1 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4 * final.c (rest_of_clean_state): Set it for the
6 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7 class when TDF_COMPARE_DEBUG is set.
9 * dwarf2out.c (gen_producer_string): Discard
12 2017-11-15 Joseph Myers <joseph@codesourcery.com>
15 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
16 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
17 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
18 (__TGMATH_REAL_2_3): Remove macros.
19 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
20 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
22 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
24 (remquo): Define using __TGMATH_REAL_3.
26 2017-11-14 Jeff Law <law@redhat.com>
28 * vr-values.c: New file with contents extracted from tree-vrp.c.
29 * Makefile.in (OBJS): Add vr-values.o
30 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
31 (set_value_range, set_and_canonicalize_value_range): Likewise.
32 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
33 (value_range_constant_singleton, symbolic_range_p): Likewise.
34 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
35 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
36 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
37 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
38 (find_case_label_range, find_case_label_index): Likewise.
39 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
40 (range_int_cst_singleton_p, value_inside_range): Likewise.
41 (get_single_symbol): Likewise.
42 (switch_update): Move structure definition here.
43 (to_remove_edges, to_update_switch_stmts): Provide externs.
44 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
45 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
46 (vrp_val_is_min, set_value_range): Likewise.
47 (set_and_canonicalize_value_range, copy_value_range): Likewise.
48 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
49 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
50 (range_is_nonnull, range_int_cst_p): Likewwise.
51 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
52 (get_single_symbol, operand_less_p): Likewise
53 (compare_values_warnv, compare_values): Likewise.
54 (value_inside_range, value_range_constant_singleton): Likewise.
55 (zero_nonzero_bitgs_from_vr): Likewise.
56 (extract_range_from_binary_expr_1): Likewise.
57 (overflow_comparison_p): Likewise.
58 (to_remove_edges, to_update_switch_stmts): Likewise.
59 (find_case_label-index, find_case_label_range): Likewise.
60 (switch_update, set_value_range_to_nonnegative): Remove.
61 (set_value_range_to_truthvalue): Likewise.
62 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
63 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
64 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
65 (find_case_label_ranges, test_for_singularity): Likewise.
66 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
67 (x_vr_values): Move to its remaining use site.
69 2017-11-10 Jeff Law <law@redhat.com>
71 * vr-values.h (VR_INITIALIZER): Move #define here.
72 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
73 * Makefile.in (OBJS): Add tree-evrp.o
74 * tree-vrp.h (assert_info): Move structure definition here.
75 (set_value_range_to_varying): Prototype.
76 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
77 (infer_value_range, register_edge_assert_for): Likewise.
78 (stmt_interesting_for_vrp): Likewise.
79 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
80 (set_value_range_to_varying): No longer static.
81 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
82 (infer_value_range, register_edge_assert_for): Likewise.
84 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
86 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
87 generate the XXBRD instruction.
89 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
90 (rs6000_builtin_type_compatible): Treat _Float128 and long double
91 as being compatible if -mabi=ieeelongdouble.
92 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
93 to setup float128 built-ins with hardware support.
94 (BU_FLOAT128_HW_2): Likewise.
95 (BU_FLOAT128_HW_3): Likewise.
96 (BU_FLOAT128_HW_VSX_1): Likewise.
97 (BU_FLOAT128_HW_VSX_2): Likewise.
98 (scalar_extract_expq): Change float128 built-in functions to
99 accommodate having both KFmode and TFmode functions. Use the
100 KFmode variant as the default.
101 (scalar_extract_sigq): Likewise.
102 (scalar_test_neg_qp): Likewise.
103 (scalar_insert_exp_q): Likewise.
104 (scalar_insert_exp_qp): Likewise.
105 (scalar_test_data_class_qp): Likewise.
106 (sqrtf128_round_to_odd): Delete processing the round to odd
107 built-in functions as special built-in functions, and define them
108 as float128 built-ins. Use the KFmode variant as the default.
109 (truncf128_round_to_odd): Likewise.
110 (addf128_round_to_odd): Likewise.
111 (subf128_round_to_odd): Likewise.
112 (mulf128_round_to_odd): Likewise.
113 (divf128_round_to_odd): Likewise.
114 (fmaf128_round_to_odd): Likewise.
115 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
116 support for KFmode and TFmode xststdcqp calls.
117 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
118 point, switch the built-in handlers for the get/set float128
119 exponent, get float128 mantissa, float128 test built-ins, and the
120 float128 round to odd built-in functions. Eliminate creating the
121 float128 round to odd built-in functions as special built-ins.
122 (rs6000_init_builtins): Eliminate special creation of the float128
123 round to odd built-in functions.
124 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
125 function insns to support both TFmode and KFmode variants.
126 (xsxsigqp_<mode>): Likewise.
127 (xsiexpqpf_<mode>): Likewise.
128 (xsiexpqp_<mode>): Likewise.
129 (xststdcqp_<mode>): Likewise.
130 (xststdcnegqp_<mode>): Likewise.
131 (xststdcqp_<mode>): Likewise.
133 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
135 * tree-ssa-threadupdate.c (compute_path_counts): Remove
136 unused path_in_freq_ptr parameter.
137 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
139 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
141 * ipa-inline.c (edge_badness): Dump sreal frequency.
142 (compute_inlined_call_time): Match natural implementaiton ...
143 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
144 forgotten division by CGRAPH_FREQ_BASE.
146 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
148 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
149 Solaris 11. Update comment.
150 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
152 * config/sol2.h (STARTFILE_SPEC): Likewise.
153 * configure: Regenerate.
155 2017-11-14 Carl Love <cel@us.ibm.com>
157 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
158 le_ and be_ prefixes to swap* variables. Remove
159 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
162 2017-11-14 Jason Merrill <jason@redhat.com>
164 Support GTY((cache)) on hash_map.
165 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
166 (ggc_cache_remove): Override it instead of ggc_mx.
167 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
168 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
169 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
170 (simple_cache_map_traits): Override maybe_mx.
171 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
172 (hash_map): Friend gt_cleare_cache.
173 (gt_cleare_cache): New.
174 * tree.h (tree_cache_traits): New hash_map traits class.
175 (tree_cache_map): New typedef.
177 2017-11-14 Richard Biener <rguenther@suse.de>
179 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
180 paramter and handling.
181 (cleanup_control_flow_bb): Likewise.
182 (cleanup_control_flow_pre): New helper performing a DFS walk
183 to call cleanup_control_flow_bb in PRE order.
184 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
185 via cleanup_control_flow_pre.
187 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
189 * config/aarch64/aarch64-simd.md
190 (aarch64_simd_bsl<mode>_internal): Remove DImode.
191 (*aarch64_simd_bsl<mode>_alt): Likewise.
192 (aarch64_simd_bsldi_internal): New.
193 (aarch64_simd_bsldi_alt): Likewise.
195 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
197 * tracer.c (better_p): Do not compare frequencies.
198 * reg-stack.c (better_edge): Likewise.
199 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
202 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
204 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
205 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
206 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
207 * lto-streamer-in.c (input_function): Use update_max_bb_count.
208 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
209 * predict.c (maybe_hot_frequency_p): Inline to ...
210 (maybe_hot_count_p): ... here; rewrite to counts.
211 (counts_to_freqs): Rename to ...
212 (update_max_bb_count): ... this one.
213 (expensive_function_p): Use counts.
214 (estimate_bb_frequencies): Update.
215 (rebuild_frequencies): Update.
216 * predict.h (counts_to_freqs): Rename to ...
217 (update_max_bb_count): ... this one.
218 * profile.c (compute_branch_probabilities): Add debug info
219 * tree-inline.c (expand_call_inline): Update debug info.
220 (optimize_inline_calls): Use update_max_bb_count..
221 (tree_function_versioning): Use update_max_bb_count..
222 * value-prof.c (gimple_value_profile_transformations):
223 Do not use update_max_bb_count.
225 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
227 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
228 always use frequencies.
230 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
232 * bb-reorder.c: Remove frequencies from comments.
233 (better_edge_p): Use profile counts.
234 (find_traces): Dump profile counts.
235 (rotate_loop): Use profile counts.
236 (find_traces_1_round): Likewise.
237 (connect_better_edge_p): Use counts instead of probabilities for
239 (copy_bb_p): Drop early check for non-0 frequency.
240 (sanitize_hot_paths): Update comments.
242 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
244 * ipa-split.c (struct split_point): Add count.
245 (consider_split): Do not compute incoming frequency; compute incoming
246 count and store it to split_point.
247 (split_function): Set count of the call to split part correctly.
249 2017-11-13 Carl Love <cel@us.ibm.com>
251 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
253 2017-11-13 Tom Tromey <tom@tromey.com>
255 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
257 2017-11-13 Carl Love <cel@us.ibm.com>
259 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
260 Add support for builtins:
261 unsigned int vec_first_{,miss}_match_{,or_eos}index,
262 vector {un,}signed {char,int,short},
263 vector {un,}signed {char,int,short}) arguments.
264 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
265 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
266 Add BU_P9V_AV_2 expansions for the builtins.
267 * config/rs6000/altivec.h (vec_first_match_index,
268 vec_first_mismatch_index, vec_first_match_or_eos_index,
269 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
270 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
271 new extern declaration.
272 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
273 * config/rs6000/vsx.md (first_match_index_<mode>,
274 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
275 first_mismatch_or_eos_index_<mode>): Add define expand.
276 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
277 * doc/extend.texi: Update the built-in documenation file for the new
280 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
282 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
283 into the min/max operations for _Float<N> and _Float<N>X types.
285 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
288 * dwarf2out.c (do_eh_frame): New static variable.
289 (dwarf2out_begin_prologue): Set it.
290 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
292 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
294 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
296 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
298 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
301 * debug.h (dwarf2out_do_eh_frame): Declare.
302 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
303 (dwarf2out_do_frame): Use it.
304 (dwarf2out_do_cfi_asm): Likewise.
305 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
306 (dwarf2out_assembly_start): Likewise.
307 (dwarf2out_begin_prologue): Fix comment.
308 * toplev.c (compile_file): Always call dwarf2out_frame_finish
309 if the target needs either debug or unwind DWARF2 info.
310 * lto-opts.c (lto_write_options): Do not save -fexceptions,
311 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
312 -ftrapping-math, -ftrapv and -fwrapv.
314 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
316 * cgraph.c (cgraph_edge::sreal_frequency): New function.
317 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
318 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
319 (estimate_edge_size_and_time): Likewise.
320 (ipa_merge_fn_summary_after_inlining): Likewise.
321 * ipa-inline.c (cgraph_freq_base_rec): Remove.
322 (compute_uninlined_call_time): Use sreal_frequency.
323 (compute_inlined_call_time): Likewise.
324 (ipa_inline): Do not initialize cgraph_freq_base_rec.
325 * profile-count.c: Include sreal.h.
326 (profile_count::to_sreal_scale): New.
327 * profile-count.h: Forward declare sreal.
328 (profile_count::to_sreal_scale): Declare.
330 2017-11-13 Nathan Sidwell <nathan@acm.org>
332 * diagnostic.c (maybe_line_and_column): New.
333 (diagnostic_get_location_text): Use it.
334 (diagnostic_report_current_module): Likewise.
335 (test_diagnostic_get_location_text): Add tests.
337 2017-11-13 Luis Machado <luis.machado@linaro.org>
339 * doc/md.texi (Specifying processor pipeline description): Fix
340 incorrect latency for the div instruction example.
342 2017-11-13 Jakub Jelinek <jakub@redhat.com>
344 PR tree-optimization/78821
345 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
346 that bit_not_p is the same.
347 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
348 (split_group): Count precisely bit_not_p bits in each statement.
349 (invert_op): New function.
350 (imm_store_chain_info::output_merged_store): Use invert_op to
351 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
352 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
354 2017-11-13 Martin Liska <mliska@suse.cz>
356 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
357 (struct source_info): Likewise.
358 (add_branch_counts): Likewise.
359 (add_line_counts): Likewise.
360 (function_summary): Likewise.
361 (output_intermediate_line): Likewise.
362 (generate_results): Likewise.
364 2017-11-13 Martin Liska <mliska@suse.cz>
366 * gcov.c (struct block_info): Remove typedef for block_t.
367 (struct line_info): Likewise.
368 (line_info::has_block): Likewise.
369 (EXIT_BLOCK): Likewise.
372 (get_cycles_count): Likewise.
373 (process_file): Likewise.
374 (read_graph_file): Likewise.
375 (solve_flow_graph): Likewise.
376 (find_exception_blocks): Likewise.
377 (add_line_counts): Likewise.
378 (accumulate_line_info): Likewise.
379 (output_line_details): Likewise.
381 2017-11-13 Martin Liska <mliska@suse.cz>
383 * gcov.c (struct arc_info): Remove typedef for arc_t.
384 (struct line_info): Likewise.
385 (add_branch_counts): Likewise.
386 (output_branch_count): Likewise.
387 (function_info::~function_info): Likewise.
389 (output_intermediate_line): Likewise.
390 (read_graph_file): Likewise.
391 (solve_flow_graph): Likewise.
392 (find_exception_blocks): Likewise.
393 (add_line_counts): Likewise.
394 (accumulate_line_info): Likewise.
395 (output_line_details): Likewise.
396 (output_function_details): Likewise.
398 2017-11-13 Martin Liska <mliska@suse.cz>
400 * gcov.c (struct function_info): Remove typedef for function_t.
401 (struct source_info): Likewise.
402 (source_info::get_functions_at_location): Likewise.
403 (solve_flow_graph): Likewise.
404 (find_exception_blocks): Likewise.
405 (add_line_counts): Likewise.
406 (output_intermediate_file): Likewise.
407 (process_file): Likewise.
408 (generate_results): Likewise.
409 (release_structures): Likewise.
410 (read_graph_file): Likewise.
411 (read_count_file): Likewise.
412 (accumulate_line_counts): Likewise.
413 (output_lines): Likewise.
415 2017-11-13 Martin Liska <mliska@suse.cz>
417 * gcov.c (function_info::function_info): Remove num_counts
418 and add vector<gcov_type>.
419 (function_info::~function_info): Use the vector.
420 (process_file): Likewise.
421 (read_graph_file): Likewise.
422 (read_count_file): Likewise.
423 (solve_flow_graph): Likewise.
425 2017-11-13 Martin Liska <mliska@suse.cz>
427 * gcov.c (function_info::is_artificial): New function.
428 (process_file): Erase all artificial early.
429 (generate_results): Skip as all artificial are already
432 2017-11-13 Martin Liska <mliska@suse.cz>
434 * gcov.c (read_graph_file): Store to global vector of functions.
435 (read_count_file): Iterate the vector.
436 (process_file): Likewise.
437 (generate_results): Likewise.
438 (release_structures): Likewise.
440 2017-11-13 Jakub Jelinek <jakub@redhat.com>
442 PR tree-optimization/82954
443 * gimple-ssa-store-merging.c
444 (imm_store_chain_info::coalesce_immediate_stores): If
445 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
447 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
449 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
450 Upddate call to ENDIAN_LANE_N.
451 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
452 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
453 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
454 and use aarch64_endian_lane_rtx.
455 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
457 2017-11-12 Tom de Vries <tom@codesourcery.com>
459 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
461 2017-11-12 Tom de Vries <tom@codesourcery.com>
463 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
465 2017-11-12 Tom de Vries <tom@codesourcery.com>
467 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
469 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
470 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
471 * defaults.h (ASM_OUTPUT_LABELREF): Same.
473 2017-11-11 Martin Sebor <msebor@redhat.com>
476 * doc/extend.texi (attribute nonstring): Remove spurious argument.
479 * prefic.c (translate_name): Replace strncpy with memcpy to
480 avoid -Wstringop-truncation.
482 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
484 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
486 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
488 * predict.c (maybe_hot_frequency_p): Do not use cfun.
490 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
492 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
495 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
497 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
498 enable generating XXBRH if the value is in a vector register.
499 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
500 value is in a vector register.
501 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
502 register to register bswap64's instead of doing the GPR sequence
503 used on previous machines.
504 (bswapdi2_xxbrd): New insn.
505 (bswapdi2_reg): Disallow on ISA 3.0.
506 (register to register bswap64 splitter): Do not split the insn on
507 ISA 3.0 systems that use XXBRD.
509 2017-11-10 Martin Sebor <msebor@redhat.com>
512 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
514 (find_subframework_file): Same.
516 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
518 * auto-profile.c (afdo_indirect_call): Drop frequency.
519 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
520 (cgraph_node::create_edge): Drop frequency argument.
521 (cgraph_node::create_indirect_edge): Drop frequency argument.
522 (cgraph_edge::make_speculative): Drop frequency arguments.
523 (cgraph_edge::resolve_speculation): Do not update frequencies
524 (cgraph_edge::dump_edge_flags): Do not dump frequency.
525 (cgraph_node::dump): Check consistency in IPA mode.
526 (cgraph_edge::maybe_hot_p): Use IPA counter.
527 (cgraph_edge::verify_count_and_frequency): Rename to ...
528 (cgraph_edge::verify_count): ... this one; drop frequency checking.
529 (cgraph_node::verify_node): Update.
530 * cgraph.h (struct cgraph_edge): Drop frequency.
531 (cgraph_edge::frequency): New function.
532 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
534 (cgraph_edge::rebuild_edges): Likewise.
535 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
536 (duplicate_thunk_for_node): Do not pass frequency.
537 (cgraph_node::create_clone): Scale only counts.
538 (cgraph_node::create_virtual_clone): Do not pass frequency.
539 (cgraph_node::create_edge_including_clones): Do not pass frequency.
540 (cgraph_node::create_version_clone): Do not pass frequency.
541 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
542 (cgraph_node::expand_thunk): Do not pass frequency.
543 (cgraph_node::create_wrapper): Do not pass frequency.
544 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
546 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
547 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
548 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
549 (gather_caller_stats): Use frequency function.
550 (ipcp_cloning_candidate_p): Use frequency function.
551 (ipcp_propagate_stage): Use frequency function.
552 (get_info_about_necessary_edges): Use frequency function.
553 (update_profiling_info): Update only IPA profile.
554 (update_specialized_profile): Use frequency functoin.
555 (perhaps_add_new_callers): Update only IPA profile.
556 * ipa-devirt.c (ipa_devirt): Use IPA profile.
557 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
558 (dump_ipa_call_summary): Use frequency function.
559 (estimate_edge_size_and_time): Use frequency function.
560 (ipa_merge_fn_summary_after_inlining): Use frequency function.
561 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
562 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
563 (update_noncloned_counts): ... ths one; scale counts only.
564 (clone_inlined_nodes): Do not scale frequency.
565 (inline_call): Do not pass frequency.
566 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
567 (compute_inlined_call_time): Use IPA profile.
568 (want_inline_small_function_p): Use IPA profile.
569 (want_inline_self_recursive_call_p): Use IPA profile.
570 (edge_badness): Use IPA profile.
571 (lookup_recursive_calls): Use IPA profile.
572 (recursive_inlining): Do not pass frequency.
573 (resolve_noninline_speculation): Do not update frequency.
574 (inline_small_functions): Collect max of IPA profile.
575 (dump_overall_stats): Dump IPA porfile.
576 (dump_inline_stats): Dump IPA porfile.
577 (ipa_inline): Collect IPA stats.
578 * ipa-inline.h (clone_inlined_nodes): Update prototype.
579 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
580 (ipa_propagate_frequency): Use frequency function.
581 (ipa_profile): Cleanup.
582 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
583 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
584 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
585 (input_node): Do not stream frequency.
586 (input_edge): Do not stream frequency.
587 (merge_profile_summaries): Scale only IPA profiles.
588 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
589 * predict.c (drop_profile): Do not recompute frequency.
590 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
591 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
592 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
593 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
594 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
595 * tree-inline.c (copy_bb): Do not scale frequency.
596 (expand_call_inline): Do not scale frequency.
597 (tree_function_versioning): Do not scale frequency.
598 * ubsan.c (ubsan_create_edge): Do not pass frequency.
600 2017-11-10 Julia Koval <julia.koval@intel.com>
602 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
603 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
604 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
605 _mm256_maskz_gf2p8affine_epi64_epi8,
606 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
607 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
608 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
609 __builtin_ia32_vgf2p8affineqb_v32qi,
610 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
611 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
613 2017-11-10 Tamar Christina <tamar.christina@arm.com>
617 (arm_option_override): Refactor.
618 (arm_option_reconfigure_globals): New.
619 (arm_options_perform_arch_sanity_checks): New.
620 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
621 (arm_options_perform_arch_sanity_checks): Likewise
623 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
625 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
627 * rs6000/rs6000.md ("type" attr): Add qmul.
628 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
629 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
630 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
632 2017-11-10 Martin Sebor <msebor@redhat.com>
635 * builtins.c (compute_objsize): Handle arrays that
636 compute_builtin_object_size likes to fail for. Make extern.
637 * builtins.h (compute_objsize): Declare.
638 (check_strncpy_sizes): New function.
639 (expand_builtin_strncpy): Call check_strncpy_sizes.
640 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
641 -Wstringop-truncation.
642 (gimple_fold_builtin_strncat): Same.
643 * gimple.c (gimple_build_call_from_tree): Set call location.
644 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
645 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
646 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
647 (handle_builtin_strlen): Use strlen_to_stridx.
648 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
650 Use strlen_to_stridx.
651 (pass_strlen::execute): Release strlen_to_stridx.
652 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
653 (-Wstringop-truncation): Document new option.
655 2017-11-10 Martin Liska <mliska@suse.cz>
657 PR gcov-profile/82702
658 * gcov.c (main): Handle intermediate files in a different
660 (get_gcov_intermediate_filename): New function.
661 (output_gcov_file): Remove support of intermediate files.
662 (generate_results): Allocate intermediate file.
663 (release_structures): Clean-up properly fn_end.
664 (output_intermediate_file): Start iterating with line 1.
666 2017-11-10 Jakub Jelinek <jakub@redhat.com>
668 PR tree-optimization/82929
669 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
670 ops_swapped_p non-static data member.
671 (store_immediate_info::store_immediate_info): Clear it.
672 (imm_store_chain_info::coalesce_immediate_stores): If swapping
673 ops set ops_swapped_p.
674 (count_multiple_uses): Handle ops_swapped_p.
676 2017-11-10 Martin Liska <mliska@suse.cz>
678 * coverage.c (coverage_init): Stream information about
679 support of has_unexecuted_blocks.
680 * doc/gcov.texi: Document that.
681 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
682 * gcov.c (read_graph_file): Likewise.
683 (output_line_beginning): Fix a small issue with
686 2017-11-10 Bin Cheng <bin.cheng@arm.com>
688 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
689 reference of trivial component.
691 2017-11-10 Jakub Jelinek <jakub@redhat.com>
694 * gimple-ssa-store-merging.c
695 (pass_store_merging::terminate_all_aliasing_chains): For
696 gimple_store_p stmts also call refs_output_dependent_p.
698 PR rtl-optimization/82913
699 * compare-elim.c (try_merge_compare): Punt if def_insn is not
702 2017-11-09 Jeff Law <law@redhat.com>
704 * vr-values.h: New file with vr_values class.
705 * tree-vrp.c: Include vr-values.h
706 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
707 data objects into the vr_values class.
708 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
709 (get_value_range): Make it a member function within vr_values class.
710 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
711 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
712 (op_with_constant_singleton_value_range): Likewise.
713 (extract_range_for_var_from_comparison_expr): Likewise.
714 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
715 (extract_range_from_binary_expr): Likewise.
716 (extract_range_from_unary_expr): Likewise.
717 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
718 (check_for_binary_op_overflow, extract_range_basic): Likewise.
719 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
720 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
721 (compare_name_with_value, compare_names): Likewise.
722 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
723 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
724 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
725 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
726 (extract_range_from_phi_node): Likewise.
727 (simplify_truth_ops_using_ranges): Likewise.
728 (simplify_div_or_mod_using_ranges): Likewise.
729 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
730 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
731 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
732 (simplify_float_conversion_using_ranges): Likewise.
733 (simplify_internal_call_using_ranges): Likewise.
734 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
735 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
736 poitner via x_vr_values for calls into gimple folder.
737 (vrp_initialize_lattice): Make this the vr_values ctor.
738 (vrp_free_lattice): Make this the vr_values dtor.
739 (set_vr_value): New function.
740 (class vrp_prop): Add vr_values data member. Add various member
741 functions as well as member functions that delegate to vr_values.
742 (check_array_ref): Make a member function within vrp_prop class.
743 (search_for_addr_array, vrp_initialize): Likewise.
744 (vrp_finalize): Likewise. Revamp to avoid direct access to
745 vr_value, values_propagated, etc.
746 (check_array_bounds): Extract vrp_prop class instance pointer from
747 walk info structure. Use it to call member functions.
748 (check_all_array_refs): Make a member function within vrp_prop class.
749 Smuggle class instance pointer via walk info structure.
750 (x_vr_values): New local static.
751 (vrp_valueize): Use x_vr_values to get class instance.
752 (vr_valueize_1): Likewise.
753 (class vrp_folder): Add vr_values data member. Add various member
754 functions as well as member functions that delegate to vr_values.
755 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
756 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
757 class instance from vr_values. Use it to call member functions.
758 (vrp_dom_walker): Add vr_values data member.
759 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
760 instance via x_vr_values.
761 (identify_jump_threads): Accept vr_values as argument. Store
762 it into the walker structure.
763 (evrp_dom_walker): Add vr_values class data member. Add various
765 (evrp_dom_walker::try_find_new_range): Use vr_values data
766 member to access the memory allocator.
767 (evrp_dom_walker::before_dom_children): Store vr_values class
768 instance into the vrp_folder class.
769 (evrp_dom_walker::push_value_range): Rework to avoid direct
770 access to num_vr_values and vr_value.
771 (evrp_dom_walker::pop_value_range): Likewise.
772 (execute_early_vrp): Remove call to vrp_initialize_lattice.
773 Use vr_values to get to dump_all_value_ranges member function.
774 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
775 and simplify_cond_using_ranges_2 via vrp_prop class instance.
776 Pass vr_values class instance down to identify_jump_threads.
777 Remove call to vrp_free_lattice.
778 (debug_all_value_ranges): Remove.
780 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
781 (vrp_folder): Likewise.
783 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
784 Get it from the existing bitmap instead.
785 (vrp_intersect_ranges_1): Likewise.
787 2017-11-09 Jakub Jelinek <jakub@redhat.com>
789 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
791 (store_immediate_info::store_immediate_info): Add bitnotp argument,
793 (imm_store_chain_info::coalesce_immediate_stores): Break group
794 if bit_not_p is different.
795 (count_multiple_uses, split_group,
796 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
797 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
798 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
799 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
800 to store_immediate_info ctor.
802 2017-11-09 Jim Wilson <jimw@sifive.com>
804 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
805 (scan_prog_file): Likewise.
807 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
809 * bb-reorder.c (max_entry_frequency): Remove.
810 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
811 connect_traces, push_to_next_round_p): Remove prototypes.
812 (find_traces_1_round): Use counts only.
813 (push_to_next_round_p): Likewise.
814 (find_traces): Likewise.
815 (rotate_loop): Likewise.
816 (find_traces_1_round): Likewise.
817 (connect_traces): Likewise.
818 (edge_order): Likewise.
820 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
822 * config/arm/arm.c (output_return_instruction): Add comments to
823 indicate requirement for cmse_nonsecure_entry return to account
824 for the size of clearing instruction output here.
825 (thumb_exit): Likewise.
826 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
827 return in hardfloat mode.
829 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
831 * config/rs6000/rs6000.c (machine_function): Add a bool,
832 "toc_is_wrapped_separately".
833 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
834 if it wasn't explicitly set or unset, we are optimizing for speed, and
835 doing separate shrink-wrapping.
836 (rs6000_get_separate_components): Enable the TOC component if
837 saving the TOC register in the prologue.
838 (rs6000_components_for_bb): Handle the TOC component.
839 (rs6000_emit_prologue_components): Store the TOC register where needed.
840 (rs6000_set_handled_components): Mark TOC as handled, if handled.
841 (rs6000_emit_prologue): Don't save the TOC if that is already done.
843 2017-11-09 Martin Jambor <mjambor@suse.cz>
845 * ipa-param-manipulation.c: New file.
846 * ipa-param-manipulation.h: Likewise.
847 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
848 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
849 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
850 (ipa_parm_adjustment): Likewise.
851 (ipa_parm_adjustment_vec): Likewise.
852 (ipa_get_vector_of_formal_parms): Moved declaration to
853 ipa-param-manipulation.h.
854 (ipa_get_vector_of_formal_parm_types): Likewise.
855 (ipa_modify_formal_parameters): Likewise.
856 (ipa_modify_call_arguments): Likewise.
857 (ipa_combine_adjustments): Likewise.
858 (ipa_dump_param_adjustments): Likewise.
859 (ipa_modify_expr): Likewise.
860 (ipa_get_adjustment_candidate): Likewise.
861 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
862 ipa-param-manipulation.c.
863 (ipa_get_vector_of_formal_parm_types): Likewise.
864 (ipa_modify_formal_parameters): Likewise.
865 (ipa_modify_call_arguments): Likewise.
866 (ipa_modify_expr): Likewise.
867 (get_ssa_base_param): Likewise.
868 (ipa_get_adjustment_candidate): Likewise.
869 (index_in_adjustments_multiple_times_p): Likewise.
870 (ipa_combine_adjustments): Likewise.
871 (ipa_dump_param_adjustments): Likewise.
872 * tree-sra.c: Also include ipa-param-manipulation.h
873 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
876 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
877 Alan Hayward <alan.hayward@arm.com>
878 David Sherwood <david.sherwood@arm.com>
880 * doc/sourcebuild.texi (vect_masked_store): Document.
882 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
883 Alan Hayward <alan.hayward@arm.com>
884 David Sherwood <david.sherwood@arm.com>
886 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
888 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
889 Alan Hayward <alan.hayward@arm.com>
890 David Sherwood <david.sherwood@arm.com>
892 * doc/sourcebuild.texi (vect_variable_length): Document.
894 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
895 Alan Hayward <alan.hayward@arm.com>
896 David Sherwood <david.sherwood@arm.com>
898 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
900 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
901 Alan Hayward <alan.hayward@arm.com>
902 David Sherwood <david.sherwood@arm.com>
904 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
906 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
907 Alan Hayward <alan.hayward@arm.com>
908 David Sherwood <david.sherwood@arm.com>
910 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
911 previously undocumented selectors.
912 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
914 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
916 * doc/rtl.texi (const_vector): Say that elements can be
918 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
919 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
920 (gen_vec_duplicate): Use it instead of CONSTANT_P.
921 * optabs.c (expand_vector_broadcast): Likewise.
923 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
924 Alan Hayward <alan.hayward@arm.com>
925 David Sherwood <david.sherwood@arm.com>
927 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
928 scaled index even if the unscaled address was invalid.
929 Don't increase the complexity of using a scale in that case.
931 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
932 Alan Hayward <alan.hayward@arm.com>
933 David Sherwood <david.sherwood@arm.com>
935 * doc/rtl.texi: Rewrite the subreg rules so that they partition
936 the inner register into REGMODE_NATURAL_SIZE bytes rather than
937 UNITS_PER_WORD bytes.
938 * emit-rtl.c (validate_subreg): Divide subregs into blocks
939 based on REGMODE_NATURAL_SIZE of the inner mode.
940 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
941 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
942 * expmed.c (lowpart_bit_field_p): Divide the value up into
943 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
944 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
945 whether something is likely to occupy more than one register.
947 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
950 * ipa-inline-transform.c (update_noncloned_frequencies): Use
951 profile_count::adjust_for_ipa_scaling.
952 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
953 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
955 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
957 2017-11-09 Jakub Jelinek <jakub@redhat.com>
959 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
960 (split_group): Add total_orig and total_new arguments, estimate the
961 number of statements related to the store group without store merging
962 and with store merging.
963 (imm_store_chain_info::output_merged_store): Adjust split_group
964 callers, punt if estimated number of statements with store merging
965 is not smaller than estimated number of statements without it.
967 (handled_load): Remove has_single_use checks.
968 (pass_store_merging::process_store): Likewise.
970 2017-11-09 Richard Biener <rguenther@suse.de>
972 PR tree-optimization/82902
973 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
975 2017-11-09 Martin Liska <mliska@suse.cz>
978 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
981 2017-11-09 Martin Liska <mliska@suse.cz>
983 PR tree-optimization/82669
984 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
986 2017-11-09 Martin Liska <mliska@suse.cz>
988 PR gcov-profile/48463
989 * coverage.c (coverage_begin_function): Output also end locus
990 of a function and information whether the function is
992 * gcov-dump.c (tag_function): Parse and print the information.
993 * gcov.c (INCLUDE_MAP): Add include.
994 (INCLUDE_SET): Likewise.
995 (struct line_info): Move earlier in the source file because
996 of vector<line_info> in function_info structure.
997 (line_info::line_info): Likewise.
998 (line_info::has_block): Likewise.
999 (struct source_info): Add new member index.
1000 (source_info::get_functions_at_location): New function.
1001 (function_info::group_line_p): New function.
1002 (output_intermediate_line): New function.
1003 (output_intermediate_file): Use the mentioned function.
1004 (struct function_start): New.
1005 (struct function_start_pair_hash): Likewise.
1006 (process_file): Add code that identifies group functions.
1007 Assign lines either to global or function scope.
1008 (generate_results): Skip artificial functions.
1009 (find_source): Assign index for each source file.
1010 (read_graph_file): Read new flag artificial and end_line.
1011 (add_line_counts): Assign it either to global of function scope.
1012 (accumulate_line_counts): Isolate core of the function to
1013 accumulate_line_info and call it for both function and global
1015 (accumulate_line_info): New function.
1016 (output_line_beginning): Fix GNU coding style.
1017 (print_source_line): New function.
1018 (output_line_details): Likewise.
1019 (output_function_details): Likewise.
1020 (output_lines): Iterate both source (global) scope and function
1022 (struct function_line_start_cmp): New class.
1023 * doc/gcov.texi: Reflect changes in documentation.
1025 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1028 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1029 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1030 and similarly for not instead of neg.
1032 2017-11-08 Andi Kleen <ak@linux.intel.com>
1034 * config/i386/i386.opt: Add -mforce-indirect-call.
1035 * config/i386/predicates.md: Check for flag_force_indirect_call.
1036 * doc/invoke.texi: Document -mforce-indirect-call
1038 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1040 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1042 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1043 config/riscv/riscv.c (predict.h): New include.
1044 (riscv_slow_unaligned_access_p): No longer static.
1045 (riscv_block_move_straight): Add require.
1046 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1048 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1051 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1052 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1053 nonimmediate_operand predicate for operand 1 instead of
1056 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1058 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1060 * config/aarch64/constraints.md (Uml): New constraint.
1061 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1064 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1066 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1067 of two vec_duplicates into a vec_concat.
1069 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1071 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1072 Simplify vec_merge of vec_duplicate and vec_concat.
1073 * config/aarch64/constraints.md (Utq): New constraint.
1074 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1077 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1079 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1080 Simplify vec_merge of vec_duplicate and const_vector.
1081 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1083 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1084 mode iterator. Update predicate on operand 1 to
1085 handle non-const_vec constants. Delete constraints.
1086 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1088 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1090 PR tree-optimization/78821
1091 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1093 (store_operand_info::store_operand_info): Initialize it to false.
1094 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1095 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1096 store in the group, and if chain_info is non-NULL, to ignore altogether
1098 (compatible_load_p): Fail if bit_not_p does not match.
1099 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1100 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1101 (pass_store_merging::process_store): Adjust
1102 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1103 call terminate_all_aliasing_chains newly when adding a store into
1104 a chain with non-NULL chain_info.
1106 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1108 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1110 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1112 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1114 (aarch64_layout_frame): Initialise emit_frame_chain.
1115 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1116 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1118 2017-11-08 Martin Liska <mliska@suse.cz>
1120 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1121 of gimple_call_internal_p.
1123 2017-11-07 Tom de Vries <tom@codesourcery.com>
1125 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1127 2017-11-07 Tom de Vries <tom@codesourcery.com>
1129 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1132 2017-11-08 Martin Liska <mliska@suse.cz>
1135 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1137 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1139 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1141 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1144 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1145 (*cmp<mode>_ccz_1): New insn with $k alternative.
1148 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1149 RTX_COMM_COMPARE as commutative as well.
1150 (ix86_binary_operator_ok): Formatting fix.
1151 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1152 *<code><mode>3<mask_name><round_saeonly_name>,
1153 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1154 *<s>mul<mode>3_highpart<mask_name>,
1155 *vec_widen_umult_even_v16si<mask_name>,
1156 *vec_widen_umult_even_v8si<mask_name>,
1157 *vec_widen_umult_even_v4si<mask_name>,
1158 *vec_widen_smult_even_v16si<mask_name>,
1159 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1160 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1161 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1162 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1163 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1164 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1165 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1166 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1167 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1168 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1169 ix86_binary_operator_ok. Formatting fixes.
1170 (*<plusminus_insn><mode>3<mask_name><round_name>,
1171 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1174 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1176 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1177 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1178 eq and ne if TARGET_ISEL.
1179 (cmp): New code_iterator.
1180 (UNS, UNSU_, UNSIK): New code_attrs.
1181 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1182 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1184 ("eq<mode>3"): ... this.
1185 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1187 ("ne<mode>3"): ... this.
1189 2017-11-07 Julia Koval <julia.koval@intel.com>
1192 * common/config/i386/i386-common.c
1193 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1194 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1195 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1196 * config/i386/i386.opt: Ditto.
