf2out_begin_prologue.
30707 (vmsdbgout_source_line): Add column argument, pass it through to
30708 dwarf2out_source_line.
30709 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30710 dbxout_source_line caller.
30711 (dbxout_source_line): Add column argument.
30713 * common.opt (gno-column-info, gcolumn-info): New options.
30714 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30715 (check_die): Also test for multiple DW_AT_decl_column attributes.
30716 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30717 DW_AT_decl_column if requested.
30718 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30720 (gen_variable_die): Likewise.
30721 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30722 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30725 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30726 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30727 (ix86_handle_option): Handle OPT_m3dnowa.
30728 * doc/invoke.texi (-m3dnowa): Document.
30729 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30730 -m3dnowa instead of -m3dnow -march=athlon.
30733 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30734 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30736 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30739 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30740 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30741 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30742 generator for vsx_xxpermdi_<mode>_be.
30743 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30744 force big-endian semantics.
30745 (vsx_xxpermdi_<mode>_be): New define_expand with same
30746 implementation as previous version of vsx_xxpermdi_<mode>.
30748 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30750 PR tree-optimization/79327
30751 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30752 variable, its initialization and use.
30754 2017-02-17 Julia Koval <julia.koval@intel.com>
30756 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30757 (OPTION_MASK_ISA_PKU_UNSET): New.
30758 (ix86_handle_option): Handle -mrdpid.
30759 * config/i386/cpuid.h (bit_RDPID): New.
30760 * config/i386/driver-i386.c (host_detect_local_cpu):
30761 Detect RDPID feature.
30762 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30763 * config/i386/i386-c.c (ix86_target_macros_internal):
30765 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30766 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30767 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30768 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30769 * config/i386/i386.md (define_insn "rdpid"): New.
30770 * config/i386/i386.opt Add -mrdpid.
30771 * config/i386/immintrin.h (_rdpid_u32): New.
30773 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30775 PR rtl-optimization/79541
30776 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30777 instead of transforming it into USE.
30779 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30781 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30782 If HONOR_SNANS (SFmode) force the input to a register.
30783 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30784 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30785 an frsp or similar insn.
30787 2017-02-17 Martin Liska <mliska@suse.cz>
30789 PR rtl-optimization/79577
30790 * params.def (selsched-max-sched-times): Increase minimum to 1.
30792 2017-02-17 Martin Liska <mliska@suse.cz>
30794 PR rtl-optimization/79574
30795 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30797 2017-02-17 Martin Liska <mliska@suse.cz>
30799 PR tree-optimization/79529
30800 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30801 ssa_defined_default_def_p to handle cases which are implicitly
30803 * tree-ssa.c (ssa_defined_default_def_p): New function.
30804 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30805 which are implicitly defined.
30806 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30808 2017-02-17 Richard Biener <rguenther@suse.de>
30810 PR middle-end/79576
30811 * params.def (max-ssa-name-query-depth): Limit to 10.
30813 2017-02-17 Richard Biener <rguenther@suse.de>
30815 PR tree-optimization/79552
30816 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30819 2017-02-17 Richard Biener <rguenther@suse.de>
30822 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30824 2017-02-17 Marek Polacek <polacek@redhat.com>
30826 PR middle-end/79536
30827 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30828 (fold_negate_expr): New wrapper.
30830 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30832 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30833 Correct terminology and de-emphasize pre-standard behavior.
30835 2017-02-16 Alan Modra <amodra@gmail.com>
30837 PR rtl-optimization/79286
30838 * ira.c (def_dominates_uses): New function.
30839 (update_equiv_regs): Don't create an equivalence for insns that
30840 may trap where the register def does not dominate the use.
30842 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30844 PR rtl-optimization/78127
30845 * lra.c (lra): Call lra_eliminate before finish the loop after
30848 2017-02-16 Richard Biener <rguenther@suse.de>
30850 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30852 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30853 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30854 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30855 (isl_val_int_from_wi): New function.
30856 (extract_affine_gmp): Rename to ...
30857 (extract_affine_wi): ... this, take a widest_int.
30858 (extract_affine_int): Just wrap extract_affine_wi.
30859 (add_param_constraints): Use isl_val_int_from_wi.
30860 (add_loop_constraints): Likewise, and extract_affine_wi.
30862 2017-02-15 Jeff Law <law@redhat.com>
30864 PR middle-end/79521
30865 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30866 ira_init_register_move_cost_if_necessary.
30868 2017-02-15 Martin Sebor <msebor@redhat.com>
30870 PR middle-end/32003
30871 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30872 removed in a prior commit.
30874 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30876 PR tree-optimization/79347
30877 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30878 counters during peeling.
30880 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30882 * Makefile.in (site.exp): Remove "set ISLVER".
30884 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30887 * real.c (real_from_integer): Call real_convert even for decimal.
30889 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30892 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30894 2017-02-14 Andrew Pinski <apinski@cavium.com>
30896 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30897 cores and change the partno/implementer to be correct.
30898 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30899 the 'B" as the implementer.
30900 * config/aarch64/aarch64-tune.md: Regenerate.
30902 2017-02-14 Carl Love <cel@us.ibm.com>
30904 * config/rs6000/rs6000.c: Add case statement entry to make the
30905 xvcvuxdsp built-in argument unsigned.
30906 * config/rs6000/vsx.md: Fix the source and return operand types so they
30907 match the instruction definitions from the ISA document. Fix typo
30908 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30911 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30914 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30915 member early_clobber_alts.
30916 * lra-lives.c (reg_early_clobber_p): New.
30917 (process_bb_lives): Use it.
30918 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30919 (debug_operand_data): Initialize early_clobber_alts.
30920 (setup_operand_alternative): Set up early_clobber_alts.
30921 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30922 alternatives to new_insn_reg.
30923 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30925 (lra_update_insn_regno_info): Pass the new arg.
30927 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30929 PR middle-end/79505
30930 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30931 (new_oacc_loop_raw): Don't clear already cleared fields.
30934 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30935 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30936 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30938 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30941 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30943 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30946 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30947 the extra instruction to the right place to store 128-bit constant
30950 2017-02-14 Martin Sebor <msebor@redhat.com>
30952 PR middle-end/79448
30953 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30954 warning for strings of unknown length.
30956 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30958 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30960 2017-02-14 Jeff Law <law@redhat.com>
30963 * ira-costs.c (scan_one_insn): Initialize register move costs
30964 for pseudos seen in USE/CLOBBER insns.
30966 PR tree-optimization/79095
30967 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30968 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30969 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30970 if the operands are known to be not equal, then the resulting range
30972 (intersect_ranges): If the new range is ~[0,0] and the old range is
30973 wide, then prefer ~[0,0].
30974 * tree-vrp.c (overflow_comparison_p_1): New function.
30975 (overflow_comparison_p): New function.
30976 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30977 if NAME is used in an overflow test.
30978 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30979 overflow check that can be expressed as an equality test, then adjust
30980 ops to be that equality test.
30982 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30984 * config/s390/s390-builtin-types.def: Remove flags argument.
30985 * config/s390/s390.c (s390_init_builtins): Likewise.
30987 2017-02-14 Martin Liska <mliska@suse.cz>
30989 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30990 vector. Fix trailing white spaces.
30992 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30994 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30997 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30999 PR rtl-optimization/68664
31000 * config/arm/arm.c (arm_sched_can_speculate_insn):
31001 New function. Declare prototype.
31002 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31004 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31006 PR rtl-optimization/68664
31007 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31009 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31011 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31013 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31014 max skip bytes for function, loop and jump.
31016 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31018 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31019 ABS_EXPR for gimple dump.
31021 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31024 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31026 PR tree-optimization/79408
31027 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31028 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31029 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31030 also if rhs1 is INTEGER_CST.
31032 2017-02-14 Richard Biener <rguenther@suse.de>
31034 PR middle-end/79432
31035 * tree-into-ssa.c (insert_phi_nodes): When the function can
31036 have abnormal edges rewrite SSA names with broken use-def
31037 dominance out of SSA and register them for PHI insertion.
31039 2017-02-13 Martin Sebor <msebor@redhat.com>
31041 PR middle-end/79496
31042 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31043 clearing info.nowrite flag when snprintf size argument is a range.
31045 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31047 * cprop.c (cprop_jump): Add missing space in string literal.
31048 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31049 (get_constraint_for_component_ref): Likewise.
31050 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31051 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31052 * lra-constraints.c (process_alt_operands): Likewise.
31053 * ipa-inline.c (inline_small_functions): Likewise.
31054 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31055 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31056 * trans-mem.c (diagnose_tm_1_op): Likewise.
31057 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31058 (grid_parallel_clauses_gridifiable): Likewise.
31060 * config/nvptx/mkoffload.c (process): Add space in between
31063 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31064 "MOD4_SSE_REGS" and "ALL_REGS".
31066 * spellcheck.c (test_data): Add , in between "foo" and "food".
31068 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31071 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31072 boundary crossing check and subsequent code generation agree.
31074 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31076 * config/aarch64/aarch64.c (has_memory_op): Delete.
31077 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31080 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31082 PR rtl-optimization/79388
31083 PR rtl-optimization/79450
31084 * combine.c (distribute_notes): When removing TEM_INSN for which
31085 corresponding dest has last value recorded, invalidate that last
31088 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31090 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31091 of explicit '@'. Add missing assembly comment marker on branch costs
31094 2017-02-13 Nathan Sidwell <nathan@acm.org>
31096 * gengtype-lex.l (<in_struct>): Add '/'.
31098 2017-02-13 Martin Liska <mliska@suse.cz>
31101 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31103 2017-02-13 Richard Biener <rguenther@suse.de>
31105 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31107 * configure: Re-generate.
31108 * config.in: Likewise.
31109 * graphite-dependences.c: Simplify as if
31110 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31111 * graphite-isl-ast-to-gimple.c: Likewise.
31112 * graphite-optimize-isl.c: Likewise.
31113 * graphite-poly.c: Likewise.
31114 * graphite-sese-to-poly.c: Likewise.
31115 * graphite.h: Likewise.
31116 * toplev.c: Include isl/version.h and use isl_version () for
31117 printing the ISL version.
31118 * doc/install.texi: Update ISL requirement.
31120 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31122 * doc/standards.texi (Standards): Update reference to
31125 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31127 * doc/extend.texi (Named Address Spaces): sourceware.org now
31129 * doc/install.texi (Binaries): Ditto.
31132 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31134 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31135 standard terminology "stringize"/"stringizing" throughout.
31136 * doc/cppinternals.texi: Likewise.
31138 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31140 * doc/extend.texi: Fix some spelling mistakes and typos.
31141 * doc/invoke.texi: Likewise.
31143 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31146 * params.def (inline-min-speedup) Change from 10 to 8.
31148 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31150 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31153 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31156 * ipa-inline-analysis.c (get_minimal_bb): New function.
31157 (record_modified): Use it.
31158 (remap_edge_change_prob): Handle also ancestor functions.
31160 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31162 * doc/contrib.texi (Contributors): Remove broken link into
31163 the Mauve CVS repository.
31165 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31167 PR middle-end/79454
31168 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31169 result computation whenever lhs doesn't have vector mode, not
31170 just when it has BLKmode.
31172 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31174 * doc/makefile.texi (profiledbootstrap): Refer to the
31175 installation instructions only in textual form.
31177 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31180 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31182 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31184 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31185 (Specific): Update mingw-w64 reference.
31187 (Specific): Remove broken link to Renesas RX processor.
31189 2017-02-10 Richard Biener <rguenther@suse.de>
31191 * toplev.c (process_options): Do not mention obsolete graphite
31192 options when printing sorry message about missing graphite support.
31193 Mention -floop-nest-optimize.
31195 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31197 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31198 (vtst_p16): Likewise.
31199 (vtstq_p8): Likewise.
31200 (vtstq_p16): Likewise.
31202 (vtstq_p64): Likewise.
31203 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31204 (vset_lane_p64): New.
31205 (vsetq_lane_p64): New.
31207 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31209 PR tree-optimization/79411
31210 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31211 stmt operands are SSA_NAMEs used in abnormal phis.
31212 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31215 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31218 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31221 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31223 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31225 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31227 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31231 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31232 not arbitrary TREE_CONSTANT.
31235 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31236 "omp declare target link" attribute unless is_global_var.
31237 * omp-offload.c (find_link_var_op): Likewise.
31239 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31240 Chung-Lin Tang <cltang@codesourcery.com>
31242 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31244 (gimplify_adjust_omp_clauses): Don't delete TILE.
31245 (gimplify_omp_for): Deal with TILE.
31246 * internal-fn.c (expand_GOACC_TILE): New function.
31247 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31249 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31250 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31252 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31253 avoid DIV for outermost collapse var.
31254 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31255 Remove out of date comments, fix whitespace.
31256 * omp-general.c (omp_extract_for_data): Deal with tiling.
31257 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31258 adjust OLF_DIM_BASE value.
31259 (struct omp_for_data): Add tiling field.
31260 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31261 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31262 for auto loops. Remove default auto determining, moved to
31263 oacc_loop_fixed_partitions.
31264 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31265 stmts, add e_mask field.
31266 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31267 (oacc_thread_numbers): Use oacc_dim_call.
31268 (oacc_xform_tile): New.
31269 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31270 (finish_oacc_loop): Adjust for ifns vector.
31271 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31272 add case for GOACC_TILE fns.
31273 (oacc_loop_xform_loop): Delete.
31274 (oacc_loop_process): Iterate over call list directly, and add
31275 handling for GOACC_TILE fns.
31276 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31278 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31279 vector partitioning to outer loops. Assign 2 partitions to loops
31280 when available. Add TILE handling.
31281 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31282 (execite_oacc_device_lower): Process GOACC_TILE fns,
31283 ignore unknown specs.
31284 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31285 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31286 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31288 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31290 * configure.ac (ACX_BUGURL): Update.
31291 * configure: Regenerate.
31293 2017-02-09 Richard Biener <rguenther@suse.de>
31295 PR tree-optimization/69823
31296 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31297 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31299 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31301 * config/arc/arc-c.def: Add __NPS400__ definition.
31302 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31303 (TARGET_NPS400): Define.
31305 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31307 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31309 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31310 pointer, arch_info.
31311 (arc_cpu_types): Fill the arch_info field with a pointer into the
31312 arc_arch_types table.
31313 (arc_selected_cpu): Declare.
31314 * config/arc/arc.c (arc_selected_cpu): Make global.
31315 (arc_selected_arch): Delete.
31316 (arc_base_cpu): Delete.
31317 (arc_override_options): Remove references to deleted variables,
31318 update access to arch information.
31319 (ARC_OPT): Update access to arch information.
31320 (ARC_OPTX): Likewise.
31321 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31322 (TARGET_ARC600): Update access to arch information.
31323 (TARGET_ARC601): Likewise.
31324 (TARGET_ARC700): Likewise.
31325 (TARGET_EM): Likewise.
31326 (TARGET_HS): Likewise.
31327 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31330 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31333 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31334 condition/operands for integer GE/LE/GEU/LEU operations.
31336 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31338 PR translation/79397
31339 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31342 2017-02-08 Martin Jambor <mjambor@suse.cz>
31345 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31346 whether allocation happened.
31347 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31348 nothing was allocated.
31350 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31352 PR tree-optimization/79408
31353 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31354 constant, but SSA_NAME with a known integer range, use the minimum
31355 of that range instead of op1 to determine if modulo can be replaced
31356 with its first operand.
31358 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31360 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31362 2017-02-08 Richard Biener <rguenther@suse.de>
31364 PR tree-optimization/71824
31365 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31366 Check all loops contained in the merged region.
31368 2017-02-07 Andrew Pinski <apinski@cavium.com>
31370 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31372 2017-02-07 Andrew Pinski <apinski@cavium.com>
31374 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31375 (thunderxt88): Likewise.
31376 (thunderxt81): Disable LSE and change v8.1 to v8.
31377 (thunderxt83): Likewise.
31379 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31380 Richard Biener <rguenther@suse.de>
31382 PR middle-end/79399
31383 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31384 type from int to size_t.
31385 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31387 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31389 PR rtl-optimization/79386
31390 * cprop.c (bypass_conditional_jumps): Initialize
31391 bypass_last_basic_block already before splitting bbs after
31392 unconditional traps...
31393 (bypass_conditional_jumps): ... rather than here.
31396 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31397 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31398 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31399 fix -masm=intel patterns.
31401 2017-02-07 Richard Biener <rguenther@suse.de>
31403 PR tree-optimization/79256
31404 PR middle-end/79278
31405 * builtins.c (get_object_alignment_2): Use min_align_of_type
31406 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31407 and ADJUST_FIELD_ALIGN.
31409 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31411 * doc/tm.texi: Regenerate.
31412 * stor-layout.c (layout_decl): Adjust.
31413 (update_alignment_for_field): Likewise.
31414 (place_field): Likewise.
31415 (min_align_of_type): Likewise.
31416 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31417 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31418 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31419 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31420 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31421 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31422 * config/i386/i386.c (x86_field_alignment): Likewise.
31423 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31424 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31425 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31426 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31427 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31428 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31432 2017-01-30 Richard Biener <rguenther@suse.de>
31434 PR tree-optimization/79256
31435 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31436 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31439 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31441 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31442 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31443 builtins to SImode and emit a zero-extend, if necessary.
31445 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31447 * docs/invoke.texi (RISC-V Options): Alphabetize.
31449 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31451 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31454 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31456 * config/riscv/riscv.c: New file.
31457 * common/config/riscv/riscv-common.c: Likewise.
31458 * config.gcc: Likewise.
31459 * config/riscv/constraints.md: Likewise.
31460 * config/riscv/elf.h: Likewise.
31461 * config/riscv/generic.md: Likewise.
31462 * config/riscv/linux.h: Likewise.
31463 * config/riscv/multilib-generator: Likewise.
31464 * config/riscv/peephole.md: Likewise.
31465 * config/riscv/pic.md: Likewise.
31466 * config/riscv/predicates.md: Likewise.
31467 * config/riscv/riscv-builtins.c: Likewise.
31468 * config/riscv/riscv-c.c: Likewise.
31469 * config/riscv/riscv-ftypes.def: Likewise.
31470 * config/riscv/riscv-modes.def: Likewise.
31471 * config/riscv/riscv-opts.h: Likewise.
31472 * config/riscv/riscv-protos.h: Likewise.
31473 * config/riscv/riscv.h: Likewise.
31474 * config/riscv/riscv.md: Likewise.
31475 * config/riscv/riscv.opt: Likewise.
31476 * config/riscv/sync.md: Likewise.
31477 * config/riscv/t-elf-multilib: Likewise.
31478 * config/riscv/t-linux: Likewise.
31479 * config/riscv/t-linux-multilib: Likewise.
31480 * config/riscv/t-riscv: Likewise.
31481 * configure.ac: Likewise.
31482 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31483 Waterman as RISC-V maintainers.
31484 * doc/install.texi: Add RISC-V entries.
31485 * doc/invoke.texi: Add RISC-V options section.
31486 * doc/md.texi: Add RISC-V constraints section.
31487 * configure: Regenerated.
31489 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31492 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31493 false values to be constant vectors with all 0 or all 1 bits set.
31494 (vcondu<mode><mode>): Likewise.
31495 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31497 (fpmask_comparison_operator): Update comment.
31498 (vecint_comparison_operator): New predicate.
31499 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31500 vector conditionals when the true and false values are constant
31501 vectors with all 0 bits or all 1 bits set.
31503 2017-02-06 Martin Sebor <msebor@redhat.com>
31505 PR tree-optimization/79376
31506 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31508 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31510 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31511 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31512 to simplify split condition.
31514 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31516 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31517 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31520 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31522 PR rtl-optimization/68664
31523 * target.def (can_speculate_insn): New hook.
31524 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31525 * doc/tm.texi: Regenerate.
31526 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31527 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31528 (rs6000_sched_can_speculate_insn): New function.
31530 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31532 PR tree-optimization/79284
31533 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31534 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31535 vectorizable_mask_load_store, vectorizable_operation,
31536 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31537 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31538 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31539 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31540 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31541 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31542 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31543 is_gimple_assign (stmt). Replace another such test with
31544 is_gimple_assign (stmt).
31546 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31549 * config/avr/avr.c (rtl-iter.h): Include it.
31550 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31551 (avr_legitimate_combined_insn): ...and implementation.
31553 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31555 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31556 * config/s390/s390.c (s390_const_operand_ok)
31557 (s390_canonicalize_comparison, s390_extract_part)
31558 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31559 (s390_contiguous_bitmask_p, s390_rtx_costs)
31560 (legitimize_pic_address): Likewise.
31561 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31562 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31563 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31564 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31565 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31567 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31569 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31570 REGNO($0) == REGNO($1).
31572 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31574 * config/s390/linux.h(SIZE_TYPE): Add comment.
31576 2017-02-06 Julian Brown <julian@codesourcery.com>
31577 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31578 Virendra Pathak <virendra.pathak@broadcom.com>
31580 * config/aarch64/aarch64-cores.def: Change the scheduler
31582 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31583 * config/aarch64/thunderx2t99.md: New file.
31585 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31587 * doc/standards.texi (Go Language): Update link to language
31590 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31592 * tree-eh.c (lower_resx): Sanitize profile.
31593 (cleanup_empty_eh_move_lp): Likewise.
31595 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31598 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31600 * cfgloopmanip.h (loop_version): Update prototype.
31601 * modulo-sched.c (sms_schedule): Update call of loop_version.
31602 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31603 * tree-parloops.c (gen_parallel_loop): Likewise.
31604 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31605 * tree-ssa-loop-split.c (split_loop): Likewise.
31606 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31607 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31609 2017-02-05 Martin Liska <mliska@suse.cz>
31612 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31614 (print_operand_address): Initialize a struct to zero.
31616 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31618 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31619 garbage collector only in textual form.
31621 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31623 * doc/extend.texi (x86 specific memory model extensions for
31624 transactional memory): Simplify a phrase.
31626 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31629 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31630 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31631 (atomic_storedi_1): Likewise.
31633 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31635 PR tree-optimization/79338
31636 * tree-parloops.c (gather_scalar_reductions): Don't call
31637 vect_analyze_loop_form for loop->inner before destroying loop's
31640 2017-02-03 Martin Sebor <msebor@redhat.com>
31642 PR tree-optimization/79327
31643 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31644 when precision has resulted in leading zeros.
31645 (format_integer): Adjust the likely counter to assume an unknown
31646 argument that may be zero is non-zero.
31648 2017-02-03 Jason Merrill <jason@redhat.com>
31651 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31652 avoid copying non-taken branch.
31654 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31656 PR tree-optimization/79340
31657 * tree-vect-loop.c (vectorizable_reduction): Release
31658 vec_defs elements after safe_splicing them into other vectors.
31661 PR tree-optimization/79327
31662 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31663 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31665 (format_integer): Use wide_int_to_tree instead of build_int_cst
31666 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31667 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31668 of shortest and longest sequence.
31670 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31672 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31673 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31675 2017-02-03 Walter Lee <walt@tilera.com>
31678 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31679 after initial stackframe link reg save.
31680 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31682 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31685 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31686 wu for stxssp alternative.
31688 2017-02-03 Martin Sebor <msebor@redhat.com>
31690 PR tree-optimization/79352
31691 * gimple-fold.c (get_range_strlen): Add argument.
31692 (get_range_strlen): Change return type to bool.
31693 (get_maxval_strlen): Pass in a dummy argument.
31694 * gimple-fold.h (get_range_strlen): Change return type to bool.
31695 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31696 * tree.h (array_at_struct_end_p): Add argument.
31697 * tree.c (array_at_struct_end_p): Handle it.
31699 2017-02-03 Martin Liska <mliska@suse.cz>
31702 * multiple_target.c (create_dispatcher_calls): Redirect edge
31703 from a caller of a dispatcher.
31704 (expand_target_clones): Make the clones local.
31705 (ipa_target_clone): Do both target clones and resolvers.
31706 (ipa_dispatcher_calls): Remove the pass.
31707 (pass_dispatcher_calls::gate): Likewise.
31708 (make_pass_dispatcher_calls): Likewise.
31709 * passes.def (pass_target_clone): Put as very first IPA early
31712 2017-02-03 Martin Liska <mliska@suse.cz>
31714 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31715 in case of a function with ifunc attribute.
31717 2017-02-03 Martin Liska <mliska@suse.cz>
31719 * cgraph.c (cgraph_node::dump): Dump function version info.
31720 * symtab.c (symtab_node::dump_base): Add missing new line.
31722 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31724 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31725 (ifcombine_ifandif): Use it.
31727 2017-02-03 Martin Liska <mliska@suse.cz>
31729 * doc/invoke.texi: Document default value for
31730 use-after-scope-direct-emission-threshold.
31732 2017-02-03 Martin Liska <mliska@suse.cz>
31734 PR tree-optimization/79339
31735 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31736 (format_floating): Likewise.
31738 2017-02-03 Martin Liska <mliska@suse.cz>
31741 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31743 (ipa_node_params_t::remove): Likewise.
31744 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31745 initialization from removed ipa_node_params_t::insert.
31746 (ipa_node_params::~ipa_node_params): Move from removed
31747 ipa_node_params_t::release.
31748 * symbol-summary.h (symbol_summary::m_released): New member.
31749 Do not release a summary twice. Do not allow to call finalizer
31750 for types of a summary that live in GGC memory.
31752 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31754 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31757 2017-02-02 Martin Sebor <msebor@redhat.com>
31759 PR middle-end/79275
31760 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31761 (format_string): Tighten up the range of output for non-constant
31762 strings and correct the expected range for wide non-constant strings.
31764 2017-02-02 Martin Sebor <msebor@redhat.com>
31766 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31768 PR middle-end/32003
31769 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31771 (-fdump-tree-@var): Add to index and document how to come up
31772 with pass-specific option and dump file names.
31773 (-fdump-passes): Clarify where to look for output.
31775 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31777 PR middle-end/77445
31778 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31779 statistics of the analyzed path; allow threading for speed when
31780 any of BBs along the path are optimized for speed.
31782 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31784 PR middle-end/78468
31785 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31786 settings of the virtual registers.
31789 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31791 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31792 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31795 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31797 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31798 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31800 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31802 * config/s390/s390.md: Add missing comments with the expanded
31804 * config/s390/vector.md: Likewise.
31805 * config/s390/vx-builtins.md: Likewise.
31807 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31810 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31811 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31812 conditions on a single line.
31814 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31816 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31817 __S390_VX__ to __VX__.
31819 2017-02-01 Andrew Pinski <apinski@cavium.com>
31821 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31822 stmt_info to record_stmt_cost.
31823 (vect_get_known_peeling_cost): Pass stmt_info if known to
31825 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31826 cpu_vector_cost field into
31827 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31828 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31829 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31830 splitting of scalar_stmt_cost and vec_stmt_cost.
31831 (thunderx_vector_cost): Likewise.
31832 (cortexa57_vector_cost): LIkewise.
31833 (exynosm1_vector_cost): Likewise.
31834 (xgene1_vector_cost): Likewise.
31835 (thunderx2t99_vector_cost): Improve after the splitting of the two
31837 (aarch64_builtin_vectorization_cost): Update for the splitting of
31838 scalar_stmt_cost and vec_stmt_cost.
31840 2017-02-01 Torvald Riegel <triegel@redhat.com>
31841 Richard Henderson <rth@redhat.com>
31843 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31844 conditional on existance of a fast atomic load.
31845 * optabs-query.c (can_atomic_load_p): New function.
31846 * optabs-query.h (can_atomic_load_p): Declare it.
31847 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31848 no fast atomic load is available for the particular size of access.
31849 (expand_atomic_compare_and_swap): Likewise.
31850 (expand_atomic_load): Likewise.
31851 (expand_atomic_store): Likewise.
31852 (expand_atomic_fetch_op): Likewise.
31853 * testsuite/lib/target-supports.exp
31854 (check_effective_target_sync_int_128): Remove x86 because it provides
31855 no fast atomic load.
31856 (check_effective_target_sync_int_128_runtime): Likewise.
31858 2017-02-01 Richard Biener <rguenther@suse.de>
31860 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31861 (graphite_transform_loops): Provide opt-info for optimized nests.
31862 * tree-parloop.c (parallelize_loops): Provide opt-info for
31863 parallelized loops.
31865 2017-02-01 Richard Biener <rguenther@suse.de>
31867 PR middle-end/79315
31868 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31869 was not set before.
31871 2017-02-01 Richard Biener <rguenther@suse.de>
31873 PR tree-optimization/71824
31874 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31875 Verify the loops are valid in the merged SESE region.