1197 * config/i386/i386.c (ix86_target_string): Ditto.
1198 (ix86_option_override_internal): Ditto.
1199 (ix86_init_mpx_builtins): Move MPX to args2.
1200 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1201 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1202 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1203 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1204 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1205 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1206 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1208 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1211 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1212 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1213 (zero-extendsidi peephole2): Remove peephole.
1215 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1218 * stor-layout.c: Include attribs.h.
1219 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1220 explicit lookup of "aligned" attribute.
1222 2017-11-07 Andrew Waterman <andrew@sifive.com>
1224 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1225 (riscv_expand_block_move): Likewise.
1226 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1228 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1229 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1230 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1231 (riscv_adjust_block_mem): Likewise.
1232 (riscv_block_move_loop): Likewise.
1233 (riscv_expand_block_move): Likewise.
1234 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1236 2017-11-07 Michael Clark <michaeljclark@mac.com>
1238 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1239 (MUSL_DYNAMIC_LINKER): Likewise.
1241 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1243 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1244 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1247 2017-11-07 Richard Biener <rguenther@suse.de>
1249 * match.pd: Fix build.
1251 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1252 Jackson Woodruff <jackson.woodruff@arm.com>
1254 PR tree-optimization/71026
1255 * match.pd: Canonicalize negate in division.
1257 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1260 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1262 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1264 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1265 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1267 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1269 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1270 non-scalar integral types.
1271 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1272 (-(A-B), -(~A)): New transformations.
1274 2017-11-07 Tom de Vries <tom@codesourcery.com>
1276 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1277 semicolon after "do {} while (0)".
1278 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1279 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1280 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1281 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1282 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1284 2017-11-07 Tom de Vries <tom@codesourcery.com>
1286 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1287 after "do {} while (0)".
1288 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1289 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1290 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1291 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1292 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1294 2017-11-07 Tom de Vries <tom@codesourcery.com>
1297 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1299 (arm_rtx_costs_internal): Add missing semicolon after
1300 HANDLE_NARROW_SHIFT_ARITH call.
1302 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1305 disable isel if it was not set explicitly.
1307 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1309 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1311 (add_type_attribute) likewise.
1313 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1315 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1316 of struct ix86_frame.
1317 (ix86_initial_elimination_offset): Likewise.
1318 (ix86_expand_split_stack_prologue): Likewise.
1320 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1322 * tree-vrp.h (enum value_range_type): Update stale comment.
1324 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1325 Alan Hayward <alan.hayward@arm.com>
1326 David Sherwood <david.sherwood@arm.com>
1328 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1329 (aarch64_expand_vec_perm_const): Take the number of units too.
1330 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1331 (aarch64_expand_vec_perm_const): Likewise.
1332 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1333 (vec_perm<mode>): Update accordingly.
1335 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1340 Take the number of units too.
1341 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1342 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1343 but check for a vector mode before rather than after the call.
1344 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1345 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1346 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1347 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1348 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1349 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1350 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1351 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1352 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1353 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1354 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1355 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1356 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1357 (aarch64_sqdmull2_n<mode>): Update accordingly.
1359 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1364 the number of units too.
1365 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1366 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1367 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1368 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1369 (vec_store_lanesxi<mode>): Update accordingly.
1371 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1372 Alan Hayward <alan.hayward@arm.com>
1373 David Sherwood <david.sherwood@arm.com>
1375 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1376 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1377 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1378 of units rather than the mode.
1379 * config/aarch64/iterators.md (nunits): New mode attribute.
1380 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1381 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1382 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1383 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1384 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1385 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1386 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1387 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1388 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1389 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1390 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1391 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1392 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1393 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1394 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1395 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1396 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1397 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1398 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1399 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1400 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1401 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1402 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1403 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1404 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1405 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1406 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1407 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1408 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1409 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1410 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1411 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1412 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1413 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1414 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1415 (aarch64_simd_vec_setv2di): Likewise.
1417 2017-11-06 Carl Love <cel@us.ibm.com>
1419 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1421 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1422 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1424 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1426 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1427 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1428 Add power 8 macro expansions.
1429 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1430 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1431 power 8 instructions. (VSX_XXBR): Add iterator.
1433 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1435 * config/arm/arm.md (predicable_short_it): Change default to "no",
1436 improve documentation, remove uses that are identical to the default.
1437 (enabled_for_depr_it): Rename to enabled_for_short_it.
1438 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1439 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1440 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1441 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1442 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1444 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1447 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1448 float128 helper macros, which are no longer used after deleting
1449 the old 'q' built-in functions, and moving the round to odd
1450 built-in functions to being special built-in functions.
1451 (BU_FLOAT128_2): Likewise.
1452 (BU_FLOAT128_1_HW): Likewise.
1453 (BU_FLOAT128_2_HW): Likewise.
1454 (BU_FLOAT128_3_HW): Likewise.
1455 (FABSQ): Delete old 'q' built-in functions.
1456 (COPYSIGNQ): Likewise.
1457 (SQRTF128_ODD): Move round to odd built-in functions to be
1458 special built-in functions, so that we can handle
1459 -mabi=ieeelongdouble.
1460 (TRUNCF128_ODD): Likewise.
1461 (ADDF128_ODD): Likewise.
1462 (SUBF128_ODD): Likewise.
1463 (MULF128_ODD): Likewise.
1464 (DIVF128_ODD): Likewise.
1465 (FMAF128_ODD): Likewise.
1466 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1467 built-in names to 'f128'.
1468 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1469 old 'q' built-in functions, as the machine independent code for
1470 'f128' built-in functions handles this.
1471 (rs6000_expand_builtin): Add expansion for float128 round to odd
1472 functions, keying off on -mabi=ieeelongdouble of whether to use
1473 the KFmode or TFmode variant.
1474 (rs6000_init_builtins): Initialize the _Float128 round to odd
1476 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1477 _Float128 'q' built-in functions are now mapped into the new
1478 'f128' built-in functions.
1480 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1482 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1483 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1485 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1488 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1489 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1490 * toplev.c: Include "ipa-fnsummary.h".
1491 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1493 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1495 PR tree-optimization/82838
1496 * gimple-ssa-store-merging.c
1497 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1498 on a separate gimple_seq which is then appended to seq.
1500 2017-11-06 Jeff Law <law@redhat.com>
1503 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1504 (get_probe_interval): New functions.
1505 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1506 (ix86_adjust_stack_and_probe): Likewise.
1507 (output_adjust_stack_and_probe): Likewise.
1508 (ix86_emit_probe_stack_range): Likewise.
1509 (ix86_expand_prologue): Likewise.
1511 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1513 PR tree-optimization/82816
1514 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1515 if the modes of the two types are the same.
1516 (convert_plusminus_to_widen): Likewise.
1518 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1520 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1522 (usadv16qi): New define_expand.
1523 (usadv8hi): New define_expand.
1525 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1528 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1530 (inline_transform): Likewise.
1531 * predict.c (counts_to_freqs): Remove useless conditional.
1532 * profile-count.h (profile_count::apply_scale): Move sanity check.
1533 * tree-inline.c (copy_bb): Always scale.
1534 (copy_cfg_body): Likewise.
1536 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1539 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1541 (*cmp_ite0): Add enabled_for_depr_it attribute.
1542 (*cmp_ite1): Likewise.
1544 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1546 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1549 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1551 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1552 return the value by pointer.
1553 (extract_range_from_multiplicative_op_1): Update accordingly.
1554 Return as soon as an operation fails.
1556 2017-11-05 Tom de Vries <tom@codesourcery.com>
1559 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1560 (DEF_SANITIZER_BUILTIN): ... here.
1561 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1562 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1564 2017-11-05 Tom de Vries <tom@codesourcery.com>
1567 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1569 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1570 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1571 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1573 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1574 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1575 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1577 2017-11-05 Tom de Vries <tom@codesourcery.com>
1580 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1583 2017-11-04 Michael Clark <michaeljclark@mac.com>
1585 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1586 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1588 (*addsi3_extended): Likewise.
1589 (*addsi3_extended2): Likewise.
1590 (<optab>si3): Likewise.
1591 (<optab>di3): Likewise.
1592 (<optab><mode>3): Likewise.
1593 (<*optabe>si3_internal): Likewise.
1594 (zero_extendqi<SUPERQI:mode>2): Likewise.
1595 (*add<mode>hi3): Likewise.
1596 (*xor<mode>hi3): Likewise.
1597 (<optab>di3): Likewise.
1598 (*<optab>si3_extend): Likewise.
1599 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1600 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1601 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1603 2017-11-04 Andrew Waterman <andrew@sifive.com>
1605 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1606 TARGET_STRICT_ALIGN based upon -mtune argument.
1608 2017-11-04 Andrew Waterman <andrew@sifive.com>
1610 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1612 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1614 config/i386/i386.c (choose_basereg): Use optional scratch
1615 register and add assertion.
1616 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1617 needed, and don't allocate stack.
1618 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1619 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1620 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1622 2017-11-03 Jeff Law <law@redhat.com>
1624 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1625 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1626 to probe at the start of a noreturn function.
1628 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1630 PR tree-optimization/78821
1631 * gimple-ssa-store-merging.c: Update the file comment.
1632 (MAX_STORE_ALIAS_CHECKS): Define.
1633 (struct store_operand_info): New type.
1634 (store_operand_info::store_operand_info): New constructor.
1635 (struct store_immediate_info): Add rhs_code and ops data members.
1636 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1637 and op1r arguments to the ctor, initialize corresponding data members.
1638 (struct merged_store_group): Add load_align_base and load_align
1640 (merged_store_group::merged_store_group): Initialize them.
1641 (merged_store_group::do_merge): Update them.
1642 (merged_store_group::apply_stores): Pick the constant for
1643 encode_tree_to_bitpos from one of the two operands, or skip
1644 encode_tree_to_bitpos if neither operand is a constant.
1645 (class pass_store_merging): Add process_store method decl. Remove
1646 bool argument from terminate_all_aliasing_chains method decl.
1647 (pass_store_merging::terminate_all_aliasing_chains): Remove
1648 var_offset_p argument and corresponding handling.
1649 (stmts_may_clobber_ref_p): New function.
1650 (compatible_load_p): New function.
1651 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1652 if there is overlap and rhs_code is not INTEGER_CST. For
1653 non-overlapping stores terminate group if rhs is not mergeable.
1654 (get_alias_type_for_stmts): Change first argument from
1655 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1656 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1657 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1659 (get_location_for_stmts): Change first argument from
1660 auto_vec<gimple *> & to vec<gimple *> &.
1661 (struct split_store): Remove orig_stmts data member, add orig_stores.
1662 (split_store::split_store): Create orig_stores rather than orig_stmts.
1663 (find_constituent_stmts): Renamed to ...
1664 (find_constituent_stores): ... this. Change second argument from
1665 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1666 to info structures rather than the statements.
1667 (split_group): Rename ALLOW_UNALIGNED argument to
1668 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1669 it. Adjust find_constituent_stores caller.
1670 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1671 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1672 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1673 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1674 (mem_valid_for_store_merging): New function.
1675 (handled_load): New function.
1676 (pass_store_merging::process_store): New method.
1677 (pass_store_merging::execute): Use process_store method. Adjust
1678 terminate_all_aliasing_chains caller.
1680 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1682 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1683 Return true for more constants, symbols and label references.
1684 (aarch64_valid_floating_const): Remove unused function.
1686 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1689 * config/aarch64/aarch64.c (aarch64_layout_frame):
1690 Undo forcing of LR at bottom of frame.
1692 2017-11-03 Jeff Law <law@redhat.com>
1695 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1696 for int_registers_saved.
1698 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1699 extracted from tree-ssa-dom.c.
1700 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1701 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1702 (record_equivalences_from_incoming_edge): Add additional argument
1703 to single_pred_edge_ignoring_loop_edges call.
1704 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1705 (uncprop_dom_walker::before_dom_children): Add additional argument
1706 to single_pred_edge_ignoring_loop_edges call.
1707 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1708 single_pred_edge_ignoring_loop_edges rather than open coding.
1709 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1711 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1713 * match.pd (-(-A)): Rewrite.
1715 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1717 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1718 (rs6000_emit_int_cmove): New declaration.
1719 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1720 (rs6000_emit_sISEL): Delete.
1721 (rs6000_emit_int_cmove): Make non-static.
1722 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1723 instead of rs6000_emit_sISEL.
1725 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1727 * asan.c (create_cond_insert_point): Maintain profile.
1728 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1730 * basic-block.h (struct basic_block_def): Remove frequency.
1731 (EDGE_FREQUENCY): Use to_frequency
1732 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1734 (find_traces): Update to use to_frequency.
1735 (find_traces_1_round): Likewise; use only IPA counts.
1736 (bb_to_key): Likewise.
1737 (connect_traces): Use IPA counts only.
1738 (copy_bb_p): Update to use to_frequency.
1739 (fix_up_crossing_landing_pad): Likewise.
1740 (sanitize_hot_paths): Likewise.
1741 * bt-load.c (basic_block_freq): Likewise.
1742 * cfg.c (init_flow): Set count_max to uninitialized.
1743 (check_bb_profile): Remove frequencies; check counts.
1744 (dump_bb_info): Do not dump frequencies.
1745 (update_bb_profile_for_threading): Update counts only.
1746 (scale_bbs_frequencies_int): Likewise.
1747 (MAX_SAFE_MULTIPLIER): Remove.
1748 (scale_bbs_frequencies_gcov_type): Update counts only.
1749 (scale_bbs_frequencies_profile_count): Update counts only.
1750 (scale_bbs_frequencies): Update counts only.
1751 * cfg.h (struct control_flow_graph): Add count-max.
1752 (update_bb_profile_for_threading): Update prototype.
1753 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1754 (find_many_sub_basic_blocks): Likewise.
1755 * cfgcleanup.c (try_forward_edges): Likewise.
1756 (try_crossjump_to_edge): Likewise.
1757 * cfgexpand.c (expand_gimple_cond): Likewise.
1758 (expand_gimple_tailcall): Likewise.
1759 (construct_init_block): Likewise.
1760 (construct_exit_block): Likewise.
1761 * cfghooks.c (verify_flow_info): Check consistency of counts.
1762 (dump_bb_for_graph): Do not dump frequencies.
1763 (split_block_1): Do not update frequencies.
1764 (split_edge): Do not update frequencies.
1765 (make_forwarder_block): Do not update frequencies.
1766 (duplicate_block): Do not update frequencies.
1767 (account_profile_record): Do not update frequencies.
1768 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1769 for global heuristics.
1770 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1771 (expected_loop_iterations_unbounded): Use counts only.
1772 * cfgloopmanip.c (scale_loop_profile): Simplify.
1773 (create_empty_loop_on_edge): Simplify
1775 (duplicate_loop_to_header_edge): Simplify
1776 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1777 (update_br_prob_note): Take care of removing note when profile
1779 (relink_block_chain): Do not dump frequency.
1780 (rtl_account_profile_record): Use to_frequency.
1781 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1782 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1783 (cgraph_update_edges_for_call_stmt_node): Likewise.
1784 (cgraph_edge::verify_count_and_frequency): Update.
1785 (cgraph_node::verify_node): Temporarily disable frequency verification.
1786 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1787 to_cgraph_frequency.
1788 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1789 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1791 (cgraph_node::expand_thunk): Update profile.
1792 * except.c (dw2_build_landing_pads): Do not update frequency.
1793 * final.c (compute_alignments): Use to_frequency.
1794 (dump_basic_block_info): Do not dump frequency.
1795 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1796 (dump_gimple_bb_header): Do not dump frequency.
1797 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1799 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1800 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1801 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1802 (init_before_recovery): Do not update frequency.
1803 (sched_create_recovery_edges): Do not update frequency.
1804 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1805 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1806 (ipa_cp_c_finalize): Set max_count to uninitialized.
1807 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1808 (param_change_prob): Use counts.
1809 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1811 * ipa-split.c (consider_split): Use to_frequency.
1812 (split_function): Use to_frequency.
1813 * ira-build.c (loop_compare_func): Likewise.
1814 (mark_loops_for_removal): Likewise.
1815 (mark_all_loops_for_removal): Likewise.
1816 * loop-doloop.c (doloop_modify): Do not update frequency.
1817 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1819 * lto-streamer-in.c (input_function): Update count_max.
1820 * omp-expand.c (expand_omp_taskreg): Update count_max.
1821 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1822 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1823 (maybe_hot_count_p): Use ipa counts only.
1824 (maybe_hot_bb_p): Simplify.
1825 (maybe_hot_edge_p): Simplify.
1826 (probably_never_executed): Do not take frequency argument.
1827 (probably_never_executed_bb_p): Do not pass frequency.
1828 (probably_never_executed_edge_p): Likewise.
1829 (combine_predictions_for_bb): Check that profile is nonzero.
1830 (propagate_freq): Do not set frequency.
1831 (drop_profile): Simplify.
1832 (counts_to_freqs): Simplify.
1833 (expensive_function_p): Use to_frequency.
1834 (propagate_unlikely_bbs_forward): Simplify.
1835 (determine_unlikely_bbs): Simplify.
1836 (estimate_bb_frequencies): Add hack to silence graphite issues.
1837 (compute_function_frequency): Use ipa counts.
1838 (pass_profile::execute): Update.
1839 (rebuild_frequencies): Use counts only.
1840 (force_edge_cold): Use counts only.
1841 * profile-count.c (profile_count::dump): Dump new count types.
1842 (profile_count::differs_from_p): Check compatiblity.
1843 (profile_count::to_frequency): New function.
1844 (profile_count::to_cgraph_frequency): New function.
1845 * profile-count.h (struct function): Declare.
1846 (enum profile_quality): Add profile_guessed_local and
1847 profile_guessed_global0.
1848 (class profile_proability): Decrease number of bits to 29;
1849 update from_reg_br_prob_note and to_reg_br_prob_note.
1850 (class profile_count: Update comment; decrease number of bits
1851 to 61. Check compatibility.
1852 (profile_count::compatible_p): New private member function.
1853 (profile_count::ipa_p): New member function.
1854 (profile_count::operator<): Handle global zero correctly.
1855 (profile_count::operator>): Handle global zero correctly.
1856 (profile_count::operator<=): Handle global zero correctly.
1857 (profile_count::operator>=): Handle global zero correctly.
1858 (profile_count::nonzero_p): New member function.
1859 (profile_count::force_nonzero): New member function.
1860 (profile_count::max): New member function.
1861 (profile_count::apply_scale): Handle IPA scalling.
1862 (profile_count::guessed_local): New member function.
1863 (profile_count::global0): New member function.
1864 (profile_count::ipa): New member function.
1865 (profile_count::to_frequency): Declare.
1866 (profile_count::to_cgraph_frequency): Declare.
1867 * profile.c (OVERLAP_BASE): Delete.
1868 (compute_frequency_overlap): Delete.
1869 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1870 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1871 * sched-ebb.c (rank): Use counts only.
1872 * shrink-wrap.c (handle_simple_exit): Use counts only.
1873 (try_shrink_wrapping): Use counts only.
1874 (place_prologue_for_one_component): Use counts only.
1875 * tracer.c (find_best_predecessor): Use to_frequency.
1876 (find_trace): Use to_frequency.
1877 (tail_duplicate): Use to_frequency.
1878 * trans-mem.c (expand_transaction): Do not update frequency.
1879 * tree-call-cdce.c: Do not update frequency.
1880 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1881 (gimple_merge_blocks): Likewise.
1882 (gimple_split_edge): Likewise.
1883 (gimple_duplicate_sese_region): Likewise.
1884 (gimple_duplicate_sese_tail): Likewise.
1885 (move_sese_region_to_fn): Likewise.
1886 (gimple_account_profile_record): Likewise.
1887 (insert_cond_bb): Likewise.
1888 * tree-complex.c (expand_complex_div_wide): Likewise.
1889 * tree-eh.c (lower_resx): Update profile.
1890 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1892 (initialize_cfun): Do not initialize frequencies
1893 (freqs_to_counts): Delete.
1894 (copy_cfg_body): Ignore count parameter.
1895 (copy_body): Update.
1896 (expand_call_inline): Update count_max.
1897 (optimize_inline_calls): Update count_max.
1898 (tree_function_versioning): Update count_max.
1899 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1900 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1902 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1903 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1904 (try_peel_loop): Likewise.
1905 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1907 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1908 (tree_transform_and_unroll_loop): Do not use frequencies
1909 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1910 Use reliable prediction only.
1911 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1912 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1913 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1914 probability scaling.
1915 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1916 not update frequency
1917 (any_remaining_duplicated_blocks): Likewise.
1918 (update_profile): Likewise.
1919 (estimated_freqs_path): Delete.
1920 (freqs_to_counts_path): Delete.
1921 (clear_counts_path): Delete.
1922 (ssa_fix_duplicate_block_edges): Likewise.
1923 (duplicate_thread_path): Likewise.
1924 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1925 * tree-tailcall.c (decrease_profile): Do not update frequency.
1926 (eliminate_tail_call): Likewise.
1927 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1928 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1929 (optimize_mask_stores): Likewise.
1930 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1931 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1932 (ubsan_expand_ptr_ifn): Update profile.
1933 * value-prof.c (gimple_ic): Simplify.
1934 * value-prof.h (gimple_ic): Update prototype.
1935 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1936 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1938 (want_inline_self_recursive_call_p): Likewise.
1939 (resolve_noninline_speculation): Only cummulate defined counts.
1940 (inline_small_functions): Use nonzero_p.
1941 (ipa_inline): Do not access freed node.
1943 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1945 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1946 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1948 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1950 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1951 non-legitimate address.
1953 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1955 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1956 (*lt0_<mode>di, *lt0_<mode>si): New.
1958 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1960 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1961 TARGET_PAIRED_FLOAT.
1963 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1964 Jim Wilson <jim.wilson@linaro.org>
1966 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1967 * config/aarch64/aarch64-tune.md: Regenerated.
1968 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1969 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1971 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1973 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1974 (arc_expand_prologue): Restore blink for millicode.
1975 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1977 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1980 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1981 gen_vec_duplicate after forcing the scalar into a register.
1983 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1985 * combine (try_combine): Print the insns input to try_combine to the
1988 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1991 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1992 Remove second argument from aarch64_process_target_attr call.
1993 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1995 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1997 (aarch64_handle_attr_arch): Remove second argument.
1998 (aarch64_handle_attr_cpu): Ditto.
1999 (aarch64_handle_attr_tune): Ditto.
2000 (aarch64_handle_attr_isa_flags): Ditto.
2001 (aarch64_process_one_target_attr): Ditto.
2002 (aarch64_process_target_attr): Ditto.
2003 (aarch64_option_valid_attribute_p): Remove second argument.
2004 on aarch64_process_target_attr call.
2006 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2008 * diagnostic.c: Include "selftest-diagnostic.h".
2009 (selftest::assert_location_text): New function.
2010 (selftest::test_diagnostic_get_location_text): New function.
2011 (selftest::diagnostic_c_tests): Call it.
2013 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2015 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2016 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2017 (class selftest::test_diagnostic_context): Move to...
2018 * selftest-diagnostic.c: New file.
2019 * selftest-diagnostic.h: New file.
2021 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2023 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2024 offset range for FT32B.
2025 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2026 * config/ft32/ft32.md: Add TARGET_NOPM.
2027 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2028 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2030 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2032 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2034 2017-11-02 Jeff Law <law@redhat.com>
2036 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2037 virtual keyword on FINAL OVERRIDE members.
2039 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2040 virtuals together. Add virtual destructor.
2041 (substitute_and_fold_engine): Similarly.
2043 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2045 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2047 2017-11-02 Richard Biener <rguenther@suse.de>
2049 PR tree-optimization/82795
2050 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2052 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2054 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2056 (gcc_GAS_CHECK_FEATURE): Remove.
2057 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2058 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2059 * configure: Regenerate.
2061 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2063 * config/arc/arc.c (hwloop_optimize): Account for empty
2066 2017-11-02 Richard Biener <rguenther@suse.de>
2069 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2070 Truncate ARRAY_REF index and element size.
2072 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2074 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2076 2017-11-01 Jeff Law <law@redhat.com>
2078 * tree-ssa-ccp.c (ccp_folder): New class derived from
2079 substitute_and_fold_engine.
2080 (ccp_folder::get_value): New member function.
2081 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2082 (ccp_fold_stmt): Remove prototype.
2083 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2084 * tree-ssa-copy.c (copy_folder): New class derived from
2085 substitute_and_fold_engine.
2086 (copy_folder::get_value): Renamed from get_value.
2087 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2088 * tree-vrp.c (vrp_folder): New class derived from
2089 substitute_and_fold_engine.
2090 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2091 (vrp_folder::get_value): New member function.
2092 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2093 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2094 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2095 provide a class interface to folder/substitute routines.
2096 (ssa_prop_fold_stmt_fn): Remove typedef.
2097 (ssa_prop_get_value_fn): Likewise.
2098 (subsitute_and_fold): Remove prototype.
2099 (replace_uses_in): Likewise.
2100 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2101 Renamed from replace_uses_in. Call the virtual member function
2102 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2103 (substitute_and_fold_dom_walker): Remove initialization of
2104 data member entries for calbacks. Add substitute_and_fold_engine
2105 member and initialize it.
2106 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2107 member functions for get_value, replace_phi_args_in c
2108 replace_uses_in, and fold_stmt calls.
2109 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2110 substitute_and_fold. Remove assert. Update ctor call.
2112 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2113 (ssa_prop_visit_phi_fn): Likewise.
2114 (class ssa_propagation_engine): New class to provide an interface
2116 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2118 (ssa_prop_visit_phi): Likewise.
2119 (ssa_propagation_engine::simulate_stmt): Moved into class.
2120 Call visit_phi/visit_stmt from the class rather than via
2121 file scoped static variables.
2122 (ssa_propagation_engine::simulate_block): Moved into class.
2123 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2124 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2125 set file scoped statics for the visit_stmt/visit_phi callbacks.
2126 * tree-complex.c (complex_propagate): New class derived from
2127 ssa_propagation_engine.
2128 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2129 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2130 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2132 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2133 ssa_propagation_engine.
2134 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2135 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2136 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2137 * tree-ssa-copy.c (copy_prop): New class derived from
2138 ssa_propagation_engine.
2139 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2140 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2141 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2142 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2143 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2144 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2145 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2147 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2149 PR rtl-optimization/82778
2150 PR rtl-optimization/82597
2151 * compare-elim.c (struct comparison): Add in_a_setter field.
2152 (find_comparison_dom_walker::before_dom_children): Remove killed
2153 bitmap and df_simulate_find_defs call, instead walk the defs.
2154 Compute last_setter and initialize in_a_setter. Merge definitions
2155 with first initialization for a few variables.
2156 (try_validate_parallel): Use insn_invalid_p instead of
2157 recog_memoized. Return insn rather than just the pattern.
2158 (try_merge_compare): Fix up comment. Don't uselessly test if
2159 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2161 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2164 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2165 Alan Hayward <alan.hayward@arm.com>
2166 David Sherwood <david.sherwood@arm.com>
2168 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2169 aarch64_hard_regno_nregs to get the number of registers
2172 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2173 Alan Hayward <alan.hayward@arm.com>
2174 David Sherwood <david.sherwood@arm.com>
2176 * config/aarch64/constraints.md (Upl): Rename to...
2178 * config/aarch64/aarch64.md
2179 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2182 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2183 Alan Hayward <alan.hayward@arm.com>
2184 David Sherwood <david.sherwood@arm.com>
2186 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2187 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayward <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2195 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2196 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2198 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2199 of a single element.
2200 * config/aarch64/iterators.md: Add a comment above the permute
2201 unspecs to say that they are generated directly by
2202 aarch64_expand_vec_perm_const.
2203 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2205 2017-11-01 Nathan Sidwell <nathan@acm.org>
2207 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2209 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2211 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2212 and medany code models, and describe what they do.
2214 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2216 Revert accidental duplicate:
2218 * combine.c (can_change_dest_mode): Reject changes in
2219 REGMODE_NATURAL_SIZE.
2221 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2223 PR rtl-optimization/64682
2224 PR rtl-optimization/69567
2225 PR rtl-optimization/69737
2226 PR rtl-optimization/82683
2227 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2228 register mentioned in the note, drop the note, unless it came from I3,
2229 in which case it should go to I3 again.
2231 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2233 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2234 and return false if not.
2235 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2237 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2239 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2240 range is known to be empty.
2242 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2243 Alan Hayward <alan.hayward@arm.com>
2244 David Sherwood <david.sherwood@arm.com>
2246 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2247 and CONST_VECTOR_NUNITS instead of computing the number of units from
2248 the byte sizes of the vector and element.
2249 (simplify_binary_operation_1): Likewise.
2250 (simplify_const_binary_operation): Likewise.
2251 (simplify_ternary_operation): Likewise.
2253 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2254 Alan Hayward <alan.hayward@arm.com>
2255 David Sherwood <david.sherwood@arm.com>
2257 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2258 (int_mem_offset): ...this new function.
2259 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2260 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2261 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2264 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2265 Alan Hayward <alan.hayward@arm.com>
2266 David Sherwood <david.sherwood@arm.com>
2268 * lower-subreg.c (interesting_mode_p): New function.
2269 (compute_costs, find_decomposable_subregs, decompose_register)
2270 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2271 (resolve_clobber, dump_choices): Use it.
2273 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2274 Alan Hayward <alan.hayward@arm.com>
2275 David Sherwood <david.sherwood@arm.com>
2277 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
2284 truncations. Check the precision rather than the size.
2286 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2287 Alan Hayward <alan.hayward@arm.com>
2288 David Sherwood <david.sherwood@arm.com>
2290 * machmode.h (is_narrower_int_mode): New function
2291 * optabs.c (expand_float, expand_fix): Use it.
2292 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
2299 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
2306 (convert_optab_handler): ...here.
2307 (widening_optab_handler): Delete.
2308 (find_widening_optab_handler): Remove permit_non_widening parameter.
2309 (find_widening_optab_handler_and_mode): Likewise. Provide an
2310 override that operates on mode class wrappers.
2311 * optabs-query.c (widening_optab_handler): Delete.
2312 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2313 parameter. Assert that the two modes are the same class and that
2314 the "from" mode is narrower than the "to" mode. Use
2315 convert_optab_handler instead of widening_optab_handler.
2316 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2317 instead of widening_optab_handler.
2318 * expr.c (expand_expr_real_2): Update calls to
2319 find_widening_optab_handler.
2320 * optabs.c (expand_widen_pattern_expr): Likewise.
2321 (expand_binop_directly): Take the insn_code as a parameter.
2322 (expand_binop): Only call find_widening_optab_handler for
2323 conversion optabs; use optab_handler otherwise. Update calls
2324 to find_widening_optab_handler and expand_binop_directly.
2325 Use convert_optab_handler instead of widening_optab_handler.
2326 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2327 find_widening_optab_handler and use scalar_mode rather than
2329 (convert_plusminus_to_widen): Likewise.
2331 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2332 Alan Hayward <alan.hayward@arm.com>
2333 David Sherwood <david.sherwood@arm.com>
2335 * machmode.h (fixed_size_mode): New class.
2336 * rtl.h (get_pool_mode): Return fixed_size_mode.
2337 * gengtype.c (main): Add fixed_size_mode.
2338 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2339 (get_raw_arg_mode): Likewise.
2340 * doc/tm.texi: Regenerate.
2341 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2342 * targhooks.c (default_get_reg_raw_mode): Likewise.
2343 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2344 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2345 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2346 (msp430_get_raw_result_mode): Likewise.
2347 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2348 * dbxout.c (dbxout_parms): Require fixed-size modes.
2349 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2350 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2351 * omp-low.c (lower_oacc_reductions): Likewise.
2352 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2353 (simplify_subreg): Update accordingly.
2354 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2355 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2356 that aren't fixed-size.
2357 (get_pool_mode): Return a fixed_size_mode.
2358 (output_constant_pool_2): Take a fixed_size_mode.
2360 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2361 Alan Hayward <alan.hayward@arm.com>
2362 David Sherwood <david.sherwood@arm.com>
2364 * doc/rtl.texi (vec_series): Document.
2365 (const): Say that the operand can be a vec_series.
2366 * rtl.def (VEC_SERIES): New rtx code.
2367 * rtl.h (const_vec_series_p_1): Declare.
2368 (const_vec_series_p): New function.
2369 * emit-rtl.h (gen_const_vec_series): Declare.
2370 (gen_vec_series): Likewise.
2371 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2372 (gen_vec_series): Likewise.
2373 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2374 * simplify-rtx.c (simplify_unary_operation): Handle negations
2376 (simplify_binary_operation_series): New function.
2377 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2378 (test_vector_ops_series): New function.
2379 (test_vector_ops): Call it.
2380 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2381 gen_const_vec_series.
2382 (altivec_lvsr): Likewise.
2383 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2386 Alan Hayward <alan.hayward@arm.com>
2387 David Sherwood <david.sherwood@arm.com>
2389 * doc/rtl.texi (const): Update description of address constants.
2390 Say that vector constants are allowed too.
2391 * common.md (E, F): Use CONSTANT_P instead of checking for
2393 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2394 checking for CONST_VECTOR.
2395 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2397 * expr.c (expand_expr_real_2): Check for vector modes instead
2398 of checking for CONST_VECTOR.
2399 * rtl.h (const_vec_p): New function.
2400 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2401 (unwrap_const_vec_duplicate): Handle them here too.
2403 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2404 David Malcolm <dmalcolm@redhat.com>
2405 Alan Hayward <alan.hayward@arm.com>
2406 David Sherwood <david.sherwood@arm.com>
2408 * rtl.h (vec_duplicate_p): New function.
2409 * selftest-rtl.c (assert_rtx_eq_at): New function.
2410 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2411 (assert_rtx_eq_at): Declare.
2412 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2413 * selftest-run-tests.c (selftest::run_tests): Call it.
2414 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2415 (simplify_unary_operation_1): Recursively handle vector duplicates.
2416 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2418 (simplify_subreg): Handle subregs of vector duplicates.
2419 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2420 (selftest::simplify_rtx_c_tests): New functions.
2422 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2423 Alan Hayward <alan.hayward@arm.com>
2424 David Sherwood <david.sherwood@arm.com>
2426 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2427 (gen_vec_duplicate): Likewise.
2428 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2430 (gen_const_vector): ...here.
2431 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2432 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2433 whose elements are all equal.
2434 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2435 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2436 (simplify_relational_operation): Likewise.
2437 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2439 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2440 (aarch64_expand_vector_init): Likewise.
2441 * config/arm/arm.c (neon_vdup_constant): Likewise.
2442 (neon_expand_vector_init): Likewise.
2443 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2444 (arm_block_set_unaligned_vect): Likewise.
2445 (arm_block_set_aligned_vect): Likewise.
2446 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2447 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2448 (expand_vec_perm_even_odd_pack): Likewise.
2449 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2450 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2451 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2452 gen_const_vec_duplicate.
2453 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2454 * config/mips/mips.c (mips_gen_const_int_vector): Use
2455 gen_const_vec_duplicate.
2456 (mips_expand_vector_init): Use CONST0_RTX.
2457 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2458 (define_split): Use gen_const_vec_duplicate.
2459 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2460 (define_split): Use gen_const_vec_duplicate.
2461 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2462 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2463 * config/spu/spu.c (spu_const): Likewise.
2465 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2466 Alan Hayward <alan.hayward@arm.com>
2467 David Sherwood <david.sherwood@arm.com>
2469 * combine.c (can_change_dest_mode): Reject changes in
2470 REGMODE_NATURAL_SIZE.
2472 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2474 * configure.ac (--enable-libssp): New.
2475 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2476 trying to determine target-specific default. Adjust indentation.
2477 * configure: Regenerated.
2478 * doc/install.texi (Configuration): Expand --disable-libssp
2481 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2483 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2486 2017-10-31 Martin Jambor <mjambor@suse.cz>
2489 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2491 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2493 * auto-profile.c (autofdo_source_profile::read): Use
2494 UNKNOWN_LOCATION rather than 0.
2495 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2496 (warning_at): ...this overload.
2497 (warning_at_rich_loc_n): Rename to...
2498 (warning_n): ...this overload.
2499 (error_at_rich_loc): Rename to...
2500 (error_at): ...this overload.
2501 (pedwarn_at_rich_loc): Rename to...
2502 (pedwarn): ...this overload.
2503 (permerror_at_rich_loc): Rename to...
2504 (permerror): ...this overload.
2505 (inform_at_rich_loc): Rename to...
2506 (inform): ...this overload.
2507 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2508 (diagnostic_n_impl_richloc): Rename to...
2509 (diagnostic_n_impl): ...this rich_location *-based decl.
2510 (inform_at_rich_loc): Rename to...
2511 (inform): ...this, and add an assertion.
2512 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2513 (warning_at_rich_loc): Rename to...
2514 (warning_at): ...this, and add an assertion.
2515 (warning_at_rich_loc_n): Rename to...
2516 (warning_n): ...this, and add an assertion.
2517 (warning_n): Update location_t-based implementation for removal of
2518 location_t-based diagnostic_n_impl.
2519 (pedwarn_at_rich_loc): Rename to...
2520 (pedwarn): ...this, and add an assertion.
2521 (permerror_at_rich_loc): Rename to...
2522 (permerror): ...this, and add an assertion.
2523 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2524 (error_at_rich_loc): Rename to...
2525 (error_at): ...this, and add an assertion.
2526 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2527 (driver::do_spec_on_infiles): Likewise.
2528 * substring-locations.c (format_warning_va): Update for renaming
2529 of inform_at_rich_loc.