31876 (scop_detection::can_represent_loop_1): Check analyzing the
31877 evolution of the number of iterations in the region succeeds.
31879 2017-01-31 Ian Lance Taylor <iant@golang.org>
31881 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31882 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31884 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31886 PR preprocessor/79210
31887 * input.c (get_substring_ranges_for_loc): Replace line_width
31888 assertion with error-handling.
31890 2017-01-31 Richard Biener <rguenther@suse.de>
31892 PR tree-optimization/77318
31893 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31894 (create_pw_aff_from_tree): Take loop parameter.
31895 (add_condition_to_pbb): Pass loop of the condition to
31896 create_pw_aff_from_tree.
31898 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31900 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31901 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31903 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31907 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31909 (convert_int_to_float128): Likewise.
31910 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31911 (convert_int_to_float128): Likewise.
31912 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31913 (UNSPEC_IEEE128_CONVERT): Likewise.
31914 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31915 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31916 Use local variables for IBM extended format.
31917 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31918 (fix_trunc<mode>si2_fprs): Likewise.
31919 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31920 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31921 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31922 to know that we can now have integers of all sizes in vector
31924 (fix<uns>_<mode>di2_hw): Likewise.
31925 (float<uns>_<mode>si2_hw): Likewise.
31926 (fix_<mode>si2_hw): Likewise.
31927 (fixuns_<mode>si2_hw): Likewise.
31928 (float<uns>_<mode>di2_hw): Likewise.
31929 (float_<mode>di2_hw): Likewise.
31930 (float_<mode>si2_hw): Likewise.
31931 (floatuns_<mode>di2_hw): Likewise.
31932 (floatuns_<mode>si2_hw): Likewise.
31933 (xscvqp<su>wz_<mode>): Delete, no longer used.
31934 (xscvqp<su>dz_<mode>): Likewise.
31935 (xscv<su>dqp_<mode>): Likewise.
31936 (ieee128_mfvsrd_64bit): Likewise.
31937 (ieee128_mfvsrd_32bit): Likewise.
31938 (ieee128_mfvsrwz): Likewise.
31939 (ieee128_mtvsrw): Likewise.
31940 (ieee128_mtvsrd_64bit): Likewise.
31941 (ieee128_mtvsrd_32bit): Likewise.
31943 2017-01-31 Martin Liska <mliska@suse.cz>
31946 * ipa-prop.c (ipa_free_all_node_params): Call release method
31947 instead of ~sumbol_summary to not to trigger double times
31950 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31952 PR tree-optimization/71691
31953 * bitmap.h (class auto_bitmap): New.
31954 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31955 is_maybe_undefined instead of ssa_undefined_value_p.
31957 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31959 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31960 __S390_ARCH_LEVEL__ to __ARCH__.
31962 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31964 PR tree-optimization/79267
31965 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31966 if should_remove_lhs_p is true.
31968 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31971 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31972 (add_alignment_attribute): New.
31973 (base_type_die): Add alignment attribute.
31974 (subrange_type_die): Likewise.
31975 (modified_type_die): Likewise.
31976 (gen_array_type_die): Likewise.
31977 (gen_descr_array_type_die: Likewise.
31978 (gen_enumeration_type_die): Likewise.
31979 (gen_subprogram_die): Likewise.
31980 (gen_variable_die): Likewise.
31981 (gen_field_die): Likewise.
31982 (gen_ptr_to_mbr_type_die): Likewise.
31983 (gen_struct_or_union_type_die): Likewise.
31984 (gen_subroutine_type_die): Likewise.
31985 (gen_typedef_die): Likewise.
31986 (base_type_cmp): Compare alignment attribute.
31988 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31991 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31992 (setb_unsigned) New pattern for setb with CCUNS.
31993 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31994 subfc./subfe sequence to avoid overflow problems. Generate a
31995 shorter sequence with cmpld/setb for power9.
31996 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31997 for generating subfc. instruction.
31998 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31999 now uses this instruction.
32001 2017-01-30 Ian Lance Taylor <iant@google.com>
32004 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32005 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32007 2017-01-30 Martin Sebor <msebor@redhat.com>
32009 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32010 Move constant to the right of a relational operator.
32011 (get_mpfr_format_length, format_character, format_string): Ditto.
32012 (should_warn_p, maybe_warn): Same.
32014 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32016 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32019 * asan.c (get_translation_unit_decl): Remove function.
32020 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32022 2017-01-30 Martin Liska <mliska@suse.cz>
32024 PR gcov-profile/79259
32025 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32026 -fprofile-generate.
32028 2017-01-30 Martin Liska <mliska@suse.cz>
32031 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32032 Initialize variables with NULL value.
32034 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32037 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32039 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32041 2017-01-30 Richard Biener <rguenther@suse.de>
32043 PR tree-optimization/79276
32044 * tree-vrp.c (process_assert_insertions): Properly adjust common
32045 when removing a duplicate.
32047 2017-01-30 Richard Biener <rguenther@suse.de>
32049 PR tree-optimization/79256
32050 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32051 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32053 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32055 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32058 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32059 ("*r<noxa>sbg_<mode>_sll_bitmask")
32060 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32061 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32062 Use contiguous_bitmask_nowrap_operand.
32064 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32067 * config/rs6000/altivec.h (vec_xl): Revise #define.
32068 (vec_xst): Likewise.
32070 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32072 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32074 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32076 PR rtl-optimization/79194
32077 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32078 traps before call to bypass_conditional_jumps.
32080 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32082 PR tree-optimization/71374
32083 * lra-constraints.c (check_conflict_input_operands): New.
32084 (match_reload): Use it.
32086 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32089 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32090 account to calculate conflict_set.
32092 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32094 PR rtl-optimization/78559
32095 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32096 other_insn in combine.
32098 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32100 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32101 uint16_type_node for BT_UINT16.
32103 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32105 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32106 "RTL Tests" to menu.
32107 (GIMPLE Tests): New node.
32108 (RTL Tests): New node.
32110 2017-01-27 Richard Biener <rguenther@suse.de>
32112 PR tree-optimization/79245
32113 * tree-loop-distribution.c (distribute_loop): Apply cost
32114 modeling also to detected patterns.
32116 2017-01-27 Richard Biener <rguenther@suse.de>
32118 PR tree-optimization/71433
32119 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32120 (compare_assert_loc): New function.
32121 (process_assert_insertions): Sort and optimize assert locations
32122 to remove duplicates and push down identical assertions on
32123 edges to their destination block.
32125 2017-01-27 Richard Biener <rguenther@suse.de>
32127 PR tree-optimization/79244
32128 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32129 out SSA names even if abnormal.
32131 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32133 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32134 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32135 instead of MPFR_RNDN.
32137 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32140 * arm.c (arm_option_override): Don't call build_target_option_node
32141 until after doing all option overrides.
32142 (arm_valid_target_attribute_tree): Likewise.
32144 2017-01-27 Martin Liska <mliska@suse.cz>
32146 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32147 for {cd}tors and C++ {cd}tors.
32149 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32151 * config/s390/s390.md ("*setmem_long_and")
32152 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32154 2017-01-26 Martin Sebor <msebor@redhat.com>
32156 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32159 2017-01-26 Martin Sebor <msebor@redhat.com>
32161 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32162 HAVE_DFmode before using XFmode or DFmode.
32163 (parse_directive): Avoid using the z length modifier to avoid
32164 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32166 PR middle-end/78703
32167 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32168 to accept adjustment as an array.
32169 (get_int_range): New function.
32170 (struct directive): Make width and prec arrays.
32171 (directive::set_width, directive::set_precision): Call get_int_range.
32172 (format_integer, format_floating): Handle width and precision ranges.
32173 (format_string, parse_directive): Same.
32175 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32178 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32179 comdat_type_p set, just clone them, but keep the children in the
32183 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32184 which have direct callers with -fvar-tracking-assignments enabled
32186 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32187 inside of type units.
32189 2017-01-26 Martin Sebor <msebor@redhat.com>
32191 PR middle-end/78703
32192 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32194 (struct format_result): Replace number_chars, number_chars_min,
32195 and number_chars_max with a single member of struct result_range.
32197 (format_result::operator+=): Adjust.
32198 (struct fmtresult): Remove bounded. Handle likely and unlikely
32200 (fmtresult::adjust_for_width_or_precision): New function.
32201 (fmtresult:type_max_digits): New function.
32202 (bytes_remaining): Handle likely and unlikely counters.
32203 (min_bytes_remaining): Remove.
32204 (format_percent): Simplify.
32205 (format_integer, format_floating): Set likely and unlikely counters.
32206 (get_string_length, format_character, format_string): Same.
32207 (format_plain, should_warn_p): New function.
32208 (maybe_warn): Call should_warn_p. Update diagnostic messages
32209 and handle those for all directives, including plain strings.
32210 (format_directive): Handle likely and unlikely counters.
32211 Remove unnecessary quoting from diagnostics. Add an informational
32213 (add_bytes): Remove.
32214 (pass_sprintf_length::compute_format_length): Simplify.
32215 (try_substitute_return_value): Handle likely and unlikely counters.
32217 2017-01-26 Carl Love <cel@us.ibm.com>
32219 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32220 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32222 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32225 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32226 endianess for subregs into account.
32227 * lra-constraints.c (lra_constraints): Do risky transformations
32228 always on the first iteration.
32229 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32230 last_call_used_reg_set.
32231 (process_bb_lives): Define and use last_call_used_reg_set.
32232 * lra.c (lra): Always continue after lra_constraints on the first
32235 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32237 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32239 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32241 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32243 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32244 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32245 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32246 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32247 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32248 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32249 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32250 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32251 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32253 2017-01-26 Marek Polacek <polacek@redhat.com>
32256 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32257 for the third operand.
32259 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32261 PR middle-end/79236
32262 * omp-low.c (struct omp_context): Add simt_stmt field.
32263 (scan_omp_for): Return omp_context *.
32264 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32265 context to the _simt_ SIMD stmt.
32266 (lower_omp_for): For combined SIMD with sibling _simt_
32267 SIMD, make sure to use the same decls in _looptemp_
32268 clauses as in the sibling.
32270 2017-01-26 David Sherwood <david.sherwood@arm.com>
32272 PR middle-end/79212
32273 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32276 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32279 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32280 emit fld b; fld a; if possible.
32282 * brig-builtins.def: Update copyright years.
32283 * config/arm/arm_acle_builtins.def: Update copyright years.
32285 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32288 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32289 constraint instead of o for the stxsd instruction.
32291 2017-01-25 Carl Love <cel@us.ibm.com>
32293 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32294 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32296 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32298 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32300 2017-01-25 Richard Biener <rguenther@suse.de>
32302 PR tree-optimization/69264
32303 * target.def (vector_alignment_reachable): Improve documentation.
32304 * doc/tm.texi: Regenerate.
32305 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32307 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32308 earlier changes with respect to TYPE_USER_ALIGN.
32309 (vector_alignment_reachable_p): Likewise. Improve dumping.
32311 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32314 * config/arm/arm.md (xordi3): Force constant operand into a register
32317 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32319 * doc/invoke.texi (-fstore-merging): Correct default optimization
32320 levels at which it is enabled.
32321 (-O): Move -fstore-merging from list to...
32324 2017-01-25 Richard Biener <rguenther@suse.de>
32327 * omp-expand.c: Include debug.h.
32328 (expand_omp_taskreg): Make sure to generate early debug before
32329 outlining anything from a function.
32330 (expand_omp_target): Likewise.
32331 (grid_expand_target_grid_body): Likewise.
32333 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32336 * asan.c (get_translation_unit_decl): New function.
32337 (asan_add_global): Extract modules file name from globals
32338 TRANSLATION_UNIT_DECL name.
32340 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32343 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32344 for long calls with APCS frame and VFP.
32346 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32348 * cfg.c (original_copy_tables_initialized_p): New function.
32349 * cfg.h (original_copy_tables_initialized_p): New decl.
32350 * cfgrtl.c (relink_block_chain): Guard the call to
32351 free_original_copy_tables with a call to
32352 original_copy_tables_initialized_p.
32353 * cgraph.h (symtab_node::native_rtl_p): New decl.
32354 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32355 (symtab_node::needed_p): Don't assert for early assembly output
32356 for __RTL functions.
32357 (cgraph_node::finalize_function): Set "force_output" for __RTL
32359 (cgraph_node::analyze): Bail out early for __RTL functions.
32360 (analyze_functions): Update assertion to support __RTL functions.
32361 (cgraph_node::expand): Bail out early for __RTL functions.
32362 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32364 * function.h (struct function): Update comment for field
32366 * gimple-expr.c: Include "tree-pass.h".
32367 (gimple_has_body_p): Return false for __RTL functions.
32368 * Makefile.in (OBJS): Add run-rtl-passes.o.
32369 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32371 (gcc::pass_manager::get_clean_slate): New accessor.
32372 * passes.c: Include "insn-addr.h".
32373 (should_skip_pass_p): Add logging. Update logic for running
32374 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32375 property-provider override so it is only done for gimple passes.
32377 (skip_pass): New function.
32378 (execute_one_pass): Call skip_pass when skipping passes.
32379 * read-md.c (md_reader::read_char): Support filtering
32380 the input to a subset of line numbers.
32381 (md_reader::md_reader): Initialize fields
32382 m_first_line and m_last_line.
32383 (md_reader::read_file_fragment): New function.
32384 * read-md.h (md_reader::read_file_fragment): New decl.
32385 (md_reader::m_first_line): New field.
32386 (md_reader::m_last_line): New field.
32387 * read-rtl-function.c (function_reader::create_function): Only
32388 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32389 curr_properties. Set DECL_INITIAL to a dummy block.
32390 (read_rtl_function_body_from_file_range): New function.
32391 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32393 * run-rtl-passes.c: New file.
32394 * run-rtl-passes.h: New file.
32396 2017-01-24 Jeff Law <law@redhat.com>
32398 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32401 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32403 PR tree-optimization/79159
32404 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32405 (record_nonwrapping_iv): Improve boundary using above function if no
32406 value range information.
32408 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32409 Martin Jambor <mjambor@suse.cz>
32411 * brig-builtins.def: New file.
32412 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32413 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32414 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32415 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32416 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32417 * builtin-types.def (BT_INT8): New.
32418 (BT_INT16): Likewise.
32419 (BT_UINT8): Likewise.
32420 (BT_UINT16): Likewise.
32421 (BT_FN_ULONG): Likewise.
32422 (BT_FN_UINT_INT): Likewise.
32423 (BT_FN_UINT_ULONG): Likewise.
32424 (BT_FN_UINT_LONG): Likewise.
32425 (BT_FN_UINT_PTR): Likewise.
32426 (BT_FN_ULONG_PTR): Likewise.
32427 (BT_FN_INT8_FLOAT): Likewise.
32428 (BT_FN_INT16_FLOAT): Likewise.
32429 (BT_FN_UINT32_FLOAT): Likewise.
32430 (BT_FN_UINT16_FLOAT): Likewise.
32431 (BT_FN_UINT8_FLOAT): Likewise.
32432 (BT_FN_UINT64_FLOAT): Likewise.
32433 (BT_FN_UINT16_UINT32): Likewise.
32434 (BT_FN_UINT32_UINT16): Likewise.
32435 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32436 (BT_FN_INT_PTR_INT): Likewise.
32437 (BT_FN_UINT_PTR_UINT): Likewise.
32438 (BT_FN_LONG_PTR_LONG): Likewise.
32439 (BT_FN_ULONG_PTR_ULONG): Likewise.
32440 (BT_FN_VOID_UINT64_UINT64): Likewise.
32441 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32442 (BT_FN_INT8_INT8_INT8): Likewise.
32443 (BT_FN_INT16_INT16_INT16): Likewise.
32444 (BT_FN_INT_INT_INT): Likewise.
32445 (BT_FN_UINT_FLOAT_UINT): Likewise.
32446 (BT_FN_FLOAT_UINT_UINT): Likewise.
32447 (BT_FN_ULONG_UINT_UINT): Likewise.
32448 (BT_FN_ULONG_UINT_PTR): Likewise.
32449 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32450 (BT_FN_UINT_UINT_UINT): Likewise.
32451 (BT_FN_VOID_UINT_PTR): Likewise.
32452 (BT_FN_UINT_UINT_PTR: Likewise.
32453 (BT_FN_UINT32_UINT64_PTR): Likewise.
32454 (BT_FN_INT_INT_UINT_UINT): Likewise.
32455 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32456 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32457 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32458 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32459 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32460 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32461 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32462 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32463 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32464 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32465 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32466 * doc/frontends.texi: List BRIG FE.
32467 * doc/install.texi (Testing): Add BRIG tesring requirements.
32468 * doc/invoke.texi (Overall Options): Mention BRIG.
32469 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32471 2017-01-24 Richard Biener <rguenther@suse.de>
32473 PR translation/79208
32474 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32476 2017-01-24 Martin Jambor <mjambor@suse.cz>
32479 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32480 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32481 and known_contexts.
32483 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32485 PR middle-end/79123
32486 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32487 casts from signed to unsigned really don't have a range.
32489 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32491 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32492 GMP_RNDx for compatiblity.
32494 2017-01-24 Martin Liska <mliska@suse.cz>
32497 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32498 that would prevent us to call alloca with -1 as argument.
32500 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32502 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32503 Avoid -Wformat-security warning.
32505 2017-01-23 Andrew Pinski <apinski@cavium.com>
32507 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32510 2017-01-23 Martin Sebor <msebor@redhat.com>
32512 PR middle-end/78703
32513 * gimple-ssa-sprintf.c (warn_level): New global.
32514 (format_integer): Use it here and throughout the rest of the file.
32515 Use the same switch to compute sign as base.
32516 (maybe_warn): New function.
32517 (format_directive): Factor out warnings into maybe_warn.
32518 Add debugging output. Use warn_level.
32519 (add_bytes): Use warn_level.
32520 (pass_sprintf_length::compute_format_length): Add debugging output.
32521 (try_substitute_return_value): Same.
32522 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32524 PR middle-end/78703
32525 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32526 (struct fmtresult, format_integer, format_floating): Adjust.
32527 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32528 (get_string_length, format_string,format_directive): Same.
32529 (pass_sprintf_length::compute_format_length): Same.
32530 (try_substitute_return_value): Simplify slightly.
32532 PR middle-end/78703
32533 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32534 (fmtresult::operator+=): Outlined.
32535 (struct fmtresult): Add ctors.
32536 (struct conversion_spec): Rename...
32537 (struct directive): ...to this. Add and remove data members.
32538 (directive::set_width, directive::set_precision): New functions.
32539 (format_percent): Use fmtresult ctor.
32540 (get_width_and_precision): Remove.
32541 (format_integer): Make naming changes. Avoid computing width and
32543 (format_floating): Same. Adjust indentation.
32544 (format_character, format_none): New functions.
32545 (format_string): Moved character handling to format_character.
32546 (format_directive): Remove arguments, change return type.
32547 (parse_directive): New function.
32548 (pass_sprintf_length::compute_format_length): Move directive
32549 parsing to parse_directive.
32551 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32553 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32554 (assign_assembler_name_if_needed): ... this.
32555 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32556 (assign_assembler_name_if_needed): ... this.
32557 (free_lang_data_in_cgraph): Adjust callers.
32558 * cgraphunit.c (cgraph_node::analyze): Likewise.
32559 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32562 2017-01-23 Richard Biener <rguenther@suse.de>
32564 PR tree-optimization/79088
32565 PR tree-optimization/79188
32566 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32567 resetting loop bounds after last path deletion. Reset loop
32568 bounds of the target loop, make code match the comments.
32569 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32570 Make sure loops need no fixups.
32572 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32574 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32575 exponent support with double type for first argument.
32576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32577 type returned by __builtin_vec_extract_sig,
32578 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32579 functions from "vector int" to "vector unsigned int" or from
32580 "vector long long int" to "vector unsigned long long int".
32581 Changed type returned by __builtin_vec_extract_exp,
32582 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32583 functions from "vector int" to "vector unsigned int" or from
32584 "vector long long int" to "vector unsigned long long int".
32585 Changed return type of __builtin_vec_test_data_class,
32586 __builtin_vec_test_data_class_sp, and
32587 __builtin_vec_test_data_class_dp from "vector int" to
32588 "vector bool int" or from "vector long long int" to "vector bool
32589 long long int" and changed second argument type from "unsigned
32590 int" to "int". Added new overloaded function forms "vector float
32591 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32592 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32593 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32594 double, vector unsigned long long int)" and "vector double
32595 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32596 long int)". Changed return type of
32597 __builtin_scalar_test_data_class and
32598 __builtin_scalar_test_data_class_sp and
32599 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32600 int" and changed second argument from "unsigned int" to "int".
32601 Changed type returned by __builtin_scalar_test_neg,
32602 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32603 from "int" to "bool int". Added new overloaded function form
32604 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32605 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32606 exponent double-precision with floating point first argument.
32607 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32608 documentation of scalar_test_data_class, scalar_test_neg,
32609 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32610 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32611 vec_test_data_class built-in functions to reflect refinements in
32612 their type signatures.
32614 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32616 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32618 (aarch64_elf_asm_destructor): Likewise.
32620 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32622 PR rtl-optimization/78634
32623 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32624 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32625 * ifcvt.c (noce_try_cmove): Add missing cost check.
32627 PR rtl-optimization/71724
32628 * combine.c (if_then_else_cond): Look for situations where it is
32629 beneficial to undo the work of one of the recursive calls.
32631 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32633 PR tree-optimization/70754
32634 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32635 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32636 combined stmt before it if not NULL.
32637 (combine_chains): Process refs reversely and compute dominance point
32640 2017-01-23 Martin Liska <mliska@suse.cz>
32642 PR tree-optimization/79196
32643 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32644 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32646 (strlen_optimize_stmt): Call the renamed function.
32648 2017-01-23 Michael Matz <matz@suse.de>
32650 PR tree-optimization/78384
32651 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32653 2017-01-23 Richard Biener <rguenther@suse.de>
32655 PR tree-optimization/79186
32656 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32657 both incoming edges before moving an assert.
32659 2017-01-23 Martin Jambor <mjambor@suse.cz>
32661 * ipa-prop.c (load_from_param_1): Removed.
32662 (load_from_unmodified_param): Bits from load_from_param_1 put back
32664 (load_from_param): Removed.
32665 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32666 with stmt. Reverted back to use of load_from_unmodified_param.
32668 2017-01-23 Martin Jambor <mjambor@suse.cz>
32671 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32672 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32673 field a pointer to garbage collected vector, mark lattices and
32674 ipcp_orig_node with GTY((skip)).
32675 (ipa_get_param_count): Adjust to descriptors being a pointer.
32676 (ipa_get_param): Likewise.
32677 (ipa_get_type): Likewise.
32678 (ipa_get_param_move_cost): Likewise.
32679 (ipa_set_param_used): Likewise.
32680 (ipa_get_controlled_uses): Likewise.
32681 (ipa_set_controlled_uses): Likewise.
32682 (ipa_is_param_used): Likewise.
32683 (ipa_node_params_t): Move into garbage collector. New methods insert
32685 (ipa_node_params_sum): Annotate wth GTY(()).
32686 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32688 (ipa_load_from_parm_agg): Adjust declaration.
32689 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32690 * ipa-profile.c (ipa_profile): Likewise.
32691 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32692 (ipa_populate_param_decls): Make descriptors parameter garbage
32694 (ipa_dump_param): Adjust to descriptors being a pointer.
32695 (ipa_alloc_node_params): Likewise.
32696 (ipa_initialize_node_params): Likewise.
32697 (load_from_param_1): Make descriptors parameter garbage collected.
32698 (load_from_unmodified_param): Likewise.
32699 (load_from_param): Likewise.
32700 (ipa_load_from_parm_agg): Likewise.
32701 (ipa_node_params::~ipa_node_params): Removed.
32702 (ipa_free_all_node_params): Remove call to delete operator.
32703 (ipa_node_params_t::insert): New.
32704 (ipa_node_params_t::remove): Likewise.
32705 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32706 copy known_csts and known_contexts vectors.
32707 (ipa_read_node_info): Adjust to descriptors being a pointer.
32708 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32710 (ipcp_transform_function): Make descriptors variable garbage
32713 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32715 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32716 * config/i386/avx512dqintrin.h: Ditto.
32717 * config/i386/avx512fintrin.h: Ditto.
32718 * config/i386/i386.c: Handle new builtins.
32719 * config/i386/i386-builtin.def: Add new builtins.
32720 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32721 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32723 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32724 Martin Liska <mliska@suse.cz>
32726 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32727 * asan.c (asan_expand_poison_ifn): Support stores and use
32728 appropriate ASAN report function.
32729 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32730 * internal-fn.def (ASAN_POISON_USE): Declare.
32731 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32732 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32733 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32734 ASAN_POISON calls w/o LHS.
32735 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32736 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32737 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32738 * gimplify.c (asan_poison_variables): Add attribute
32739 use_after_scope_memory to variables that really needs to live
32741 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32742 having the attribute.
32744 2017-01-23 Martin Liska <mliska@suse.cz>
32746 * asan.c (create_asan_shadow_var): New function.
32747 (asan_expand_poison_ifn): Likewise.
32748 * asan.h (asan_expand_poison_ifn): New declaration.
32749 * internal-fn.c (expand_ASAN_POISON): Likewise.
32750 * internal-fn.def (ASAN_POISON): New builtin.
32751 * sanopt.c (pass_sanopt::execute): Expand
32752 asan_expand_poison_ifn.
32753 * tree-inline.c (copy_decl_for_dup_finish): Make function
32755 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32756 * tree-ssa.c (is_asan_mark_p): New function.
32757 (execute_update_addresses_taken): Rewrite local variables
32758 (identified just by use-after-scope as addressable) into SSA.
32760 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32762 * doc/install.texi (Specific): opensource.apple.com uses https
32763 now. Remove trailing slash.
32765 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32767 * README.Portability: Remove note on an Irix compatibility issue.
32769 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32771 * gcov.c (INCLUDE_ALGORITHM): Define.
32772 (INCLUDE_VECTOR): Define.
32773 No longer include <vector> and <algorithm> directly.
32775 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32777 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32779 * doc/invoke.texi (Code Gen Options): Ditto.
32781 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32784 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32786 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32788 rtl-optimization/79125
32789 * cprop.c (local_cprop_pass): Handle cases where we make an
32790 unconditional trap.
32792 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32796 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32797 read from, for big endian.
32799 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32801 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32802 register pauth builtins for LP64 only.
32804 2017-01-20 Marek Polacek <polacek@redhat.com>
32807 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32810 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32812 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32814 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32815 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32816 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32818 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32821 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32822 in big-endian mode when they are not a single duplicated value.
32824 2017-01-20 Richard Biener <rguenther@suse.de>
32826 * BASE-VER: Bump to 7.0.1.
32828 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32830 * omp-low.c (omplow_simd_context): New struct. Use it...
32831 (lower_rec_simd_input_clauses): ...here and...
32832 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32833 references to idx, lane, max_vf, is_simt.
32835 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32837 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32840 2017-01-20 Martin Jambor <mjambor@suse.cz>
32842 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32843 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32845 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32846 (GTFILES): Rename hsa.c to hsa-common.c.
32847 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32848 * hsa-dump.c: Likewise.
32849 * hsa-gen.c: Likewise.
32850 * hsa-regalloc.c: Likewise.
32851 * ipa-hsa.c: Likewise.
32852 * omp-expand.c: Likewise.
32853 * omp-low.c: Likewise.
32854 * toplev.c: Likewise.
32856 2017-01-20 Marek Polacek <polacek@redhat.com>
32859 * doc/invoke.texi: Document -Wduplicated-branches.
32860 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32861 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32862 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32863 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32864 return 0 only when not OEP_LEXICOGRAPHIC.
32865 (fold_build_cleanup_point_expr): Use the expression
32866 location when building CLEANUP_POINT_EXPR.
32867 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32868 * tree.c (add_expr): Handle error_mark_node.
32870 2017-01-20 Martin Liska <mliska@suse.cz>
32873 * tree-profile.c (init_ic_make_global_vars): Do not call
32875 (gimple_init_gcov_profiler): Likewise.
32877 2017-01-20 Martin Liska <mliska@suse.cz>
32880 * cgraph.h (maybe_create_reference): Remove argument and
32882 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32884 * ipa-cp.c (create_specialized_node): Likewise.
32885 * symtab.c (symtab_node::maybe_create_reference): Handle
32886 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32888 2017-01-20 Martin Liska <mliska@suse.cz>
32890 * read-rtl-function.c (function_reader::create_function): Use
32891 build_decl instread of build_decl_stat.