2531 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2533 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2534 _Float<N>X built-in functions so that the variant without the
2535 "__builtin_" prefix is only enabled for the GNU C and Objective C
2536 languages when they are in non-strict ANSI/ISO mode.
2537 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2538 * target.def (floatn_builtin_p): Add a target hook to control
2539 whether _Float<N> and _Float<N>X built-in functions without the
2540 "__builtin_" prefix are enabled, and return true for C and
2541 Objective C in the default hook. Include langhooks.h in
2543 * targhooks.h (default_floatn_builtin_p): Likewise.
2544 * targhooks.c (default_floatn_builtin_p): Likewise.
2545 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2546 floatn_builtin_p target hook.
2547 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2549 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2550 Eric Botcazou <ebotcazou@adacore.com>
2552 PR rtl-optimization/81803
2553 * lra-constraints.c (curr_insn_transform): Also reload the whole
2554 register for a strict subreg no wider than a word if this is for
2555 a WORD_REGISTER_OPERATIONS target.
2557 2017-10-31 Jason Merrill <jason@redhat.com>
2559 * gdbinit.in: Skip over inlines from timevar.h.
2561 2017-10-31 Martin Liska <mliska@suse.cz>
2563 * doc/gcov.texi: Document new option.
2564 * gcov.c (print_usage): Likewise print it.
2565 (process_args): Support the argument.
2566 (format_count): New function.
2567 (format_gcov): Use the function.
2569 2017-10-31 Martin Liska <mliska@suse.cz>
2571 * gcov.c (struct name_map): do not use typedef.
2572 Define operator== and operator<.
2573 (name_search): Remove.
2574 (name_sort): Remove.
2575 (main): Do not allocate names.
2576 (process_file): Add vertical space.
2577 (generate_results): Use std::find.
2578 (release_structures): Do not release memory.
2579 (find_source): Use std::find.
2581 2017-10-31 Martin Liska <mliska@suse.cz>
2583 * gcov.c (struct line_info): Remove it's typedef.
2584 (line_info::line_info): Add proper ctor.
2585 (line_info::has_block): Do not use a typedef.
2586 (struct source_info): Do not use typedef.
2587 (circuit): Likewise.
2588 (get_cycles_count): Likewise.
2589 (output_intermediate_file): Iterate via vector iterator.
2590 (add_line_counts): Use std::vector methods.
2591 (accumulate_line_counts): Likewise.
2592 (output_lines): Likewise.
2594 2017-10-31 Martin Liska <mliska@suse.cz>
2596 * gcov.c (struct source_info): Remove typedef.
2597 (source_info::source_info): Add proper ctor.
2598 (accumulate_line_counts): Use struct, not it's typedef.
2599 (output_gcov_file): Likewise.
2600 (output_lines): Likewise.
2601 (main): Do not allocate an array.
2602 (output_intermediate_file): Use size of vector container.
2603 (process_file): Resize the vector.
2604 (generate_results): Do not preallocate, use newly added vector
2606 (release_structures): Do not release sources.
2607 (find_source): Use vector methods.
2608 (add_line_counts): Do not use typedef.
2610 2017-10-31 Martin Liska <mliska@suse.cz>
2612 * doc/gcov.texi: Document that.
2613 * gcov.c (add_line_counts): Mark lines with a non-executed
2615 (output_line_beginning): Handle such lines.
2616 (output_lines): Pass new argument.
2617 (output_intermediate_file): Print it in intermediate format.
2619 2017-10-31 Martin Liska <mliska@suse.cz>
2621 * color-macros.h: New file.
2622 * diagnostic-color.c: Factor out color related to macros to
2624 * doc/gcov.texi: Document -k option.
2625 * gcov.c (INCLUDE_STRING): Include string.h.
2626 (print_usage): Add -k option.
2627 (process_args): Parse it.
2628 (pad_count_string): New function.
2629 (output_line_beginning): Likewise.
2630 (DEFAULT_LINE_START): New macro.
2631 (output_lines): Support color output.
2633 2017-10-31 Martin Liska <mliska@suse.cz>
2635 PR gcov-profile/82633
2636 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2637 their interaction with GCOV infrastructure.
2638 * configure.ac: Add -fkeep-{inline,static}-functions to
2640 * configure: Regenerate.
2642 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2645 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2647 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2650 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2651 into a register if it does not fit into an immediate offset field.
2653 2017-10-31 Olivier Hainque <hainque@adacore.com>
2655 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2657 2017-10-31 Julia Koval <julia.koval@intel.com>
2659 * config.gcc: Add gfniintrin.h.
2660 * config/i386/gfniintrin.h: New.
2661 * config/i386/i386-builtin-types.def
2662 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2663 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2664 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2665 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2666 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2667 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2668 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2669 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2670 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2671 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2672 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2673 * config/i386/immintrin.h: Include gfniintrin.h.
2674 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2676 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2678 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2680 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2682 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2683 (arm_ashldi3_1bit): Remove pattern.
2684 (ashrdi3): Remove shift by 1 expansion.
2685 (arm_ashrdi3_1bit): Remove pattern.
2686 (lshrdi3): Remove shift by 1 expansion.
2687 (arm_lshrdi3_1bit): Remove pattern.
2688 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2689 cost of ashldi3 by 1.
2690 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2691 (<shift>di3_neon): Likewise.
2693 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2695 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2696 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2697 and (*aarch64_simd_mov<VQ:mode>).
2698 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2699 pattern alternative.
2700 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2701 attributes to match pattern alternative.
2703 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2705 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2706 * config/rs6000/emmintrin.h: New file.
2707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2709 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2711 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2712 * (movdi_vfp_cortexa8): Remove pattern.
2714 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2716 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2718 (Specific, alpha*-dec-osf5.1): Remove.
2719 (Specific, mips-sgi-irix5): Remove.
2720 (Specific, mips-sgi-irix6): Remove.
2722 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2725 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2726 arguments to clear_bit_region_be.
2728 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2730 * gimplify.c: Include memmodel.h.
2732 2017-10-30 Martin Jambor <mjambor@suse.cz>
2734 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2735 condition whether loop should be executed at all.
2737 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2739 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2740 gimple folding of vec_madd() intrinsics.
2741 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2742 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2743 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2745 2017-10-30 Richard Biener <rguenther@suse.de>
2747 PR tree-optimization/82762
2749 2017-10-23 Richard Biener <rguenther@suse.de>
2751 PR tree-optimization/82129
2753 2017-08-01 Richard Biener <rguenther@suse.de>
2755 PR tree-optimization/81181
2756 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2757 (compute_antic): ... end of iteration here.
2759 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2761 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2762 -std=iso9899:2017 and -std=gnu17.
2763 * doc/standards.texi (C Language): Document C17 support.
2764 * doc/cpp.texi (Overview): Mention -std=c17.
2765 (Standard Predefined Macros): Document C11 and C17 values of
2766 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2767 * doc/extend.texi (Inline): Do not list individual options for
2768 standards newer than C99.
2769 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2771 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2774 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2776 * asan.c (asan_finish_file): Align asan globals array by shadow
2779 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2782 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2783 (struct store_immediate_info): Add bitregion_start and bitregion_end
2785 (store_immediate_info::store_immediate_info): Add brs and bre
2786 arguments and initialize bitregion_{start,end} from those.
2787 (struct merged_store_group): Add bitregion_start, bitregion_end,
2788 align_base and mask fields. Drop unnecessary struct keyword from
2789 struct store_immediate_info. Add do_merge method.
2790 (clear_bit_region_be): Use memset instead of loop storing zeros.
2791 (merged_store_group::do_merge): New method.
2792 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2793 stores as long as the surrounding bitregions have no gaps.
2794 (merged_store_group::merge_overlapping): Use do_merge.
2795 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2796 is byte aligned, rather than requiring that start and width are
2797 byte aligned. Drop unnecessary struct keyword from
2798 struct store_immediate_info. Allocate and populate also mask array.
2799 Make start of the arrays relative to bitregion_start rather than
2800 start and size them according to bitregion_{end,start} difference.
2801 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2802 struct store_immediate_info.
2803 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2804 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2805 struct keyword from struct store_immediate_info.
2806 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2807 between stores as long as the surrounding bitregions have no gaps.
2809 (struct split_store): Add orig non-static data member.
2810 (split_store::split_store): Initialize orig to false.
2811 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2812 if there is exactly a single original stmt. Change stmts argument
2813 to pointer from reference, if NULL, don't push anything to it. Add
2814 first argument, use it to optimize skipping over orig stmts that
2815 are known to be before bitpos already. Simplify.
2816 (split_group): Return unsigned int count how many stores are or
2817 would be needed rather than a bool. Add allow_unaligned argument.
2818 Change split_stores argument from reference to pointer, if NULL,
2819 only do a dry run computing how many stores would be produced.
2820 Rewritten algorithm to use both alignment and misalign if
2821 !allow_unaligned and handle bitfield stores with gaps.
2822 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2823 from bitregion_start instead of start. Compute allow_unaligned
2824 here, if true, do 2 split_group dry runs to compute which one
2825 produces fewer stores and prefer aligned if equal. Punt if
2826 new count is bigger or equal than original before emitting any
2827 statements, rather than during that. Remove no longer needed
2828 new_ssa_names tracking. Replace num_stmts with
2829 split_stores.length (). Use 32-bit stack allocated entries
2830 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2831 if possible. Handle bitfields with gaps.
2832 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2833 Compute bitregion_{start,end} for the stores and construct
2834 store_immediate_info with that. Formatting fixes.
2836 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2839 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2840 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2842 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2844 * gimplify.c: Include tm_p.h.
2846 * common.opt (gcoff): Re-add as ignored option.
2847 (gcoff1, gcoff2, gcoff3): Likewise.
2849 * Makefile.in (OBJS): Delete sdbout.o.
2850 (GTFILES): Delete $(srcdir)/sdbout.c.
2851 * debug.h: Delete sdb_debug_hooks.
2852 * final.c: Delete sdbout.h include.
2853 (final_scan_insn): Delete SDB_DEBUG check.
2854 (rest_of_clean_state): Likewise.
2855 * output.h: Delete sdb_begin_function_line.
2858 * toplev.c: Delete sdbout.h include.
2859 (process_options): Delete SDB_DEBUG check.
2860 * tree-core.h (tree_type_common): Delete pointer field of
2862 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2863 TYPE_SYMTAB_POINTER.
2864 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2865 (TYPE_SYMTAB_IS_POINTER): Delete.
2866 (TYPE_SYMTAB_IS_DIE): Renumber.
2867 * xcoffout.c: Refer to former sdbout.c file.
2868 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2870 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2871 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2872 (Debugging Options): Delete -gcoff.
2873 (-gstabs): Delete SDB reference.
2875 (-gcoff@var{level}): Delete.
2876 * doc/passes.texi (Debugging information output): Delete SDB and
2877 sdbout.c references.
2878 * doc/tm.texi: Regenerate.
2879 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2880 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2882 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2883 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2884 (SDB_DEBUGGING_INFO): Delete.
2885 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2886 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2887 * target.def (output_source_filename): Delete COFF reference.
2889 * common.opt (gcoff): Delete.
2890 (gxcoff+): Update Negative chain.
2891 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2893 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2894 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2895 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2897 (expand_function_start): Change sdb reference to past tense.
2898 (expand_function_end): Change sdb reference to past tense.
2899 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2900 * opts.c (debug_type_names): Delete coff entry.
2901 (common_handle_option): Delete OPT_gcoff case.
2902 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2904 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2905 * config/cris/cris.h: Delete SDB reference in comment.
2906 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2907 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2908 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2909 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2911 (ix86_expand_prologue): Likewise.
2912 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2913 * config/ia64/ia64.h: Likewise.
2914 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2915 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2917 * config/mmix/mmix.h: Likewise.
2918 * config/nds32/nds32.c: Likewise.
2919 * config/stormy/storym16.h: Likewise.
2920 * config/visium/visium.h: Likewise.
2921 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2923 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2925 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2926 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2927 FRAME_POINTER_REGNUM point at high end of local var area.
2929 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2931 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2932 Move comment around. Do not reset best_edge for a copiable
2933 destination if the copy would cause a partition change.
2934 (better_edge_p): Remove redundant check.
2936 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2938 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2940 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2942 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2943 for math functions that have _Float<N> and _Float<N>X variants.
2944 (mathfn_built_in_2): Add support for math functions that have
2945 _Float<N> and _Float<N>X variants.
2946 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2947 (expand_builtin_mathfn_ternary): Add support for fma with
2948 _Float<N> and _Float<N>X variants.
2949 (expand_builtin): Likewise.
2950 (fold_builtin_3): Likewise.
2951 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2952 create math function _Float<N> and _Float<N>X variants as external
2954 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2955 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2956 the __builtin_ prefix and if not strict ansi, without the prefix.
2957 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2958 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2959 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2960 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2961 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2962 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2963 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2964 function signatures for fma _Float<N> and _Float<N>X variants.
2965 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2966 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2967 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2968 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2969 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2970 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2971 * gencfn-macros.c (print_case_cfn): Add support for math functions
2972 that have _Float<N> and _Float<N>X variants.
2973 (print_define_operator_list): Likewise.
2974 (fltfn_suffixes): Likewise.
2976 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2977 for math functions that have _Float<N> and _Float<N>X variants.
2978 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2979 and _Float<N>X variants.
2980 (COPYSIGN): Likewise.
2983 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2984 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2986 (integer_valued_read_call_p): Likewise.
2987 * fold-const-call.c (fold_const_call_ss): Likewise.
2988 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2989 _Float<N> and _Float<N>X variants.
2990 (fold_const_call_ssss): Add support for fma _Float<N> and
2991 _Float<N>X variants.
2992 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2993 support for copysign and fma _Float<N> and _Float<N>X variants.
2994 (backprop::process_builtin_call_use): Likewise.
2995 * tree-call-cdce.c (can_test_argument_range); Add support for
2996 sqrt _Float<N> and _Float<N>X variants.
2997 (edom_only_function): Likewise.
2998 (get_no_error_domain): Likewise.
2999 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3000 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3001 copysign _Float<N> and _Float<N>X variants.
3002 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3003 handled by machine independent code.
3004 (FMAF128): Likewise.
3005 * doc/cpp.texi (Common Predefined Macros): Document defining
3006 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3007 fma _Float<N> and _Float<N>X variants.
3009 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3012 * config/i386/i386-modes.def (CCFPU): Remove definition.
3013 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3014 (ix86_cc_modes_compatible): Ditto.
3015 (ix86_expand_carry_flag_compare): Ditto.
3016 (ix86_expand_int_movcc): Ditto.
3017 (ix86_expand_int_addcc): Ditto.
3018 (ix86_reverse_condition): Ditto.
3019 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3020 Return true/false for unordered/ordered fp comparisons.
3021 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3022 (ix86_prepare_fp_compare_args): Update for rename.
3023 (ix86_expand_fp_compare): Update for rename. Generate unordered
3024 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3025 (ix86_expand_sse_compare_and_jump): Ditto.
3026 * config/i386/predicates.md (fcmov_comparison_operator):
3027 Remove CCFPU mode handling.
3028 (ix86_comparison_operator): Ditto.
3029 (ix86_carry_flag_operator): Ditto.
3030 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3031 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3032 (*cmpu<mode>_cc_i387): Ditto.
3033 (FPCMP): Remove mode iterator.
3034 (unord): Remove mode attribute.
3035 (unord_subst): New define_subst transformation
3036 (unord): New define_subst attribute.
3038 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3039 (*cmpi<unord>xf_i387): Ditto.
3040 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3041 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3042 using unord_subst transformation.
3043 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3044 (round_saeonly): Also handle CCFP mode.
3045 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3046 Remove UNSPEC_SAHF unspec handling.
3048 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3050 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3052 2017-10-27 Jeff Law <law@redhat.com>
3054 * gimple-ssa-sprintf.c: Include domwalk.h.
3055 (class sprintf_dom_walker): New class, derived from dom_walker.
3056 (sprintf_dom_walker::before_dom_children): New function.
3057 (struct call_info): Moved into sprintf_dom_walker class
3058 (compute_formath_length, handle_gimple_call): Likewise.
3059 (sprintf_length::execute): Call the dominator walker rather
3060 than walking the statements.
3062 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3063 gimple statement locations.
3064 (check_array_bounds): Corresponding changes. Get the statement's
3065 location directly from wi->stmt.
3067 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3070 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3072 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3074 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3075 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3077 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3080 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3081 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3082 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3083 maybe_get_pool_constant.
3084 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3087 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3089 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3090 2.26 caveat. Update gas and gld versions.
3091 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3094 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3096 * cgraph.h (set_malloc_flag): Declare.
3097 * cgraph.c (set_malloc_flag_1): New function.
3098 (set_malloc_flag): Likewise.
3099 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3100 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3102 (read_ipa_call_summary): Add support for reading is_return_callee.
3103 (write_ipa_call_summary): Stream is_return_callee.
3104 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3105 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3106 ipa-prop.h, ipa-fnsummary.h.
3107 (pure_const_names): Change to static.
3108 (malloc_state_e): Define.
3109 (malloc_state_names): Define.
3110 (funct_state_d): Add field malloc_state.
3111 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3112 (check_retval_uses): New function.
3113 (malloc_candidate_p): Likewise.
3114 (analyze_function): Add support for malloc attribute.
3115 (pure_const_write_summary): Stream malloc_state.
3116 (pure_const_read_summary): Add support for reading malloc_state.
3117 (dump_malloc_lattice): New function.
3118 (propagate_malloc): New function.
3119 (warn_function_malloc): New function.
3120 (ipa_pure_const::execute): Call propagate_malloc and
3121 ipa_free_fn_summary.
3122 (pass_local_pure_const::execute): Add support for malloc attribute.
3123 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3124 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3126 2017-10-27 Martin Liska <mliska@suse.cz>
3128 PR gcov-profile/82457
3129 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3130 for fork-like functions to be properly instrumented.
3132 2017-10-27 Richard Biener <rguenther@suse.de>
3135 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3136 info when we redirected EH.
3138 2017-10-26 Michael Collison <michael.collison@arm.com>
3140 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3142 (<optab>_trunchf<GPI:mode>2: New pattern.
3143 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3144 * config/aarch64/iterators.md (wv): New mode attribute.
3145 (vf, VF): New mode attributes.
3146 (vgp, VGP): New mode attributes.
3147 (s): Update attribute with SImode and DImode prefixes.
3149 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3151 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3152 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3153 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3154 (nios2_option_overide): Initialize it. Don't allow R0-relative
3155 addressing with PIC.
3156 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3157 (nios2_symbolic_constant_p): Likewise.
3158 (nios2_legitimate_address_p): Likewise.
3159 (nios2_r0rel_section_name_p): New.
3160 (nios2_symbol_ref_in_r0rel_data_p): New.
3161 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3162 (r0rel_constant_p): New.
3163 (nios2_print_operand_address): Handle r0rel_constant_p.
3164 (nios2_cdx_narrow_form_p): Likewise.
3165 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3166 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3167 (Nios II Options): Document -mr0rel-sec.
3169 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3171 * config/nios2/nios2.c: Include xregex.h.
3172 (nios2_gprel_sec_regex): New.
3173 (nios2_option_overide): Initialize it. Don't allow GP-relative
3174 addressing with PIC.
3175 (nios2_small_section_name_p): Check for regex match.
3176 * config/nios2/nios2.opt (mgprel-sec=): New option.
3177 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3178 (Nios II Options): Document -mgprel-sec.
3180 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3182 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3184 2017-10-26 Tom de Vries <tom@codesourcery.com>
3186 PR tree-optimization/82707
3187 * gimple.c (gimple_copy): Fix unsharing of
3188 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3190 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3192 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3193 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3194 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3195 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3196 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3197 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3198 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3199 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3200 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3201 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3202 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3203 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3204 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3205 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3206 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3207 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3208 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3210 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3212 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3214 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3215 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3216 warning to rs6000.c. Remove the Undocumented flag, since it has
3218 (-mabi=ibmlongdouble): Likewise.
3219 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3220 already set, set the default format for long double.
3221 (rs6000_debug_reg_global): Print whether long double is IBM or
3223 (rs6000_option_override_internal): Rework setting long double
3224 format. Only warn if the user is changing the long double default
3225 and they did not use -Wno-psabi.
3226 * doc/invoke.texi (PowerPC options): Update the documentation for
3227 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3229 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3230 Alan Hayward <alan.hayward@arm.com>
3231 David Sherwood <david.sherwood@arm.com>
3233 * rtl.h (wider_subreg_mode): New function.
3234 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3235 rather than an unsigned int *.
3236 * ira-color.c (regno_max_ref_width): Replace with...
3237 (regno_max_ref_mode): ...this new variable.
3238 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3239 Use wider_subreg_mode.
3240 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3241 rather than an unsigned int *.
3242 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3243 (process_alt_operands): Likewise.
3244 (invariant_p): Likewise.
3245 * lra-spills.c (assign_mem_slot): Likewise.
3246 (add_pseudo_to_slot): Likewise.
3247 * lra.c (collect_non_operand_hard_regs): Likewise.
3248 (add_regs_to_insn_regno_info): Likewise.
3249 * reload1.c (regno_max_ref_width): Replace with...
3250 (regno_max_ref_mode): ...this new variable.
3251 (reload): Update accordingly. Update call to
3252 ira_sort_regnos_for_alter_reg.
3253 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3254 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3255 (scan_paradoxical_subregs): Likewise.
3257 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3259 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3260 (aarch64_frame): Add emit_frame_chain boolean.
3261 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3262 Move eh_return case to aarch64_layout_frame.
3263 (aarch64_layout_frame): Initialize emit_frame_chain.
3264 (aarch64_expand_prologue): Use emit_frame_chain.
3266 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3268 * config/aarch64/aarch64.c (aarch64_layout_frame):
3269 Ensure LR is always stored at the bottom of the callee-saves.
3270 Remove rarely used frame layout which saves callee-saves at top of
3271 frame, so the store of LR can be used as a valid probe in all cases.
3273 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3275 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3276 Improve unaligned TImode/TFmode base/offset split.
3278 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3279 Alan Hayward <alan.hayward@arm.com>
3280 David Sherwood <david.sherwood@arm.com>
3282 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3283 * combine.c (find_single_use_1): Likewise.
3284 (expand_field_assignment): Likewise.
3285 (move_deaths): Likewise.
3286 * lra-constraints.c (simplify_operand_subreg): Likewise.
3287 (curr_insn_transform): Likewise.
3288 * lra.c (collect_non_operand_hard_regs): Likewise.
3289 (add_regs_to_insn_regno_info): Likewise.
3290 * rtlanal.c (reg_referenced_p): Likewise.
3291 (covers_regno_no_parallel_p): Likewise.
3293 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3295 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3296 Don't print any bits outside the precision of the value.
3297 * wide-int.cc (test_printing): Add some new tests.
3299 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3301 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3302 supports -xbrace_comment option.
3303 * configure: Regenerate.
3304 * config.in: Regenerate.
3305 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3306 (ASM_CPU_SPEC): Use it.
3308 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3310 * target.def (static_rtx_alignment): New hook.
3311 * targhooks.h (default_static_rtx_alignment): Declare.
3312 * targhooks.c (default_static_rtx_alignment): New function.
3313 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3314 * doc/tm.texi: Regenerate.
3315 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3316 instead of targetm.constant_alignment. Remove call to
3318 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3319 (cris_preferred_mininum_alignment): New function, split out from...
3320 (cris_constant_alignment): ...here.
3321 (cris_static_rtx_alignment): New function.
3322 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3324 (ix86_constant_alignment): ...here.
3325 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3326 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3327 (mmix_static_rtx_alignment): New function.
3328 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3329 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3331 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3334 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3335 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3337 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3339 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3340 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3341 mark as a sign-extending load.
3342 (local_pic_load_u): Define.
3344 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3347 * fold-const.c (operand_equal_for_comparison_p): Also return true
3348 if ARG0 is a simple variant of ARG1 with narrower precision.
3349 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3351 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3353 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3355 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3356 scatter_static, scatter_per_elt.
3357 * x86-tune-costs.h: Add new cost entries.
3359 2017-10-25 Richard Biener <rguenther@suse.de>
3361 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3362 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3363 class pass_fre): Move to ...
3364 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3365 class pass_fre): ... here and adjust for statistics.
3367 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3370 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3371 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3372 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3373 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3376 2017-10-25 Richard Biener <rguenther@suse.de>
3378 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3379 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3380 eliminate_push_avail, eliminate_insert): Move inside...
3381 (class eliminate_dom_walker): ... this class in preparation
3383 (fini_eliminate): Remove by merging with ...
3384 (eliminate): ... this function. Adjust for class changes.
3385 (pass_pre::execute): Remove fini_eliminate call.
3386 (pass_fre::execute): Likewise.
3388 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3391 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3392 (VPERMI2, VPERMI2I): New mode iterators.
3393 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3394 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3396 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3397 mode iterator. Remove 3 old define_insn patterns.
3398 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3399 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3400 VPERMI2 mode iterator, remove the other two expanders.
3401 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3402 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3403 instructions, remove the other two patterns.
3404 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3405 mode iterator, remove the other two patterns.
3406 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3407 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3408 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3409 and adjust argument order accordingly.
3410 (ix86_expand_vec_perm): Adjust caller.
3411 (expand_vec_perm_1): Likewise.
3412 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3413 (expand_vec_perm_vpermt2_vpshub2): ... this.
3414 (ix86_expand_vec_perm_const_1): Adjust caller.
3415 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3418 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3419 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3420 (vec_shl_<mode>): Remove unused expander.
3421 (avx512bw_<shift_insn><mode>3): New define_insn.
3422 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3423 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3425 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3428 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3431 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3433 PR rtl-optimization/82396
3434 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3435 (autopref_multipass_init): Simplify initialization.
3436 (autopref_rank_data): Simplify sort order.
3437 * gcc/sched-int.h (autopref_multipass_data_): Remove
3438 multi_mem_insn_p, min_offset and max_offset.
3440 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3443 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3444 Check special value of flag_omit_frame_pointer.
3445 (aarch64_can_eliminate): Likewise.
3446 (aarch64_override_options_after_change_1): Simplify handling of
3447 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3449 2017-10-24 Richard Biener <rguenther@suse.de>
3451 PR tree-optimization/82697
3452 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3453 zero for conditional load and unconditional store.
3455 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3457 * doc/install.texi: Document bootstrap-cet.
3459 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3462 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3463 ENDBR instruction at function entrance if function is only
3466 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3469 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3470 patterns to better describe from which operation the CF is computed.
3471 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3472 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3473 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3474 is 0, use _0 suffixed expanders instead of emitting a comparison
3477 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3479 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3480 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3482 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3485 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3486 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3487 * loop-iv.c (iv_get_reaching_def): Likewise.
3488 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3489 variable is promoted and the partition contains undefined values.
3491 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3493 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3495 (nios2_address_cost): Define.
3496 (nios2_legitimize_address): Recognize (exp + constant) directly.
3497 (TARGET_ADDRESS_COST): Define.
3499 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3501 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3502 (nios2_symbolic_memory_operand_p): Declare.
3503 (nios2_split_large_constant): Declare.
3504 (nios2_split_symbolic_memory_operand): Declare.
3505 * config/nios2/nios2.c: Adjust includes.
3506 (nios2_symbolic_constant_allowed): New.
3507 (nios2_symbolic_constant_p): New.
3508 (nios2_plus_symbolic_constant_p): New.
3509 (nios2_valid_addr_expr_p): Recognize addresses involving
3511 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3512 (nios2_symbolic_memory_operand_p): New.
3513 (nios2_large_constant_p): New.
3514 (nios2_split_large_constant): New.
3515 (nios2_split_plus_large_constant): New.
3516 (nios2_split_symbolic_memory_operand): New.
3517 (nios2_legitimize_address): Code refactoring. Handle addresses
3518 involving symbolic constants.
3519 (nios2_emit_move_sequence): Likewise.
3520 (nios2_print_operand): Improve error output.
3521 (nios2_print_operand_address): Handle LO_SUM.
3522 (nios2_cdx_narrow_form_p): Likewise.
3523 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3524 operands involving symbolic constants.
3525 (movhi_internal, movsi_internal): Likewise.
3526 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3527 (extendhisi2, extendqi<mode>2): Likewise.
3529 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3531 * tree-pass.h (PROP_rtl_split_insns): Define.
3532 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3534 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3536 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3538 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3541 * target.def (const_not_ok_for_debug_p): Default to
3542 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3543 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3544 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3545 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3546 which targetm.const_not_ok_for_debug_p returned true.
3547 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3549 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3551 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3552 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3553 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3555 (ix86_const_not_ok_for_debug_p): New function.
3556 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3557 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3559 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3562 * system.h: Conditionally include "unique-ptr.h" if
3563 INCLUDE_UNIQUE_PTR is defined.
3564 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3565 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3567 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3569 * config/rl78/rl78.md: New define_expand "subdi3".
3571 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3574 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3575 DF_REF_INSN if DF_REF_INSN_INFO is false.
3577 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3579 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3580 xmm_move instead of sse_move.
3581 (sse_store_index): New function.
3582 (ix86_register_move_cost): Be more sensible about mismatch stall;
3583 model AVX moves correctly; make difference between sse->integer and
3585 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3586 moves; make difference between SSE and AVX.
3587 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3588 and zmm_move. Increase size of sse load and store tables;
3589 add unaligned load and store tables; add ssemmx_to_integer.
3590 * x86-tune-costs.h: Update all entries according to real
3591 move latencies from Agner Fog's manual and chip documentation.
3593 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3596 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3597 * config/i386/constraints.md (Wf): New constraint.
3598 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3599 (cmp<dwi>_doubleword): Removed.
3600 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3601 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3602 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3603 expand with cmp<dwi>_doubleword. For LTU and GEU use
3604 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3606 * common.opt (gcolumn-info): Enable by default.
3607 * doc/invoke.texi (gcolumn-info): Document new default.
3609 2017-10-23 Richard Biener <rguenther@suse.de>
3611 PR tree-optimization/82672
3612 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3613 Fold the stmt if we propagated into it.
3615 2017-10-23 Richard Biener <rguenther@suse.de>
3617 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3618 (bitmap_remove_expr_from_set): ... this. All callers call this
3619 for non-constant values.
3620 (bitmap_set_subtract): Rename to...
3621 (bitmap_set_subtract_expressions): ... this. Adjust and
3623 (bitmap_set_contains_value): Remove superfluous check.
3624 (bitmap_set_replace_value): Inline into single caller ...
3625 (bitmap_value_replace_in_set): ... here and simplify.
3626 (dependent_clean): Merge into ...
3627 (clean): ... this using an overload. Adjust.
3628 (prune_clobbered_mems): Adjust.
3629 (compute_antic_aux): Likewise.
3630 (compute_partial_antic_aux): Likewise.
3632 2017-10-23 Richard Biener <rguenther@suse.de>
3634 PR tree-optimization/82129
3636 2017-08-01 Richard Biener <rguenther@suse.de>
3638 PR tree-optimization/81181
3639 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3640 (compute_antic): ... end of iteration here.
3642 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3644 * target.def (starting_frame_offset): New hook.
3645 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3646 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3647 * doc/tm.texi.in: Regenerate.
3648 * hooks.h (hook_hwi_void_0): Declare.
3649 * hooks.c (hook_hwi_void_0): New function.
3650 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3651 STARTING_FRAME_OFFSET.
3652 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3653 * reload1.c (reload): Likewise.
3654 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3655 instead of STARTING_FRAME_OFFSET.
3656 * function.c (try_fit_stack_local): Likewise.
3657 (assign_stack_local_1): Likewise
3658 (instantiate_virtual_regs): Likewise.
3659 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3660 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3661 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3662 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3663 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3664 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3665 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3666 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3667 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3668 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3669 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3670 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3671 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3672 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3673 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3674 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3675 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3676 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3677 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3678 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3679 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3680 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3681 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3682 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3683 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3684 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3685 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3686 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3687 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3688 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3689 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3690 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3691 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3692 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3693 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3694 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3695 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3696 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3697 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3698 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3699 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3700 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3701 return a HOST_WIDE_INT.
3702 (avr_builtin_setjmp_frame_value): Use it instead of
3703 STARTING_FRAME_OFFSET.
3704 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3705 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3706 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3708 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3709 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3710 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3711 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3712 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3713 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3714 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3715 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3716 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3717 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3718 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3719 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3721 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3722 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3723 * config/mips/mips.c (mips_compute_frame_info): Refer to
3724 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3725 (mips_starting_frame_offset): New function.
3726 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3727 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3728 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3729 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3730 and return a HOST_WIDE_INT.
3731 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3732 (mmix_initial_elimination_offset): Refer to
3733 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3734 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3735 * config/pa/pa.c (pa_starting_frame_offset): New function.
3736 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3737 (pa_expand_prologue): Likewise.
3738 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3739 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3740 !FRAME_GROWS_DOWNWARD handling to...
3741 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3742 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3743 !FRAME_GROWS_DOWNWARD handling to...
3744 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3745 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3746 !FRAME_GROWS_DOWNWARD handling to...
3747 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3748 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3750 (rs6000_starting_frame_offset): New function.
3751 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3752 !FRAME_GROWS_DOWNWARD handling to...
3753 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3754 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3755 !FRAME_GROWS_DOWNWARD handling to...
3756 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3757 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3758 !FRAME_GROWS_DOWNWARD handling to...
3759 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3760 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3761 (rs6000_starting_frame_offset): New function.
3762 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3763 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3764 * config/vax/vax.c (vax_starting_frame_offset): New function.
3765 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3766 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3767 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3768 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3769 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3770 * system.h (STARTING_FRAME_OFFSET): Poison.
3772 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3774 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3775 SCALAR_TYPE_MODE instead of TYPE_MODE.
3777 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3778 Alan Hayward <alan.hayward@arm.com>
3779 David Sherwood <david.sherwood@arm.com>
3781 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3783 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3784 Alan Hayward <alan.hayward@arm.com>
3785 David Sherwood <david.sherwood@arm.com>
3787 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3789 2017-10-23 Richard Biener <rguenther@suse.de>
3791 PR tree-optimization/82129
3792 * tree-ssa-pre.c (bitmap_set_and): Remove.
3793 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3794 canonicalizing expressions in the set to those with lowest
3795 ID rather than taking that from the first edge.
3797 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3799 * combine.c (rtx_equal_for_field_assignment_p): Use
3800 byte_lowpart_offset.
3802 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3803 Alan Hayward <alan.hayward@arm.com>
3804 David Sherwood <david.sherwood@arm.com>
3806 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3807 to a SUBREG_PROMOTED_VAR.
3809 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3810 Alan Hayward <alan.hayward@arm.com>
3811 David Sherwood <david.sherwood@arm.com>
3813 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3814 (expand_debug_source_expr): Likewise.
3815 * combine.c (combine_simplify_rtx): Likewise.
3816 * cse.c (fold_rtx): Likewise.
3817 * optabs.c (expand_float): Likewise.
3818 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3819 (simplify_binary_operation_1): Likewise.
3821 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3822 Alan Hayward <alan.hayward@arm.com>
3823 David Sherwood <david.sherwood@arm.com>
3825 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3826 (record_promoted_value): Likewise.
3827 * expr.c (expand_expr_real_2): Likewise.
3828 * ree.c (update_reg_equal_equiv_notes): Likewise.
3829 (combine_set_extension): Likewise.
3830 * rtlanal.c (low_bitmask_len): Likewise.
3831 * simplify-rtx.c (neg_const_int): Likewise.
3832 (simplify_binary_operation_1): Likewise.
3834 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3835 Alan Hayward <alan.hayward@arm.com>
3836 David Sherwood <david.sherwood@arm.com>
3838 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3839 * regcprop.c (maybe_mode_change): Likewise.
3840 * reload1.c (alter_reg): Likewise.
3842 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3844 * inchash.h (inchash::hash::add_wide_int): New function.
3845 * lto-streamer-out.c (hash_tree): Use it.
3847 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3849 * inchash.h (inchash::hash::add_wide_int): Rename to...
3850 (inchash::hash::add_hwi): ...this.
3851 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3852 (polymorphic_call_target_hasher::hash): Likewise.
3853 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3854 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3855 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3856 * lto-streamer-out.c (hash_tree): Likewise.
3857 * optc-save-gen.awk: Likewise.
3858 * tree.c (add_expr): Likewise.
3860 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3863 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3864 for ordered inequality comparisons even with TARGET_IEEE_FP.
3866 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3869 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3870 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3871 Expand with cmp<dwi>_doubleword.
3873 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3875 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3876 List CET intrinsics.
3877 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3878 specific to -fcf-protection option.
3880 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3882 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3883 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3884 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3885 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3886 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3887 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3888 (extra_objs): Add cet.o for Linux/x86 targets.
3889 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3890 * config/i386/cet.c: New file.
3891 * config/i386/cetintrin.h: Likewise.
3892 * config/i386/t-cet: Likewise.
3893 * config/i386/cpuid.h (bit_SHSTK): New.
3894 (bit_IBT): Likewise.
3895 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3896 pass IBT and SHSTK bits.
3897 * config/i386/i386-builtin-types.def
3898 (VOID_FTYPE_UNSIGNED_PVOID): New.
3899 (VOID_FTYPE_UINT64_PVOID): Likewise.
3900 * config/i386/i386-builtin.def: Add CET intrinsics.
3901 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3902 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3903 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3904 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3906 * config/i386/i386.c (rest_of_insert_endbranch): New.
3907 (pass_data_insert_endbranch): Likewise.
3908 (pass_insert_endbranch): Likewise.
3909 (make_pass_insert_endbranch): Likewise.