32893 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32895 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32896 * config/i386/avx512dqintrin.h: Ditto.
32897 * config/i386/avx512fintrin.h: Ditto.
32898 * config/i386/i386-builtin-types.def: Add new types.
32899 * config/i386/i386.c: Handle new types.
32900 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32901 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32902 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32903 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32904 (__builtin_ia32_kshiftridi): New.
32905 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32907 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32911 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32912 define to rs6000_init_stack_protect_guard.
32913 (rs6000_init_stack_protect_guard): New function.
32915 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32916 Yunqiang Su <yunqiang.su@imgtec.com>
32918 * config.gcc (supported_defaults): Add madd4.
32919 (with_madd4): Add validation.
32920 (all_defaults): Add madd4.
32921 * config/mips/mips.opt (mmadd4): New option.
32922 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32924 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32926 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32927 (ISA_HAS_FUSED_MADD4): Likewise.
32928 * doc/invoke.texi (-mmadd4): Document the new option.
32929 * doc/install.texi (--with-madd4): Document the new option.
32931 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32933 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32934 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32935 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32936 (aarch64_init_pauth_hint_builtins): New.
32937 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32938 (aarch64_expand_builtin): Expand new builtins.
32940 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32942 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32943 * combine-stack-adj.c (no_unhandled_cfa): Handle
32944 REG_CFA_TOGGLE_RA_MANGLE.
32945 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32946 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32947 info for return address signing.
32948 (aarch64_expand_epilogue): Likewise.
32950 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32952 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32953 * config/aarch64/aarch64-protos.h
32954 (aarch64_return_address_signing_enabled): New declaration.
32955 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32957 (aarch64_expand_prologue): Sign return address before it's pushed onto
32959 (aarch64_expand_epilogue): Authenticate return address fetched from
32961 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32962 (aarch64_attributes): New function attributes for "sign-return-address".
32963 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32964 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32965 ("*do_return"): Generate combined instructions according to key index.
32966 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32967 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32969 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32970 * config/aarch64/aarch64.opt (msign-return-address=): New.
32971 * doc/extend.texi (AArch64 Function Attributes): Documents
32972 "sign-return-address=".
32973 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32975 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32977 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32978 overall option summary.
32980 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32982 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32983 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32984 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32985 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32987 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32989 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32990 -mpower9-minmax by default for -mcpu=power9.
32991 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32992 128-bit floating point.
32994 2017-01-20 Alan Modra <amodra@gmail.com>
32996 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32997 optimizing for size.
32999 2017-01-20 Alan Modra <amodra@gmail.com>
33002 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33003 for strcmp and strncmp from corresponding builtin decl.
33005 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33007 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33008 instead of i386/rtems-64.h.
33009 * config/i386/rtems-64.h: Remove.
33011 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33015 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33017 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33019 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33021 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33022 Change int to HOST_WIDE_INT.
33023 * config/aarch64/aarch64-protos.h
33024 (aarch64_simd_gen_const_vector_dup): Likewise.
33025 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33027 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33029 * langhooks-def.h (lhd_type_for_size): New decl.
33030 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33031 * langhooks.c (lhd_type_for_size): New function, taken from
33034 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33036 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33037 define_bypass for CR latency.
33038 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33039 (power9-alu2): Add define_bypass for CR latency.
33041 (power9-mul): Update insn latency.
33042 (power9-mul-compare): Update insn latency, bypass latency and remove
33045 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33047 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33049 * config/aarch64/aarch64.md
33050 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33051 aarch64_nopcrelative_literal_loads.
33052 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33054 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33056 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33057 TARGET_LOONGSON_3A.
33058 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33060 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33063 * config.gcc (supported_defaults): Add lxc1-sxc1.
33064 (with_lxc1_sxc1): Add validation.
33065 (all_defaults): Add lxc1-sxc1.
33066 * config/mips/mips.opt (mlxc1-sxc1): New option.
33067 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33069 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33070 __mips_no_lxc1_sxc1.
33071 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33072 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33073 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33075 2017-01-19 Richard Biener <rguenther@suse.de>
33077 PR tree-optimization/72488
33078 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33079 sure to restore SSA info.
33080 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33082 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33084 PR rtl-optimization/79121
33085 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33086 of the inner type when shifting an extended value.
33088 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33091 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33092 interposable aliases.
33094 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33097 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33098 Use the evmergelohi instruction.
33099 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33100 (mov_sitf_e500_subreg8_2_be): Likewise.
33101 (mov_sitf_e500_subreg12_2_le): Likewise.
33102 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33103 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33104 (mov_sitf_e500_subreg8_2_le): Likewise.
33105 (mov_sitf_e500_subreg12_2_be): Likewise.
33107 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33109 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33110 attribute from vecsimple to vecperm.
33111 (altivec_vbpermq2): Likewise.
33113 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33116 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33118 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33119 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33120 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33121 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33122 case where N arg is SIZE_MAX.
33123 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33124 (cmpstrsi): Add pattern.
33126 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33129 __builtin_vec_revb builtins.
33130 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33131 built-in functions to support generation of the ISA 3.0 XXBR<x>
33132 vector byte reverse instructions.
33133 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33134 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33135 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33136 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33137 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33138 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33139 (P9V_BUILTIN_VEC_REVB): Likewise.
33140 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33141 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33142 (p9_xxbrq_v16qi): Likewise.
33143 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33144 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33145 (p9_xxbrh_v8hi): Likewise.
33146 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33147 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33148 vec_revb built-in functions.
33150 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33152 PR rtl-optimization/78952
33153 * config/i386/i386.md (any_extract): New code iterator.
33154 (*insvqi_2): Use any_extract for source operand.
33155 (*insvqi_3): Use any_shiftrt for source operand.
33157 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33159 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33161 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33163 2017-01-18 Matthias Klose <doko@ubuntu.com>
33165 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33167 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33169 * config/rs6000/altivec.h (vec_bperm): Change #define.
33170 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33171 (altivec_vbpermq2): New define_insn.
33172 (altivec_vbpermd): Likewise.
33173 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33174 function interface.
33175 (VBPERMD): Likewise.
33176 (VBPERM): New polymorphic function interface.
33177 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33178 Add entries for P9V_BUILTIN_VEC_VBPERM.
33179 * doc/extend.texi: Add interfaces for vec_bperm.
33181 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33183 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33184 first letter of error messages.
33185 (s390_resolve_overloaded_builtin): Likewise.
33186 * config/s390/s390.c (s390_expand_builtin): Likewise.
33187 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33188 (s390_valid_target_attribute_inner_p): Likewise.
33189 * config/s390/s390.md ("tabort"): Likewise.
33191 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33193 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33194 (ISA_AVOID_DIV_HILO): New macro.
33195 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33196 (ISA_HAS_DDIV): Likewise.
33198 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33200 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33202 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33204 * doc/invoke.texi (fabi-version): Spelling fix.
33206 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33209 * doc/invoke.texi (fabi-version): Mention mangling fix for
33212 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33215 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33217 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33220 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33221 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33223 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33225 (stack_protect_test): Ditto.
33226 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33227 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33229 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33230 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33231 -mstack-protector-guard-offset=.
33232 (RS/6000 and PowerPC Options): Ditto.
33234 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33236 * config/i386/i386.h (MASK_CLASS_P): New define.
33237 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33238 there are no registers from different register sets also when
33239 mask registers are used. Update function comment.
33240 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33241 to (*k/*r) and (*k/*km) alternatives.
33243 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33245 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33246 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33247 (EH_RETURN_HANDLER_RTX): New define.
33248 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33249 Force frame pointer in EH return functions.
33250 (aarch64_expand_epilogue): Add barrier for eh_return.
33251 (aarch64_final_eh_return_addr): Remove.
33252 (aarch64_eh_return_handler_rtx): New function.
33253 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33255 (aarch64_eh_return_handler_rtx): New prototype.
33257 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33259 * config/rs6000/altivec.h (vec_rlmi): New #define.
33260 (vec_vrlnm): Likewise.
33261 (vec_rlnm): Likewise.
33262 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33263 (UNSPEC_VRLNM): Likewise.
33264 (VIlong): New mode iterator.
33265 (altivec_vrl<VI_char>mi): New define_insn.
33266 (altivec_vrl<VI_char>nm): Likewise.
33267 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33269 (VRLDNM): Likewise.
33270 (RLNM): New polymorphic function entry.
33271 (VRLWMI): New monomorphic function entry.
33272 (VRLDMI): Likewise.
33273 (RLMI): New polymorphic function entry.
33274 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33275 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33276 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33279 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33282 * dwarf2out.c (field_byte_offset): Restore the
33283 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33284 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33285 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33288 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33291 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33293 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33296 * dwarf2out.c (add_data_member_location_attribute): For constant
33297 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33298 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33299 DW_AT_byte_size attributes.
33301 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33303 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33304 after forcing to constant memory when the code model is medium.
33306 2017-01-17 Julia Koval <julia.koval@intel.com>
33309 * config/i386/avx512fintrin.h
33310 (_mm512_i32gather_ps): Change __addr type to void const*.
33311 (_mm512_mask_i32gather_ps): Ditto.
33312 (_mm512_i32gather_pd): Ditto.
33313 (_mm512_mask_i32gather_pd): Ditto.
33314 (_mm512_i64gather_ps): Ditto.
33315 (_mm512_mask_i64gather_ps): Ditto.
33316 (_mm512_i64gather_pd): Ditto.
33317 (_mm512_mask_i64gather_pd): Ditto.
33318 (_mm512_i32gather_epi32): Ditto.
33319 (_mm512_mask_i32gather_epi32): Ditto.
33320 (_mm512_i32gather_epi64): Ditto.
33321 (_mm512_mask_i32gather_epi64): Ditto.
33322 (_mm512_i64gather_epi32): Ditto.
33323 (_mm512_mask_i64gather_epi32): Ditto.
33324 (_mm512_i64gather_epi64): Ditto.
33325 (_mm512_mask_i64gather_epi64): Ditto.
33326 (_mm512_i32scatter_ps): Change __addr type to void*.
33327 (_mm512_mask_i32scatter_ps): Ditto.
33328 (_mm512_i32scatter_pd): Ditto.
33329 (_mm512_mask_i32scatter_pd): Ditto.
33330 (_mm512_i64scatter_ps): Ditto.
33331 (_mm512_mask_i64scatter_ps): Ditto.
33332 (_mm512_i64scatter_pd): Ditto.
33333 (_mm512_mask_i64scatter_pd): Ditto.
33334 (_mm512_i32scatter_epi32): Ditto.
33335 (_mm512_mask_i32scatter_epi32): Ditto.
33336 (_mm512_i32scatter_epi64): Ditto.
33337 (_mm512_mask_i32scatter_epi64): Ditto.
33338 (_mm512_i64scatter_epi32): Ditto.
33339 (_mm512_mask_i64scatter_epi32): Ditto.
33340 (_mm512_i64scatter_epi64): Ditto.
33341 (_mm512_mask_i64scatter_epi64): Ditto.
33342 * config/i386/avx512pfintrin.h
33343 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33344 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33345 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33346 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33347 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33348 (_mm512_prefetch_i32scatter_ps): Ditto.
33349 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33350 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33351 (_mm512_prefetch_i64scatter_pd): Ditto.
33352 (_mm512_prefetch_i64scatter_ps): Ditto.
33353 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33354 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33355 * config/i386/avx512vlintrin.h
33356 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33357 (_mm_mmask_i32gather_ps): Ditto.
33358 (_mm256_mmask_i32gather_pd): Ditto.
33359 (_mm_mmask_i32gather_pd): Ditto.
33360 (_mm256_mmask_i64gather_ps): Ditto.
33361 (_mm_mmask_i64gather_ps): Ditto.
33362 (_mm256_mmask_i64gather_pd): Ditto.
33363 (_mm_mmask_i64gather_pd): Ditto.
33364 (_mm256_mmask_i32gather_epi32): Ditto.
33365 (_mm_mmask_i32gather_epi32): Ditto.
33366 (_mm256_mmask_i32gather_epi64): Ditto.
33367 (_mm_mmask_i32gather_epi64): Ditto.
33368 (_mm256_mmask_i64gather_epi32): Ditto.
33369 (_mm_mmask_i64gather_epi32): Ditto.
33370 (_mm256_mmask_i64gather_epi64): Ditto.
33371 (_mm_mmask_i64gather_epi64): Ditto.
33372 (_mm256_i32scatter_ps): Change __addr type to void*.
33373 (_mm256_mask_i32scatter_ps): Ditto.
33374 (_mm_i32scatter_ps): Ditto.
33375 (_mm_mask_i32scatter_ps): Ditto.
33376 (_mm256_i32scatter_pd): Ditto.
33377 (_mm256_mask_i32scatter_pd): Ditto.
33378 (_mm_i32scatter_pd): Ditto.
33379 (_mm_mask_i32scatter_pd): Ditto.
33380 (_mm256_i64scatter_ps): Ditto.
33381 (_mm256_mask_i64scatter_ps): Ditto.
33382 (_mm_i64scatter_ps): Ditto.
33383 (_mm_mask_i64scatter_ps): Ditto.
33384 (_mm256_i64scatter_pd): Ditto.
33385 (_mm256_mask_i64scatter_pd): Ditto.
33386 (_mm_i64scatter_pd): Ditto.
33387 (_mm_mask_i64scatter_pd): Ditto.
33388 (_mm256_i32scatter_epi32): Ditto.
33389 (_mm256_mask_i32scatter_epi32): Ditto.
33390 (_mm_i32scatter_epi32): Ditto.
33391 (_mm_mask_i32scatter_epi32): Ditto.
33392 (_mm256_i32scatter_epi64): Ditto.
33393 (_mm256_mask_i32scatter_epi64): Ditto.
33394 (_mm_i32scatter_epi64): Ditto.
33395 (_mm_mask_i32scatter_epi64): Ditto.
33396 (_mm256_i64scatter_epi32): Ditto.
33397 (_mm256_mask_i64scatter_epi32): Ditto.
33398 (_mm_i64scatter_epi32): Ditto.
33399 (_mm_mask_i64scatter_epi32): Ditto.
33400 (_mm256_i64scatter_epi64): Ditto.
33401 (_mm256_mask_i64scatter_epi64): Ditto.
33402 (_mm_i64scatter_epi64): Ditto.
33403 (_mm_mask_i64scatter_epi64): Ditto.
33404 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33405 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33406 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33407 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33408 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33409 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33410 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33411 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33412 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33413 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33414 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33415 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33416 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33417 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33418 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33419 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33420 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33421 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33422 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33423 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33424 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33425 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33426 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33427 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33428 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33429 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33430 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33431 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33432 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33433 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33434 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33435 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33436 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33437 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33438 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33439 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33440 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33441 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33442 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33443 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33444 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33445 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33446 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33447 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33448 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33449 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33450 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33451 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33452 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33453 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33454 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33455 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33456 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33457 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33458 definitions accordingly.
33460 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33461 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33464 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33467 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33470 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33471 hard regs for inner regno.
33473 2017-01-17 Martin Liska <mliska@suse.cz>
33476 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33477 assumption and add comment.
33479 2017-01-17 Nathan Sidwell <nathan@acm.org>
33481 * ipa-visibility.c (localize_node): New function, broken out of ...
33482 (function_and_variable_visibility): ... here. Call it.
33484 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33486 PR middle-end/77445
33487 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33488 correctly set frequency of oudgoing edge.
33489 (duplicate_thread_path): Fix profile updating.
33491 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33494 * configure.ac: Add GCC_BASE_VER.
33495 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33496 version from BASE-VER file.
33497 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33498 (gcc.o): Depend on $(BASEVER).
33499 * common.opt (dumpfullversion): New option.
33500 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33501 * doc/invoke.texi: Document -dumpfullversion.
33502 * doc/install.texi: Document --with-gcc-major-version-only.
33503 * configure: Regenerated.
33505 2017-01-17 Richard Biener <rguenther@suse.de>
33507 PR tree-optimization/71433
33508 * tree-vrp.c (register_new_assert_for): Merge same asserts
33509 on all incoming edges.
33510 (process_assert_insertions_for): Handle insertions at the
33513 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33515 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33516 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33518 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33521 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33524 2017-01-17 Alan Modra <amodra@gmail.com>
33527 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33528 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33529 symbolic stack limit when pic.
33531 2017-01-16 Martin Sebor <msebor@redhat.com>
33533 PR tree-optimization/78608
33534 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33536 2017-01-16 Jeff Law <law@redhat.com>
33539 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33540 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33541 for several include directories that may be relative to sysroot.
33542 * config/i386/x-mingw32 (gplus_includedir): Define.
33543 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33544 (native_system_includedir): Likewise.
33545 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33546 override if TARGET_SYSTEM_ROOT is defined.
33547 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33549 PR tree-optimization/79090
33550 PR tree-optimization/33562
33551 PR tree-optimization/61912
33552 PR tree-optimization/77485
33553 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33554 and computed trims into the dump file.
33556 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33558 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33560 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33563 * gimplify.c (gimplify_init_constructor): If want_value and
33564 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33568 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33569 sequence. Formatting fixes.
33570 (doloop_optimize): Formatting fixes.
33573 * gcc.c (debug_level_greater_than_spec_func): New function.
33574 (static_spec_functions): Add debug-level-gt spec function.
33575 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33577 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33578 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33579 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33580 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33581 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33582 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33584 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33586 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33587 QImode fixups to general and mask registers only.
33589 2017-01-16 Carl Love <cel@us.ibm.com>
33591 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33592 for built-in functions
33593 vector signed char vec_nabs (vector signed char)
33594 vector signed short vec_nabs (vector signed short)
33595 vector signed int vec_nabs (vector signed int)
33596 vector signed long long vec_nabs (vector signed long long)
33597 vector float vec_nabs (vector float)
33598 vector double vec_nabs (vector double)
33599 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33601 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33602 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33603 * doc/extend.texi: Update the documentation file for the new built-in
33606 2017-01-16 Martin Sebor <msebor@redhat.com>
33608 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33611 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33613 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33614 UNSPEC_VSX__XXSPLTD to require special splat handling.
33616 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33619 * system.h: Poison strndup.
33621 2017-01-16 Alan Modra <amodra@gmail.com>
33624 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33627 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33629 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33631 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33633 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33634 call recog here. Assert that INSN_CODE (insn) is non-negative.
33636 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33639 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33641 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33642 in the currently scheduled RTL fragment.
33644 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33646 PR rtl-optimization/78751
33647 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33650 2017-01-14 Jeff Law <law@redhat.com>
33652 PR tree-optimization/79090
33653 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33654 variable length stores.
33655 (compute_trims): Delete dead assignment to *trim_tail.
33656 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33659 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33661 PR rtl-optimization/78626
33662 PR rtl-optimization/78727
33663 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33664 of a block, and split such blocks after everything else is finished.
33666 2017-01-14 Alan Modra <amodra@gmail.com>
33669 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33670 target legitimate_combined_insn.
33671 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33672 (rs6000_legitimate_combined_insn): New function.
33673 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33675 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33676 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33677 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33679 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33681 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33683 2017-01-13 Jeff Law <law@redhat.com>
33685 PR tree-optimization/33562
33686 PR tree-optimization/61912
33687 PR tree-optimization/77485
33688 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33690 (delete_dead_assignment): Likewise.
33691 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33692 statement to delete_dead_call and delete_dead_assignment.
33694 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33697 * substring-locations.c (format_warning_va): Strengthen case 1 so
33698 that both endpoints of the substring must be within the format
33699 range for just the substring to be printed.
33701 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33703 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33704 * config/i386/i386.c (ix86_target_string): Add missing options
33705 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33706 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33707 flags_other and ix86_target_other to flags2_other. Display unknown
33709 (ix86_valid_target_attribute_inner_p): Add missing options and
33710 reorder options by implied ISAs, as in ix86_target_string.
33712 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33714 * hash-table.h (hash_table::too_empty_p): New function.
33715 (hash_table::expand): Use it.
33716 (hash_table::traverse): Likewise.
33717 (hash_table::empty_slot): Use sizeof (value_type) instead of
33718 sizeof (PTR) to convert bytes to elements. Shrink the table
33719 if the current size is excessive for the current number of
33722 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33724 * ira-costs.c (record_reg_classes): Break from the inner loop
33725 early once alt_fail is known to be true. Update outer loop
33726 handling accordingly.
33728 2017-01-13 Jeff Law <law@redhat.com>
33730 * tree-ssa-dse.c (decrement_count): New function.
33731 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33732 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33733 when we know the partially dead statement is a mem* function.
33735 PR tree-optimization/61912
33736 PR tree-optimization/77485
33737 * tree-ssa-dse.c: Include expr.h.
33738 (maybe_trim_constructor_store): New function.
33739 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33741 PR tree-optimization/33562
33742 PR tree-optimization/61912
33743 PR tree-optimization/77485
33744 * doc/invoke.texi: Document new dse-max-object-size param.
33745 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33746 * tree-ssa-dse.c: Include params.h.
33747 (dse_store_status): New enum.
33748 (initialize_ao_ref_for_dse): New, partially extracted from
33750 (valid_ao_ref_for_dse, normalize_ref): New.
33751 (setup_live_bytes_from_ref, compute_trims): Likewise.
33752 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33753 (maybe_trim_partially_dead_store): Likewise.
33754 (maybe_trim_complex_store): Likewise.
33755 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33756 Track what bytes live from the original store. Return tri-state
33757 for dead, partially dead or live.
33758 (dse_dom_walker): Add constructor, destructor and new private members.
33759 (delete_dead_call, delete_dead_assignment): New extracted from
33761 (dse_optimize_stmt): Make a member of dse_dom_walker.
33762 Use initialize_ao_ref_for_dse.
33764 PR tree-optimization/33562
33765 PR tree-optimization/61912
33766 PR tree-optimization/77485
33767 * sbitmap.h (bitmap_count_bits): Prototype.
33768 (bitmap_clear_range, bitmap_set_range): Likewise.
33769 * sbitmap.c (bitmap_clear_range): New function.
33770 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33772 2017-01-13 Martin Liska <mliska@suse.cz>
33775 * function.c (set_cfun): Add new argument force.
33776 * function.h (set_cfun): Likewise.
33777 * ipa-inline-transform.c (inline_call): Use the function when
33778 strict alising from is dropped for function we inline to.
33780 2017-01-13 Richard Biener <rguenther@suse.de>
33782 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33783 for dumping GIMPLE INTEGER_CSTs.
33785 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33787 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33788 to 201112L since C++17.
33790 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33793 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33794 if -fsanitize=kernel-address is present.
33796 2017-01-13 Richard Biener <rguenther@suse.de>
33798 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33799 as _Literal ( type ) number in case usual suffixes do not
33800 preserve all information.
33802 2017-01-13 Richard Biener <rguenther@suse.de>
33804 PR tree-optimization/77283
33805 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33806 and ssa-iterators.h.
33807 (is_feasible_trace): Implement a cost model based on joiner
33810 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33813 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33814 char or short to __float128/_Float128 directly.
33816 2017-01-12 Martin Sebor <msebor@redhat.com>
33818 to -Wformat-overflow.
33819 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33820 (min_bytes_remaining): Same.
33821 (get_string_length): Same.
33822 (format_string): Same.
33823 (format_directive): Same.
33825 (pass_sprintf_length::handle_gimple_call): Same.
33827 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33829 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33830 info.nowrite calls with no lhs that can't throw. Return bool
33831 whether gsi_remove has been called or not.
33832 (pass_sprintf_length::handle_gimple_call): Return bool whether
33833 try_substitute_return_value called gsi_remove. Formatting fix.
33834 (pass_sprintf_length::execute): Don't use gsi_remove if
33835 handle_gimple_call returned true.
33838 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33839 be removed due to side-effects, don't remove following barrier nor
33840 turn the successor edge into fallthru edge.
33842 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33845 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33846 element-reversing loads and stores as not swappable.
33848 2017-01-12 Nathan Sidwell <nathan@acm.org>
33849 Nicolai Stange <nicstange@gmail.com>
33851 * combine.c (try_combine): Don't ignore result of overlap checking
33852 loop. Combine overlap & asm check into single loop.
33854 2017-01-12 Richard Biener <rguenther@suse.de>
33856 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33857 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33859 2017-01-12 Richard Biener <rguenther@suse.de>
33861 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33862 and TS_TARGET_OPTION directly derive from TS_BASE.
33863 * tree-core.h (tree_optimization_option): Derive from tree_base.
33864 (tree_target_option): Likewise.
33866 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33868 * config/i386/i386.c (memory_address_length): Increase len
33869 only when rip_relative_addr_p returns false.
33871 2017-01-11 Julia Koval <julia.koval@intel.com>
33873 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33874 (OPTION_MASK_ISA_SGX_SET): New.
33875 (ix86_handle_option): Handle OPT_msgx.
33876 * config.gcc: Added sgxintrin.h.
33877 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33878 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33879 * config/i386/i386.c (ix86_target_string): Add -msgx.
33881 (ix86_option_override_internal): Handle new options.
33882 (ix86_valid_target_attribute_inner_p): Add sgx.
33883 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33884 * config/i386/i386.opt: Add msgx.
33885 * config/i386/sgxintrin.h: New file.
33886 * config/i386/x86intrin.h: Add sgxintrin.h.
33888 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33891 * fold-const.c (maybe_nonzero_address): Return 1 for function
33893 (tree_single_nonzero_warnv_p): Don't handle function local objects
33897 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33900 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33903 * opts.c: Include "spellcheck.h"
33904 (struct string_fragment): New struct.
33905 (struct edit_distance_traits<const string_fragment &>): New
33907 (get_closest_sanitizer_option): New function.
33908 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33910 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33912 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33914 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33915 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33916 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33917 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33918 for initial die_offset if dwarf_split_debug_info.
33919 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33920 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33922 (output_skeleton_debug_sections): Formatting fix. Use
33923 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33924 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33926 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33928 * config/arm/cortex-a53.md: Add bypasses for
33929 cortex_a53_r2f_cvt.
33930 (cortex_a53_r2f): Only use for transfers.
33931 (cortex_a53_f2r): Likewise.
33932 (cortex_a53_r2f_cvt): Add reservation for conversions.
33933 (cortex_a53_f2r_cvt): Likewise.
33935 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33937 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33938 to all inlined functions, change static to extern.
33940 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33943 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33945 (arm_assemble_integer): Likewise.
33947 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33949 * config.gcc: Use new awk script to check CPU, FPU and architecture
33950 parameters for --with-... options.
33951 * config/arm/parsecpu.awk: New file
33952 * config/arm/arm-cpus.in: New file.
33953 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33955 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33957 * config/arm/t-arm: Update dependency rules.
33958 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33959 of processing .def files.
33960 * config/arm/genopt.sh: Deleted.
33961 * config/arm/gentune.sh: Deleted.
33962 * config/arm/arm-cores.def: Deleted.
33963 * config/arm/arm-arches.def: Deleted.
33964 * config/arm/arm-fpus.def: Deleted.
33965 * config/arm/arm-tune.md: Regenerated.
33966 * config/arm/arm-tables.opt: Regenerated.
33967 * config/arm/arm-cpu.h: New generated file.
33968 * config/arm/arm-cpu-data.h: New generated file.
33969 * config/arm/arm-cpu-cdata.h: New generated file.
33971 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33974 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33976 (input_varpool_node): Unpack dynamically_initialized bit.
33978 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33980 PR rtl-optimization/79032
33981 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33982 the alignment of the adjusted memory reference against that of MODE,
33983 instead of the alignment of the original memory reference.
33985 2017-01-11 Martin Jambor <mjambor@suse.cz>
33987 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33989 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33990 decorated functions.
33992 2017-01-11 Richard Biener <rguenther@suse.de>
33994 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33995 set range/nonnull info for PHI results. Do not set it on
33996 stmts marked for removal.
33998 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34000 * expr.c (store_field): In the bitfield case, fetch the return value
34001 from the registers before applying a single big-endian adjustment.
34002 Always do a final load for a BLKmode value not larger than a word.
34004 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34007 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34008 that we correctly handle column numbers greater than
34009 LINE_MAP_MAX_COLUMN_NUMBER.
34011 2017-01-10 Martin Sebor <msebor@redhat.com>
34013 PR middle-end/78245
34014 * gimple-ssa-sprintf.c (get_destination_size): Call
34015 {init,fini}object_sizes.
34016 * tree-object-size.c (addr_object_size): Adjust.
34017 (pass_through_call): Adjust.
34018 (pass_object_sizes::execute): Adjust.
34019 * tree-object-size.h (fini_object_sizes): Declare.