3910 (ix86_notrack_prefixed_insn_p): Likewise.
3911 (ix86_target_string): Add -mibt, -mshstk flags.
3912 (ix86_option_override_internal): Add flag_cf_protection
3914 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3915 (ix86_print_operand): Add 'notrack' prefix output.
3916 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3917 (ix86_expand_builtin): Expand CET intrinsics.
3918 (x86_output_mi_thunk): Add 'endbranch' instruction.
3919 * config/i386/i386.h (TARGET_IBT): New.
3920 (TARGET_IBT_P): Likewise.
3921 (TARGET_SHSTK): Likewise.
3922 (TARGET_SHSTK_P): Likewise.
3923 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3924 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3925 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3926 (builtin_setjmp_setup): New pattern.
3927 (builtin_longjmp): Likewise.
3928 (rdssp<mode>): Likewise.
3929 (incssp<mode>): Likewise.
3930 (saveprevssp): Likewise.
3931 (rstorssp): Likewise.
3932 (wrss<mode>): Likewise.
3933 (wruss<mode>): Likewise.
3934 (setssbsy): Likewise.
3935 (clrssbsy): Likewise.
3936 (nop_endbr): Likewise.
3937 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3939 * config/i386/immintrin.h: Include <cetintrin.h>.
3940 * config/i386/linux-common.h
3941 (file_end_indicate_exec_stack_and_cet): New prototype.
3942 (TARGET_ASM_FILE_END): New.
3944 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3946 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3947 latencies instead of having separate table; make difference between
3948 integer and float costs.
3949 * i386.h (processor_costs): Remove scalar_stmt_cost,
3950 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3951 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3953 * x86-tune-costs.h: Remove entries which has been removed in
3954 procesor_costs from all tables; make cond_taken_branch_cost
3955 and cond_not_taken_branch_cost COST_N_INSNS based.
3957 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3959 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3961 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3963 * config/i386/i386.md (isa): Remove fma_avx512f.
3964 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3965 <avx512>_fmadd_<mode>_mask3<round_name>,
3966 <avx512>_fmsub_<mode>_mask<round_name>,
3967 <avx512>_fmsub_<mode>_mask3<round_name>,
3968 <avx512>_fnmadd_<mode>_mask<round_name>,
3969 <avx512>_fnmadd_<mode>_mask3<round_name>,
3970 <avx512>_fnmsub_<mode>_mask<round_name>,
3971 <avx512>_fnmsub_<mode>_mask3<round_name>,
3972 <avx512>_fmaddsub_<mode>_mask<round_name>,
3973 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3974 <avx512>_fmsubadd_<mode>_mask<round_name>,
3975 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3976 (*vec_widen_umult_even_v16si<mask_name>,
3977 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3978 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3980 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3982 * extend.texi: Add 'nocf_check' documentation.
3983 * gimple.texi: Add second parameter to
3984 gimple_build_call_from_tree.
3985 * invoke.texi: Add -fcf-protection documentation.
3986 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3988 2017-10-20 Richard Biener <rguenther@suse.de>
3990 PR tree-optimization/82473
3991 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3992 the largest input type.
3994 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3996 * c-attribs.c (handle_nocf_check_attribute): New function.
3997 (c_common_attribute_table): Add 'nocf_check' handling.
3998 * gimple-parser.c: Add second argument NULL to
3999 gimple_build_call_from_tree.
4000 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4001 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4003 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4004 * common.opt: Add fcf-protection flag.
4005 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4006 * flag-types.h: Add enum cf_protection_level.
4007 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4008 Add 'nocf_check' attribute propagation to gimple call.
4009 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4010 (gimple_build_call_from_tree): Update prototype.
4011 (gimple_call_nocf_check_p): New function.
4012 (gimple_call_set_nocf_check): Likewise.
4013 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4014 * ipa-icf.c: Add nocf_check attribute in statement hash.
4015 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4016 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4017 * toplev.c (process_options): Add flag_cf_protection handling.
4019 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4021 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4023 2017-10-20 Richard Biener <rguenther@suse.de>
4025 PR tree-optimization/82603
4026 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4027 remove false predicated stores.
4029 2017-10-20 Richard Biener <rguenther@suse.de>
4031 * graphite-isl-ast-to-gimple.c
4032 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4033 Remove return value and simplify, dump copied stmt after lhs
4035 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4036 Reduce dump verbosity.
4037 (gsi_insert_earliest): Likewise.
4038 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4039 * graphite.c (print_global_statistics): Adjust dumping.
4040 (print_graphite_scop_statistics): Likewise.
4041 (print_graphite_statistics): Do not dump loops here.
4042 (graphite_transform_loops): But here.
4044 2017-10-20 Nicolas Roche <roche@adacore.com>
4046 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4047 * configure: Regenerate.
4049 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4052 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4053 functions when optimizing replace GIMPLE_RETURN stmts with
4054 calls to __builtin_unreachable ().
4057 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4058 for -fsanitize=thread link of executables.
4059 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4060 link of executables.
4063 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4065 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4066 define_insns for logical vector shifts to use VI248_AVX512BW
4067 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4068 condition, useless isa and prefix attributes. Change the first
4069 2 of these define_insns to ...
4070 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4071 define_insn for avx512vl.
4072 (<shift_insn><mode>3): ... and this, new define_insn without
4073 masking for non-avx512vl.
4076 * config/i386/sse.md (*andnot<mode>3,
4077 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4078 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4079 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4080 not applied use empty suffix even for TARGET_AVX512VL.
4081 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4082 is applied, supply evex,evex or evex,evex,evex instead of just
4085 2017-10-20 Julia Koval <julia.koval@intel.com>
4087 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4088 (OPTION_MASK_ISA_GFNI_UNSET): New.
4089 (ix86_handle_option): Handle OPT_mgfni.
4090 * config/i386/cpuid.h (bit_GFNI): New.
4091 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4092 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4093 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4094 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4095 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4096 * config/i386/i386.opt: Add mgfni.
4098 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4100 * config/msp430/msp430.c (msp430_option_override): Disable
4101 -fdelete-null-pointer-checks.
4102 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4104 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4106 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4107 of x87 and SSE instructions.
4109 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4111 * asan.c (create_cond_insert_point): Do not update edge count.
4112 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4113 (afdo_propagate_circuit): Likewise.
4114 (afdo_calculate_branch_prob): Likewise.
4115 (afdo_annotate_cfg): Likewise.
4116 * basic-block.h (struct edge_def): Remove count.
4117 (edge_def::count): New accessor.
4118 * bb-reorder.c (rotate_loop): Update.
4119 (find_traces_1_round): Update.
4120 (connect_traces): Update.
4121 (sanitize_hot_paths): Update.
4122 * cfg.c (unchecked_make_edge): Update.
4123 (make_single_succ_edge): Update.
4124 (check_bb_profile): Update.
4125 (dump_edge_info): Update.
4126 (update_bb_profile_for_threading): Update.
4127 (scale_bbs_frequencies_int): Update.
4128 (scale_bbs_frequencies_gcov_type): Update.
4129 (scale_bbs_frequencies_profile_count): Update.
4130 (scale_bbs_frequencies): Update.
4131 * cfganal.c (connect_infinite_loops_to_exit): Update.
4132 * cfgbuild.c (compute_outgoing_frequencies): Update.
4133 (find_many_sub_basic_blocks): Update.
4134 * cfgcleanup.c (try_forward_edges): Update.
4135 (try_crossjump_to_edge): Update
4136 * cfgexpand.c (expand_gimple_cond): Update
4137 (expand_gimple_tailcall): Update
4138 (construct_exit_block): Update
4139 * cfghooks.c (verify_flow_info): Update
4140 (redirect_edge_succ_nodup): Update
4141 (split_edge): Update
4142 (make_forwarder_block): Update
4143 (duplicate_block): Update
4144 (account_profile_record): Update
4145 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4146 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4147 * cfgloopmanip.c (scale_loop_profile): Update.
4149 (lv_adjust_loop_entry_edge): Update.
4150 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4151 (force_nonfallthru_and_redirect): Update.
4152 (purge_dead_edges): Update.
4153 (rtl_flow_call_edges_add): Update.
4154 * cgraphunit.c (init_lowered_empty_function): Update.
4155 (cgraph_node::expand_thunk): Update.
4156 * gimple-pretty-print.c (dump_probability): Update.
4157 (dump_edge_probability): Update.
4158 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4159 * haifa-sched.c (sched_create_recovery_edges): Update.
4160 * hsa-gen.c (convert_switch_statements): Update.
4161 * ifcvt.c (dead_or_predicable): Update.
4162 * ipa-inline-transform.c (inline_transform): Update.
4163 * ipa-split.c (split_function): Update.
4164 * ipa-utils.c (ipa_merge_profiles): Update.
4165 * loop-doloop.c (add_test): Update.
4166 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4167 * lto-streamer-in.c (input_cfg): Update.
4168 (input_function): Update.
4169 * lto-streamer-out.c (output_cfg): Update.
4170 * modulo-sched.c (sms_schedule): Update.
4171 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4172 * predict.c (maybe_hot_edge_p): Update.
4173 (unlikely_executed_edge_p): Update.
4174 (probably_never_executed_edge_p): Update.
4175 (dump_prediction): Update.
4176 (drop_profile): Update.
4177 (propagate_unlikely_bbs_forward): Update.
4178 (determine_unlikely_bbs): Update.
4179 (force_edge_cold): Update.
4180 * profile.c (compute_branch_probabilities): Update.
4181 * reg-stack.c (better_edge): Update.
4182 * shrink-wrap.c (handle_simple_exit): Update.
4183 * tracer.c (better_p): Update.
4184 * trans-mem.c (expand_transaction): Update.
4185 (split_bb_make_tm_edge): Update.
4186 * tree-call-cdce.c: Update.
4187 * tree-cfg.c (gimple_find_sub_bbs): Update.
4188 (gimple_split_edge): Update.
4189 (gimple_duplicate_sese_region): Update.
4190 (gimple_duplicate_sese_tail): Update.
4191 (gimple_flow_call_edges_add): Update.
4192 (insert_cond_bb): Update.
4193 (execute_fixup_cfg): Update.
4194 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4195 * tree-complex.c (expand_complex_div_wide): Update.
4196 * tree-eh.c (lower_resx): Update.
4197 (unsplit_eh): Update.
4198 (cleanup_empty_eh_move_lp): Update.
4199 * tree-inline.c (copy_edges_for_bb): Update.
4200 (freqs_to_counts): Update.
4201 (copy_cfg_body): Update.
4202 * tree-ssa-dce.c (remove_dead_stmt): Update.
4203 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4204 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4205 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4206 (unloop_loops): Update.
4207 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4208 * tree-ssa-loop-split.c (connect_loops): Update.
4209 (split_loop): Update.
4210 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4211 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4212 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4213 * tree-ssa-reassoc.c (branch_fixup): Update.
4214 * tree-ssa-tail-merge.c (replace_block_by): Update.
4215 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4216 (compute_path_counts): Update.
4217 (update_profile): Update.
4218 (recompute_probabilities): Update.
4219 (update_joiner_offpath_counts): Update.
4220 (estimated_freqs_path): Update.
4221 (freqs_to_counts_path): Update.
4222 (clear_counts_path): Update.
4223 (ssa_fix_duplicate_block_edges): Update.
4224 (duplicate_thread_path): Update.
4225 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4226 (case_bit_test_cmp): Update.
4227 (collect_switch_conv_info): Update.
4228 (gen_inbound_check): Update.
4229 (do_jump_if_equal): Update.
4230 (emit_cmp_and_jump_insns): Update.
4231 * tree-tailcall.c (decrease_profile): Update.
4232 (eliminate_tail_call): Update.
4233 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4234 (vect_do_peeling): Update.
4235 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4236 * ubsan.c (ubsan_expand_null_ifn): Update.
4237 (ubsan_expand_ptr_ifn): Update.
4238 * value-prof.c (gimple_divmod_fixed_value): Update.
4239 (gimple_mod_pow2): Update.
4240 (gimple_mod_subtract): Update.
4241 (gimple_ic): Update.
4242 (gimple_stringop_fixed_value): Update.
4244 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4247 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4249 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4251 PR rtl-optimization/82395
4252 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4253 based on non_spilled_static_chain_regno_p.
4255 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4257 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4258 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4259 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4261 2017-10-19 Martin Sebor <msebor@redhat.com>
4263 PR tree-optimization/82596
4264 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4266 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4268 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4269 (get_mem_refs_of_builtin_call): Likewise.
4270 * builtins.c (expand_builtin_apply): Adjust call to
4271 allocate_dynamic_stack_space.
4272 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4273 the third argument to allocate_dynamic_stack_space, otherwise -1.
4274 (expand_builtin): Deal with all alloca variants.
4275 (is_inexpensive_builtin): Likewise.
4276 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4277 * calls.c (special_function_p): Deal with all alloca variants.
4278 (initialize_argument_information): Adjust call to
4279 allocate_dynamic_stack_space.
4280 (expand_call): Likewise.
4281 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4282 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4283 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4284 use it for the stack usage computation.
4285 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4286 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4287 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4288 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4289 (in_loop_p): Remove first argument and useless check.
4290 (pass_walloca::execute): Remove useless test and adjust call to above.
4291 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4292 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4293 (gimplify_call_expr): Deal with all alloca variants.
4294 * hsa-gen.c (gen_hsa_alloca): Likewise.
4295 (gen_hsa_insns_for_call): Likewise.
4296 * ipa-pure-const.c (special_builtin_state): Likewise.
4297 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4298 * tree-object-size.c (alloc_object_size): Likewise.
4299 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4300 (call_may_clobber_ref_p_1): Likewise.
4301 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4302 (ccp_fold_stmt): Likewise.
4303 (optimize_stack_restore): Likewise.
4304 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4305 (mark_all_reaching_defs_necessary_1): Likewise.
4306 (propagate_necessity): Likewise.
4307 (eliminate_unnecessary_stmts): Likewise.
4308 * tree.c (build_common_builtin_nodes): Build
4309 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4310 (build_alloca_call_expr): New function.
4311 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4312 (CASE_BUILT_IN_ALLOCA): Likewise.
4313 (build_alloca_call_expr): Declare.
4314 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4316 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4319 * dwarf2out.c (new_die_raw): New static inline function.
4320 (new_die): Use it to create the DIE.
4321 (add_AT_external_die_ref): Likewise.
4322 (clone_die): Likewise.
4323 (clone_as_declaration): Likewise.
4324 (dwarf2out_vms_debug_main_pointer): Likewise.
4325 (base_type_die): Likewise. Remove early return for corner cases.
4326 Do not call add_pubtype on the DIE here.
4327 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4328 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4329 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4330 native order exists for base types, attach the DIE manually and call
4331 add_pubtype on it. Do not equate a reverse order DIE to the type.
4333 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4335 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4336 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4338 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4340 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4342 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4343 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4344 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4345 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4346 * opts.c (sanitizer_opts): Add builtin.
4347 * ubsan.c (instrument_builtin): New function.
4348 (pass_ubsan::execute): Call it.
4349 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4350 * doc/invoke.texi: Document -fsanitize=builtin.
4352 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4353 builtins, store max (log2 (align), 0) into uchar field instead of
4354 align into uptr field.
4355 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4356 store uchar 0 field instead of uptr 0 field.
4357 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4358 instead of passing one address of struct with 2 locations pass
4359 two addresses of structs with 1 location each.
4360 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4361 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4362 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4363 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4364 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4365 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4366 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4367 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4369 2017-10-19 Martin Liska <mliska@suse.cz>
4372 * file-find.c (remove_prefix): Remove.
4373 * file-find.h (remove_prefix): Likewise.
4374 * gcc-ar.c: Remove smartness of lookup.
4376 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4378 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4379 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4380 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4382 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4385 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4386 (setcc + and to xor + setcc): New peephole2.
4388 2017-10-19 Tom de Vries <tom@codesourcery.com>
4390 * doc/sourcebuild.texi (Test Directives, Variants of
4391 dg-require-support): Add dg-require-stack-size.
4393 2017-10-19 Martin Liska <mliska@suse.cz>
4396 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4397 that have a large alignment.
4398 (gimplify_target_expr): Likewise.
4400 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4402 PR rtl-optimization/82602
4403 * ira.c (rtx_moveable_p): Return false for volatile asm.
4405 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4408 * config/i386/i386-modes.def (CCGZ): New CC mode.
4409 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4410 * config/i386/predicates.md (ix86_comparison_operator):
4412 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4413 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4414 with double-word subtraction.
4415 (put_condition_code): Handle CCGZmode.
4417 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4419 * wide-int.cc (debug (const wide_int &)): New.
4420 (debug (const wide_int *)): New.
4421 (debug (const widest_int &)): New.
4422 (debug (const widest_int *)): New.
4424 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4427 * lra-constraints.c (curr_insn_transform): Use non-input operand
4428 instead of output one for matched reload.
4430 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4432 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4433 (tree-ssa-loop-ivopts.h): New header file.
4434 (struct builtin_info): New fields.
4435 (classify_builtin_1): Compute and record base and offset parts for
4436 memset builtin partition by calling strip_offset.
4437 (offset_cmp, fuse_memset_builtins): New functions.
4438 (finalize_partitions): Fuse adjacent memset partitions by calling
4440 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4441 Expose the interface.
4442 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4444 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4446 PR tree-optimization/82574
4447 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4448 that data reference must be executed exactly once per iteration
4449 against the outermost loop in nest.
4450 (classify_partition): Update call to above function.
4452 2017-10-18 Richard Biener <rguenther@suse.de>
4454 PR tree-optimization/82591
4455 * graphite.c (graphite_transform_loops): Move code gen message
4457 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4458 Here. Handle scop_to_isl_ast failing.
4459 (scop_to_isl_ast): Limit the number of ISL operations.
4461 2017-10-18 Richard Biener <rguenther@suse.de>
4463 * graphite-isl-ast-to-gimple.c
4464 (translate_isl_ast_to_gimple::set_rename): Simplify.
4465 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4466 (graphite_copy_stmts_from_block): ... here.
4467 (copy_bb_and_scalar_dependences): Simplify.
4468 (add_parameters_to_ivs_params): Canonicalize.
4469 (generate_entry_out_of_ssa_copies): Simplify.
4470 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4471 by passing in ISL dimension.
4472 (parameter_index_in_region_1): Rename to ...
4473 (parameter_index_in_region): ... this.
4474 (extract_affine): Adjust assert, pass down parameter index.
4475 (add_param_constraints): Use range-info when available.
4476 (build_scop_context): Adjust.
4477 * sese.c (new_sese_info): Adjust.
4478 (free_sese_info): Likewise.
4479 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4480 Remove unused typedefs.
4481 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4483 2017-10-18 Martin Liska <mliska@suse.cz>
4485 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4487 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4489 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4490 (s390_sched_init): Do not reset s390_sched_state if we entered the
4491 current basic block via a fallthru edge and all others are unlikely.
4493 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4495 * config/s390/s390.c (NUM_SIDES): New variable.
4496 (LONGRUNNING_THRESHOLD): New variable.
4497 (LATENCY_FACTOR): New variable.
4498 (s390_sched_score): Decrease score for long-running instructions on
4500 (s390_sched_variable_issue): Perform bookkeeping for long-running
4503 2017-10-18 Richard Biener <rguenther@suse.de>
4505 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4506 Simplify with removal of the parameter rename map.
4507 (set_rename): Likewise.
4508 (should_copy_to_new_region): Likewise.
4509 (graphite_copy_stmts_from_block): Likewise.
4510 (copy_bb_and_scalar_dependences): Remove initialization of
4511 unused copied_bb_map.
4513 (copy_internal_parameters): Likewise.
4514 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4515 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4516 Use INTEGRAL_TYPE_P.
4517 (parameter_index_in_region_1): Rename to ...
4518 (assign_parameter_index_in_region): ... this. Assert we have
4519 a parameter we handle.
4520 (scan_tree_for_params): Adjust.
4521 * sese.h (parameter_rename_map_t): Remove.
4522 (struct sese_info_t): Remove unused parameter_rename_map and
4523 copied_bb_map members.
4524 * sese.c (new_sese_info): Adjust.
4525 (free_sese_info): Likewise.
4527 2017-10-18 Martin Liska <mliska@suse.cz>
4530 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4531 on an abnormal edge.
4533 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4535 * doc/invoke.texi (ffunction-sections and fdata-sections):
4538 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4540 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4541 the use statement can throw internally.
4543 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4545 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4546 any RTX present on the RHS of a SET.
4547 * compare-elim.c (try_eliminate_compare): Restore comment.
4549 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4551 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4552 may be also called on tcc_type nodes.
4553 * langhooks.c (lhd_tree_size): Likewise.
4555 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4557 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4558 format_warning_at_substring.
4559 (maybe_warn): Convert source_range * param to a location_t. Pass
4560 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4561 (format_directive): Remove code to extract source_ranges and
4562 source_range * in favor of just a location_t.
4563 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4565 * substring-locations.c (format_warning_va): Convert
4566 source_range * param to a location_t.
4567 (format_warning_at_substring): Likewise.
4568 * substring-locations.h (format_warning_va): Likewise.
4569 (format_warning_at_substring): Likewise.
4571 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4573 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4575 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4576 and scatter/gather ops.
4578 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4579 vec_gather_load and vec_scatter_store.
4580 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4581 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4582 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4583 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4584 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4585 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4587 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4589 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4590 Detect FTST instruction and handle its register pops. Only pop
4591 second operand if can_pop_second_op is true.
4592 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4593 set can_pop_second_op to false in the compare_for_stack_reg call.
4595 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4596 output_fp_compare for stack register operands.
4597 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4598 instructions here. Do not emit stack register pops here. Assert
4599 that FCOMPP pops next to top stack register. Rewrite function.
4601 2017-10-17 Nathan Sidwell <nathan@acm.org>
4604 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4605 use DECL_ASSEMBLER_NAME_RAW.
4608 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4611 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4612 Wilco Dijkstra <wilco.dijkstra@arm.com>
4614 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4615 converstion to Pmode from the buf_addr.
4617 2017-10-17 Richard Biener <rguenther@suse.de>
4619 * graphite-dependences.c (scop_get_reads_and_writes): Change
4620 output parameters to references.
4622 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4624 PR 71026/tree-optimization
4625 * fold-const.c (distribute_real_division): Removed.
4626 (fold_binary_loc): Remove calls to distribute_real_divison.
4628 2017-10-17 Richard Biener <rguenther@suse.de>
4630 * graphite-scop-detection.c
4631 (scop_detection::stmt_has_simple_data_refs_p): Always use
4632 the full nest as region.
4633 (try_generate_gimple_bb): Likewise.
4634 * sese.c (scalar_evolution_in_region): Simplify now that
4635 SCEV can handle instantiation in regions.
4636 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4637 in the non-loop part of a function if requested.
4639 2017-10-17 Richard Biener <rguenther@suse.de>
4641 PR tree-optimization/82563
4642 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4644 (graphite_regenerate_ast_isl): Call it.
4645 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4647 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4649 PR tree-optimization/82549
4650 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4651 Formatting fixes. Instead of calling make_bit_field_ref with negative
4654 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4656 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4657 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4658 _mm_maskz_reduce_ss): New.
4659 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4660 __builtin_ia32_reducess_mask): Ditto..
4661 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4662 * config/i386/sse.md (reduces<mode>): Renamed to ...
4663 (reduces<mode><mask_scalar_name>): ... this.
4664 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4666 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4667 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4669 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4671 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4672 * selftest-run-tests.c (selftest::run_tests): Call
4673 selftest::unique_ptr_tests_cc_tests.
4674 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4675 * unique-ptr-tests.cc: New file.
4677 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4680 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4682 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4683 (make_hard_regno_born, make_hard_regno_dead): Update
4684 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4686 2017-10-16 Jeff Law <law@redhat.com>
4688 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4690 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4692 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4694 2017-10-16 Olivier Hainque <hainque@adacore.com>
4696 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4697 with_cpu if we were configured for an e500v2 target cpu name.
4699 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4701 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4702 * doc/invoke.texi: Document +nodsp as a valid extension for
4705 2017-10-16 Martin Liska <mliska@suse.cz>
4707 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4708 (test_set_range): Likewise.
4709 (test_range_functions): Rename to ...
4710 (test_bit_in_range): ... this.
4711 (sbitmap_c_tests): Add new test.
4713 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4715 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4717 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4718 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4720 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4722 * config/aarch64/aarch64-builtins.c
4723 (aarch64_types_quadopu_lane_qualifiers): New.
4724 (TYPES_QUADOPU_LANE): New.
4725 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4726 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4727 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4728 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4729 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4730 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4731 (Vdottype, DOTPROD): New.
4732 (sur): Add SDOT and UDOT.
4734 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4736 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4737 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4738 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4740 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4741 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4742 Enable TARGET_DOTPROD.
4743 (cortex-a75.cortex-a55): Likewise.
4744 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4746 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4748 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4749 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4751 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4753 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4754 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4755 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4756 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4757 * config/arm/types.md (neon_dot, neon_dot_q): New.
4758 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4760 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4762 * config/arm/arm.h (TARGET_DOTPROD): New.
4763 * config/arm/arm.c (arm_arch_dotprod): New.
4764 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4765 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4766 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4767 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4768 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4769 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4770 * doc/invoke.texi (armv8.2-a): Document dotprod
4772 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4774 * i386.c (ix86_vec_cost): New function.
4775 (ix86_rtx_costs): Handle vector operations better.
4776 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4777 * x86-tune-costs.h: Add new costs to all tables.
4779 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4781 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4783 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4784 divsd, sqrtss and sqrtsd
4785 * x86-tune-costs.h: Add new entries to all costs.
4786 (znver1_cost): Fix to match real instruction latencies.
4788 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4789 Michael Collison <michael.collison@arm.com>
4791 * compare-elim.c: Include emit-rtl.h.
4792 (can_merge_compare_into_arith): New function.
4793 (try_validate_parallel): Likewise.
4794 (try_merge_compare): Likewise.
4795 (try_eliminate_compare): Call the above when no previous clobber
4797 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4800 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4804 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4805 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4809 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4810 to be any operand_equal_p operands. For & (B - 1) require
4811 B to be power of 2. Recognize
4812 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4814 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4817 * optabs.c (expand_memory_blockage): Fix call of
4818 targetm.have_memory_blockage.
4820 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4823 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4824 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4825 objects to extra_objs instead of overwriting it.
4827 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4829 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4830 Use any_fp_register_operand as operand[3] predicate. Simplify
4831 equality test for operands[2] and operands[4] memory location.
4832 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4833 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4834 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4835 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4836 any_fp_register_operand as operand[1] predicate. Simplify
4837 equality test for operands[0] and operands[3] memory location.
4838 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4839 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4840 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4842 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4844 * target-insns.def: Add memory_blockage.
4845 * optabs.c (expand_memory_blockage): New function.
4846 (expand_asm_memory_barrier): Rename ...
4847 (expand_asm_memory_blockage): ... to this.
4848 (expand_mem_thread_fence): Call expand_memory_blockage
4849 instead of expand_asm_memory_barrier.
4850 (expand_mem_singnal_fence): Ditto.
4851 (expand_atomic_load): Ditto.
4852 (expand_atomic_store): Ditto.
4853 * doc/md.texi (Standard Pattern Names For Generation):
4854 Document memory_blockage instruction pattern.
4856 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4858 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4859 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4860 * config/rl78/rl78.md: New define_expand "adddi3".
4862 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4864 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4867 2017-10-13 Jeff Law <law@redhat.com>
4869 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4871 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4874 * internal-fn.c (expand_mul_overflow): If both operands have
4875 the same highpart of -1 or 0 and the topmost bit of lowpart
4876 is different, overflow is if res <= 0 rather than res < 0.
4878 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4880 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4881 TARGET_P9_VECTOR code for unaligned_load case.
4883 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4885 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4887 2017-10-13 Nathan Sidwell <nathan@acm.org>
4889 * tree-core.h (tree_contains_struct): Make bool.
4890 * tree.c (tree_contains_struct): Likewise.
4891 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4892 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4893 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4894 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4895 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4897 2017-10-13 Richard Biener <rguenther@suse.de>
4899 * graphite-isl-ast-to-gimple.c
4900 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4901 parameters and dominance check.
4902 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4903 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4904 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4905 Do not update SSA form here or do intermediate IL verification.
4906 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4907 (graphite_initialize): Remove check on the number of loops in
4908 the function and inline into graphite_transform_loops.
4909 (graphite_finalize): Inline into graphite_transform_loops.
4910 (graphite_transform_loops): Perform SSA update and IL verification
4912 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4914 2017-10-13 Richard Biener <rguenther@suse.de>
4916 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4917 graphite_expression_type_precision): Avoid global constructor
4919 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4920 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4921 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4922 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4923 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4924 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4926 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4929 * config/i386/i386.h (ix86_red_zone_size): New.
4930 * config/i386/i386.md (push peephole2s): Replace
4931 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4933 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4934 Alan Hayward <alan.hayward@arm.com>
4935 David Sherwood <david.sherwood@arm.com>
4937 * combine.c (can_change_dest_mode): Reject changes in
4938 REGMODE_NATURAL_SIZE.
4940 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4941 Alan Hayward <alan.hayward@arm.com>
4942 David Sherwood <david.sherwood@arm.com>
4944 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4945 (expand_debug_source_expr): Likewise.
4946 * combine.c (combine_simplify_rtx): Likewise.
4947 * cse.c (fold_rtx): Likewise.
4948 * fwprop.c (canonicalize_address): Likewise.
4949 * targhooks.c (default_shift_truncation_mask): Likewise.
4951 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4952 Alan Hayward <alan.hayward@arm.com>
4953 David Sherwood <david.sherwood@arm.com>
4955 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4956 (widened_mode): Likewise.
4957 (expand_unop): Likewise.
4958 * ree.c (transform_ifelse): Likewise.
4959 (merge_def_and_ext): Likewise.
4960 (combine_reaching_defs): Likewise.
4961 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4963 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4964 Alan Hayward <alan.hayward@arm.com>
4965 David Sherwood <david.sherwood@arm.com>
4967 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4968 * combine.c (gen_lowpart_for_combine): Likewise.
4969 * dwarf2out.c (rtl_for_decl_location): Likewise.
4970 * final.c (alter_subreg): Likewise.
4971 * rtlhooks.c (gen_lowpart_general): Likewise.
4972 (gen_lowpart_if_possible): Likewise.
4974 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4975 Alan Hayward <alan.hayward@arm.com>
4976 David Sherwood <david.sherwood@arm.com>
4978 * calls.c (expand_call): Use subreg_lowpart_offset.
4979 * cse.c (cse_insn): Likewise.
4980 * regcprop.c (copy_value): Likewise.
4981 (copyprop_hardreg_forward_1): Likewise.
4983 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4986 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4987 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4988 =Q constraints to +Q and into insn condition add check
4989 that operands[0] and operands[1] are equal.
4990 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4991 =Q constraints to +Q and into insn condition add check
4992 that operands[0] is equal to either operands[1] or operands[2].
4995 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4996 instead of handling MINUS_EXPR twice (once for each argument),
4997 canonicalize operand order and handle just once, use rtype where
4998 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5001 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5002 any values of __C while still being pattern recognizable as a simple
5005 2017-10-13 Richard Biener <rguenther@suse.de>
5007 PR tree-optimization/82451
5009 2017-10-02 Richard Biener <rguenther@suse.de>
5011 PR tree-optimization/82355
5012 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5013 a mapping for the enclosing loop but avoid generating one for
5015 (copy_bb_and_scalar_dependences): Remove premature codegen
5016 error on PHIs in blocks duplicated into multiple places.
5017 * graphite-scop-detection.c
5018 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5019 in the region use it as loop and nest to analyze the DR in.
5020 (try_generate_gimple_bb): Likewise.
5021 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5022 (add_loop_constraints): For blocks in a loop not in the region
5023 create a dimension with a single iteration.
5024 * sese.h (gbb_loop_at_index): Remove assert.
5026 * cfgloop.c (loop_preheader_edge): For the loop tree root
5027 return the single successor of the entry block.
5028 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5029 Reset the SCEV hashtable and niters.
5030 * graphite-scop-detection.c
5031 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5032 assert that we only have POLYNOMIAL_CHREC that vary in loops
5033 contained in the region.
5034 (scop_detection::graphite_can_represent_expr): Adjust.
5035 (scop_detection::stmt_has_simple_data_refs_p): For loops
5036 not in the region set loop to NULL. The nest is now the
5037 entry edge to the region.
5038 (try_generate_gimple_bb): Likewise.
5039 * sese.c (scalar_evolution_in_region): Adjust for
5040 instantiate_scev change.
5041 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5042 Make nest parameter the edge into the region.
5043 (create_data_ref): Likewise.
5044 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5045 entry edge into a region and adjust instantiate_scev calls.
5046 (create_data_ref): Likewise.
5047 (graphite_find_data_references_in_stmt): Likewise.
5048 (find_data_references_in_stmt): Pass the loop preheader edge
5049 from the nest argument.
5050 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5051 parameter the edge into the region.
5052 (instantiate_parameters): Use the loop preheader edge as entry.
5053 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5055 (get_instantiated_value_entry): Make instantiate_below parameter
5056 the edge into the region.
5057 (instantiate_scev_name): Likewise. Adjust dominance checks,
5058 when we cannot use loop-based instantiation instantiate by
5059 walking use-def chains.
5060 (instantiate_scev_poly): Adjust.
5061 (instantiate_scev_binary): Likewise.
5062 (instantiate_scev_convert): Likewise.
5063 (instantiate_scev_not): Likewise.
5064 (instantiate_array_ref): Remove.
5065 (instantiate_scev_3): Likewise.
5066 (instantiate_scev_2): Likewise.
5067 (instantiate_scev_1): Likewise.
5068 (instantiate_scev_r): Do not blindly handle N-operand trees.
5069 Do not instantiate array-refs. Handle all constants and invariants.
5070 (instantiate_scev): Make instantiate_below parameter
5071 the edge into the region.
5072 (resolve_mixers): Use the loop preheader edge for the region
5073 parameter to instantiate_scev_r.
5074 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5076 2017-10-13 Richard Biener <rguenther@suse.de>
5078 PR tree-optimization/82525
5079 * graphite-isl-ast-to-gimple.c
5080 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5082 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5083 Fail code generation when we cannot represent the isl integer.
5084 (binary_op_to_tree): Elide modulo operations that are no-ops
5085 in the type we code generate. Remove now superfluous code
5088 2017-10-13 Richard Biener <rguenther@suse.de>
5090 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5091 (scop_detection::harmful_loop_in_region): Remove premature
5092 IV type restriction.
5093 (scop_detection::graphite_can_represent_scev): We can handle
5094 pointer IVs just fine.
5096 2017-10-13 Alan Modra <amodra@gmail.com>
5098 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5099 "Clobbers and Scratch Registers". Add paragraph on
5100 alternative to clobbers for scratch registers and OpenBLAS
5103 2017-10-13 Alan Modra <amodra@gmail.com>
5105 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5106 example of a memory input for a string of known length. Move
5107 commentary out of table. Add a number of new examples
5108 covering array memory inputs.
5110 2017-10-12 Martin Liska <mliska@suse.cz>
5112 PR tree-optimization/82493
5113 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5114 (test_range_functions): New function.
5115 (sbitmap_c_tests): Likewise.
5116 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5117 * selftest.h (sbitmap_c_tests): New function.
5119 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5121 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5123 * config/rs6000/amo.h: Fix spacing issue.
5125 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5128 * config/i386/i386.md (*ashl<mode>3_mask_1,
5129 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5130 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5133 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5135 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5136 (profile_probability): Set max_probability
5137 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5139 * profile-count.c (profile_probability::differs_from_p): Do not
5140 rely on max_probaiblity == 10000
5142 2017-10-12 Jeff Law <law@redhat.com>
5144 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5147 2017-10-12 Martin Sebor <msebor@redhat.com>
5151 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5152 (handle_alias_pairs): Call it.
5153 * common.opt (-Wattribute-alias): New option.
5154 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5155 * doc/invoke.texi (-Wattribute-alias): Document.
5157 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5160 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5162 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5164 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5165 (make_hard_regno_born, make_hard_regno_dead): Update
5166 bb_killed_pseudos and bb_gen_pseudos.
5168 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5170 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5172 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5174 * config/alpha/alpha.c (alpha_split_conditional_move):
5175 Use std::swap instead of manually swapping.
5176 (alpha_stdarg_optimize_hook): Ditto.
5177 (alpha_canonicalize_comparison): Ditto.
5179 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5181 * tree-loop-distribution.c (struct builtin_info): New struct.
5182 (struct partition): Refactor fields into struct builtin_info.
5183 (partition_free): Free struct builtin_info.
5184 (build_size_arg_loc, build_addr_arg_loc): Delete.
5185 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5186 information from struct builtin_info.
5187 (find_single_drs): New function refactored from classify_partition.
5188 Also moved builtin validity checks to this function.
5189 (compute_access_range, alloc_builtin): New functions.
5190 (classify_builtin_st, classify_builtin_ldst): New functions.
5191 (classify_partition): Refactor code into functions find_single_drs,
5192 classify_builtin_st and classify_builtin_ldst.
5193 (distribute_loop): Don't do runtime alias check when distributing
5195 (find_seed_stmts_for_distribution): New function.
5196 (pass_loop_distribution::execute): Refactor code finding seed
5197 stmts into above function. Support distribution for the innermost
5198 two-level loop nest. Adjust dump information.
5200 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5202 * tree-loop-distribution.c: Adjust the general comment.
5203 (NUM_PARTITION_THRESHOLD): New macro.