34021 2017-01-10 Martin Sebor <msebor@redhat.com>
34023 PR tree-optimization/78775
34024 * builtins.c (get_size_range): Move...
34025 * calls.c: ...to here.
34026 (alloc_max_size): Accept zero argument.
34027 (operand_signed_p): Remove.
34028 (maybe_warn_alloc_args_overflow): Call get_size_range.
34029 * calls.h (get_size_range): Declare.
34031 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34033 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34034 from TI's devices.csv file as of September 2016.
34035 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34037 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34039 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34040 * doc/invoke.texi: Likewise.
34041 * doc/md.texi: Likewise.
34042 * doc/objc.texi: Likewise.
34044 2017-01-10 Joshua Conner <joshconner@google.com>
34046 * config/arm/fuchsia-elf.h: New file.
34047 * config/fuchsia.h: New file.
34048 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34049 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34051 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34053 2016-01-10 Richard Biener <rguenther@suse.de>
34055 PR tree-optimization/79034
34056 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34057 Propagate out degenerate PHIs in the joiner.
34059 2017-01-10 Martin Liska <mliska@suse.cz>
34061 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34062 (sort_congruence_classes_by_decl_uid): Likewise.
34063 (sort_congruence_class_groups_by_decl_uid): Likewise.
34064 (sem_item_optimizer::merge_classes): Sort class, groups in these
34065 classes and members in the groups by DECL_UID of declarations.
34066 This would make merge operations stable.
34068 2017-01-10 Martin Liska <mliska@suse.cz>
34070 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34071 usage of m_classes_vec.
34072 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34073 (sem_item_optimizer::get_group_by_hash): Likewise.
34074 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34075 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34076 (sem_item_optimizer::verify_classes): Likewise.
34077 (sem_item_optimizer::process_cong_reduction): Likewise.
34078 (sem_item_optimizer::dump_cong_classes): Likewise.
34079 (sem_item_optimizer::merge_classes): Likewise.
34080 * ipa-icf.h (congruence_class_hash): Rename from
34081 congruence_class_group_hash. Remove declaration of m_classes_vec.
34083 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34085 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34086 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34087 * config.gcc: Add avx512vpopcntdqintrin.h.
34088 * config/i386/avx512vpopcntdqintrin.h: New.
34089 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34090 * config/i386/i386-builtin-types.def: Add new types.
34091 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34092 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34093 __builtin_ia32_vpopcountq_v8di_mask): New.
34094 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34095 __AVX512VPOPCNTDQ__.
34096 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34097 (PTA_AVX512VPOPCNTDQ): Define.
34098 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34099 TARGET_AVX512VPOPCNTDQ_P): Define.
34100 * config/i386/i386.opt: Add mavx512vpopcntdq.
34101 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34102 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34104 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34106 PR middle-end/77484
34107 * predict.def (PRED_CALL): Set to 67.
34109 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34111 * expr.c (store_field): In the bitfield case, if the value comes from
34112 a function call and is of an aggregate type returned in registers, do
34113 not modify the field mode; extract the value in all cases if the mode
34114 is BLKmode and the size is not larger than a word.
34116 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34119 * config/i386/cpuid.h: Fix undefined behavior.
34121 2017-01-04 Jeff Law <law@redhat.com>
34123 PR tree-optimization/79007
34124 PR tree-optimization/67955
34125 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34126 conservative for pt.null when flag_non_call_exceptions is on.
34128 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34130 PR translation/79019
34131 PR translation/79020
34132 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34133 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34134 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34136 * config/avr/avr.opt (maccumulate-args): Likewise.
34137 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34138 * common.opt (freport-bug): Likewise.
34139 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34140 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34141 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34142 translatable string.
34143 * config/i386/i386.c (function_value_32): Likewise.
34144 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34145 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34147 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34148 * common/config/msp430/msp430-common.c (msp430_handle_option):
34150 * symtab.c (symtab_node::verify_base): Likewise.
34151 * opts.c (set_debug_level): Likewise.
34152 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34153 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34154 missing whitespace to translatable strings.
34155 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34156 * config/sh/superh.h: Likewise.
34157 * config/i386/xopintrin.h: Likewise.
34158 * config/i386/znver1.md: Likewise.
34159 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34160 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34161 * double-int.h (struct double_int): Likewise.
34162 * double-int.c (div_and_round_double): Likewise.
34163 * wide-int.cc: Likewise.
34164 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34165 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34166 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34167 (crossjumps_occurred): ... this.
34168 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34171 PR tree-optimization/78899
34172 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34173 returning bool return struct loop *, NULL for failure and the new
34175 (versionable_outer_loop_p): Don't version outer loop if it has
34176 dont_vectorized bit set.
34177 (tree_if_conversion): When versioning outer loop, ensure
34178 tree_if_conversion is performed also on the inner loop of the
34179 non-vectorizable outer loop copy.
34180 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34181 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34182 prevent vectorization of it.
34183 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34184 the outer loop vectorization of the non-scalar version is attempted
34185 before vectorization of the inner loop in scalar version. If
34186 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34187 vectorization of its inner loop.
34188 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34189 has 2 inner loops, rename also on edges from bb whose single pred
34190 is outer_loop->header. Fix typo in function comment.
34192 2017-01-09 Martin Sebor <msebor@redhat.com>
34195 * asan.c (asan_emit_stack_protection): Increase local buffer size
34196 to avoid snprintf truncation warning.
34198 2017-01-09 Andrew Pinski <apinski@cavium.com>
34200 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34201 to reference thunderx2t99 for the tuning structure
34202 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34204 (thunderx2t99_extra_costs): This.
34205 * config/aarch64/aarch64-tune.md: Regenerate.
34206 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34207 (vulcan_addrcost_table): This.
34208 (vulcan_regmove_cost): Rename to ...
34209 (thunderx2t99_regmove_cost): This.
34210 (vulcan_vector_cost): Rename to ...
34211 (thunderx2t99_vector_cost): this.
34212 (vulcan_branch_cost): Rename to ...
34213 (thunderx2t99_branch_cost): This.
34214 (vulcan_tunings): Rename to ...
34215 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34216 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34218 2017-01-09 Martin Jambor <mjambor@suse.cz>
34222 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34223 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34224 (propagate_vr_accross_jump_function): Use the above function for all
34225 value range computations for pass-through jump functions and type
34226 converasion from explicit value range values.
34227 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34228 parameters from TYPE_ARG_TYPES.
34229 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34230 (ipa_write_node_info): Stream type of the actual argument.
34231 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34233 2017-01-09 Martin Liska <mliska@suse.cz>
34236 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34237 (lookup_compiler): Do not show error message with have_E.
34239 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34241 PR tree-optimization/78938
34242 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34243 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34244 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34245 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34248 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34250 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34253 2017-01-09 Richard Biener <rguenther@suse.de>
34255 PR tree-optimization/78997
34256 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34257 name condition properly.
34259 2017-01-09 Richard Biener <rguenther@suse.de>
34262 * dwarf2out.c (is_cxx): New overload with context.
34263 (is_naming_typedef_decl): Use it.
34265 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34267 * invoke.texi (Option Summary): Correct spacing in option lists
34268 and add line breaks to fix over-long lines.
34270 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34272 PR middle-end/17660
34274 * extend.texi (Common Variable Attributes): Add xref to GCC
34275 Internals manual to explain mode attribute keywords.
34277 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34280 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34281 and Preprocessor Options.
34282 (Options for Linking): Document -pthread here....
34283 (RS/6000 and PowerPC Options): ...not here.
34284 (Solaris 2 Options): ...or here.
34285 * doc/cppopts.texi: Document -pthread.
34287 2017-01-08 Martin Sebor <msebor@redhat.com>
34289 PR middle-end/77708
34290 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34291 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34292 New member functions.
34293 (format_directive): Used them.
34295 (pass_sprintf_length::handle_gimple_call): Same.
34296 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34297 to avoid truncation for any argument.
34298 (extract_affine_mul): Same.
34299 * tree.c (get_file_function_name): Same.
34301 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34303 PR middle-end/77484
34304 * predict.def (PRED_INDIR_CALL): Set to 86.
34306 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34308 PR preprocessor/54124
34309 * doc/cppopts.texi: Reformat -d subtable to list the full name
34310 of the options. Add cross-reference to the docs for the general
34311 compiler -d options.
34312 * doc/invoke.texi (Developer Options): Add cross-reference to the
34313 preprocessor-specific -d option documentation.
34315 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34317 PR preprocessor/13498
34318 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34319 redudant material, and reflect new command-line options.
34320 (System Headers): Likewise.
34322 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34324 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34325 -isystem, and -idirafter. Copy-edit.
34326 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34327 default for -ftrack-macro-expansion. Delete obsolete and
34328 badly-formatted implementation details about -fdebug-cpp output.
34329 * doc/cppwarnopts.texi: Copy-edit.
34331 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34334 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34335 that the transition from a max line width >= 1<<10 to narrower
34336 lines works correctly.
34338 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34340 * doc/options.texi (PerFunction): New.
34341 * opt-functions.awk (switch_flags): Map both Optimization and
34342 PerFunction to CL_OPTIMIZATION.
34343 * opth-gen.awk: Test for PerFunction flag along with
34345 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34346 it only when the latter is present. Skip those that don't in
34347 the hash function generator.
34348 * common.opt (fvar-tracking): Mark as PerFunction instead of
34350 (fvar-tracking-assignments): Likewise.
34351 (fvar-tracking-assignments-toggle): Likewise.
34352 (fvar-tracking-uninit): Likewise.
34354 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34356 PR translation/79018
34357 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34360 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34363 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34364 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34365 TARGET_LONG_JUMP_TABLE_OFFSETS.
34366 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34367 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34368 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34369 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34370 * config/m68k/m68k.md (tablejump expander): Likewise.
34371 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34372 TARGET_LONG_JUMP_TABLE_OFFSETS.
34373 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34374 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34376 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34377 David Holsgrove <david.holsgrove@xilinx.com>
34379 * common/config/microblaze/microblaze-common.c
34380 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34381 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34383 * config/microblaze/microblaze.c (microblaze_must_save_register)
34384 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34386 (microblaze_eh_return): New function.
34387 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34388 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34389 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34390 * config/microblaze/microblaze.md (eh_return): New pattern.
34392 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34394 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34395 GCC_DIAGNOSTIC_STRINGIFY): Define.
34397 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34401 * config/arm/arm.md (<mcrr>): New.
34403 * config/arm/arm.c (arm_arch5te): New.
34404 (arm_option_override): Set arm_arch5te.
34405 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34407 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34408 (arm_mcrr_qualifiers): ... this. New.
34409 (MRRC_QUALIFIERS): Define to...
34410 (arm_mrrc_qualifiers): ... this. New.
34411 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34413 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34414 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34415 (MRRCI, mrrc, MRRC): New.
34416 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34417 VUNSPEC_MRRC2): New.
34419 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34421 * config/arm/arm.md (<mcr>): New.
34423 * config/arm/arm.c (arm_coproc_builtin_available): Add
34424 support for mcr, mrc, mcr2 and mrc2.
34425 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34426 (arm_mcr_qualifiers): ... this. New.
34427 (MRC_QUALIFIERS): Define to ...
34428 (arm_mrc_qualifiers): ... this. New.
34429 (MCR_QUALIFIERS): Define to ...
34430 (arm_mcr_qualifiers): ... this. New.
34431 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34433 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34434 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34435 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34436 VUNSPEC_MRC2): New.
34438 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34440 * config/arm/arm.md (*ldc): New.
34444 * config/arm/arm.c (arm_coproc_builtin_available): Add
34445 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34446 (arm_coproc_ldc_stc_legitimate_address): New.
34447 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34448 'qualifier_const_pointer'.
34449 (LDC_QUALIFIERS): Define to...
34450 (arm_ldc_qualifiers): ... this. New.
34451 (STC_QUALIFIERS): Define to...
34452 (arm_stc_qualifiers): ... this. New.
34453 * config/arm/arm-protos.h
34454 (arm_coproc_ldc_stc_legitimate_address): New.
34455 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34456 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34457 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34458 stc2, stcl, stc2l): New.
34459 * config/arm/constraints.md (Uz): New.
34460 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34461 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34462 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34463 VUNSPEC_STC2L): New.
34465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34467 * config/arm/arm.md (<cdp>): New.
34468 * config/arm/arm.c (neon_const_bounds): Rename this ...
34469 (arm_const_bounds): ... this.
34470 (arm_coproc_builtin_available): New.
34471 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34472 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34473 (CDP_QUALIFIERS): Define to...
34474 (arm_cdp_qualifiers): ... this. New.
34476 (arm_expand_builtin_args): Add case for 6 arguments.
34477 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34478 (arm_const_bounds): ... this.
34479 (arm_coproc_builtin_available): New.
34480 * config/arm/arm_acle.h (__arm_cdp): New.
34482 * config/arm/arm_acle_builtins.def (cdp): New.
34484 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34485 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34486 'arm_const_bounds'.
34487 * config/arm/types.md (coproc): New.
34488 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34489 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34490 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34491 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34493 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34495 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34496 (UBINOP_QUALIFIERS): New.
34498 (acle_builtin_data): New. Change comment.
34499 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34500 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34501 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34502 arm_acle_builtins.def.
34503 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34504 (arm_init_acle_builtins): New.
34505 (CRC32_BUILTIN): Remove.
34506 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34507 crc32cb, crc32ch and crc32cw.
34508 (arm_init_crc32_builtins): Remove.
34509 (arm_init_builtins): Use arm_init_acle_builtins rather
34510 than arm_init_crc32_builtins.
34511 (arm_expand_acle_builtin): New.
34512 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34513 * config/arm/arm_acle_builtins.def: New.
34515 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34517 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34518 (arm_builtin_datum): ... this.
34519 (arm_init_neon_builtin): Rename to ...
34520 (arm_init_builtin): ... this. Add a new parameters PREFIX
34521 and USE_SIG_IN_NAME.
34522 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34523 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34524 'arm_builtin_datum'.
34525 (arm_init_vfp_builtins): Likewise.
34526 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34527 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34528 (arm_expand_neon_args): Rename to ...
34529 (arm_expand_builtin_args): ... this. Rename builtin_arg
34530 enum values and differentiate between ARG_BUILTIN_MEMORY
34531 and ARG_BUILTIN_NEON_MEMORY.
34532 (arm_expand_neon_builtin_1): Rename to ...
34533 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34534 values, arm_expand_builtin_args and add bool parameter NEON.