5204 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5205 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5206 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5207 in call to build_partition_graph.
5208 (finalize_partitions): New parameter. Make loop distribution more
5209 conservative by fusing more partitions.
5210 (distribute_loop): Don't do runtime alias check in case of loop nest
5212 (find_seed_stmts_for_distribution): New function.
5213 (prepare_perfect_loop_nest): New function.
5214 (pass_loop_distribution::execute): Refactor code finding seed stmts
5215 and loop nest into above functions. Support loop nest distribution.
5216 Adjust dump information accordingly.
5218 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5220 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5221 and set PTYPE_SEQUENTIAL for merged partition.
5223 2017-10-12 Richard Biener <rguenther@suse.de>
5225 PR tree-optimization/69728
5227 2017-09-19 Richard Biener <rguenther@suse.de>
5229 PR tree-optimization/69728
5230 * graphite-sese-to-poly.c (schedule_error): New global.
5231 (add_loop_schedule): Handle empty domain by failing the
5233 (build_original_schedule): Handle schedule_error.
5235 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5236 domain by returning an unchanged schedule.
5238 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5240 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5241 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5243 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5245 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5248 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5251 * expr.c (store_field): Don't optimize away bitsize == 0 store
5252 from CALL_EXPR with addressable return type.
5254 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5256 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5257 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5258 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5259 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5260 TARGET_ISEL instead of TARGET_ISEL<sel>.
5262 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5264 * config/rs6000/rs6000.c
5265 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5267 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5269 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5271 (reg_or_cint_operand): Fix comment.
5272 (reg_or_zero_operand): New predicate.
5273 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5274 * config/rs6000/rs6000.c (output_isel): Delete.
5275 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5276 instead of reg_or_cint_operand. Output instruction directly (not via
5278 (isel_unsigned_<mode>): Ditto.
5279 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5280 gpc_reg_operand. Add an instruction alternative for this. Output
5281 instruction directly.
5282 (*isel_reversed_unsigned_<mode>): Ditto.
5284 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5286 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5287 (TARGET_CANONICALIZE_COMPARISON): Define.
5289 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5292 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5293 Check whether the dest is REG before adding REG_EQUIV note.
5295 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5298 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5300 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5301 (make_hard_regno_born, make_hard_regno_dead): Update
5302 bb_killed_pseudos and bb_gen_pseudos.
5304 2017-10-11 Nathan Sidwell <nathan@acm.org>
5306 * incpath.h (enum incpath_kind): Name enum, prefix values.
5307 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5308 * incpath.c (heads, tails): Use INC_MAX.
5309 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5310 (merge_include_chains, split_quote_chain,
5311 register_include_chains): Update incpath_kind names.
5312 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5313 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5315 (add_framework_path, darwin_register_objc_includes): Likewise.
5316 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5318 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5320 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5321 Do not use float_operator operator predicate.
5322 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5323 * config/i386/predicates.md (float_operator): Remove predicate.
5325 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5327 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5328 (*jccxf_i387): Ditto.
5329 (*jcc<mode>_i387): Ditto.
5330 (*jccu<mode>_i387): Ditto.
5331 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5332 (*jcc_*_i387 splitters): Remove.
5333 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5334 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5335 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5338 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5340 * profile-count.h (slow_safe_scale_64bit): New function.
5341 (safe_scale_64bit): New inline.
5342 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5343 * profile-count.c: Include wide-int.h
5344 (slow_safe_scale_64bit): New.
5346 2017-10-11 Nathan Sidwell <nathan@acm.org>
5348 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5349 HAS_DECL_ASSEMBLER_NAME_P.
5350 * gimple-expr.c (gimple_decl_printable_name: Check
5351 HAS_DECL_ASSEMBLER_NAME_P too.
5352 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5353 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5354 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5355 * passes.c (rest_of_decl_compilation): Check
5356 HAS_DECL_ASSEMBLER_NAME_P too.
5357 * recog.c (verify_changes): Likewise.
5358 * tree-pretty-print.c (dump_decl_name): Likewise.
5359 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5361 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5362 (SET_DECL_ASSEMBLER_NAME): Use it.
5363 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5364 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5365 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5367 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5369 * config.gcc (i386, x86_64): Add extra objects.
5370 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5371 (ix86_min_insn_size): Declare.
5372 (ix86_issue_rate): Declare.
5373 (ix86_adjust_cost): Declare.
5374 (ia32_multipass_dfa_lookahead): Declare.
5375 (ix86_macro_fusion_p): Declare.
5376 (ix86_macro_fusion_pair_p): Declare.
5377 (ix86_bd_has_dispatch): Declare.
5378 (ix86_bd_do_dispatch): Declare.
5379 (ix86_core2i7_init_hooks): Declare.
5380 (ix86_atom_sched_reorder): Declare.
5381 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5382 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5383 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5384 (rip_relative_addr_p): Rename to ...
5385 (ix86_rip_relative_addr_p): ... this one; export.
5386 (memory_address_length): Update.
5387 (ix86_issue_rate): Move to x86-tune-sched.c.
5388 (ix86_flags_dependent): Move to x86-tune-sched.c.
5389 (ix86_agi_dependent): Move to x86-tune-sched.c.
5390 (exact_dependency_1): Move to x86-tune-sched.c.
5391 (exact_store_load_dependency): Move to x86-tune-sched.c.
5392 (ix86_adjust_cost): Move to x86-tune-sched.c.
5393 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5394 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5395 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5396 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5397 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5398 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5399 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5400 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5401 (min_insn_size): Rename to ...
5402 (ix86_min_insn_size): ... this one; export.
5403 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5404 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5405 (core2i7_first_cycle_multipass_backtrack): Move to
5406 x86-tune-sched-core.c.
5407 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5408 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5409 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5410 (ix86_avoid_jump_mispredicts): Update.
5411 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5412 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5413 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5414 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5415 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5416 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5417 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5418 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5419 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5420 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5421 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5422 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5423 (BIG): Move to ix86-tune-sched-bd.c.
5424 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5425 (enum insn_path): Move to ix86-tune-sched-bd.c.
5426 (get_mem_group): Move to ix86-tune-sched-bd.c.
5427 (is_cmp): Move to ix86-tune-sched-bd.c.
5428 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5429 (is_branch): Move to ix86-tune-sched-bd.c.
5430 (is_prefetch): Move to ix86-tune-sched-bd.c.
5431 (init_window): Move to ix86-tune-sched-bd.c.
5432 (allocate_window): Move to ix86-tune-sched-bd.c.
5433 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5434 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5435 (process_end_window): Move to ix86-tune-sched-bd.c.
5436 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5437 (find_constant): Move to ix86-tune-sched-bd.c.
5438 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5439 (has_immediate): Move to ix86-tune-sched-bd.c.
5440 (get_insn_path): Move to ix86-tune-sched-bd.c.
5441 (get_insn_group): Move to ix86-tune-sched-bd.c.
5442 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5443 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5444 (add_insn_window): Move to ix86-tune-sched-bd.c.
5445 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5446 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5447 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5448 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5449 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5450 (do_dispatch): Move to ix86-tune-sched-bd.c.
5451 (has_dispatch): Move to ix86-tune-sched-bd.c.
5452 * config/i386/t-i386: Add new object files.
5453 * config/i386/x86-tune-costs.h: New file.
5454 * config/i386/x86-tune-sched-atom.c: New file.
5455 * config/i386/x86-tune-sched-bd.c: New file.
5456 * config/i386/x86-tune-sched-core.c: New file.
5457 * config/i386/x86-tune-sched.c: New file.
5459 2017-10-11 Liu Hao <lh_mouse@126.com>
5461 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5462 the generic version below instead.
5463 (should_colorize): Recognize Windows consoles as terminals
5465 * pretty-print.c [__MINGW32__] (write_all): New function.
5466 [__MINGW32__] (find_esc_head): Likewise.
5467 [__MINGW32__] (find_esc_terminator): Likewise.
5468 [__MINGW32__] (eat_esc_sequence): Likewise.
5469 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5471 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5474 2017-10-11 Richard Biener <rguenther@suse.de>
5476 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5477 Properly call analyze_scalar_evolution with the loop of the stmt.
5479 2017-10-11 Richard Biener <rguenther@suse.de>
5481 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5482 * tree-core.h (tree_base): Add chrec_var union member.
5483 * tree.h (CHREC_VAR): Remove.
5484 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5485 * tree-chrec.h (build_polynomial_chrec): Adjust.
5486 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5487 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5489 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5491 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5492 * match.pd: ... here.
5493 ((T) X == (T) Y): Relax condition.
5495 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5497 PR tree-optimization/82472
5498 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5500 (break_alias_scc_partitions): Update postorder number.
5502 2017-10-11 Martin Liska <mliska@suse.cz>
5505 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5507 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5509 2017-10-11 Martin Liska <mliska@suse.cz>
5511 * print-rtl.c (print_insn): Move declaration of idbuf
5512 to same scope as name.
5514 2017-10-11 Martin Liska <mliska@suse.cz>
5519 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5522 2017-10-11 Martin Liska <mliska@suse.cz>
5525 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5528 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5530 * config.gcc (default_gnu_indirect_function): Default to yes for
5531 arm*-*-linux* with glibc.
5533 2017-10-11 Richard Biener <rguenther@suse.de>
5535 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5536 default-defs and types we do not want to analyze.
5537 (interpret_loop_phi): Replace unreachable code with an assert.
5538 (compute_scalar_evolution_in_loop): Remove and inline ...
5539 (analyze_scalar_evolution_1): ... here, replacing condition with
5540 what makes the intent clearer. Remove handling of cases
5541 get_scalar_evolution now handles.
5543 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5545 PR rtl-optimization/81434
5546 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5547 comment for main loop. In sched_group_found if, also add checks for
5548 pass and min_cost_group.
5550 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5552 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5553 (rs6000_insn_cost): New function.
5554 * config/rs6000/rs6000.md (cost): New attribute.
5556 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5557 H.J. Lu <hongjiu.lu@intel.com>
5561 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5562 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5563 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5564 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5565 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5566 OPTION_MASK_ISA_MMX and some other bit set require both
5567 mmx and the ISAs without the mmx bit.
5568 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5569 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5570 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5571 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5572 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5573 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5574 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5575 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5576 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5577 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5578 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5579 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5580 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5581 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5582 Add OPTION_MASK_ISA_MMX.
5584 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5586 * config.gcc (armv7*-*-freebsd*): New target.
5587 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5589 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5591 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5592 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5595 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5597 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5598 a template parameter.
5599 (WIDE_INT_REF_FOR): Update accordingly.
5600 * tree.h (wi::int_traits <const_tree>): Delete.
5601 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5602 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5603 (wi::tree_to_wide_ref): New typedef.
5604 (wi::to_wide): New function.
5605 * calls.c (get_size_range): Use wi::to_wide when operating on
5607 * cgraph.c (cgraph_node::create_thunk): Likewise.
5608 * config/i386/i386.c (ix86_data_alignment): Likewise.
5609 (ix86_local_alignment): Likewise.
5610 * dbxout.c (stabstr_O): Likewise.
5611 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5612 * expr.c (const_vector_from_tree): Likewise.
5613 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5614 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5615 (fold_negate_expr_1, int_const_binop_1, const_binop)
5616 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5617 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5618 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5619 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5620 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5621 (fold_not_const, round_up_loc): Likewise.
5622 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5623 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5624 (alloca_call_type): Likewise.
5625 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5626 * godump.c (go_output_typedef): Likewise.
5627 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5628 * internal-fn.c (get_min_precision): Likewise.
5629 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5630 * ipa-polymorphic-call.c
5631 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5632 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5633 (ipa_modify_call_arguments): Likewise.
5634 * match.pd: Likewise.
5635 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5636 * print-tree.c (print_node_brief, print_node): Likewise.
5637 * stmt.c (expand_case): Likewise.
5638 * stor-layout.c (layout_type): Likewise.
5639 * tree-affine.c (tree_to_aff_combination): Likewise.
5640 * tree-cfg.c (group_case_labels_stmt): Likewise.
5641 * tree-data-ref.c (dr_analyze_indices): Likewise.
5642 (prune_runtime_alias_test_list): Likewise.
5643 * tree-dump.c (dequeue_and_dump): Likewise.
5644 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5645 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5646 * tree-pretty-print.c (dump_generic_node): Likewise.
5647 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5648 (simple_iv_with_niters): Likewise.
5649 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5650 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5651 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5652 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5653 (refine_value_range_using_guard, number_of_iterations_ne_max)
5654 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5655 (get_cst_init_from_scev, record_nonwrapping_iv)
5656 (scev_var_range_cant_overflow): Likewise.
5657 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5658 * tree-ssa-pre.c (compute_avail): Likewise.
5659 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5660 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5661 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5662 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5663 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5664 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5665 (dump_case_nodes, try_switch_expansion): Likewise.
5666 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5667 (vect_do_peeling): Likewise.
5668 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5669 * tree-vect-stmts.c (vectorizable_load): Likewise.
5670 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5671 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5672 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5673 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5674 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5675 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5676 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5677 (evrp_dom_walker::before_dom_children): Likewise.
5678 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5679 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5680 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5681 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5682 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5683 (get_range_pos_neg): Likewise.
5684 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5685 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5686 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5687 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5688 * config/avr/avr.c (avr_fold_builtin): Likewise.
5689 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5690 * config/msp430/msp430.c (msp430_attr): Likewise.
5691 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5692 * config/powerpcspe/powerpcspe-c.c
5693 (altivec_resolve_overloaded_builtin): Likewise.
5694 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5695 (rs6000_expand_ternop_builtin): Likewise.
5696 * config/rs6000/rs6000-c.c
5697 (altivec_resolve_overloaded_builtin): Likewise.
5698 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5699 (rs6000_expand_ternop_builtin): Likewise.
5700 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5702 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5704 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5705 when copying loop nest with only one inner loop.
5707 2017-10-10 Richard Biener <rguenther@suse.de>
5709 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5710 blocks if SCEV is active.
5711 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5713 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5714 (scev_initialize): Assert we are not yet initialized.
5716 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5718 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5719 inner loop's exit stmt by making it always exit the loop, otherwise
5720 we would generate an infinite empty loop.
5722 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5724 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5725 renaming variables in new preheader if it's deleted.
5727 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5729 * tree-loop-distribution.c (struct partition): Remove unused field
5730 loops of the structure.
5731 (partition_alloc, partition_free): Ditto.
5732 (build_rdg_partition_for_vertex): Ditto.
5734 2017-10-09 Jeff Law <law@redhat.com>
5736 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5737 return type to match prototype and documentation.
5739 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5741 * config/rs6000/rs6000.c (processor_costs): Move to ...
5742 * config/rs6000/rs6000.h: ... here.
5743 (rs6000_cost): Declare.
5745 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5747 * except.c (setjmp_fn): New global variable.
5748 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5749 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5750 if DONT_USE_BUILTIN_SETJMP is defined.
5752 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5754 * target.def (insn_cost): New hook.
5755 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5756 * doc/tm.texi: Regenerate.
5757 * rtlanal.c (insn_cost): Use the new hook.
5759 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5761 * combine.c (combine_validate_cost): Compute the new insn_cost,
5762 not just pattern_cost.
5763 (try_combine): Adjust comment.
5765 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5767 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5769 * combine.c (uid_insn_cost): Adjust comment.
5770 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5772 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5773 * dse.c (find_shift_sequence): Ditto.
5774 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5775 (bb_valid_for_noce_process_p): Use pattern_cost.
5776 * rtl.h (insn_rtx_cost): Delete.
5777 (pattern_cost): New prototype.
5778 (insn_cost): New prototype.
5779 * rtlanal.c (insn_rtx_cost): Rename to...
5780 (pattern_cost): ... this.
5783 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5785 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5786 (*jcc<mode>_0_r_i387): Ditto.
5787 (*jccxf_r_i387): Ditto.
5788 (*jcc<mode>_r_i387): Ditto.
5789 (*jccu<mode>_r_i387): Ditto.
5790 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5791 (*jcc): Rename from *jcc_1.
5793 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5795 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5796 deferred rescans after the lvx/stvx recombination pre-pass.
5798 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5800 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5801 memory operation instruction support.
5802 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5803 (rs6000-ibm-aix[789]*): Likewise.
5804 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5805 Document new functions.
5807 2017-10-09 Richard Biener <rguenther@suse.de>
5809 PR tree-optimization/82397
5810 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5811 equality only for semantically equal trees.
5813 2017-10-09 Richard Biener <rguenther@suse.de>
5815 PR tree-optimization/82449
5816 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5817 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5818 allow constant addresses.
5819 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5822 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5824 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5827 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5830 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5833 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5836 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5838 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5841 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5842 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5844 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5846 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5847 (WI_BINARY_PREDICATE_RESULT): Likewise.
5848 (wi::binary_traits::operator_result): New type.
5849 (wi::binary_traits::predicate_result): Likewise.
5850 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5851 (generic_wide_int::operator==, generic_wide_int::operator!=)
5852 (generic_wide_int::operator&, generic_wide_int::and_not)
5853 (generic_wide_int::operator|, generic_wide_int::or_not)
5854 (generic_wide_int::operator^, generic_wide_int::operator+
5855 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5857 (operator~, unary operator-, operator==, operator!=, operator&)
5858 (operator|, operator^, operator+, binary operator-, operator*): New
5860 * expr.c (get_inner_reference): Use wi::bit_and_not.
5861 * fold-const.c (fold_binary_loc): Likewise.
5862 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5863 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5864 (bit_value_binop): Likewise.
5865 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5866 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5867 (extract_range_from_binary_expr_1): Likewise.
5868 (masked_increment): Likewise.
5869 (simplify_bit_ops_using_ranges): Likewise.
5871 2017-10-09 Martin Jambor <mjambor@suse.cz>
5874 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5875 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5876 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5878 (hsa_fixup_mov_insn_type): New function.
5879 (hsa_op_with_type::get_in_type): Use it.
5880 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5881 immediates in an assert.
5882 (hsa_op_with_type::extend_int_to_32bit): New method.
5883 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5884 types. Convert to dest type if necessary.
5885 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5886 (reg_for_gimple_ssa): Pass false as min32int to
5887 hsa_type_for_scalar_tree_type.
5888 (gen_hsa_addr): Fixup type when creating addresable temporary.
5889 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5890 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5891 necessary. Call hsa_fixup_mov_insn_type.
5892 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5893 extend operands and convert to dest type if necessary.
5894 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5895 to dest type if necessary.
5896 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5897 if conversion nt necessary and size matches.
5898 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5899 to dest type if necessary.
5900 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5901 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5903 (gen_hsa_clrsb): Likewise.
5904 (gen_hsa_ffs): Likewise.
5905 (gen_hsa_divmod): Extend operands and convert to dest type if
5907 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5909 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5911 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5912 Remove empty default arguments. Use a brace block as output
5914 (conditional return): Ditto.
5916 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5917 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5919 (group_ending_nop): Ditto.
5920 (doloop_end): Ditto.
5921 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5922 (splitters for those): Ditto.
5924 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5926 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5927 a conditional jump (and the compare for it) so that pc_rtx is the
5929 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5930 for the deleted and renamed ctr<mode>_internal[234] patterns.
5931 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5932 Delete second conditional return pattern.
5933 (ctr<mode>_internal2): Delete this second bdnz pattern.
5934 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5935 (ctr<mode>_internal4): Delete this second bdz pattern.
5937 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5939 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5940 (always_initialized_rtx_for_ssa_name_p): New predicate.
5941 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5942 (finish_out_of_ssa): Free new field of SA.
5943 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5944 * tree-ssa-coalesce.c: Include tree-ssa.h.
5945 (get_parm_default_def_partitions): Remove extern keyword.
5946 (get_undefined_value_partitions): New function.
5947 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5948 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5950 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5952 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5954 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5956 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5958 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5959 for targets that preffer 128bit.
5961 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5963 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5965 2017-10-08 Olivier Hainque <hainque@adacore.com>
5967 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5968 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5970 (thumb_set_return_address): Likewise.
5972 2017-10-08 Olivier Hainque <hainque@adacore.com>
5974 * common/config/arm/arm-common.c (arm_except_unwind_info):
5975 Handle DWARF2_UNWIND_INFO.
5977 2017-10-07 Michael Collison <michael.collison@arm.com>
5979 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5982 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5984 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5985 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5986 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5987 defined, force the creation of a new block for a dispatch label.
5989 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5991 * invoke.texi (Wsuggest-attribute=cold): Document.
5992 * common.opt (Wsuggest-attribute=cold): New
5993 * ipa-pure-const.c (warn_function_cold): New function.
5994 * predict.c (compute_function_frequency): Use it.
5995 * predict.h (warn_function_cold): Declare.
5997 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5999 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6002 2017-10-06 Martin Liska <mliska@suse.cz>
6004 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6005 keyword for member functions.
6006 (struct sanopt_tree_couple): New struct.
6007 (struct sanopt_tree_couple_hash): New function.
6008 (struct sanopt_ctx): Add new hash_map.
6009 (has_dominating_ubsan_ptr_check): New function.
6010 (record_ubsan_ptr_check_stmt): Likewise.
6011 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6012 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6013 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6015 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6018 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6019 aarch64_simd_valid_immediate on CONST_VECTORs.
6020 (aarch64_reg_or_bic_imm): Likewise.
6022 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6024 PR rtl-optimization/82396
6025 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6027 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6029 * graphite-dependences.c (scop_get_reads): Move code to...
6030 (scop_get_must_writes): Move code to...
6031 (scop_get_may_writes): Move code to...
6032 (scop_get_reads_and_writes): ... here.
6033 (scop_get_dependences): Call scop_get_reads_and_writes.
6035 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6037 PR tree-optimization/82434
6038 * fold-const.h (can_native_encode_type_p,
6039 can_native_encode_string_p): Remove.
6040 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6041 don't encode anything, just return what would be otherwise returned.
6042 (native_encode_fixed, native_encode_complex, native_encode_vector):
6044 (native_encode_string): Likewise. Inline by hand
6045 can_native_encode_string_p.
6046 (can_native_encode_type_p): Remove.
6047 (can_native_encode_string_p): Remove.
6048 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6049 STRING_CSTs using can_native_encode_string_p, test all
6050 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6051 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6052 argument from native_encode_expr.
6053 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6054 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6057 2017-10-06 Richard Biener <rguenther@suse.de>
6059 PR tree-optimization/82397
6060 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6061 operand_equal_p but rely on data_ref_compare_tree for detecting
6063 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6064 to match up with dr_group_sort_cmp.
6066 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6069 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6071 * config/s390/s390-builtin-types.def: Regenerate.
6073 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6076 * config/s390/s390-builtin-types.def: Regenerate.
6077 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6078 Change flag from B_VXE to B_VX.
6079 (s390_vec_min_dbl): Remove B_VXE flag.
6081 2017-10-06 Richard Biener <rguenther@suse.de>
6083 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6084 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6085 translate_isl_ast_to_gimple::is_valid_rename,
6086 translate_isl_ast_to_gimple::get_rename,
6087 translate_isl_ast_to_gimple::get_def_bb_for_const,
6088 translate_isl_ast_to_gimple::get_new_name,
6089 translate_isl_ast_to_gimple::collect_all_ssa_names,
6090 translate_isl_ast_to_gimple::copy_loop_phi_args,
6091 translate_isl_ast_to_gimple::collect_all_ssa_names,
6092 translate_isl_ast_to_gimple::copy_loop_phi_args,
6093 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6094 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6095 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6096 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6097 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6098 translate_isl_ast_to_gimple::copy_cond_phi_args,
6099 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6100 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6101 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6102 translate_isl_ast_to_gimple::rename_uses,
6103 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6104 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6105 (set_rename_for_each_def): Likewise.
6106 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6107 here. Handle rewriting SCEV analyzable uses here.
6108 (copy_bb_and_scalar_dependences): Generate code for PHI
6110 (graphite_regenerate_ast_isl): Adjust.
6111 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6112 (add_write, add_read): New functions.
6113 (build_cross_bb_scalars_def): Use it and simplify.
6114 (build_cross_bb_scalars_use): Likewise.
6115 (graphite_find_cross_bb_scalar_vars): Inline into...
6116 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6117 simulating out-of-SSA. Compute liveout and add dependencies.
6118 (build_scops): Force an empty entry block.
6119 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6121 (sese_build_liveouts): Declare.
6122 (sese_trivially_empty_bb_p): Likewise.
6123 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6124 compute liveout and debug_liveout.
6125 (sese_bad_liveouts_use): Remove.
6126 (sese_reset_debug_liveouts_bb): Likewise.
6127 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6128 (sese_build_liveouts): Build liveout and debug_liveout and store
6130 (new_sese_info): Adjust.
6131 (free_sese_info): Likewise.
6132 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6133 do not build liveout here.
6134 (move_sese_in_condition): Adjust region entry.
6135 (scev_analyzable_p): Match up with chrec_apply requirements.
6136 (sese_trivially_empty_bb_p): New.
6137 * tree-into-ssa.c (get_reaching_def): Properly support generating
6138 default-defs for incremental rewrite of anonymous names.
6140 2017-10-06 Richard Biener <rguenther@suse.de>
6142 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6143 precision do not perform modulo reduction.
6145 2017-10-06 Richard Biener <rguenther@suse.de>
6147 PR tree-optimization/82436
6148 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6149 conservatively choose the vectorization factor when checking
6150 whether we can perform the required load permutation.
6151 (vect_transform_slp_perm_load): Assert when we may not fail.
6153 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6155 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6156 message for incompatible -msdata=* and -mcall-* options.
6158 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6160 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6161 rate for post-reload scheduling.
6163 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6165 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6167 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6169 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6170 to improve monte carlo in scimark.
6172 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6174 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6175 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6176 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6177 pentium4_cost, nocona_cost): Set reassociation width to 1.
6178 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6179 width to 2 for fp operations and 1 otherwise.
6180 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6182 (atom_cost): Set reassociation width to 2.
6183 (slm_cost, generic_cost): Set fp reassociation width
6184 to 2 and 1 otherwise.
6185 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6186 (core_cost): Set fp reassociation width to 4 and vector to 2.
6187 (ix86_reassociation_width): Rewrite using cost table; special case
6188 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6189 and TARGET_AVX128_OPTIMAL.
6190 * config/i386/i386.h (processor_costs): Add
6191 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6192 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6193 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6194 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6195 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6196 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6198 2017-10-05 Nathan Sidwell <nathan@acm.org>
6200 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6202 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6204 * config/arm/arm.c (arm_test_fpu_data): New.
6205 (arm_run_selftests): Call arm_test_fpu_data.
6207 2017-10-04 Nathan Sidwell <nathan@acm.org>
6209 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6211 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6213 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6215 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6216 check type for aarch64_simd_valid_immediate.
6217 (aarch64_output_simd_mov_immediate): Update prototype.
6218 (aarch64_simd_valid_immediate): Update prototype.
6219 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6220 support for ORR-immediate.
6221 (and<mode>3): modified pattern to add support for BIC-immediate.
6222 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6223 now checks for valid immediate for BIC and ORR based on new enum
6225 (aarch64_output_simd_mov_immediate): Function now used to output
6226 BIC/ORR imm as well based on new enum argument.
6227 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6229 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6230 (aarch64_reg_or_bic_imm): Likewise.
6232 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6234 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6235 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6237 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6241 PR rtl-optimization/82396
6242 * haifa-sched.c (autopref_multipass_init): Simplify
6244 (autopref_rank_data): Simplify sort order.
6245 * sched-int.h (autopref_multipass_data_): Remove
6246 multi_mem_insn_p, min_offset and max_offset.
6248 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6250 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6252 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6254 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6257 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6259 * doc/sourcebuild.texi: Document vect_long_mult.
6261 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6263 PR tree-optimization/82413
6264 * fold-const.c (build_range_check): Use widest_int when comparing
6265 the maximum ETYPE value with HIGH.
6267 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6269 PR rtl-optimization/82396
6270 * haifa-sched.c (autopref_multipass_init): Simplify
6272 (autopref_rank_data): Simplify sort order.
6273 * sched-int.h (autopref_multipass_data_): Remove
6274 multi_mem_insn_p, min_offset and max_offset.
6276 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6278 PR tree-optimization/82381
6279 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6280 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6283 PR tree-optimization/82374
6284 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6285 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6286 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6287 current_function_decl to the new decl.
6289 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6291 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6292 helper macro for IEEE float128 hardware built-in functions.
6293 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6295 (TRUNCF128_ODD): Likewise.
6296 (ADDF128_ODD): Likewise.
6297 (SUBF128_ODD): Likewise.
6298 (MULF128_ODD): Likewise.
6299 (DIVF128_ODD): Likewise.
6300 (FMAF128_ODD): Likewise.
6301 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6302 UNSPEC_TRUNC_ROUND_TO_ODD.
6303 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6304 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6305 floating point round to odd instructions.
6306 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6307 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6308 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6309 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6310 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6311 (trunc<mode>sf2_hw): Change the truncate with round to odd
6312 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6313 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6314 to odd hardware instructions.
6315 (sub<mode>3_odd): Likewise.
6316 (mul<mode>3_odd): Likewise.
6317 (div<mode>3_odd): Likewise.
6318 (sqrt<mode>2_odd): Likewise.
6319 (fma<mode>4_odd): Likewise.
6320 (fms<mode>4_odd): Likewise.
6321 (nfma<mode>4_odd): Likewise.
6322 (nfms<mode>4_odd): Likewise.
6323 (trunc<mode>df2_odd): Change the truncate with round to odd
6324 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6326 * doc/extend.texi (PowerPC built-in functions): Update documentation
6327 for existing IEEE float128-bit built-in functions. Add built-in
6328 functions that generate the IEEE 128-bit floating point round to
6331 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6333 PR rtl-optimization/77729
6334 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6335 to (X&(C1&~C2))|C2 transformations.
6337 2017-10-03 Martin Jambor <mjambor@suse.cz>
6339 PR tree-optimization/82363
6340 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6341 mismatch, mark lacc written regardless of racc.
6343 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6345 PR tree-optimization/82381
6346 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6347 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6348 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6349 fallthrough into reassoc_stmt_dominates_stmt_p.
6352 * combine.c (combine_instructions): Don't combine in unreachable
6355 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6358 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6359 function to not use the have_cpu variable. Do not set cpu_index,
6360 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6362 (rs6000_valid_attribute_p): Remove duplicate initializations of
6363 old_optimize and func_optimize.
6364 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6365 (rs6000_activate_target_options): Make global.
6366 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6369 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6371 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6372 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6373 Set *poffset to 0 and *psize and *pmax_size to -1 if
6374 *poffset + *psize overflows in HOST_WIDE_INT.
6376 PR tree-optimization/82387
6377 PR tree-optimization/82388
6378 PR tree-optimization/82389
6379 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6380 instead of live_bytes non-NULL.
6382 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6385 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6388 2017-10-02 Richard Biener <rguenther@suse.de>
6390 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6391 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6392 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6394 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6396 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6397 requested precision matches the type's.
6398 * calls.c (alloc_max_size): Calculate the new candidate size as
6399 a widest_int and use wi::to_widest when comparing it with the
6400 current candidate size.
6401 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6402 zero rather than integer_zero_node.
6403 * match.pd: Check for a no-op conversion before using wi::add
6404 rather than after. Use tree_to_uhwi when summing small shift
6405 counts into an unsigned int.
6407 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6408 Alan Hayward <alan.hayward@arm.com>
6409 David Sherwood <david.sherwood@arm.com>
6412 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6413 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6414 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6415 POINTER_AND_FP_REGS.
6417 2017-10-02 Richard Biener <rguenther@suse.de>
6419 PR tree-optimization/82355
6420 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6421 a mapping for the enclosing loop but avoid generating one for
6423 (copy_bb_and_scalar_dependences): Remove premature codegen
6424 error on PHIs in blocks duplicated into multiple places.
6425 * graphite-scop-detection.c
6426 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6427 in the region use it as loop and nest to analyze the DR in.
6428 (try_generate_gimple_bb): Likewise.
6429 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6430 (add_loop_constraints): For blocks in a loop not in the region
6431 create a dimension with a single iteration.
6432 * sese.h (gbb_loop_at_index): Remove assert.
6434 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6436 * omp-expand.c (adjust_context_scope): New function.
6437 (expand_parallel_call): Call adjust_context_scope.
6439 2017-10-01 Jeff Law <law@redhat.com>
6441 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6442 dom_opt_dom_walker class with direct access to private members.
6443 Add comments. Call test_for_singularity.
6444 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6445 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6446 m_dummy_cond anymore.
6447 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6449 (pass_dominator:execute): Build the dummy_cond here and pass it
6450 to the dom_opt_dom_walker ctor.
6451 (test_for_singularity): New function.
6453 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6454 Maya Rashish <coypu@sdf.org>
6456 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6457 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6458 (alpha*-*-netbsd*) Use nbsd_tm_file.
6459 (arm*-*-netbsdelf*) Likewise.
6460 (i[34567]86-*-netbsdelf*) Likewise.
6461 (x86_64-*-netbsd*) Likewise.
6462 (mips*-*-netbsd*) Likewise.
6463 (powerpc-*-netbsd*) Likewise.
6464 (sh*-*-netbsd*) Likewise.
6465 (sparc-*-netbsdelf*) Likewise.
6466 (sparc64-*-netbsd*) Likewise.
6467 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6469 (vax-*-netbsdelf*) Likewise.
6470 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6471 (UINT_FAST8_TYPE) Likewise.
6472 (INT_FAST16_TYPE) Check CHAR_FAST16.
6473 (UINT_FAST16_TYPE) Likewise.
6475 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6478 * config/i386/i386.md
6479 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6480 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6481 *divmodsi4_zext_2): New define_insn_and_split.
6482 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6483 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6484 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6485 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6486 New define_insn_and_split.
6487 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6488 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6489 operands[1] having DImode when mode is SImode.
6491 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6492 always SImode for DIV and MOD in REG_EQUAL notes.
6494 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6497 * match.pd: Fix handling of NaNs in pattern.
6499 2017-09-29 Jeff Law <law@redhat.com>
6501 * sbitmap.c (bitmap_bit_in_range_p): New function.
6502 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6503 * tree-ssa-dse.c (live_bytes_read): New function.
6504 (dse_classify_store): Ignore reads of dead bytes.
6506 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6507 typos and whitespace errors.
6508 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6509 * config/s390/s390.c (s390_emit_prologue): Likewise.
6511 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6514 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6515 with a symbol for LRA.
6517 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6519 PR rtl-optimization/82338
6520 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6522 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6524 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6525 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6526 (qsort_chk): Declare.
6527 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6528 (qsort_chk): New function.
6530 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6532 PR tree-optimization/82337
6533 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6534 phi definition if the PHI result appears in an abnormal PHI.
6535 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6536 basis appears in an abnormal PHI.
6538 2017-09-29 Richard Biener <rguenther@suse.de>
6540 * graphite-isl-ast-to-gimple.c
6541 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6542 (binary_op_to_tree): Use it.
6543 (get_rename_from_scev): Likewise.
6544 (copy_loop_phi_nodes): Likewise.
6545 (copy_bb_and_scalar_dependences): Likewise.
6546 (translate_pending_phi_nodes): Likewise.
6548 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6551 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6552 for movabsq $(i32 << shift), r64.
6554 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6556 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6557 index when encoding %esp as %rsp to avoid 0x67 prefix.
6559 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6561 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6562 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6564 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6566 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6567 Extensions with more than 16 double VFP registers.
6568 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6569 to_clear_mask and all code related to it. Replace the remaining
6570 entry by a sbitmap and adapt code accordingly.
6572 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6574 * brig-builtins.def: Change pure attributes to const.
6576 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6578 * config.gcc (default_gnu_indirect_function): Default to yes for
6579 sparc*-*-linux* with glibc.
6581 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6583 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6584 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6585 when creating .init_array and .fini_array sections with priority
6588 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6591 * config/aarch64/aarch64.c
6592 (aarch64_builtin_support_vector_misalignment): Always return false
6593 when misalignment is unknown.
6595 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6597 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6598 this function to return false if the definition used by the swap
6599 instruction is artificial, or if the memory address from which the
6600 constant value is loaded is not represented by a base address held
6601 in a register or if the base address register is a frame or stack
6602 pointer. Additionally, return false if the base address of the
6603 loaded constant is a SYMBOL_REF but is not considered to be a
6605 (replace_swapped_load_constant): New function.
6606 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6607 loaded constant vector with a load of a swapped constant vector.
6609 2017-09-27 Carl Love <cel@us.ibm.com>
6611 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6612 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6613 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6616 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6618 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6619 first, always call autopref_rank_data otherwise.
6621 2017-09-27 Richard Biener <rguenther@suse.de>
6623 * graphite-scop-detection.c (find_scop_parameters): Move
6624 loop bound handling ...
6625 (gather_bbs::before_dom_children): ... here, avoiding the need
6626 to build scop_info->loop_nest.
6627 (record_loop_in_sese): Remove.
6628 * sese.h (sese_info_t::loop_nest): Remove.
6629 * sese.c (new_sese_info): Do not allocate loop_nest.
6630 (free_sese_info): Do not free loop_nest.
6632 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6635 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6636 lhs from calls if the lhs has addressable type.
6638 2017-09-27 Richard Biener <rguenther@suse.de>
6640 * graphite.h (scop::max_alias_set): New member.
6641 * graphite-scop-detection.c: Remove references to non-existing
6642 --param in comments.
6643 (build_alias_sets): Record the maximum alias set used for drs.
6644 (build_scops): Support zero as unlimited for
6645 --param graphite-max-arrays-per-scop.
6646 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6648 (build_poly_sr_1): ... here. Compute alias set based on the
6649 maximum alias set used for drs rather than
6650 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6652 2017-09-27 Richard Biener <rguenther@suse.de>
6654 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6655 --param loop-block-tile-size=0 to disable tiling.
6657 2017-09-27 Richard Biener <rguenther@suse.de>
6659 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6660 (graphite-max-nb-scop-params): Document special value zero.
6661 * domwalk.h (dom_walker::STOP): New symbolical constant.
6662 (dom_walker::dom_walker): Add optional parameter for bb to
6664 (dom_walker::~dom_walker): Declare.
6665 (dom_walker::before_dom_children): Document STOP return value.
6666 (dom_walker::m_user_bb_to_rpo): New member.
6667 (dom_walker::m_bb_to_rpo): Likewise.
6668 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6669 mapping here if not provided by the user.
6670 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6671 provided by the user.
6672 (dom_walker::STOP): Define.
6673 (dom_walker::walk): Do not compute bb to RPO mapping here.
6674 Support STOP return value from before_dom_children to stop
6676 * graphite-optimize-isl.c (optimize_isl): If the schedule
6677 is the same still generate code if -fgraphite-identity
6678 or -floop-parallelize-all are given.
6679 * graphite-scop-detection.c: Include cfganal.h.
6680 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6682 (gather_bbs::before_dom_children): Return STOP for BBs
6684 (build_scops): Compute bb to RPO mapping and pass it to
6685 the domwalk. Treat --param graphite-max-nb-scop-params=0
6686 as not limiting the number of params.
6687 * graphite.c (graphite_initialize): Remove limit on the
6688 number of basic-blocks in a function.
6689 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6690 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6691 default value of 10.
6693 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6695 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6696 Adjust code to eliminate needing to do the shift right 32-bits
6697 operation after XSCVDPSPN.
6699 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6701 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6702 ((X / Y) != 0 -> X >= Y): Likewise.
6704 2017-09-26 Carl Love <cel@us.ibm.com>
6706 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6707 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6708 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6709 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6710 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6711 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6712 definitions and overloading.
6713 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6714 statement for P9V_BUILTIN_XST_LEN_R.
6715 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6716 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6717 define_expand and define_insn for the instructions and builtins.
6718 * doc/extend.texi: Update the built-in documentation file for the new
6720 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6721 define_insn for the instructions
6723 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6726 * gcc/config/netbsd-protos.h: New file.
6727 * gcc/config/netbsd.c: New file.
6728 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6729 * gcc/config/t-netbsd: New file.
6730 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6731 (tmake_file) Add t-netbsd.
6732 (extra_objs) Add netbsd.o.
6734 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6738 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6740 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6742 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6743 sign extension from a vector register to a GPR by doing a 32-bit
6744 direct move and then an EXTSW.
6745 (extendsi<mode>2 splitter): Likewise.
6746 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6747 right or vector extract after doing XSCVDPSPN. Use
6748 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6750 (movdi_from_sf_zero_ext): Likewise.
6751 (reload_gpr_from_vsxsf): Likewise.
6752 (p8_mfvsrd_4_disf): Delete, no longer used.
6753 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6754 then needing to move the SFmode to a GPR to use the XSCVDPSP
6755 instruction instead of FRSP and XSCVDPSPN.
6756 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6757 it is adjacent to the other XSCVSPDP insns.
6758 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6759 SFmode to be in traditional Altivec registers.
6760 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6761 (vsx_xscvspdpn): Likewise.
6762 (vsx_xscvspdpn_scalar): Likewise.
6764 2017-09-26 Martin Jambor <mjambor@suse.cz>
6766 * tree-sra.c (compare_access_positions): Put integral types first,
6767 stabilize sorting of integral types, remove conditions putting
6768 non-full-precision integers last.
6769 (sort_and_splice_var_accesses): Disable scalarization if a
6770 non-integert would be represented by a non-full-precision integer.
6772 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6774 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6775 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6776 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6777 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6778 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6779 conditionals inside the function instead of around it. Call
6780 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6781 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6783 2017-09-26 Richard Biener <rguenther@suse.de>
6785 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6787 (scop_detection::build_scop_breadth): ... this. Removed.
6788 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6789 (scop_detection::harmful_stmt_in_bb): Likewise.
6790 (scop_detection::graphite_can_represent_stmt): Likewise.
6791 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6792 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6793 (scop_detection::can_represent_loop_1): ... this. Removed.
6794 (scop_detection::harmful_loop_in_region): Simplify after inlining
6795 the above and remove more quadraticness.
6796 (build_scops): Adjust.
6797 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6800 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6803 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6804 REGNO (base) == SP_REG if base is a REG.
6807 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6808 if it is different SSA_NAME.
6809 (optimize_range_tests_cmp_bitwise): New function.
6810 (optimize_range_tests): Call it.
6812 2017-09-26 Richard Biener <rguenther@suse.de>
6814 PR tree-optimization/82321
6815 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6816 for the def being inside the loop.
6818 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6820 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6822 * config/s390/s390-builtins.def: Fix constraint on op4.
6824 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6826 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6827 independent expanders.
6828 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6829 ("vec_ordered", "vec_unordered"): New expanders.
6831 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6833 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6836 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6838 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6839 vec_unpacks_lo_v16qi.
6840 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6842 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6844 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6845 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6846 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6848 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6850 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6852 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6854 ("*vec_slb<mode>"): New insn pattern.
6855 ("vec_shr_<mode>"): New expander.
6856 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6857 and force the shift count operand to V16QImode.
6858 ("vec_srb<mode>"): Set shift count mode to V16QI.
6860 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6862 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6863 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6864 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6866 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6869 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6871 2017-09-26 Richard Biener <rguenther@suse.de>
6873 PR tree-optimization/82320
6874 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6877 2017-09-25 Jeff Law <law@redhat.com>
6879 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6880 prototype for new argument.
6881 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6882 mostly extracted from rs6000_emit_allocate_stack.
6883 (rs6000_emit_probe_stack_range_stack_clash): New function.
6884 (rs6000_emit_allocate_stack): Call
6885 rs6000_emit_probe_stack_range_stack_clash as needed.
6886 (rs6000_emit_probe_stack_range): Add additional argument
6887 to call to gen_probe_stack_range{si,di}.
6888 (output_probe_stack_range): New.
6889 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6890 (output_probe_stack_range_stack_clash): New.
6891 (rs6000_emit_prologue): Emit notes into dump file as requested.
6892 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6893 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6894 Add additional operand and pass it to output_probe_stack_range.
6896 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6898 PR tree-optimization/82163
6899 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6900 (checking_verify_loop_closed_ssa): New parameter.
6901 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6902 (check_loop_closed_ssa_stmt): Delete.
6903 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6904 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6905 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6908 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6910 * brig-builtins.def: Treat HSAIL barrier builtins as
6911 setjmp/longjump style functions.
6913 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6915 * target.def (constant_alignment): New hook.
6916 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6917 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6918 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6919 * doc/tm.texi: Regenerate.
6920 * targhooks.h (default_constant_alignment): Declare.
6921 (constant_alignment_word_strings): Likewise.
6922 * targhooks.c (default_constant_alignment): New function.
6923 (constant_alignment_word_strings): Likewise.
6924 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6925 instead of CONSTANT_ALIGNMENT.
6926 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6927 (force_const_mem): Likewise.
6928 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6929 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6930 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6931 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6932 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6934 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6935 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6936 constant_alignment_word_strings.
6937 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6938 (CONSTANT_ALIGNMENT): Likewise.
6939 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6940 (arm_constant_alignment): New function.
6941 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6942 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6943 constant_alignment_word_strings.
6944 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6945 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6946 constant_alignment_word_strings.
6947 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6948 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6949 (cris_constant_alignment): New function.
6950 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6951 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6952 (epiphany_constant_alignment): New function.
6953 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6954 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6955 constant_alignment_word_strings.
6956 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6957 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6958 constant_alignment_word_strings.
6959 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6960 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6961 constant_alignment_word_strings.
6962 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6963 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6964 * config/i386/i386.c (ix86_constant_alignment): Make static.
6965 Use the same interface as the target hook.
6966 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6967 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6968 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6969 constant_alignment_word_strings.
6970 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6971 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6972 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6973 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6974 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6975 constant_alignment_word_strings.
6976 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6977 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6978 constant_alignment_word_strings.
6979 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6980 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6981 constant_alignment_word_strings.
6982 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6983 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6985 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6986 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6987 * config/mips/mips.c (mips_constant_alignment): New function.
6988 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6989 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6990 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6991 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6992 (mmix_constant_alignment): Make static. Use the same interface
6994 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6995 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6996 constant_alignment_word_strings.
6997 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6998 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6999 constant_alignment_word_strings.
7000 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7001 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7002 constant_alignment_word_strings.
7003 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7004 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7005 (rs6000_constant_alignment): New function.
7006 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7007 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7008 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7009 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7010 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7011 (rs6000_constant_alignment): New function.
7012 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7013 * config/s390/s390.c (s390_constant_alignment): New function.
7014 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7015 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7016 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7017 constant_alignment_word_strings.
7018 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7019 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7020 (sparc_constant_alignment): New function.
7021 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7022 * config/spu/spu.c (spu_constant_alignment): New function.
7023 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7024 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7025 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7026 constant_alignment_word_strings.
7027 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7028 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7029 constant_alignment_word_strings.
7030 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7031 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7032 constant_alignment_word_strings.
7033 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7034 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7035 (visium_constant_alignment): New function.
7036 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7037 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7038 (xtensa_constant_alignment): New function.
7039 * system.h (CONSTANT_ALIGNMENT): Poison.
7041 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7043 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7044 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7045 (rs6000_builtin_valid_without_lhs): New helper function.
7046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7047 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7049 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7051 * target.h (vec_perm_indices): Use unsigned short rather than
7053 (auto_vec_perm_indices): Likewise.
7054 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7055 Use unsigned int rather than unsigned char.
7056 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7058 2017-09-25 Richard Biener <rguenther@suse.de>
7060 * cfgloop.h (sort_sibling_loops): Declare.
7061 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7062 (sort_sibling_loops): New function sorting the sibling loop list
7064 * graphite.c (graphite_transform_loops): Sort sibling loops.
7066 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7068 * target.def (vec_perm_const_ok): Change sel parameter to
7070 * optabs-query.c (can_vec_perm_p): Update accordingly.
7071 * doc/tm.texi: Regenerate.
7072 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7073 auto_vec_perm_indices and remove separate nelt field.
7074 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7075 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7076 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7077 (aarch64_expand_vec_perm_const): Update accordingly.
7078 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7079 to vec_perm_indices.
7080 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7081 auto_vec_perm_indices and remove separate nelt field.
7082 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7083 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7084 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7086 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7087 to vec_perm_indices.
7088 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7089 sel to vec_perm_indices.
7090 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7091 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7092 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7094 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7097 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7100 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7101 on the FUNCTION_DECL function context if it has a DIE that is a
7104 2017-09-25 Richard Biener <rguenther@suse.de>
7106 PR tree-optimization/82285
7107 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7110 2017-09-25 Tom de Vries <tom@codesourcery.com>
7114 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7117 2017-09-25 Richard Biener <rguenther@suse.de>
7119 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7120 ISL errors other than isl_error_quota happen. Dump if the
7121 schedule is the same.
7122 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7123 errors instead of aborting inside ISL.
7125 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7128 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7129 of libgcc_eh for m64.
7130 * config/i386/darwin64.h: Likewise.
7132 2017-09-25 Richard Biener <rguenther@suse.de>
7135 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7136 attribute for incomplete types nor twice for complete ones.
7138 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7141 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7142 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7145 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7148 * config/i386/i386.opt (mprefer-avx256): Use
7149 ix86_target_flags variable.
7150 * config/i386/i386.c (ix86_target_string): Move
7151 -mprefer-avx256 to flag2_opts.
7153 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7156 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7157 and x != -1 | y != -1 into (x & y) != -1.
7159 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7161 * config.gcc: Add new case statement to set
7162 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7163 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7164 s390x-*-linux* case statements. Added aarch64 to the list of
7165 supported architectures.
7167 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7169 PR tree-optimization/82289
7170 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7171 STMT_VINFO_RELEVANT_P.
7173 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7174 Alan Hayward <alan.hayward@arm.com>
7175 David Sherwood <david.sherwood@arm.com>
7177 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7178 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7179 (extract_range_from_binary_expr_1): Don't call
7180 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7182 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7183 Alan Hayward <alan.hayward@arm.com>
7184 David Sherwood <david.sherwood@arm.com>
7186 * target.def (preferred_vector_alignment): New hook.
7187 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7189 * doc/tm.texi: Regenerate.
7190 * targhooks.h (default_preferred_vector_alignment): Declare.
7191 * targhooks.c (default_preferred_vector_alignment): New function.
7192 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7194 (DR_TARGET_ALIGNMENT): New macro.
7195 (aligned_access_p): Update commentary.
7196 (vect_known_alignment_in_bytes): New function.
7197 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7199 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7200 Calculate the misalignment based on the target alignment rather than
7202 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7203 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7204 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7205 the target alignment, rather than masking the element misalignment
7206 with the number of elements in a vector. Also use the target
7207 alignment when calculating the maximum number of peels.
7208 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7209 instead of TYPE_ALIGN_UNIT.
7210 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7211 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7212 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7213 (vect_create_data_ref_ptr): Likewise.
7214 (vect_setup_realignment): Realign by ANDing with
7215 -DR_TARGET_MISALIGNMENT.
7216 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7217 the number of peels based on DR_TARGET_ALIGNMENT.
7218 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7219 with the guaranteed alignment boundary when deciding whether
7221 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7222 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7223 (ensure_base_align): Remove stmt_info parameter. Get the
7224 target base alignment from DR_TARGET_ALIGNMENT.
7225 (vectorizable_store): Update call accordingly. Interpret
7226 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7228 (vectorizable_load): Likewise.
7230 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7231 Alan Hayward <alan.hayward@arm.com>
7232 David Sherwood <david.sherwood@arm.com>
7234 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7235 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7236 (vect_enhance_data_refs_alignment): Likewise.
7238 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7240 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7241 error. Only quit immediately if parsing is complete.
7242 (BEGIN): Initialize fatal_err and parse_done.
7243 (begin fpu, end fpu): Check number of arguments.
7244 (begin arch, end arch): Likewise.
7245 (begin cpu, end cpu): Likewise.
7246 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7247 (optalias): Likewise.
7249 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7251 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7252 * config/arm/arm-isa.h: Delete. Move definitions to ...
7253 * arm-cpus.in: ... here. Use new feature and fgroup values.
7254 * config/arm/arm.c (arm_option_override): Use lower case for feature
7256 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7257 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7258 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7260 (print_isa_bits_for): New function.
7261 (gen_isa): New function.
7262 (gen_comm_data): Use print_isa_bits_for.
7263 (define feature): New keyword.
7264 (define fgroup): New keyword.
7265 * config/arm/t-arm (TM_H): Remove.
7266 (GTM_H): Add arm-isa.h.
7267 (arm-isa.h): Add rule to generate file.
7268 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7269 case for feature bit names.
7271 2017-09-22 Richard Biener <rguenther@suse.de>
7273 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7275 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7276 print of no dependency loops ...
7277 * graphite.c (graphite_transform_loops): ... here.
7278 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7280 (same_close_phi_node, remove_duplicate_close_phi,
7281 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7282 (canonicalize_loop_closed_ssa): ... here and simplify.
7283 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7284 (optimize_isl): Use dump_printf_loc to tell when we stopped
7285 optimizing because of an ISL timeout.
7287 2017-09-22 Richard Biener <rguenther@suse.de>
7289 PR tree-optimization/82291
7290 * tree-if-conv.c (predicate_mem_writes): Make sure to
7291 remove writes in blocks predicated with false.
7293 2017-09-22 Richard Biener <rguenther@suse.de>
7295 * sese.c: Include cfganal.h.
7296 (if_region_set_false_region): Remove.
7297 (create_if_region_on_edge): Likewise.
7298 (move_sese_in_condition): Re-implement without destroying
7301 2017-09-22 Richard Biener <rguenther@suse.de>
7303 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7304 Verify both BBs contain loop PHI nodes before dispatching to
7306 (graphite_regenerate_ast_isl): Do not recompute dominators,
7307 do not verify three times. Restructure for clarity.
7308 * graphite-scop-detection.c (same_close_phi_node,
7309 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7310 defined_in_loop_p, canonicalize_loop_closed_ssa,
7311 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7312 checking and SSA rewrite, move to ...
7313 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7314 (graphite_initialize): Do not pass in ctx, do not reset the
7315 SCEV cache, compute only dominators.
7316 (graphite_transform_loops): Allocate ISL ctx after
7317 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7318 Maintain post-dominators only around build_scops.
7319 * sese.c (if_region_set_false_region): Make static. Free
7320 and recompute dominators.
7321 (move_sese_in_condition): Assert we don't get called with
7322 post-dominators computed.
7323 * sese.h (if_region_set_false_region): Remove.
7325 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7327 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7328 mode attribute for TARGET_AVX512VL.
7330 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7332 * config/i386/i386.opt (mprefer-avx256): New option.
7333 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7335 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7336 for TARGET_PREFER_AVX256.
7337 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7339 2017-09-21 Jeff Law <law@redhat.com>
7341 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7342 Fix dump output if the only stack space is for pushed registers.
7344 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7346 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7349 2017-09-21 Martin Sebor <msebor@redhat.com>
7352 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7353 code (in C++) or code that triggers warnings.
7355 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7357 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7359 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7361 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7362 * sched-rgn.c: Ditto.
7363 * sel-sched-ir.c: Ditto.
7365 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7367 * toplev.h (set_random_seed): Adjust return type.
7368 * toplev.c (init_local_tick): Move eager initialization of random_seed
7369 to get_random_seed. Adjust comment.
7370 (init_random_seed): Inline to get_random_seed, delete.
7371 (get_random_seed): Initialize random_seed lazily.
7372 (set_random_seed): Do not return previous value.
7373 (print_switch_value): Do not call get_random_seed.
7375 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7377 * cgraph.c (delete_function_version): New, broken out from...
7378 (cgraph_node::delete_function_version): ...here. Rename to
7379 cgraph_node::delete_function_version_by_decl. Update all uses.
7380 (cgraph_node::remove): Call delete_function_version.
7382 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7385 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7386 VAR_DECLs to which addressable non-volatile parameters are mapped
7387 and for id->retvar after the return value assignment. Clear
7388 id->retval and id->retbnd after inlining.
7390 2017-09-21 Richard Biener <rguenther@suse.de>
7392 PR tree-optimization/82276
7393 PR tree-optimization/82244
7394 * tree-vrp.c (build_assert_expr_for): Set
7395 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7397 (remove_range_assertions): Revert earlier change.
7399 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7402 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7404 2017-09-21 Richard Biener <rguenther@suse.de>
7406 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7407 Restore valid IL after code generation errors.
7408 * graphite.c (graphite_transform_loops): Diagnose code
7409 generation issues as MSG_MISSED_OPTIMIZATION and continue
7410 with processing SCOPs.
7412 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7413 Alan Hayward <alan.hayward@arm.com>
7414 David Sherwood <david.sherwood@arm.com>
7416 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7417 rather than choosing between plus_constant and gen_rtx_<CODE>.
7418 * expr.c (emit_push_insn): Likewise.
7419 (expand_expr_real_2): Likewise.
7421 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7422 Alan Hayward <alan.hayward@arm.com>
7423 David Sherwood <david.sherwood@arm.com>
7425 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7432 calling tree_to_uhwi.
7434 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7435 Alan Hayward <alan.hayward@arm.com>
7436 David Sherwood <david.sherwood@arm.com>
7438 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7439 INTEGER_CST rather than a negative test for ADDR_EXPR.
7441 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7442 Alan Hayward <alan.hayward@arm.com>
7443 David Sherwood <david.sherwood@arm.com>
7445 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7446 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7448 2017-09-21 Richard Biener <rguenther@suse.de>
7450 PR tree-optimization/71351
7451 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7452 graphite_create_new_loop_guard): Remove, fold remaining parts
7454 (translate_isl_ast_node_for): ... here and simplify.
7456 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7459 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7460 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7461 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7462 alternative always use QI mode, for -Os imov (=R,R) alternative
7463 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7466 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7467 Jeff Law <law@redhat.com>
7469 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7470 (allocate_stack_space): New function, partially extracted from
7472 (s390_emit_prologue): Track offset to most recent stack probe.
7473 Code to allocate space moved into allocate_stack_space.
7474 Dump actions when no stack is allocated.
7475 (s390_prologue_plus_offset): New function.
7476 (s390_emit_stack_probe): Likewise.
7478 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7480 * common.opt (Wa, Wl, Wp, g, gz=): Add
7482 (gno-column-info): Remove.
7483 (gcolumn-info): Drop RejectNegative.
7485 (gno-record-gcc-switches): Remove.
7486 (grecord-gcc-switches): Drop RejectNegative.
7487 (gno-split-dwarf): Remove.
7488 (gsplit-dwarf): Drop RejectNegative.
7489 (gno-strict-dwarf): Remove.
7490 (gstrict-dwarf): Drop RejectNegative.
7491 * config/darwin.opt (gfull, gused): Add RejectNegative.
7492 * dwarf2out.c (gen_producer_string): Drop
7493 gno-record-gcc-switches handler.
7494 * optc-gen.awk: Add g to prefixes with negative forms.
7495 * opts-common.c (remapping_prefix_p): New.
7496 (find_opt): Check it.
7497 (generate_canonical_option): Test g prefix.
7498 (option_map): Add -gno- mapping.
7499 (add_misspelling_candidates): Check remapping_prefix_p.
7501 2017-09-20 Jeff Law <law@redhat.com>
7503 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7504 thinko in stack clash protection support.
7506 * explow.c (compute_stack_clash_protection_loop_data): Use
7507 CONST_INT_P instead of explicit test. Verify object is a
7508 CONST_INT_P before looking at INTVAL.
7509 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7510 instead of explicit test.
7512 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7515 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7516 address instead of to r1 and r11.
7518 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7520 * config.gcc: Support "knm".
7521 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7522 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7524 * config/i386/i386.c (m_KNM): Define.
7525 (processor_target_table): Add "knm".
7527 (ix86_option_override_internal): Add "knm".
7528 (ix86_issue_rate): Add PROCESSOR_KNM.
7529 (ix86_adjust_cost): Ditto.
7530 (ia32_multipass_dfa_lookahead): Ditto.
7531 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7532 (fold_builtin_cpu): Add M_INTEL_KNM.
7533 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7534 (processor_type): Add PROCESSOR_KNM.
7535 * config/i386/x86-tune.def: Add m_KNM.
7536 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7538 2017-09-20 Richard Biener <rguenther@suse.de>
7540 PR tree-optimization/80213
7541 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7542 are allowed in empty BBs as well.
7543 (canonicalize_loop_closed_ssa): Also look for other complex
7545 (scop_detection::get_sese): Include the loop-closed PHI block
7547 (scop_detection::merge_sese): Remove code adding extra blocks.
7548 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7549 (build_scops): Assert the final returned scop is invalid.
7551 2017-09-20 Richard Biener <rguenther@suse.de>
7553 PR tree-optimization/82264
7554 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7556 (vn_phi_lookup): Likewise.
7557 (vn_phi_insert): Likewise.
7559 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7561 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7562 that fits into uhwi or shwi, add DW_AT_const_value regardless
7563 of early_dwarf without going through RTL, using add_AT_unsigned
7566 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7567 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7568 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7569 (DEBUG_MACRO_SECTION): Likewise.
7570 (DEBUG_LTO_MACRO_SECTION): Likewise.
7571 (DEBUG_STR_DWO_SECTION): Likewise.
7572 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7573 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7574 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7575 (DEBUG_LTO_LINE_STR_SECTION): Define.
7576 (init_sections_and_labels): Initialize debug_line_str_section
7577 variable. Initialize debug_loc_section for -gdwarf-5 to
7578 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7580 2017-09-20 Richard Biener <rguenther@suse.de>
7582 * graphite-sese-to-poly.c (extract_affine): Properly handle
7583 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7585 2017-09-20 Richard Biener <rguenther@suse.de>
7587 PR tree-optimization/81373
7588 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7589 Force SESE live-out defs to be handled even if they are
7592 2017-09-19 Jeff Law <law@redhat.com>
7594 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7595 nothing for stack adjustments with REG_STACK_CHECK.
7596 * sched-deps.c (parse_add_or_inc): Reject insns with
7597 REG_STACK_CHECK from dependency breaking.
7598 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7599 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7600 * reg-notes.def (STACK_CHECK): New note.
7602 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7603 (ix86_expand_prologue): Dump stack clash info as needed.
7604 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7606 * function.c (dump_stack_clash_frame_info): New function.
7607 * function.h (dump_stack_clash_frame_info): Prototype.
7608 (enum stack_clash_probes): New enum.
7610 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7611 flag_stack_clash_protection.
7612 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7613 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7614 (arm_frame_pointer_required): Likewise.
7615 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7616 (ia64_expand_prologue): Likewise.
7617 * config/mips/mips.c (mips_expand_prologue): Likewise.
7618 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7619 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7620 (sparc_flat_expand_prologue): Likewise.
7621 * config/spu/spu.c (spu_expand_prologue): Likewise.
7623 * explow.c: Include "params.h".
7624 (anti_adjust_stack_and_probe_stack_clash): New function.
7625 (get_stack_check_protect): Likewise.
7626 (compute_stack_clash_protection_loop_data): Likewise.
7627 (emit_stack_clash_protection_loop_start): Likewise.
7628 (emit_stack_clash_protection_loop_end): Likewise.
7629 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7630 Use anti_adjust_stack_and_probe_stack_clash.
7631 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7632 (emit_stack_clash_protection_loop_start): Likewise.
7633 (emit_stack_clash_protection_loop_end): Likewise.
7634 * rtl.h (get_stack_check_protect): Prototype.
7635 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7636 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7637 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7639 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7641 * doc/tm.texi: Rebuilt.
7642 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7643 get_stack_check_protect.
7644 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7645 * config/arm/arm.c (arm_expand_prologue): Likewise.
7646 (arm_frame_pointer_required): Likewise.
7647 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7648 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7649 * config/mips/mips.c (mips_expand_prologue): Likewise.
7650 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7651 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7652 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7653 (sparc_flat_expand_prologue): Likewise.
7655 * common.opt (-fstack-clash-protection): New option.
7656 * flag-types.h (enum stack_check_type): Note difference between
7657 -fstack-check= and -fstack-clash-protection.
7658 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7659 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7660 * toplev.c (process_options): Issue warnings/errors for cases
7661 not handled with -fstack-clash-protection.
7662 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7663 (-fstack-check): Note additional problem with -fstack-check=generic.
7664 Note that -fstack-check is primarily for Ada and refer users
7665 to -fstack-clash-protection for stack-clash-protection.
7666 Document new params for stack clash protection.
7668 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7670 * config/i386/i386.c (ix86_split_long_move): Do not handle
7671 address used for LEA in a special way.
7673 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7675 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7677 2017-09-19 Martin Sebor <msebor@redhat.com>
7680 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7681 of incompatible types.
7683 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7685 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7686 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7687 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7688 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7690 2017-09-19 Richard Biener <rguenther@suse.de>
7692 PR tree-optimization/82244
7693 * tree-vrp.c (remove_range_assertions): Do not propagate
7694 a constant to abnormals but replace the assert with a copy.
7696 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7698 PR rtl-optimization/57878
7699 PR rtl-optimization/68988
7700 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7701 avoidance test involving non_reload_pseudos. Move frequency test
7702 below the general fragmentation avoidance test.
7704 2017-09-19 Richard Biener <rguenther@suse.de>
7706 PR tree-optimization/69728
7707 * graphite-sese-to-poly.c (schedule_error): New global.
7708 (add_loop_schedule): Handle empty domain by failing the
7710 (build_original_schedule): Handle schedule_error.
7712 2017-09-19 Richard Biener <rguenther@suse.de>
7714 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7715 Do not iterate to sibling loops but only to siblings of inner
7718 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7721 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7722 that the registers are different.
7724 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7726 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7727 to processor_model and "amdfam17h" to arch_names_table.
7728 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7730 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7733 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7735 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7736 Alan Hayward <alan.hayward@arm.com>
7737 David Sherwood <david.sherwood@arm.com>
7739 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7741 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7743 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7744 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7745 vect_schedule_slp_instance.
7746 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7747 Update call to vect_slp_analyze_node_operations. Simplify return
7749 (vect_slp_analyze_bb_1): Update call accordingly.
7750 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7751 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7752 (vect_schedule_slp): Update call accordingly.
7754 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7755 Alan Hayward <alan.hayward@arm.com>
7756 David Sherwood <david.sherwood@arm.com>
7758 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7759 with types that aren't in fact scalar.
7761 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7763 * tree-vect-slp.c (vect_record_max_nunits): New function,
7765 (vect_build_slp_tree_1): ...here.
7766 (vect_build_slp_tree_2): Call it for phis too.
7768 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7770 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7771 to vect_get_vec_def_for_operand when getting the mask operand.
7773 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7774 Alan Hayward <alan.hayward@arm.com>
7775 David Sherwood <david.sherwood@arm.com>
7777 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7780 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7781 Alan Hayward <alan.hayward@arm.com>
7782 David Sherwood <david.sherwood@arm.com>
7784 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7785 calculation for vector booleans.
7787 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7788 Alan Hayward <alan.hayward@arm.com>
7789 David Sherwood <david.sherwood@arm.com>
7791 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7793 (vect_transform_stmt): ...here.
7794 (vect_analyze_stmt): Use it instead of calling
7795 vectorizable_live_operation directly.
7797 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7799 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7800 non-SIMT targets in acc vector loops.
7802 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7804 * configure.ac: Add arc and check if assembler supports gdwarf2.
7805 * configure: Regenerate.
7807 2017-09-18 Richard Biener <rguenther@suse.de>
7809 PR tree-optimization/82220
7810 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7811 epilogue niters from the min_profitable_iters compute.
7813 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7816 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7817 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7818 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7820 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7823 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7824 switching to a new text section.
7826 2017-09-18 Richard Biener <rguenther@suse.de>
7828 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7830 (build_alias_set): Reject aliases with no access function.
7832 2017-09-18 Richard Biener <rguenther@suse.de>
7834 PR tree-optimization/79622
7835 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7837 (build_cross_bb_scalars_use): Likewise.
7839 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7841 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7843 2017-09-18 Alan Modra <amodra@gmail.com>
7846 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7847 stack_pointer_rtx for count 0. Update comments. Break up
7848 large rtl expression.
7850 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7852 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7853 Increase to 20 bytes.
7854 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7855 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7856 or avx version of the stub.
7858 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7861 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7862 compute the minimum stack alignment. Also update preferred stack
7863 boundary for leaf functions.
7865 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7867 PR tree-optimization/82228
7868 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7871 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7873 * common/config/nds32/nds32-common.c
7874 (nds32_option_optimization_table): Refine formatting.
7875 (nds32_option_optimization_table): Use -fsched-pressure and
7876 -fomit-frame-pointer for specific optimization level.
7878 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7880 * config/nds32/nds32.c: Refine formatting and comments.
7881 * config/nds32/nds32.h: Likewise.
7882 * config/nds32/nds32.md: Likewise.
7883 * config/nds32/nds32-cost.c: Likewise.
7884 * config/nds32/nds32-isr.c: Likewise.
7885 * config/nds32/nds32-md-auxiliary.c: Likewise.
7886 * config/nds32/nds32-multiple.md: Likewise.
7887 * config/nds32/nds32-predicates.c: Likewise.
7889 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7890 Jakub Jelinek <jakub@redhat.com>
7892 Add support for -std=c++2a.
7893 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7895 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7897 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7900 * doc/extend.texi (Common Function Attributes): Add
7901 references to ARM, AArch64, and S/390 specific attributes.
7902 (Function Specific Option Pragmas): Add AArch64 and S/390
7903 to list of back ends that support the target pragma.
7905 2017-09-15 Nathan Sidwell <nathan@acm.org>
7907 * doc/standards.texi: Fix C++17 description. Update URLs for
7910 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7912 * common.opt (Wcast-align=strict): New warning option.
7913 * doc/invoke.texi: Document -Wcast-align=strict.
7915 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7917 * cgraph.h (cgraph_thunk_info): Add comments.
7918 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7919 assert for VIRTUAL_* arguments stricter.
7921 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7923 PR tree-optimization/71026
7924 * match.pd: Move RDIV patterns from fold-const.c
7925 * fold-const.c (distribute_real_division): Removed.
7926 (fold_binary_loc): Remove calls to distribute_real_divison.
7928 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7930 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7931 c++1z and gnu++1z as deprecated. Change other references to
7932 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7933 Change -Wc++1z-compat to -Wc++17-compat.
7934 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7935 * dwarf2out.c (highest_c_language): Handle C++17.
7936 (gen_compile_unit_die): Likewise.
7938 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7940 PR rtl-optimization/82192
7941 * combine.c (make_extraction): Don't look through non-paradoxical
7942 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7945 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7946 Alan Hayward <alan.hayward@arm.com>
7947 David Sherwood <david.sherwood@arm.com>
7949 * target.def (function_arg_offset): New hook.
7950 * targhooks.h (default_function_arg_offset): Declare.
7951 * targhooks.c (default_function_arg_offset): New function.
7952 * function.c (locate_and_pad_parm): Use
7953 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7954 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7955 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7956 * doc/tm.texi: Regenerate.
7957 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7958 * config/spu/spu.c (spu_function_arg_offset): New function.
7959 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7960 * system.h (FUNCTION_ARG_OFFSET): Poison.
7962 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7963 Alan Hayard <alan.hayward@arm.com>
7964 David Sherwood <david.sherwood@arm.com>
7966 * target.def (truly_noop_truncation): New hook.
7967 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7968 than TRULY_NOOP_TRUNCATION.
7969 * hooks.h (hook_bool_uint_uint_true): Declare.
7970 * hooks.c (hook_bool_uint_uint_true): New function.
7971 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7972 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7973 * doc/tm.texi: Regenerate.
7974 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7975 rather than TRULY_NOOP_TRUNCATION in comments.
7976 (simplify_comparison): Likewise.
7977 (record_truncated_value): Likewise.
7978 * expmed.c (extract_bit_field_1): Likewise.
7979 (extract_split_bit_field): Likewise.
7980 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7981 instead of TRULY_NOOP_TRUNCATION.
7982 * function.c (assign_parm_setup_block): Likewise.
7983 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7984 * rtlhooks.c: Include target.h.
7985 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7986 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7987 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7988 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7989 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7990 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7991 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7992 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7993 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7994 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7995 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7996 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7997 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7998 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7999 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8000 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8001 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8002 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8003 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8004 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8005 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8006 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8007 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8008 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8009 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8010 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8011 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8012 rather than TRULY_NOOP_TRUNCATION in comments.
8013 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8014 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8015 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8016 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8017 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8018 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8019 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8020 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8021 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8022 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8023 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8024 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8025 rather than TRULY_NOOP_TRUNCATION in comments.
8026 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8027 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8028 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8029 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8030 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8031 TRULY_NOOP_TRUNCATION condition.
8032 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8033 (TRULY_NOOP_TRUNCATION): Delete.
8034 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8035 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8036 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8037 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8038 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8039 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8040 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8041 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8042 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8043 rather than TRULY_NOOP_TRUNCATION in comments.
8044 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8045 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8046 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8047 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8048 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8049 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8051 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8054 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8055 (*cmp_ior): Likewise.
8056 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8057 (*ior_scc_scc_cmp): Likewise.
8058 (*and_scc_scc): Likewise.
8059 (*and_scc_scc_cmp): Likewise.
8061 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8062 Alan Hayard <alan.hayward@arm.com>
8063 David Sherwood <david.sherwood@arm.com>
8065 * target.def (can_change_mode_class): New hook.
8066 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8067 (hard_regno_nregs): Likewise.
8068 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8069 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8070 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8071 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8072 (LOAD_EXTEND_OP): Update accordingly.
8073 * doc/tm.texi: Regenerate.
8074 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8075 CANNOT_CHANGE_MODE_CLASS.
8076 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8077 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8078 * combine.c (simplify_set): Update accordingly.
8079 * emit-rtl.c (validate_subreg): Likewise.
8080 * recog.c (general_operand): Likewise.
8081 * regcprop.c (mode_change_ok): Likewise.
8082 * reload1.c (choose_reload_regs): Likewise.
8083 (inherit_piecemeal_p): Likewise.
8084 * rtlanal.c (simplify_subreg_regno): Likewise.
8085 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8086 instead of CANNOT_CHANGE_MODE_CLASS.
8087 (reload_cse_simplify_operands): Likewise.
8088 * reload.c (push_reload): Use targetm.can_change_mode_class
8089 instead of CANNOT_CHANGE_MODE_CLASS.
8090 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8091 REG_CANNOT_CHANGE_MODE_P.
8092 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8093 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8094 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8095 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8096 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8097 (arm_can_change_mode_class): New function.
8098 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8099 than CANNOT_CHANGE_MODE_CLASS in comments.
8100 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8101 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8102 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8103 (ix86_can_change_mode_class): ...this new function, inverting the
8104 sense of the return value.
8105 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8106 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8107 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8108 (ia64_can_change_mode_class): New function.
8109 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8110 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8111 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8112 (m32c_can_change_mode_class): ...this new function, inverting the
8113 sense of the return value.
8114 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8115 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8116 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8117 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8118 (mips_can_change_mode_class): ...this new function, inverting the
8119 sense of the return value.