34535 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34536 (arm_expand_vfp_builtin): Likewise.
34537 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34539 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34541 PR middle-end/77484
34542 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34543 * predict.c (tree_estimate_probability_bb): Reverse direction of
34544 polymorphic call predictor.
34546 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34548 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34549 (determine_pass_name_match): ...this new function and...
34550 (should_skip_pass_p): ...this new function.
34552 2017-01-06 Nathan Sidwell <nathan@acm.org>
34554 * ipa-visibility.c (function_and_variable_visibility): Reformat
34555 comments and long lines. Remove extrneous if.
34556 * symtab.c (symtab_node::make_decl_local): Fix code format.
34557 (symtab_node::set_section_for_node): Fix comment typo.
34559 2017-01-06 Martin Liska <mliska@suse.cz>
34562 * lra-constraints.c: Rename invariant to lra_invariant.
34563 * predict.c (set_even_probabilities): Initialize e to NULL.
34565 2017-01-05 Martin Sebor <msebor@redhat.com>
34567 PR tree-optimization/78910
34568 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34569 (format_integer): Correct off-by-one error in the handling
34570 of precision with negative numbers in signed conversions..
34572 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34574 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34576 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34578 PR tree-optimization/71016
34579 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34580 factor_out_conditional_conversion. Formatting fix.
34581 (factor_out_conditional_conversion): Add cond_stmt argument.
34582 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34583 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34586 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34588 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34589 read-rtl-function.o, and selftest-rtl.o.
34590 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34591 (selftest::aarch64_test_loading_full_dump): New function.
34592 (selftest::aarch64_run_selftests): New function.
34593 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34594 selftest::aarch64_run_selftests.
34595 * config/i386/i386.c
34596 (selftest::ix86_test_loading_dump_fragment_1): New function.
34597 (selftest::ix86_test_loading_call_insn): New function.
34598 (selftest::ix86_test_loading_full_dump): New function.
34599 (selftest::ix86_test_loading_unspec): New function.
34600 (selftest::ix86_run_selftests): Call the new functions.
34601 * emit-rtl.c (maybe_set_max_label_num): New function.
34602 * emit-rtl.h (maybe_set_max_label_num): New decl.
34603 * function.c (instantiate_decls): Guard call to
34604 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34605 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34607 * gensupport.c (gen_reader::gen_reader): Pass "false"
34608 for new "compact" param of rtx_reader.
34609 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34610 rather than an empty string for NULL strings.
34611 * read-md.c: Potentially include config.h rather than bconfig.h.
34612 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34613 (have_error): New global, copied from errors.c.
34614 (md_reader::read_name): Rename to...
34615 (md_reader::read_name_1): ...this, adding "out_loc" param,
34616 and converting "missing name or number" to returning false, rather
34618 (md_reader::read_name): Reimplement in terms of read_name_1.
34619 (md_reader::read_name_or_nil): New function.
34620 (md_reader::read_string): Handle "(nil)" by returning NULL.
34621 (md_reader::md_reader): Add new param "compact".
34622 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34623 (md_reader::read_file): New method.
34624 * read-md.h (md_reader::md_reader): Add new param "compact".
34625 (md_reader::read_file): New method.
34626 (md_reader::is_compact): New accessor.
34627 (md_reader::read_name): Convert return type from void to file_location.
34628 (md_reader::read_name_or_nil): New decl.
34629 (md_reader::read_name_1): New decl.
34630 (md_reader::m_compact): New field.
34631 (noop_reader::noop_reader): Pass "false" for new "compact" param
34633 (rtx_reader::rtx_reader): Add new "compact" param.
34634 (rtx_reader::read_rtx_operand): Make virtual and convert return
34635 type from void to rtx.
34636 (rtx_reader::read_until): New decl.
34637 (rtx_reader::handle_any_trailing_information): New virtual function.
34638 (rtx_reader::postprocess): New virtual function.
34639 (rtx_reader::finalize_string): New virtual function.
34640 (rtx_reader::m_in_call_function_usage): New field.
34641 (rtx_reader::m_reuse_rtx_by_id): New field.
34642 * read-rtl-function.c: New file.
34643 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34644 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34645 (selftest::verify_three_block_rtl_cfg): New decl.
34646 * read-rtl-function.h: New file.
34647 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34648 For host, include function.h, memmodel.h, and emit-rtl.h.
34649 (one_time_initialization): New function.
34650 (struct compact_insn_name): New struct.
34651 (compact_insn_names): New array.
34652 (find_code): Handle insn codes in compact dumps.
34653 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34654 (bind_subst_iter_and_attr): Likewise.
34655 (add_condition_to_string): Likewise.
34656 (add_condition_to_rtx): Likewise.
34657 (apply_attribute_uses): Likewise.
34658 (add_current_iterators): Likewise.
34659 (apply_iterators): Likewise.
34660 (initialize_iterators): Guard usage of apply_subst_iterator with
34661 #ifdef GENERATOR_FILE.
34662 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34663 (md_reader::read_mapping): Likewise.
34664 (add_define_attr_for_define_subst): Likewise.
34665 (add_define_subst_attr): Likewise.
34666 (read_subst_mapping): Likewise.
34667 (check_code_iterator): Likewise.
34668 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34670 (one_time_initialization): New function.
34671 (rtx_reader::read_until): New method.
34672 (read_flags): New function.
34673 (parse_reg_note_name): New function.
34674 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34675 Handle reuse_rtx ids.
34676 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34677 Add parsing support for RTL dumps, mirroring the special-cases in
34678 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34679 values, and calling handle_any_trailing_information.
34680 (rtx_reader::read_rtx_operand): Convert return type from void
34681 to rtx, returning return_rtx. Handle case 'e'. Call
34682 finalize_string on XSTR and XTMPL fields.
34683 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34684 "(nil)" values were omitted. Call the postprocess vfunc on the
34686 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34687 class ctor. Initialize m_in_call_function_usage. Call
34688 one_time_initialization.
34689 * rtl-tests.c (selftest::test_uncond_jump): Call
34690 set_new_first_and_last_insn.
34691 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34692 * selftest-rtl.c: New file.
34693 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34694 (selftest::get_insn_by_uid): New decl.
34695 * selftest-run-tests.c (selftest::run_tests): Call
34696 read_rtl_function_c_tests.
34697 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34698 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34701 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34703 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34704 operands in a special way. Assert that pos+len <= mode precision.
34706 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34708 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34709 3 argument Alias with unlimited for the negative form.
34710 (fno-vect-cost-model): Removed.
34712 2017-01-05 Martin Liska <mliska@suse.cz>
34714 * hsa-gen.c (gen_hsa_divmod): New function.
34715 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34717 2017-01-05 Martin Liska <mliska@suse.cz>
34720 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34723 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34725 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34726 small constant length operands.
34728 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34730 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34731 between loop iterations.
34733 2017-01-05 Martin Liska <mliska@suse.cz>
34736 * gimplify.c (gimplify_decl_expr): Compare to
34737 asan_poisoned_variables instread of checking flags.
34738 (gimplify_target_expr): Likewise.
34739 (gimplify_expr): Likewise.
34740 (gimplify_function_tree): Conditionally initialize
34741 asan_poisoned_variables.
34743 2017-01-04 Jeff Law <law@redhat.com>
34745 PR tree-optimizatin/78812
34746 * rtl.h (contains_mem_rtx_p): Prototype.
34747 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34748 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34749 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34750 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34751 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34753 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34755 * input.c (assert_char_at_range): Default-initialize actual_range.
34757 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34759 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34760 to match the caller.
34762 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34764 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34765 insns after final jump in test to emit dummy move.
34767 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34769 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34770 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34772 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34774 * multiple_target.c (create_dispatcher_calls): Init e_next.
34775 * tree-ssa-loop-split.c (split_loop): Init border.
34776 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34779 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34784 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34785 (altivec_register_operand): Do not return true if the operand
34786 contains a SUBREG mixing SImode and SFmode.
34787 (vsx_register_operand): Likewise.
34788 (vsx_reg_sfsubreg_ok): New predicate.
34789 (vfloat_operand): Do not return true if the operand contains a
34790 SUBREG mixing SImode and SFmode.
34791 (vint_operand): Likewise.
34792 (vlogical_operand): Likewise.
34793 (gpc_reg_operand): Likewise.
34794 (int_reg_operand): Likewise.
34795 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34796 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34797 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34799 (rs6000_emit_move_si_sf_subreg): New helper function.
34800 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34801 fixup SUBREGs involving SImode and SFmode.
34802 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34803 numbers for the new peephole2 optimization.
34804 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34805 the GLIBC math library that do AND/IOR/XOR operations on single
34806 precision floating point.
34807 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34808 target macros to say whether we need to avoid SUBREGs mixing
34810 (TARGET_ALLOW_SF_SUBREG): Likewise.
34811 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34812 (UNSPEC_SI_FROM_SF): Likewise.
34813 (iorxor): Change spacing.
34814 (and_ior_xor): New iterator for AND, IOR, and XOR.
34815 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34816 (movdi_from_sf_zero_ext): Likewise.
34817 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34818 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34819 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34820 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34821 (fms<mode>4): Likewise.
34822 (fnma<mode>4): Likewise.
34823 (fnms<mode>4): Likewise.
34824 (nfma<mode>4): Likewise.
34825 (nfms<mode>4): Likewise.
34827 2017-01-04 Marek Polacek <polacek@redhat.com>
34830 * doc/invoke.texi: Document -Wpointer-compare.
34832 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34834 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34837 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34838 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34841 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34844 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34845 documentation of the powerpc_popcntb_ok attribute.
34846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34847 code to issue warning messages if a requested CPU configuration is
34848 not supported by the binary (assembler and loader) toolchain.
34849 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34850 made to define a built-in function that has been disabled.
34851 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34852 made to define a built-in function that has been disabled.
34853 (altivec_init_builtins): Add comment explaining why definition
34854 of the DST built-in functions is not preceded by an assertion
34855 check. Add assertions to prevent ICE if attempts are made to
34856 define an altivec predicate or an abs* built-in function that has
34858 (htm_init_builtins): Add comment explaining why definition of the
34859 htm built-in functions is not preceded by an assertion check.
34861 2017-01-04 Jeff Law <law@redhat.com>
34863 PR tree-optimizatin/67955
34864 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34865 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34866 the points-to solution does not include pt_null. Use DECL_PT_UID
34869 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34871 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34872 Use gen_int_mode instead of gen_lopwart for const_int operands.
34874 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34876 PR tree-optimization/71563
34877 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34878 out of range value - has low bits known to be zero.
34880 2017-01-04 Alan Modra <amodra@gmail.com>
34882 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34883 * configure: Regenerate.
34884 * config.in: Regenerate.
34886 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34889 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34890 a substring of the message, but strcmp with the whole message. Ifdef
34891 ENABLE_NLS, translate the message first using dgettext.
34893 2017-01-03 Jeff Law <law@redhat.com>
34895 PR tree-optimizatin/78856
34896 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34897 (mark_threaded_blocks): Remove code to truncate thread paths that
34898 cross multiple loop headers. Instead invalidate the cached loop
34899 iteration information and handle case of a thread path walking
34900 into an irreducible region.
34902 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34905 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34906 assertions. Add support for doing the signbit if the IEEE 128-bit
34907 floating point value is in a GPR.
34908 * config/rs6000/rs6000.md (Fsignbit): Delete.
34909 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34910 Update the length attribute if the value is in a GPR.
34911 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34912 the sign or zero extension instruction, since the value is always 0/1.
34913 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34916 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34917 extracting SImode to a GPR register so that we can generate a
34918 store, limit the vector to be in a traditional Altivec register
34919 for the vextuwrx instruction.
34921 2017-01-03 Ian Lance Taylor <iant@google.com>
34923 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34925 2017-01-03 Martin Sebor <msebor@redhat.com>
34927 PR tree-optimization/78696
34928 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34929 precision. Use MPFR for %f for greater fidelity. Correct handling
34931 (pass_sprintf_length::compute_format_length): Set width and precision
34932 specified by asrerisk to void_node for vararg functions.
34933 (try_substitute_return_value): Adjust dump output.
34935 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34937 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34939 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34941 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34942 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34943 -mlra/-mno-lra was passed to the compiler.
34945 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34947 PR rtl-optimization/65618
34948 * emit-rtl.c (try_split): Move initialization of "before" and
34949 "after" to just before the call to emit_insn_after_setloc.
34951 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34953 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34955 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34957 * dwarf2out.c (gen_enumeration_type_die): When
34958 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34960 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34962 PR tree-optimization/78965
34963 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34964 Change first argument from const call_info & to call_info &. For %n
34965 set info.nowrite to false.
34967 PR middle-end/78901
34968 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34969 possibly throwing calls.
34971 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34972 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34973 and fns handling, rather than in a separate case SSA_NAME.
34975 2017-01-02 Jeff Law <law@redhat.com>
34977 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34978 fixes for first_period and second_period variables.
34980 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34983 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34984 (*insvqi_1): New insn pattern.
34985 (*insvqi_1_mem_rex64): Ditto.
34986 (*insvqi_2): Ditto.
34987 (*insvqi_3): Rename from *insvqi.
34989 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34991 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34993 * doc/cfg.texi (Edges): Remove reference to Java.
34994 (Maintaining the CFG): Ditto.
34996 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34998 PR middle-end/77674
34999 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35000 transparent aliases.
35002 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35004 PR middle-end/77484
35005 * predict.def (PRED_CALL): Update hitrate.
35006 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35007 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35008 into direct/indirect/polymorphic variants.
35010 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35012 Update copyright years.
35014 * gcc.c (process_command): Update copyright notice dates.
35015 * gcov-dump.c (print_version): Ditto.
35016 * gcov.c (print_version): Ditto.
35017 * gcov-tool.c (print_version): Ditto.
35018 * gengtype.c (create_file): Ditto.
35019 * doc/cpp.texi: Bump @copying's copyright year.
35020 * doc/cppinternals.texi: Ditto.
35021 * doc/gcc.texi: Ditto.
35022 * doc/gccint.texi: Ditto.
35023 * doc/gcov.texi: Ditto.
35024 * doc/install.texi: Ditto.
35025 * doc/invoke.texi: Ditto.
35027 Copyright (C) 2017 Free Software Foundation, Inc.
35029 Copying and distribution of this file, with or without modification,
35030 are permitted in any medium without royalty provided the copyright
35031 notice and this notice are preserved.