8120 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8121 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8122 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8123 (msp430_can_change_mode_class): New function.
8124 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8125 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8126 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8127 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8128 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8129 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8130 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8131 (pa_cannot_change_mode_class): Replace with...
8132 (pa_can_change_mode_class): ...this new function, inverting the
8133 sense of the return value.
8134 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8135 than CANNOT_CHANGE_MODE_CLASS in comments.
8136 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8137 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8138 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8139 (pdp11_cannot_change_mode_class): Replace with...
8140 (pdp11_can_change_mode_class): ...this new function, inverting the
8141 sense of the return value.
8142 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8143 * config/powerpcspe/powerpcspe-protos.h
8144 (rs6000_cannot_change_mode_class_ptr): Delete.
8145 * config/powerpcspe/powerpcspe.c
8146 (rs6000_cannot_change_mode_class_ptr): Delete.
8147 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8148 (rs6000_option_override_internal): Assign to
8149 targetm.can_change_mode_class instead of
8150 rs6000_cannot_change_mode_class_ptr.
8151 (rs6000_cannot_change_mode_class): Replace with...
8152 (rs6000_can_change_mode_class): ...this new function, inverting the
8153 sense of the return value.
8154 (rs6000_debug_cannot_change_mode_class): Replace with...
8155 (rs6000_debug_can_change_mode_class): ...this new function.
8156 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8157 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8158 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8159 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8160 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8162 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8163 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8164 (rs6000_option_override_internal): Assign to
8165 targetm.can_change_mode_class instead of
8166 rs6000_cannot_change_mode_class_ptr.
8167 (rs6000_cannot_change_mode_class): Replace with...
8168 (rs6000_can_change_mode_class): ...this new function, inverting the
8169 sense of the return value.
8170 (rs6000_debug_cannot_change_mode_class): Replace with...
8171 (rs6000_debug_can_change_mode_class): ...this new function.
8172 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8173 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8174 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8175 (s390_can_change_mode_class): ...this new function, inverting the
8176 sense of the return value.
8177 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8178 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8179 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8180 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8181 (sh_cannot_change_mode_class): Replace with...
8182 (sh_can_change_mode_class): ...this new function, inverting the
8183 sense of the return value.
8184 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8185 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8186 (sparc_can_change_mode_class): New function.
8187 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8188 * config/spu/spu.c (spu_can_change_mode_class): New function.
8189 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8190 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8191 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8192 (visium_can_change_mode_class): New function.
8193 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8195 2017-09-15 Richard Biener <rguenther@suse.de>
8197 PR tree-optimization/82217
8198 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8199 but not undefined case.
8201 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8204 * postreload.c (reload_cse_simplify_operands): Skip
8205 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8207 2017-09-15 Richard Biener <rguenther@suse.de>
8209 PR tree-optimization/68823
8210 * graphite-scop-detection.c (build_alias_set): If we have a
8211 possible dependence check whether we can handle them by just
8212 looking at the DRs DR_ACCESS_FNs.
8213 (build_scops): If build_alias_set fails, fail the SCOP.
8215 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8217 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8218 to support float128 built-in functions that require the ISA 3.0
8220 (BU_FLOAT128_3_HW): Likewise.
8221 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8223 (FMAF128): Likewise.
8225 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8226 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8227 floating point instructions.
8228 (rs6000_invalid_builtin): Likewise.
8229 (rs6000_builtin_mask_names): Likewise.
8230 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8231 (RS6000_BTM_FLOAT128_HW): Likewise.
8232 (RS6000_BTM_COMMON): Likewise.
8233 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8235 * doc/extend.texi (RS/6000 built-in functions): Document the
8236 IEEE 128-bit floating point square root and fused multiply-add
8239 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8241 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8242 reg (r2) isn't in the set of registers defined in the prologue.
8244 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8245 Alan Hayward <alan.hayward@arm.com>
8246 David Sherwood <david.sherwood@arm.com>
8248 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8249 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8250 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8251 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8252 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8254 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8255 max_vectorization_factor.
8256 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8258 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8259 Alan Hayward <alan.hayward@arm.com>
8260 David Sherwood <david.sherwood@arm.com>
8262 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8263 (vect_worthwhile_without_simd_p): Declare.
8264 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8265 (vectorizable_reduction): Use it.
8266 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8267 (vectorizable_operation): Likewise.
8269 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8270 Alan Hayward <alan.hayward@arm.com>
8271 David Sherwood <david.sherwood@arm.com>
8273 * tree-vectorizer.h (vect_get_num_copies): New function.
8274 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8275 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8276 (vectorizable_induction): Likewise.
8277 (vectorizable_live_operation): Likewise.
8278 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8279 (vectorizable_bswap): Likewise.
8280 (vectorizable_call): Likewise.
8281 (vectorizable_conversion): Likewise.
8282 (vectorizable_assignment): Likewise.
8283 (vectorizable_shift): Likewise.
8284 (vectorizable_operation): Likewise.
8285 (vectorizable_store): Likewise.
8286 (vectorizable_load): Likewise.
8287 (vectorizable_condition): Likewise.
8288 (vectorizable_comparison): Likewise.
8289 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8291 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8292 Alan Hayward <alan.hayward@arm.com>
8293 David Sherwood <david.sherwood@arm.com>
8295 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8296 of vect_init_vector.
8298 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8299 Alan Hayward <alan.hayward@arm.com>
8300 David Sherwood <david.sherwood@arm.com>
8302 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8303 an inline wrapper that provides a location.
8304 (gimple_build_vector): Likewise.
8305 * gimple-fold.c (gimple_build_vector_from_val): New function.
8306 (gimple_build_vector): Likewise.
8307 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8308 functions to build the initial value. Always return a gimple value.
8309 (get_initial_defs_for_reduction): Likewise. Only compute
8311 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8312 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8313 (vectorizable_induction): Use gimple_build_vector rather than
8316 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8317 Alan Hayward <alan.hayward@arm.com>
8318 David Sherwood <david.sherwood@arm.com>
8320 * target.h (vec_perm_indices): New typedef.
8321 (auto_vec_perm_indices): Likewise.
8322 * optabs-query.h: Include target.h
8323 (can_vec_perm_p): Take a vec_perm_indices *.
8324 * optabs-query.c (can_vec_perm_p): Likewise.
8325 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8326 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8327 * tree-vect-generic.c (lower_vec_perm): Likewise.
8328 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8329 (vect_grouped_load_supported): Likewise.
8330 (vect_shift_permute_load_chain): Likewise.
8331 (vect_permute_store_chain): Use auto_vec_perm_indices.
8332 (vect_permute_load_chain): Likewise.
8333 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8334 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8335 Update uses of can_vec_perm_p.
8336 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8337 mode with a number of elements. Take a vec_perm_indices *.
8338 (vect_create_epilog_for_reduction): Update accordingly.
8339 Use auto_vec_perm_indices.
8340 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8341 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8342 (vect_transform_slp_perm_load): Likewise.
8343 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8344 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8345 (vect_gen_perm_mask_checked): Likewise.
8346 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8347 (vect_gen_perm_mask_checked): Likewise.
8348 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8349 (vectorizable_store): Likewise.
8350 (vectorizable_load): Likewise.
8351 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8352 (vectorizable_bswap): Likewise.
8354 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8355 Alan Hayward <alan.hayward@arm.com>
8356 David Sherwood <david.sherwood@arm.com>
8358 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8359 * tree.c (build_vector): Likewise.
8360 (build_vector_from_ctor): Update accordingly.
8361 (build_vector_from_val): Likewise.
8362 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8363 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8364 * tree-vect-generic.c (add_rshift): Likewise.
8365 (expand_vector_divmod): Likewise.
8366 (optimize_vector_constructor): Likewise.
8367 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8368 (vect_transform_slp_perm_load): Likewise.
8369 (vect_schedule_slp_instance): Likewise.
8370 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8371 (vectorizable_call): Likewise.
8372 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8373 * expmed.c (make_tree): Likewise.
8374 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8375 a vector passed to build_vector.
8376 (fold_convert_const): Likewise.
8377 (exact_inverse): Likewise.
8378 (fold_ternary_loc): Likewise.
8379 (fold_relational_const): Likewise.
8380 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8381 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8382 (const_unop): Likewise. Store the reduction accumulator in a
8383 variable rather than an array.
8384 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8385 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8386 the new vector, rather than constructing it after the input arrays.
8387 (native_interpret_vector): Use auto_vec<tree> when building
8388 a vector passed to build_vector. Add elements in order.
8389 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8390 auto_vec<tree> when building a vector passed to build_vector.
8391 (vect_create_epilog_for_reduction): Likewise.
8392 (vectorizable_induction): Likewise.
8393 (get_initial_def_for_reduction): Likewise. Fix indentation of
8395 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8397 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8398 passed to build_vector.
8400 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8401 Alan Hayward <alan.hayward@arm.com>
8402 David Sherwood <david.sherwood@arm.com>
8404 * tree-core.h (tree_base::u): Add an "nelts" field.
8405 (tree_vector): Use VECTOR_CST_NELTS as the length.
8406 * tree.c (tree_size): Likewise.
8407 (make_vector): Initialize VECTOR_CST_NELTS.
8408 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8409 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8410 TYPE_VECTOR_SUBPARTS.
8411 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8412 as the number of units, setting it from VECTOR_CST_NELTS.
8413 (const_vector_from_tree): Likewise.
8414 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8415 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8416 (fold_negate_expr_1): Likewise.
8417 (fold_convert_const): Likewise.
8418 (const_binop): Likewise. Differentiate the number of output and
8420 (const_unop): Likewise.
8421 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8422 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8423 in cases that did the opposite.
8425 2017-09-14 Richard Biener <rguenther@suse.de>
8427 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8430 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8432 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8434 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8437 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8438 if and where to split a bb, except for splitting before debug insn
8439 sequences followed by non-label real insn. Delete debug insns
8440 in between basic blocks.
8442 * combine.c (make_compound_operation_int): Formatting fixes.
8444 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8445 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8446 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8447 * config/sol2.h (LINK_EH_SPEC): Likewise.
8448 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8449 * config/s390/linux.h (LINK_SPEC): Likewise.
8450 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8451 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8452 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8453 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8454 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8455 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8456 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8457 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8459 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8461 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8463 (ENDFILE_LINUX_SPEC): Likewise.
8464 (LINK_EH_SPEC): Likewise.
8465 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8466 (LINK_OS_LINUX_SPEC32): Likewise.
8467 (LINK_OS_LINUX_SPEC64): Likewise.
8468 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8469 (LINK_OS_LINUX_SPEC): Likewise.
8471 2017-09-13 Martin Liska <mliska@suse.cz>
8474 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8475 CASE_HIGH is NULL_TREE.
8477 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8478 Alan Hayward <alan.hayward@arm.com>
8479 David Sherwood <david.sherwood@arm.com>
8481 * target.def (secondary_memory_needed): New hook.
8482 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8483 instead of SECONDARY_MEMORY_NEEDED.
8484 (secondary_memory_needed_mode): Likewise.
8485 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8486 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8487 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8488 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8489 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8490 * doc/tm.texi: Regenerate.
8491 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8492 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8493 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8494 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8495 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8496 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8497 mode argument first and change the reg_class arguments to reg_class_t.
8498 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8499 Make static. Update the call to inline_secondary_memory_needed.
8500 (ix86_register_move_cost): Update the call to
8501 inline_secondary_memory_needed.
8502 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8503 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8505 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8506 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8508 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8509 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8510 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8511 and match hook interface. Add comment from mips.h.
8512 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8513 * config/mmix/mmix.md (truncdfsf2): Refer to
8514 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8516 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8517 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8518 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8519 (pa_secondary_memory_needed): New function.
8520 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8521 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8522 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8523 (pdp11_secondary_memory_needed): Make static and match hook interface.
8524 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8525 * config/powerpcspe/powerpcspe-protos.h
8526 (rs6000_secondary_memory_needed_ptr): Delete.
8527 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8529 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8530 (rs6000_option_override_internal): Assign to
8531 targetm.secondary_memory_needed rather than
8532 rs6000_secondary_memory_needed_ptr.
8533 (rs6000_secondary_memory_needed): Match hook interface.
8534 (rs6000_debug_secondary_memory_needed): Likewise.
8535 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8536 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8537 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8538 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8539 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8540 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8542 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8543 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8544 (rs6000_option_override_internal): Assign to
8545 targetm.secondary_memory_needed rather than
8546 rs6000_secondary_memory_needed_ptr.
8547 (rs6000_secondary_memory_needed): Match hook interface.
8548 (rs6000_debug_secondary_memory_needed): Likewise.
8549 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8550 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8552 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8553 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8554 (sparc_secondary_memory_needed): New function.
8555 * lra-constraints.c (check_and_process_move): Refer to
8556 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8558 (curr_insn_transform): Likewise.
8559 (process_alt_operands): Use targetm.secondary_memory_needed
8560 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8561 (check_secondary_memory_needed_p): Likewise.
8562 (choose_split_class): Likewise.
8563 * reload.c: Unconditionally include code that was previously
8564 conditional on SECONDARY_MEMORY_NEEDED.
8565 (push_secondary_reload): Use targetm.secondary_memory_needed
8566 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8567 (push_reload): Likewise.
8568 * reload1.c: Unconditionally include code that was previously
8569 conditional on SECONDARY_MEMORY_NEEDED.
8570 (choose_reload_regs): Use targetm.secondary_memory_needed
8571 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8572 (gen_reload): Likewise.
8573 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8575 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8576 Alan Hayward <alan.hayward@arm.com>
8577 David Sherwood <david.sherwood@arm.com>
8579 * target.def (secondary_memory_needed_mode): New hook:
8580 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8581 * targhooks.h (default_secondary_memory_needed_mode): New function.
8582 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8583 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8584 * doc/tm.texi: Regenerate.
8585 * lra-constraints.c (check_and_process_move): Use
8586 targetm.secondary_memory_needed_mode instead of
8587 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8588 (curr_insn_transform): Likewise.
8589 * reload.c (get_secondary_mem): Likewise.
8590 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8591 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8593 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8594 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8595 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8596 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8597 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8599 * config/powerpcspe/powerpcspe-protos.h
8600 (rs6000_secondary_memory_needed_mode): Delete.
8601 * config/powerpcspe/powerpcspe.c
8602 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8603 (rs6000_secondary_memory_needed_mode): Make static.
8604 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8605 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8607 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8609 (rs6000_secondary_memory_needed_mode): Make static.
8610 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8611 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8612 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8613 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8614 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8616 (sparc_secondary_memory_needed_mode): New function.
8617 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8619 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8621 * config/aarch64/constraints.md (Umq): New constraint.
8622 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8624 (mov<mode>): Update condition.
8626 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8628 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8629 when bitposition is the same.
8631 2017-09-13 Richard Biener <rguenther@suse.de>
8633 * dwarf2out.c (output_die_symbol): Remove.
8634 (output_die): Do not output a DIEs symbol.
8636 2017-09-13 Richard Biener <rguenther@suse.de>
8639 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8640 default-def to avoid breaking iterator update with the weird
8641 interaction with cgraph_update_edges_for_call_stmt_node.
8643 2017-09-13 Richard Biener <rguenther@suse.de>
8645 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8646 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8647 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8648 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8650 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8652 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8653 Disable pc relative literal load irrespective of
8654 TARGET_FIX_ERR_A53_84341 for default.
8656 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8658 * config/sparc/sparc.c (output_return): Output the source location of
8659 the insn in the delay slot, if any.
8660 (output_sibcall): Likewise.
8662 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8665 * common.opt (-static-pie): New alias.
8666 (shared): Negate static-pie.
8667 (-no-pie): Update help text.
8669 (static-pie): New option.
8670 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8671 -static-pie support.
8672 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8673 (LINK_EH_SPEC): Likewise.
8674 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8675 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8676 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8677 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8678 (init_gcc_specs): Likewise.
8679 (init_spec): Likewise.
8680 (display_help): Update help message for -pie.
8681 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8684 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8686 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8687 (movdi_aarch64): Likewise.
8688 (movti_aarch64): Likewise.
8690 2017-09-12 Simon Wright <simon@pushface.org>
8693 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8694 calculation of the minor version, always output as 0.
8696 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8699 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8700 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8701 on it early, rather than manual conversion late. For
8702 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8703 instead of performing manual conversion.
8705 2017-09-12 Carl Love <cel@us.ibm.com>
8707 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8708 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8710 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8711 VMULOSW): Add definitions.
8712 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8713 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8714 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8715 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8716 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8718 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8720 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8722 (movti_aarch64): Likewise.
8723 (movdf_aarch64): Likewise.
8724 (movtf_aarch64): Likewise.
8725 (load_pairdi): Likewise.
8726 (store_pairdi): Likewise.
8727 (load_pairdf): Likewise.
8728 (store_pairdf): Likewise.
8729 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8730 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8731 (ldr_got_small_<mode>): Likewise.
8732 (ldr_got_small_28k_<mode>): Likewise.
8733 (ldr_got_tiny): Likewise.
8734 * config/aarch64/iterators.md (ldst_sz): New.
8735 (ldpstp_sz): Likewise.
8736 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8738 (thunderx_load): Split load_8 to load_16.
8739 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8741 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8742 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8743 (xgene1_store_pair): Split store_8 to store_16.
8744 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8745 (falkor_st_0_st_sd): Split store_8 to store_16.
8747 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8749 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8750 and store1/2/3/4 to store_4/8/12/16.
8751 * config/aarch64/aarch64.md: Update for rename.
8752 * config/arm/arm.md: Likewise.: Likewise.
8753 * config/arm/arm.c: Likewise.
8754 * config/arm/thumb1.md: Likewise.
8755 * config/arm/thumb2.md: Likewise.
8756 * config/arm/vfp.md: Likewise.
8757 * config/arm/arm-generic.md: Likewise.
8758 * config/arm/arm1020e.md: Likewise.
8759 * config/arm/arm1026ejs.md: Likewise.
8760 * config/arm/arm1136jfs.md: Likewise.
8761 * config/arm/arm926ejs.md: Likewise.
8762 * config/arm/cortex-a15.md: Likewise.
8763 * config/arm/cortex-a17.md: Likewise.
8764 * config/arm/cortex-a5.md: Likewise.
8765 * config/arm/cortex-a53.md: Likewise.
8766 * config/arm/cortex-a57.md: Likewise.
8767 * config/arm/cortex-a7.md: Likewise.
8768 * config/arm/cortex-a8.md: Likewise.
8769 * config/arm/cortex-a9.md: Likewise.
8770 * config/arm/cortex-m4.md: Likewise.
8771 * config/arm/cortex-m7.md: Likewise.
8772 * config/arm/cortex-r4.md: Likewise.
8773 * config/arm/exynos-m1.md: Likewise.
8774 * config/arm/fa526.md: Likewise.
8775 * config/arm/fa606te.md: Likewise.
8776 * config/arm/fa626te.md: Likewise.
8777 * config/arm/fa726te.md: Likewise.
8778 * config/arm/fmp626.md: Likewise.
8779 * config/arm/iwmmxt.md: Likewise.
8780 * config/arm/ldmstm.md: Likewise.
8781 * config/arm/marvell-pj4.md: Likewise.
8782 * config/arm/xgene1.md: Likewise.
8783 * config/aarch64/thunderx.md: Likewise.
8784 * config/aarch64/thunderx2t99.md: Likewise.
8785 * config/aarch64/falkor.md: Likewise.
8787 2017-09-12 Martin Liska <mliska@suse.cz>
8789 * attribs.c (private_lookup_attribute): New function.
8790 * attribs.h (private_lookup_attribute): Declared here.
8791 (lookup_attribute): Called from this place.
8793 2017-09-12 Richard Biener <rguenther@suse.de>
8795 PR tree-optimization/82157
8796 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8797 stmts with side-effects.
8799 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8800 Alan Hayward <alan.hayward@arm.com>
8801 David Sherwood <david.sherwood@arm.com>
8803 * target.def (hard_regno_nregs): New hook.
8804 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8805 * targhooks.h (default_hard_regno_nregs): Declare.
8806 * targhooks.c (default_hard_regno_nregs): New function.
8807 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8808 (TARGET_HARD_REGNO_NREGS): ...this hook.
8809 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8810 (CLASS_MAX_NREGS): Likewise.
8811 * doc/tm.texi: Regenerate.
8812 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8813 instead of HARD_REGNO_NREGS.
8814 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8815 HARD_REGNO_NREGS in the comment.
8816 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8817 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8818 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8819 Return an unsigned int.
8820 (TARGET_HARD_REGNO_NREGS): Redefine.
8821 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8822 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8823 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8824 (arc_hard_regno_nregs): New function.
8825 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8826 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8827 (arm_hard_regno_nregs): New function.
8828 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8829 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8830 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8831 (TARGET_HARD_REGNO_NREGS): Redefine.
8832 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8833 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8834 (HARD_REGNO_NREGS): Delete.
8835 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8836 (cr16_hard_regno_nregs): New function.
8837 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8838 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8839 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8840 (cris_hard_regno_nregs): New function.
8841 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8842 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8843 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8844 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8845 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8846 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8847 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8848 (frv_hard_regno_nregs): Make static. Take and return an
8850 (frv_class_max_nregs): Remove outdated copy of documentation.
8851 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8852 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8853 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8854 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8855 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8856 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8857 (TARGET_HARD_REGNO_NREGS): Redefine.
8858 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8859 (CLASS_MAX_NREGS): Update comment.
8860 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8861 (ia64_hard_regno_nregs): New function.
8862 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8863 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8864 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8865 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8866 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8868 (m32c_hard_regno_nregs): Likewise. Make static.
8869 (TARGET_HARD_REGNO_NREGS): Redefine.
8870 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8871 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8872 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8873 (m68k_hard_regno_nregs): New function.
8874 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8875 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8876 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8877 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8878 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8879 Take and return an unsigned int.
8880 (TARGET_HARD_REGNO_NREGS): Redefine.
8881 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8882 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8883 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8884 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8885 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8886 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8887 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8888 (msp430_hard_regno_nregs): Make static. Take and return an
8890 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8891 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8892 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8893 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8894 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8895 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8896 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8897 (TARGET_HARD_REGNO_NREGS): Redefine.
8898 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8899 (PA_HARD_REGNO_NREGS): ...this.
8900 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8901 (PA_HARD_REGNO_NREGS): ...this.
8902 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8903 (pa_hard_regno_nregs): New function.
8904 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8905 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8906 (pdp11_hard_regno_nregs): New function.
8907 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8908 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8909 (rs6000_hard_regno_nregs_hook): New function.
8910 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8911 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8912 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8913 Take and return an unsigned int. Move earlier in file.
8914 (TARGET_HARD_REGNO_NREGS): Redefine.
8915 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8916 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8917 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8918 (rl78_hard_regno_nregs): Make static. Take and return an
8920 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8921 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8922 (rs6000_hard_regno_nregs_hook): New function.
8923 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8924 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8925 (TARGET_HARD_REGNO_NREGS): Redefine.
8926 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8927 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8928 instead of HARD_REGNO_NREGS.
8929 (s390_hard_regno_nregs): New function.
8930 (s390_hard_regno_mode_ok): Add comment from s390.h.
8931 (TARGET_HARD_REGNO_NREGS): Redefine.
8932 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8933 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8934 (sh_hard_regno_nregs): New function.
8935 (sh_pass_in_reg_p): Use it.
8936 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8937 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8938 (sparc_hard_regno_nregs): New function.
8939 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8940 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8941 (spu_function_arg_advance): Use it, supplying a valid register number.
8942 (TARGET_HARD_REGNO_NREGS): Redefine.
8943 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8944 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8945 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8946 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8947 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8948 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8949 (CLASS_MAX_NREGS): Remove copy of old documentation.
8950 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8951 (visium_hard_regno_nregs): New function.
8952 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8953 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8954 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8955 xtensa_hard_regno_nregs): New function.
8956 * system.h (HARD_REGNO_NREGS): Poison.
8958 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8960 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8961 hard_regno_nregs instead of HARD_REGNO_NREGS.
8962 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8963 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8964 (c6x_expand_epilogue): Likewise.
8965 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8966 (frv_read_iacc_argument): Likewise.
8967 * config/sh/sh.c: Include regs.h.
8968 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8969 (regs_used): Likewise.
8970 (output_stack_adjust): Likewise.
8971 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8972 * expmed.c: Include regs.h.
8973 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8974 * ree.c: Include regs.h.
8975 (combine_reaching_defs): Use hard_regno_nregs instead of
8977 (add_removable_extension): Likewise.
8979 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8981 * regs.h (hard_regno_nregs): Turn into a function.
8982 (end_hard_regno): Update accordingly.
8983 * caller-save.c (setup_save_areas): Likewise.
8984 (save_call_clobbered_regs): Likewise.
8985 (replace_reg_with_saved_mem): Likewise.
8986 (insert_restore): Likewise.
8987 (insert_save): Likewise.
8988 * combine.c (can_change_dest_mode): Likewise.
8989 (move_deaths): Likewise.
8990 (distribute_notes): Likewise.
8991 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8992 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8993 (rs6000_split_multireg_move): Likewise.
8994 (rs6000_register_move_cost): Likewise.
8995 (rs6000_memory_move_cost): Likewise.
8996 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8997 (rs6000_split_multireg_move): Likewise.
8998 (rs6000_register_move_cost): Likewise.
8999 (rs6000_memory_move_cost): Likewise.
9000 * cselib.c (cselib_reset_table): Likewise.
9001 (cselib_lookup_1): Likewise.
9002 * emit-rtl.c (set_mode_and_regno): Likewise.
9003 * function.c (aggregate_value_p): Likewise.
9004 * ira-color.c (setup_profitable_hard_regs): Likewise.
9005 (check_hard_reg_p): Likewise.
9006 (calculate_saved_nregs): Likewise.
9007 (assign_hard_reg): Likewise.
9008 (improve_allocation): Likewise.
9009 (calculate_spill_cost): Likewise.
9010 * ira-emit.c (modify_move_list): Likewise.
9011 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9012 (ira_hard_reg_in_set_p): Likewise.
9013 * ira.c (setup_reg_mode_hard_regset): Likewise.
9014 (clarify_prohibited_class_mode_regs): Likewise.
9015 (check_allocation): Likewise.
9016 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9017 (lra_setup_reg_renumber): Likewise.
9018 (setup_try_hard_regno_pseudos): Likewise.
9019 (spill_for): Likewise.
9020 (assign_hard_regno): Likewise.
9021 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9022 * lra-constraints.c (in_class_p): Likewise.
9023 (lra_constraint_offset): Likewise.
9024 (simplify_operand_subreg): Likewise.
9025 (lra_constraints): Likewise.
9026 (split_reg): Likewise.
9027 (split_if_necessary): Likewise.
9028 (invariant_p): Likewise.
9029 (inherit_in_ebb): Likewise.
9030 * lra-lives.c (process_bb_lives): Likewise.
9031 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9032 (get_hard_regs): Likewise.
9033 (do_remat): Likewise.
9034 * lra-spills.c (assign_spill_hard_regs): Likewise.
9035 * mode-switching.c (create_pre_exit): Likewise.
9036 * postreload.c (reload_combine_recognize_pattern): Likewise.
9037 * recog.c (peep2_find_free_register): Likewise.
9038 * regcprop.c (kill_value_regno): Likewise.
9039 (set_value_regno): Likewise.
9040 (copy_value): Likewise.
9041 (maybe_mode_change): Likewise.
9042 (find_oldest_value_reg): Likewise.
9043 (copyprop_hardreg_forward_1): Likewise.
9044 * regrename.c (check_new_reg_p): Likewise.
9045 (regrename_do_replace): Likewise.
9046 * reload.c (push_reload): Likewise.
9047 (combine_reloads): Likewise.
9048 (find_dummy_reload): Likewise.
9049 (operands_match_p): Likewise.
9050 (find_reloads): Likewise.
9051 (find_equiv_reg): Likewise.
9052 (reload_adjust_reg_for_mode): Likewise.
9053 * reload1.c (count_pseudo): Likewise.
9054 (count_spilled_pseudo): Likewise.
9055 (find_reg): Likewise.
9056 (clear_reload_reg_in_use): Likewise.
9057 (free_for_value_p): Likewise.
9058 (allocate_reload_reg): Likewise.
9059 (choose_reload_regs): Likewise.
9060 (reload_adjust_reg_for_temp): Likewise.
9061 (emit_reload_insns): Likewise.
9062 (delete_output_reload): Likewise.
9063 * rtlanal.c (subreg_get_info): Likewise.
9064 * sched-deps.c (sched_analyze_reg): Likewise.
9065 * sel-sched.c (init_regs_for_mode): Likewise.
9066 (mark_unavailable_hard_regs): Likewise.
9067 (choose_best_reg_1): Likewise.
9068 (verify_target_availability): Likewise.
9069 * valtrack.c (dead_debug_insert_temp): Likewise.
9070 * var-tracking.c (track_loc_p): Likewise.
9071 (emit_note_insn_var_location): Likewise.
9072 * varasm.c (make_decl_rtl): Likewise.
9073 * reginfo.c (choose_hard_reg_mode): Likewise.
9074 (init_reg_modes_target): Refer directly to
9075 this_target_regs->x_hard_regno_nregs.
9077 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9079 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9080 instead of hard_regno_nregs.
9082 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9084 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9085 end_hard_regno instead of hard_regno_nregs.
9086 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9087 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9088 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9089 * ira-color.c (improve_allocation): Likewise.
9090 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9091 * lra-lives.c (mark_regno_live): Likewise.
9092 (mark_regno_dead): Likewise.
9093 * lra-remat.c (operand_to_remat): Likewise.
9094 * lra.c (collect_non_operand_hard_regs): Likewise.
9095 * postreload.c (reload_combine_note_store): Likewise.
9096 (move2add_valid_value_p): Likewise.
9097 * reload.c (regno_clobbered_p): Likewise.
9099 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9101 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9103 * config/v850/v850.c (v850_reorg): Likewise.
9104 * reload.c (refers_to_regno_for_reload_p): Likewise.
9105 (find_equiv_reg): Likewise.
9106 * reload1.c (reload_reg_reaches_end_p): Likewise.
9108 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9110 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9112 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9113 * config/arm/arm.c (output_move_neon): Likewise.
9114 (arm_attr_length_move_neon): Likewise.
9115 (neon_split_vcombine): Likewise.
9116 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9117 (c6x_mark_reg_written): Likewise.
9118 (c6x_dwarf_register_span): Likewise.
9119 * config/i386/i386.c (ix86_save_reg): Likewise.
9120 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9121 (rws_access_reg): Likewise.
9122 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9123 * mode-switching.c (create_pre_exit): Likewise.
9124 * ree.c (combine_reaching_defs): Likewise.
9125 (add_removable_extension): Likewise.
9126 * regcprop.c (find_oldest_value_reg): Likewise.
9127 (copyprop_hardreg_forward_1): Likewise.
9128 * reload.c (reload_inner_reg_of_subreg): Likewise.
9129 (push_reload): Likewise.
9130 (combine_reloads): Likewise.
9131 (find_dummy_reload): Likewise.
9132 (reload_adjust_reg_for_mode): Likewise.
9133 * reload1.c (find_reload_regs): Likewise.
9134 (forget_old_reloads_1): Likewise.
9135 (reload_reg_free_for_value_p): Likewise.
9136 (reload_adjust_reg_for_temp): Likewise.
9137 (emit_reload_insns): Likewise.
9138 (delete_output_reload): Likewise.
9139 * sel-sched.c (choose_best_reg_1): Likewise.
9140 (choose_best_pseudo_reg): Likewise.
9142 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9143 Alan Hayward <alan.hayward@arm.com>
9144 David Sherwood <david.sherwood@arm.com>
9146 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9147 * target.def (slow_unaligned_access): New hook.
9148 * targhooks.h (default_slow_unaligned_access): Declare.
9149 * targhooks.c (default_slow_unaligned_access): New function.
9150 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9151 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9152 * doc/tm.texi: Regenerate.
9153 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9154 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9155 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9157 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9158 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9160 (rs6000_slow_unaligned_access): New function.
9161 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9162 (expand_block_compare): Likewise.
9163 (expand_strn_compare): Likewise.
9164 (rs6000_rtx_costs): Likewise.
9165 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9166 (riscv_slow_unaligned_access): Likewise.
9167 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9168 (riscv_slow_unaligned_access_p): ...this and make static.
9169 (riscv_option_override): Update accordingly.
9170 (riscv_slow_unaligned_access): New function.
9171 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9172 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9173 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9174 (rs6000_slow_unaligned_access): New function.
9175 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9176 (rs6000_rtx_costs): Likewise.
9177 * config/rs6000/rs6000-string.c (expand_block_compare)
9178 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9179 of SLOW_UNALIGNED_ACCESS.
9180 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9181 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9182 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9183 of SLOW_UNALIGNED_ACCESS.
9184 * expmed.c (simple_mem_bitfield_p): Likewise.
9185 * expr.c (alignment_for_piecewise_move): Likewise.
9186 (emit_group_load_1): Likewise.
9187 (emit_group_store): Likewise.
9188 (copy_blkmode_from_reg): Likewise.
9189 (emit_push_insn): Likewise.
9190 (expand_assignment): Likewise.
9191 (store_field): Likewise.
9192 (expand_expr_real_1): Likewise.
9193 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9194 * lra-constraints.c (simplify_operand_subreg): Likewise.
9195 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9196 * gimple-ssa-store-merging.c: Likewise in block comment at start
9198 * tree-ssa-strlen.c: Include target.h.
9199 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9200 of SLOW_UNALIGNED_ACCESS.
9201 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9203 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9205 PR rtl-optimization/82185
9206 * expmed.c (emit_store_flag_int): Only test tem if it has been
9209 2017-09-12 Richard Biener <rguenther@suse.de>
9212 * match.pd ((FTYPE) N CMP CST): Fix typo.
9214 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9216 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9218 (mips_near_type_p): Add 'short_call' attribute as a synonym
9220 * doc/extend.texi (short_call): Document new function attribute.
9222 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9225 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9226 assertion check that in the condition.
9227 (get_atomic_generic_size): Likewise. Before testing if parameter
9228 has pointer type, if it has array type, call for C++
9229 default_conversion to perform array-to-pointer conversion.
9231 2017-09-12 Richard Biener <rguenther@suse.de>
9233 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9234 for operations we cannot scalarize.
9236 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9238 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9239 vectors heap vectors. Clean up comments.
9240 Make visited_bbs a reference.
9241 (profitable_jump_thread_path): Make GC
9242 vectors heap vectors. Clean up comments.
9244 (convert_and_register_jump_thread_path): Make GC vectors heap
9246 (check_subpath_and_update_thread_path): Same. Clean up comments.
9247 Make visited_bbs a reference.
9248 (handle_phi): Abstract common code to to
9249 register_jump_thread_path_if_profitable.
9250 Rename VAR_BB to DEF_BB.
9252 Make GC vectors heap vectors.
9253 Make visited_bbs a reference.
9254 (handle_assignment): Same.
9255 (register_jump_thread_path_if_profitable): New.
9256 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9258 Make GC vectors heap vectors. Clean up comments.
9259 Make visited_bbs a reference.
9260 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9261 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9264 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9267 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9268 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9270 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9272 Revert r251800 and r251799.
9274 2017-09-11 Martin Jambor <mjambor@suse.cz>
9277 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9278 arguments in advance.
9279 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9280 use it to find predecessor edges.
9281 (naive_outof_ssa): Collect vector of predecessors.
9283 2017-09-08 Jason Merrill <jason@redhat.com>
9285 PR c++/70029 - ICE with ref-qualifier and -flto
9286 * langhooks.h (struct lang_hooks_for_types): Add
9287 copy_lang_qualifiers.
9288 * attribs.c (build_type_attribute_qual_variant): Use it.
9289 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9291 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9292 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9294 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9297 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9298 (*mulsi3_sp64): New instruction.
9299 (mulsi3): New expander.
9301 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9303 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9305 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9307 * sancov.c: Include memmodel.h.
9309 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9312 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9315 2017-09-07 Carl Love <cel@us.ibm.com>
9317 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9318 the sldi instruction.
9320 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9322 * sancov.c: Include tm_p.h.
9324 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9327 * output.h (switch_to_other_text_partition): New declaration.
9328 * varasm.c (switch_to_other_text_partition): New function.
9329 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9330 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9331 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9332 to the other text partition before emitting LCL label and switch back
9333 after emitting the word after it.
9335 2017-09-07 Richard Biener <rguenther@suse.de>
9337 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9338 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9339 critical edges here, after loop init.
9340 (pass_data_pre): Remove PROP_no_crit_edges flags.
9341 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9342 for valueization of call args to avoid leaking VN_TOP.
9343 (visit_use): Assert we do not visit default defs.
9344 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9345 Use error_mark_node to more easily detect leaking VN_TOP.
9346 All default-defs are varying, not VN_TOP. Mark them visited.
9347 (run_scc_vn): Make code match comment.
9349 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9351 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9352 OPTION_MASK_FLOAT128_KEYWORD.
9353 (POWERPC_MASKS): Likewise.
9354 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9355 support for the -mfloat128-type option, and make -mfloat128
9356 default on PowerPC Linux systems. Define or undefine
9357 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9358 Define __float128 to be __ieee128 if IEEE 128-bit support is
9359 enabled, or undefine it.
9360 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9361 Delete defining __FLOAT128_TYPE__.
9362 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9363 -mfloat128-type option and make -mfloat128 default on PowerPC
9365 (TARGET_FLOAT128_TYPE): Likewise.
9366 (-mfloat128-type): Likewise.
9367 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9368 Delete the -mfloat128-type option and make -mfloat128 default on
9369 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9370 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9371 128-bit floating point is enabled. Change tests from using
9372 -mfloat128-type to -mfloat128.
9373 (rs6000_mangle_type): Use the correct mangling for the __float128
9374 type even if normal long double is restricted to 64-bits.
9375 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9377 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9378 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9379 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9381 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9382 documentation for -mfloat128.
9384 2017-09-06 Olivier Hainque <hainque@adacore.com>
9386 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9388 2017-09-06 Wish Wu <wishwu007@gmail.com>
9389 Jakub Jelinek <jakub@redhat.com>
9391 * asan.c (initialize_sanitizer_builtins): Add
9392 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9393 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9394 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9395 BT_FN_VOID_UINT64_PTR variables.
9396 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9397 (BT_FN_VOID_UINT16_UINT16): Likewise.
9398 (BT_FN_VOID_UINT32_UINT32): Likewise.
9399 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9400 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9401 (BT_FN_VOID_UINT64_PTR): Likewise.
9402 * common.opt (flag_sanitize_coverage): New variable.
9403 (fsanitize-coverage=trace-pc): Remove.
9404 (fsanitize-coverage=): Add.
9405 * flag-types.h (enum sanitize_coverage_code): New enum.
9406 * fold-const.c (fold_range_test): Disable non-short-circuit
9407 optimization if flag_sanitize_coverage.
9408 (fold_truth_andor): Likewise.
9409 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9410 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9411 (coverage_sanitizer_opts): New array.
9412 (get_closest_sanitizer_option): Add OPTS argument, handle also
9413 OPT_fsanitize_coverage_.
9414 (parse_sanitizer_options): Adjusted to also handle
9415 OPT_fsanitize_coverage_.
9416 (common_handle_option): Add OPT_fsanitize_coverage_.
9417 * sancov.c (instrument_comparison, instrument_switch): New function.
9418 (sancov_pass): Add trace-cmp support.
9419 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9420 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9421 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9422 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9423 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9424 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9425 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9426 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9427 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9428 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9430 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9432 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9433 error. Only quit immediately if parsing is complete.
9434 (BEGIN): Initialize fatal_err and parse_done.
9435 (begin fpu, end fpu): Check number of arguments.
9436 (begin arch, end arch): Likewise.
9437 (begin cpu, end cpu): Likewise.
9438 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9439 (optalias): Likewise.
9441 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9443 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9444 * config/arm/arm-isa.h: Delete. Move definitions to ...
9445 * arm-cpus.in: ... here. Use new feature and fgroup values.
9446 * config/arm/arm.c (arm_option_override): Use lower case for feature
9448 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9449 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9450 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9452 (print_isa_bits_for): New function.
9453 (gen_isa): New function.
9454 (gen_comm_data): Use print_isa_bits_for.
9455 (define feature): New keyword.
9456 (define fgroup): New keyword.
9457 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9458 (arm-isa.h): Add rule to generate file.
9459 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9460 case for feature bit names.
9462 2017-09-06 Richard Biener <rguenther@suse.de>
9464 * tree-ssa-pre.c (NECESSARY): Remove.
9465 (create_expression_by_pieces): Do not touch pass-local flags.
9466 (insert_into_preds_of_block): Likewise.
9467 (do_pre_regular_insertion): Likewise.
9468 (eliminate_insert): Likewise.
9469 (eliminate_dom_walker::before_dom_children): Likewise.
9470 (fini_eliminate): Do not look at inserted_exprs.
9471 (mark_operand_necessary): Remove.
9472 (remove_dead_inserted_code): Replace with simple work-list
9473 algorithm based on inserted_exprs and SSA uses.
9474 (pass_pre::execute): Re-order fini_eliminate and
9475 remove_dead_inserted_code.
9477 2017-09-06 Olivier Hainque <hainque@adacore.com>
9479 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9480 for VxWorks 7. Adjust surrounding comments.
9482 2017-09-06 Richard Biener <rguenther@suse.de>
9484 * gimple-ssa-strength-reduction.c
9485 (find_candidates_dom_walker::before_dom_children): Also allow
9488 2017-09-06 Richard Biener <rguenther@suse.de>
9490 PR tree-optimization/82108
9491 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9492 for gap in the non-permutation SLP case.
9494 2017-09-06 Martin Jambor <mjambor@suse.cz>
9496 PR tree-optimization/82078
9497 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9498 add_access_to_work_queue...
9499 (build_accesses_from_assign): ...here.
9500 (propagate_all_subaccesses): Make sure racc is the group
9501 representative, if there is one.
9503 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9506 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9509 2017-09-06 Richard Biener <rguenther@suse.de>
9511 * gimple-ssa-strength-reduction.c
9512 (find_candidates_dom_walker::before_doom_children): Use a
9513 type and not a mode check.
9515 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9518 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9520 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9522 2017-09-05 Jeff Law <law@redhat.com>
9524 PR tree-optimization/64910
9525 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9526 cases where we have 3 or more operands.
9528 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9531 * omp-low.c (lower_omp_for): Recompute tree invariant if
9532 gimple_omp_for_initial/final is ADDR_EXPR.
9535 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9536 into gimple val before gimplification fo the COND_EXPR.
9538 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9540 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9541 REGION_COPY argument.
9542 (thread_through_all_blocks): Remove unused argument to
9543 duplicate_thread_path.
9545 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9546 Alan Hayward <alan.hayward@arm.com>
9547 David Sherwood <david.sherwood@arm.com>
9549 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9550 Take a scalar_mode rather than a machine_mode.
9551 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9552 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9553 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9554 (aarch64_gen_adjusted_ldpstp): Likewise.
9555 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9557 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9558 Alan Hayward <alan.hayward@arm.com>
9559 David Sherwood <david.sherwood@arm.com>
9561 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9562 Take a scalar_int_mode instead of a machine_mode.
9563 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9564 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9565 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9566 (aarch64_simd_attr_length_rglist): Delete.
9567 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9568 a scalar_int_mode instead of a machine_mode.
9569 (aarch64_add_offset): Likewise.
9570 (aarch64_internal_mov_immediate): Likewise
9571 (aarch64_add_constant_internal): Likewise.
9572 (aarch64_add_constant): Likewise.
9573 (aarch64_movw_imm): Likewise.
9574 (aarch64_rtx_arith_op_extract_p): Likewise.
9575 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9576 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9577 Remove assert that the mode isn't a vector.
9578 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9579 (aarch64_expand_mov_immediate): Update calls after above changes.
9580 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9581 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9582 (aarch64_move_imm): Likewise.
9583 (aarch64_can_const_movi_rtx_p): Likewise.
9584 (aarch64_strip_extend): Likewise.
9585 (aarch64_extr_rtx_p): Likewise.
9586 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9587 a CONST_INT when the mode parameter is VOIDmode.
9588 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9590 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9592 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9593 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9594 (bitwise_type_for_mode): Update accordingly.
9596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9598 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9599 * stor-layout.c (mode_for_size_tree): Likewise.
9600 (mode_for_array): Update accordingly.
9601 (layout_decl): Likewise.
9602 (compute_record_mode): Likewise. Only set the mode once.
9604 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9606 * target.def (get_mask_mode): Change return type to opt_mode.
9608 * doc/tm.texi: Regenerate.
9609 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9610 * targhooks.c (default_get_mask_mode): Likewise.
9611 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9612 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9613 targetm.get_mask_mode.
9614 * tree.c (build_truth_vector_type): Likewise.
9616 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9618 * machmode.h (mode_for_vector): Return an opt_mode.
9619 * stor-layout.c (mode_for_vector): Likewise.
9620 (mode_for_int_vector): Update accordingly.
9621 (layout_type): Likewise.
9622 * config/i386/i386.c (emit_memmov): Likewise.
9623 (ix86_expand_set_or_movmem): Likewise.
9624 (ix86_expand_vector_init): Likewise.
9625 (ix86_get_mask_mode): Likewise.
9626 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9628 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9629 * expmed.c (extract_bit_field_1): Likewise.
9630 * expr.c (expand_expr_real_2): Likewise.
9631 * optabs-query.c (can_vec_perm_p): Likewise.
9632 (can_vec_mask_load_store_p): Likewise.
9633 * optabs.c (expand_vec_perm): Likewise.
9634 * targhooks.c (default_get_mask_mode): Likewise.
9635 * tree-vect-stmts.c (vectorizable_store): Likewise.
9636 (vectorizable_load): Likewise.
9637 (get_vectype_for_scalar_type_and_size): Likewise.
9639 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9641 * machmode.h (mode_for_int_vector): New function.
9642 * stor-layout.c (mode_for_int_vector): Likewise.
9643 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9644 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9645 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9646 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9647 (s390_expand_vcond): Likewise.
9649 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9651 * machmode.h (opt_machine_mode): New type.
9652 (opt_mode<T>): Allow construction from anything that can be
9654 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9655 (mode_for_size): Return an opt_machine_mode.
9656 * stor-layout.c (mode_for_size): Likewise.
9657 (mode_for_size_tree): Update call accordingly.
9658 (bitwise_mode_for_mode): Likewise.
9659 (make_fract_type): Likewise.
9660 (make_accum_type): Likewise.
9661 * caller-save.c (replace_reg_with_saved_mem): Update call
9663 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9664 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9665 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9666 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9667 * expmed.c (extract_bit_field_1): Likewise.
9668 * reload.c (get_secondary_mem): Likewise.
9669 * varasm.c (assemble_integer): Likewise.
9670 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9673 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9675 * machmode.h (decimal_float_mode_for_size): New function.
9676 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9677 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9678 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9679 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9680 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9681 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9683 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9685 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9686 (get_builtin_sync_mode): Likewise.
9687 (expand_ifn_atomic_compare_exchange): Likewise.
9688 (expand_builtin_atomic_clear): Likewise.
9689 (expand_builtin_atomic_test_and_set): Likewise.
9690 (fold_builtin_atomic_always_lock_free): Likewise.
9691 * calls.c (compute_argument_addresses): Likewise.
9692 (emit_library_call_value_1): Likewise.
9693 (store_one_arg): Likewise.
9694 * combine.c (combine_instructions): Likewise.
9695 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9696 * config/arm/arm.c (arm_function_value): Likewise.
9697 (aapcs_allocate_return_reg): Likewise.
9698 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9699 * config/i386/i386.c (construct_container): Likewise.
9700 (ix86_gimplify_va_arg): Likewise.
9701 (ix86_expand_sse_cmp): Likewise.
9702 (emit_memmov): Likewise.
9703 (emit_memset): Likewise.
9704 (expand_small_movmem_or_setmem): Likewise.
9705 (ix86_expand_pextr): Likewise.
9706 (ix86_expand_pinsr): Likewise.
9707 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9708 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9710 * config/mips/mips.c (mips_function_value_1) Likewise.
9711 (mips_block_move_straight): Likewise.
9712 (mips_expand_ins_as_unaligned_store): Likewise.
9713 * config/powerpcspe/powerpcspe.c
9714 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9715 (rs6000_darwin64_record_arg_flush): Likewise.
9716 * config/rs6000/rs6000.c
9717 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9718 (rs6000_darwin64_record_arg_flush): Likewise.
9719 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9720 (sparc_function_value_1): Likewise.
9721 * config/spu/spu.c (adjust_operand): Likewise.
9722 (spu_emit_branch_or_set): Likewise.
9723 (arith_immediate_p): Likewise.
9724 * emit-rtl.c (gen_lowpart_common): Likewise.
9725 * expr.c (expand_expr_real_1): Likewise.
9726 * function.c (assign_parm_setup_block): Likewise.
9727 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9728 * reload1.c (alter_reg): Likewise.
9729 * stor-layout.c (mode_for_vector): Likewise.
9730 (layout_type): Likewise.
9732 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9734 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9735 (spu_convert_move): Likewise.
9736 * lower-subreg.c (resolve_simple_move): Likewise.
9738 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9741 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9742 define_insn to a define_expand.
9743 (altivec_vsum2sws_direct): New define_insn.
9744 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9746 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9748 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9751 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9756 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9757 realign stack if stack alignment needed is less than incoming
9760 2017-09-05 Marek Polacek <polacek@redhat.com>
9763 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9766 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9768 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9770 2017-09-05 Richard Biener <rguenther@suse.de>
9772 PR tree-optimization/82084
9773 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9775 2017-09-05 Richard Biener <rguenther@suse.de>
9777 PR tree-optimization/82102
9778 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9780 2017-09-05 Martin Liska <mliska@suse.cz>
9782 PR tree-optimization/82032
9783 * tree-cfg.c (generate_range_test): New function.
9784 * tree-cfg.h (generate_range_test): Declared here.
9785 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9786 (cleanup_control_expr_graph): Use it.
9787 * tree-switch-conversion.c (try_switch_expansion): Remove
9789 (emit_case_nodes): Use generate_range_test.
9791 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9794 * config/i386/i386.md (*<btsc><mode>_mask): Add
9795 TARGET_USE_BT to insn constraint.
9796 (*btr<mode>_mask): Ditto.
9798 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9800 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9801 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9803 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9806 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9807 TARGET_NEON and TARGET_IWMMXT.
9808 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9809 TARGET_NEON and TARGET_IWMMXT.
9810 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9812 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9814 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9815 (ix86_rewrite_tls_address): Ditto.
9816 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9817 (ix86_rewrite_tls_address_1): Ditto.
9818 (ix86_rewrite_tls_address): Ditto.
9819 * config/i386/predicates.md (tls_address_pattern): New predicate.
9820 * config/i386/i386.md (TLS address splitter): New splitter.
9822 2017-09-04 Richard Biener <rguenther@suse.de>
9824 PR tree-optimization/82084
9825 * fold-const.h (can_native_encode_string_p): Declare.
9826 * fold-const.c (can_native_encode_string_p): Factor out from ...
9827 (native_encode_string): ... here.
9828 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9829 vectorizing stores from constants we later cannot handle.
9831 2017-09-04 Marek Polacek <polacek@redhat.com>
9834 * doc/invoke.texi: Update -Wtautological-compare documentation.
9836 2017-09-04 Jeff Law <law@redhat.com>
9838 PR tree-optimization/64910
9839 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9840 swap the first and last operand if the last is a constant.
9842 2017-09-04 Marek Polacek <polacek@redhat.com>
9845 * convert.c (do_narrow): When sanitizing signed integer overflows,
9846 bail out for signed types.
9847 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9849 2017-09-04 Richard Biener <rguenther@suse.de>
9851 PR tree-optimization/82060
9852 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9853 Move devirtualization after stmt folding and before EH/AB/noreturn
9854 cleanup to get the stmt refs canonicalized. Use a bool instead
9855 of gimple_modified_p since that doesn't work for NOPs. Schedule
9856 NOPs generated by folding for removal.
9858 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9859 Alan Hayward <alan.hayward@arm.com>
9860 David Sherwood <david.sherwood@arm.com>
9862 * coretypes.h (pad_direction): New enum.
9863 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9864 (FUNCTION_ARG_PADDING): Likewise.
9865 * target.def (function_arg_padding): New hook.
9866 * targhooks.h (default_function_arg_padding): Declare.
9867 * targhooks.c (default_function_arg_padding): New function.
9868 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9869 (TARGET_FUNCTION_ARG_PADDING): ...this.
9870 * doc/tm.texi: Regenerate.
9871 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9872 instead of direction.
9873 (compute_argument_addresses): Likewise.
9874 (load_register_parameters): Likewise.
9875 (emit_library_call_value_1): Likewise.
9876 (store_one_arg): Use targetm.calls.function_arg_padding instead
9877 of FUNCTION_ARG_PADDING.
9878 (must_pass_in_stack_var_size_or_pad): Likewise.
9879 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9880 (emit_group_store): Likewise.
9881 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9882 instead of FUNCTION_ARG_PADDING.
9883 (emit_push_insn): Likewise, and propagate enum change throughout
9885 * function.h (direction): Delete.
9886 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9888 * function.c (assign_parm_find_stack_rtl): Likewise.
9889 (assign_parm_setup_block_p): Likewise.
9890 (assign_parm_setup_block): Likewise.
9891 (gimplify_parameters): Likewise.
9892 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9893 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9895 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9896 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9897 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9898 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9899 (aarch64_function_arg_padding): ...this new function.
9900 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9901 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9902 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9903 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9904 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9905 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9906 (arm_pad_arg_upward): Replace with...
9907 (arm_function_arg_padding): ...this new function.
9908 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9910 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9911 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9912 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9913 (ia64_hpux_function_arg_padding): Replace with...
9914 (ia64_function_arg_padding): ...this new function. Use pad_direction
9915 instead of direction. Check for TARGET_HPUX.
9916 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9917 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9918 (iq2000_function_arg_padding): New function.
9919 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9920 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9921 (mips_function_arg_padding): ...this new function.
9922 (mips_pad_reg_upward): Update accordingly.
9923 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9924 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9925 targetm.calls.function_arg_padding.
9926 (FUNCTION_ARG_PADDING): Delete.
9927 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9928 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9929 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9930 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9931 (nios2_block_reg_padding): Return pad_direction instead of direction.
9932 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9933 instead of direction.
9934 (nios2_function_arg_padding): Likewise. Make static.
9935 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9936 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9937 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9938 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9939 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9940 (pa_function_arg_padding): Make static. Return pad_direction instead
9942 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9943 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9944 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9945 instead of direction. Use targetm.calls.function_arg_padding.
9946 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9947 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9948 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9949 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9950 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9952 (function_arg_padding): Rename to...
9953 (rs6000_function_arg_padding): ...this. Make static. Return
9954 pad_direction instead of direction.
9955 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9956 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9957 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9958 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9959 instead of direction. Use targetm.calls.function_arg_padding.
9960 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9961 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9962 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9963 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9964 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9965 (function_arg_padding): Rename to...
9966 (rs6000_function_arg_padding): ...this. Make static. Return
9967 pad_direction instead of direction.
9968 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9969 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9970 * config/s390/s390.c (s390_function_arg_padding): New function.
9971 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9972 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9973 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9974 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9975 (function_arg_padding): Rename to...
9976 (sparc_function_arg_padding): ...this. Make static. Return
9977 pad_direction instead of direction.
9978 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9979 * config/spu/spu.c (spu_function_arg_padding): New function.
9980 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9981 * system.h (FUNCTION_ARG_PADDING): Poison.
9983 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9984 Alan Hayward <alan.hayward@arm.com>
9985 David Sherwood <david.sherwood@arm.com>
9987 * target.def (modes_tieable_p): New hook.
9988 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9989 (TARGET_MODES_TIEABLE_P): ...this.
9990 * doc/tm.texi.in: Regenerate.
9991 * hooks.h (hook_bool_mode_mode_true): Declare.
9992 * hooks.c (hook_bool_mode_mode_true): New function.
9993 * combine.c (subst): Use targetm.modes_tieable_p instead of
9995 * dse.c (find_shift_sequence): Likewise.
9996 * expmed.c (extract_low_bits): Likewise.
9997 * lower-subreg.c: Include target.h.
9998 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10000 * rtlanal.c (rtx_cost): Likewise.
10001 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10002 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10003 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10004 (TARGET_MODES_TIEABLE_P): Redefine.
10005 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10006 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10007 (TARGET_MODES_TIEABLE_P): Redefine.
10008 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10009 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10010 (arc_modes_tieable_p): New function.
10011 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10012 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10013 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10014 (arm_modes_tieable_p): Make static.
10015 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10016 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10017 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10018 (TARGET_MODES_TIEABLE_P): Redefine.
10019 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10020 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10021 (TARGET_MODES_TIEABLE_P): Redefine.
10022 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10023 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10024 (cr16_modes_tieable_p): New function.
10025 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10026 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10027 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10028 (TRULY_NOOP_TRUNCATION): Update comment.
10029 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10030 (TRULY_NOOP_TRUNCATION): Update comment.
10031 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10032 (frv_modes_tieable_p): New function.
10033 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10034 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10035 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10036 (TARGET_MODES_TIEABLE_P): Redefine.
10037 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10038 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10039 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10040 (TARGET_MODES_TIEABLE_P): Redefine.
10041 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10042 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10043 (ia64_modes_tieable_p): New function.
10044 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10045 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10046 (iq2000_modes_tieable_p): New function.
10047 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10048 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10049 (lm32_modes_tieable_p): New function.
10050 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10051 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10052 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10053 (TARGET_MODES_TIEABLE_P): Redefine.
10054 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10055 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10056 (m32r_modes_tieable_p): New function.
10057 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10058 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10059 (m68k_modes_tieable_p): New function.
10060 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10061 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10062 (mcore_modes_tieable_p): New function.
10063 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10064 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10066 (TARGET_MODES_TIEABLE_P): Redefine.
10067 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10068 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10069 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10070 (TARGET_MODES_TIEABLE_P): Redefine.
10071 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10072 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10073 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10074 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10075 (mn10300_modes_tieable_p): ...this and make static.
10076 (TARGET_MODES_TIEABLE_P): Redefine.
10077 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10078 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10079 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10080 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10081 (msp430_modes_tieable_p): Make static.
10082 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10083 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10084 (TARGET_MODES_TIEABLE_P): Redefine.
10085 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10086 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10087 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10088 (TARGET_MODES_TIEABLE_P): Redefine.
10089 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10090 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10091 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10092 (TARGET_MODES_TIEABLE_P): Redefine.
10093 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10094 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10095 (pdp11_modes_tieable_p): New function.
10096 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10097 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10098 (rs6000_modes_tieable_p): New function.
10099 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10100 * config/powerpcspe/powerpcspe.md: Update comment.
10101 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10102 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10103 (TARGET_MODES_TIEABLE_P): Redefine.
10104 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10105 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10106 (rl78_modes_tieable_p): New function.
10107 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10108 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10109 (rs6000_modes_tieable_p): New function.
10110 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10111 * config/rs6000/rs6000.md: Update comment.
10112 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10113 * config/rx/rx.c (rx_modes_tieable_p): New function.
10114 (TARGET_MODES_TIEABLE_P): Redefine.
10115 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10116 * config/s390/s390.c (s390_modes_tieable_p): New function.
10117 (TARGET_MODES_TIEABLE_P): Redefine.
10118 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10119 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10120 (sh_modes_tieable_p): New function.
10121 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10122 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10123 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10124 (sparc_modes_tieable_p): Make static.
10125 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10126 * config/spu/spu.c (spu_modes_tieable_p): New function.
10127 (TARGET_MODES_TIEABLE_P): Redefine.
10128 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10129 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10130 (TARGET_MODES_TIEABLE_P): Redefine.
10131 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10132 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10133 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10134 * config/v850/v850.c (v850_modes_tieable_p): New function.
10135 (TARGET_MODES_TIEABLE_P): Redefine.
10136 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10137 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10138 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10139 (visium_modes_tieable_p): New function.
10140 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10141 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10142 (xtensa_modes_tieable_p): New function.
10143 * system.h (MODES_TIEABLE_P): Poison.
10145 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10146 Alan Hayward <alan.hayward@arm.com>
10147 David Sherwood <david.sherwood@arm.com>
10149 * target.def (hard_regno_mode_ok): New hook.
10150 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10151 (TARGET_HARD_REGNO_MODE_OK): ...this.
10152 * doc/tm.texi.in: Regenerate.
10153 * hooks.h (hook_bool_uint_mode_true): Declare.
10154 * hooks.c (hook_bool_uint_mode_true): New function.
10155 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10156 HARD_REGNO_MODE_OK.
10157 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10158 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10159 instead of HARD_REGNO_MODE_OK.
10160 * caller-save.c: Include target.h.
10161 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10162 HARD_REGNO_MODE_OK.
10163 * combine.c (can_combine_p): Likewise.
10164 (combinable_i3pat): Likewise.
10165 (can_change_dest_mode): Likewise.
10166 * expr.c (init_expr_target): Likewise.
10167 (convert_move): Likewise.
10168 (convert_modes): Likewise.
10169 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10170 (setup_prohibited_mode_move_regs): Likewise.
10171 * ira.h (target_ira): Likewise.
10172 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10173 * lra-constraints.c (process_alt_operands): Likewise.
10174 (split_reg): Likewise.
10175 * recog.c (peep2_find_free_register): Likewise.
10176 * ree.c (combine_reaching_defs): Likewise.
10177 * regcprop.c (maybe_mode_change): Likewise.
10178 * reginfo.c (init_reg_sets_1): Likewise.
10179 (choose_hard_reg_mode): Likewise.
10180 (simplifiable_subregs): Likewise.
10181 * regrename.c (check_new_reg_p): Likewise.
10182 * reload.c (find_valid_class): Likewise.
10183 (find_valid_class_1): Likewise.
10184 (reload_inner_reg_of_subreg): Likewise.
10185 (push_reload): Likewise.
10186 (combine_reloads): Likewise.
10187 (find_dummy_reload): Likewise.
10188 (find_reloads): Likewise.
10189 * reload1.c (find_reg): Likewise.
10190 (set_reload_reg): Likewise.
10191 (allocate_reload_reg): Likewise.
10192 (choose_reload_regs): Likewise.
10193 (reload_adjust_reg_for_temp): Likewise.
10194 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10195 (simplify_subreg_regno): Likewise.
10196 * sel-sched.c (init_regs_for_mode): Likewise.
10197 * varasm.c (make_decl_rtl): Likewise.
10198 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10199 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10200 HARD_REGNO_MODE_OK.
10201 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10202 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10203 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10204 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10205 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10206 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10207 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10208 (arc_mode_class): Delete.
10209 (HARD_REGNO_MODE_OK): Delete.
10210 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10211 (arc_hard_regno_mode_ok): Rename old array to...
10212 (arc_hard_regno_mode_ok_modes): ...this.
10213 (arc_conditional_register_usage): Update accordingly.
10214 (arc_mode_class): Make static.
10215 (arc_hard_regno_mode_ok): New function.
10216 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10217 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10218 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10219 (arm_hard_regno_mode_ok): Make static.
10220 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10221 HARD_REGNO_MODE_OK.
10222 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10223 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10224 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10226 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10227 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10228 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10229 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10230 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10231 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10232 * config/bfin/predicates.md (valid_reg_operand): Use
10233 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10234 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10235 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10236 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10237 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10238 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10239 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10240 (cr16_hard_regno_mode_ok): Make static and return a bool.
10241 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10242 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10243 (cris_hard_regno_mode_ok): New function.
10244 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10245 (epiphany_mode_class): Delete.
10246 (HARD_REGNO_MODE_OK): Delete.
10247 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10248 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10249 (hard_regno_mode_ok): Rename to...
10250 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10251 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10252 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10253 HARD_REGNO_MODE_OK.
10254 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10255 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10256 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10257 (frv_hard_regno_mode_ok): Make static and return a bool.
10258 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10259 HARD_REGNO_MODE_OK.
10260 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10261 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10262 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10263 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10266 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10267 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10268 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10271 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10272 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10273 (ia64_hard_regno_mode_ok): New function.
10274 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10275 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10276 (iq2000_hard_regno_mode_ok): New function.
10277 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10278 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10279 (lm32_hard_regno_mode_ok): New function.
10280 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10281 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10282 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10283 instead of HARD_REGNO_MODE_OK.
10284 (m32c_hard_regno_ok): Rename to...
10285 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10286 (m32c_cannot_change_mode_class): Update accordingly.
10287 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10288 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10289 (m32r_mode_class): Delete.
10290 (HARD_REGNO_MODE_OK): Delete.
10291 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10292 (m32r_hard_regno_mode_ok): Rename to...
10293 (m32r_hard_regno_modes): ...this.
10294 (m32r_mode_class): Make static.
10295 (m32r_hard_regno_mode_ok): New function.
10296 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10297 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10298 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10299 (m68k_hard_regno_mode_ok): Make static.
10300 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10301 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10302 (mcore_hard_regno_mode_ok): New function.
10303 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10304 (HARD_REGNO_MODE_OK): Delete.
10305 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10307 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10308 (microblaze_hard_regno_mode_ok): New function.
10309 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10310 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10311 (mips_hard_regno_mode_ok): Delete.
10312 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10313 (mips_hard_regno_mode_ok_p): ...this and make static.
10314 (mips_hard_regno_mode_ok_p): Rename to...
10315 (mips_hard_regno_mode_ok_uncached): ...this.
10316 (mips_hard_regno_mode_ok): New function.
10317 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10318 of HARD_REGNO_MODE_OK.
10319 (mips_option_override): Update after above name changes.
10320 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10321 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10322 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10323 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10324 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10325 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10326 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10327 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10328 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10329 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10330 (msp430_hard_regno_mode_ok): Make static and return a bool.
10331 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10332 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10333 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10336 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10337 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10338 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10339 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10340 (PA_HARD_REGNO_MODE_OK): ...this
10341 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10342 (PA_HARD_REGNO_MODE_OK): ...this.
10343 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10344 (pa_hard_regno_mode_ok): New function.
10345 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10346 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10347 (pdp11_hard_regno_mode_ok): New function.
10348 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10349 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10351 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10353 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10354 (rs6000_hard_regno_mode_ok): Rename to...
10355 (rs6000_hard_regno_mode_ok_uncached): ...this.
10356 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10357 (rs6000_hard_regno_mode_ok): New function.
10358 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10359 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10360 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10361 (riscv_hard_regno_mode_ok): ...this and make static.
10362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10363 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10364 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10365 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10366 (rl78_hard_regno_mode_ok): Make static and return bool.
10367 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10368 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10370 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10371 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10372 (rs6000_hard_regno_mode_ok): Rename to...
10373 (rs6000_hard_regno_mode_ok_uncached): ...this.
10374 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10375 (rs6000_hard_regno_mode_ok): New function.
10376 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10377 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10378 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10379 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10380 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10381 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10382 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10383 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10384 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10385 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10386 (sh_hard_regno_mode_ok): Make static.
10387 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10388 instead of HARD_REGNO_MODE_OK.
10389 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10390 (sparc_mode_class): Delete.
10391 (HARD_REGNO_MODE_OK): Delete.
10392 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10393 (hard_regno_mode_classes): Make static.
10394 (sparc_mode_class): Likewise.
10395 (sparc_hard_regno_mode_ok): New function.
10396 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10397 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10398 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10401 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10402 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10403 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10404 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10406 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10407 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10408 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10409 (visium_hard_regno_mode_ok): New function.
10410 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10411 instead of HARD_REGNO_MODE_OK.
10412 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10413 (HARD_REGNO_MODE_OK): Delete.
10414 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10415 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10416 (xtensa_option_override): Update accordingly.
10417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10418 (xtensa_hard_regno_mode_ok): New function.
10419 * system.h (HARD_REGNO_MODE_OK): Poison.
10421 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10422 Alan Hayward <alan.hayward@arm.com>
10423 David Sherwood <david.sherwood@arm.com>
10425 * target.def (hard_regno_call_part_clobbered): New hook.
10426 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10427 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10428 * doc/tm.texi: Regenerate.
10429 * hooks.h (hook_bool_uint_mode_false): Declare.
10430 * hooks.c (hook_bool_uint_mode_false): New function.
10431 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10432 * cselib.c (cselib_process_insn): Use
10433 targetm.hard_regno_call_part_clobbered instead of
10434 HARD_REGNO_CALL_PART_CLOBBERED.
10435 * ira-conflicts.c (ira_build_conflicts): Likewise.
10436 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10437 * lra-constraints.c (need_for_call_save_p): Likewise.
10438 * lra-lives.c: Include target.h.
10439 (check_pseudos_live_through_calls): Use
10440 targetm.hard_regno_call_part_clobbered instead of
10441 HARD_REGNO_CALL_PART_CLOBBERED.
10442 * regcprop.c: Include target.h.
10443 (copyprop_hardreg_forward_1): Use
10444 targetm.hard_regno_call_part_clobbered instead of
10445 HARD_REGNO_CALL_PART_CLOBBERED.
10446 * reginfo.c (choose_hard_reg_mode): Likewise.
10447 * regrename.c (check_new_reg_p): Likewise.
10448 * reload.c (find_equiv_reg): Likewise.
10449 * reload1.c (emit_reload_insns): Likewise.
10450 * sched-deps.c (deps_analyze_insn): Likewise.
10451 * sel-sched.c (init_regs_for_mode): Likewise.
10452 (mark_unavailable_hard_regs): Likewise.
10453 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10454 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10455 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10457 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10458 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10459 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10461 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10463 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10464 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10465 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10467 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10468 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10469 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10471 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10472 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10474 * config/powerpcspe/powerpcspe.c
10475 (rs6000_hard_regno_call_part_clobbered): New function.
10476 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10477 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10478 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10480 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10481 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10482 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10484 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10485 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10486 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10488 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10489 Alan Hayward <alan.hayward@arm.com>
10490 David Sherwood <david.sherwood@arm.com>
10492 * rtl.h (subreg_memory_offset): Declare.
10493 * emit-rtl.c (subreg_memory_offset): New function.
10494 * expmed.c (store_bit_field_1): Use it.
10495 * expr.c (undefined_operand_subword_p): Likewise.
10496 * simplify-rtx.c (simplify_subreg): Likewise.
10498 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10500 PR rtl-optimization/57448
10503 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10504 using atomic_load pattern.
10505 (expand_atomic_store): Likewise.
10507 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10510 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10511 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10512 handling. Use replace_call_with_value with NULL instead of
10513 gsi_replace, unlink_stmt_vdef and release_defs.
10515 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10518 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10520 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10523 * rtl.h (emit_library_call_value_1): Declare.
10524 (emit_library_call): Replace declaration with a series of overloads.
10525 Remove the parameter count argument.
10526 (emit_library_call_value): Likewise.
10527 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10528 with an "rtx_mode_t *".
10529 (emit_library_call_value): Delete.
10530 (emit_library_call): Likewise.
10531 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10532 (asan_emit_allocas_unpoison): Likewise.
10533 * builtins.c (expand_builtin_powi): Likewise.
10534 (expand_asan_emit_allocas_unpoison): Likewise.
10535 * cfgexpand.c (expand_main_function): Likewise.
10536 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10537 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10538 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10539 * config/arm/arm.c (arm_trampoline_init): Likewise.
10540 (arm_call_tls_get_addr): Likewise.
10541 (arm_expand_divmod_libfunc): Likewise.
10542 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10543 (smulsi3_highpart): Likewise.
10544 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10545 (c6x_expand_compare): Likewise.
10546 (c6x_expand_movmem): Likewise.
10547 * config/frv/frv.c (frv_trampoline_init): Likewise.
10548 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10549 (ix86_expand_divmod_libfunc): Likewise.
10550 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10551 (ia64_expand_compare): Likewise.
10552 (ia64_profile_hook): Likewise.
10553 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10554 (nonlocal_goto): Likewise.
10555 (restore_stack_nonlocal): Likewise.
10556 * config/m32r/m32r.c (block_move_call): Likewise.
10557 (m32r_trampoline_init): Likewise.
10558 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10559 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10560 (m68k_call_m68k_read_tp): Likewise.
10561 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10562 (microblaze_expand_divide): Likewise.
10563 * config/mips/mips.h (mips_args): Likewise.
10564 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10565 (MIPS_ICACHE_SYNC): Likewise.
10566 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10567 (nios2_trampoline_init): Likewise.
10568 * config/pa/pa.c (hppa_tls_call): Likewise.
10569 (pa_trampoline_init): Likewise.
10570 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10571 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10572 (expand_strn_compare): Likewise.
10573 (rs6000_generate_compare): Likewise.
10574 (rs6000_expand_float128_convert): Likewise.
10575 (output_profile_hook): Likewise.
10576 (rs6000_trampoline_init): Likewise.
10577 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10578 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10579 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10580 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10581 (rs6000_generate_compare): Likewise.
10582 (rs6000_expand_float128_convert): Likewise.
10583 (output_profile_hook): Likewise.
10584 (rs6000_trampoline_init): Likewise.
10585 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10586 * config/sh/sh.c (sh_trampoline_init): Likewise.
10587 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10588 (sparc_emit_float_lib_cmp): Likewise.
10589 (sparc32_initialize_trampoline): Likewise.
10590 (sparc64_initialize_trampoline): Likewise.
10591 (sparc_profile_hook): Likewise.
10592 * config/spu/spu.c (ea_load_store): Likewise.
10593 * config/spu/spu.md (floatunssidf2): Likewise.
10594 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10595 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10596 * config/visium/visium.c (expand_block_move_4): Likewise.
10597 (expand_block_move_2): Likewise.
10598 (expand_block_move_1): Likewise.
10599 (expand_block_set_4): Likewise.
10600 (expand_block_set_2): Likewise.
10601 (expand_block_set_1): Likewise.
10602 (visium_trampoline_init): Likewise.
10603 (visium_profile_hook): Likewise.
10604 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10605 (xtensa_setup_frame_addresses): Likewise.
10606 (xtensa_trampoline_init): Likewise.
10607 * except.c (sjlj_emit_function_enter): Likewise.
10608 (sjlj_emit_function_exit): Likewise.
10609 * explow.c (allocate_dynamic_stack_space): Likewise.
10610 (probe_stack_range): Likewise.
10611 * expr.c (convert_mode_scalar): Likewise.
10612 * optabs.c (expand_binop): Likewise.
10613 (expand_twoval_binop_libfunc): Likewise.
10614 (expand_unop): Likewise.
10615 (prepare_cmp_insn): Likewise.
10616 (prepare_float_lib_cmp): Likewise.
10617 (expand_float): Likewise.
10618 (expand_fix): Likewise.
10619 (expand_fixed_convert): Likewise.
10620 (maybe_emit_sync_lock_test_and_set): Likewise.
10621 (expand_atomic_compare_and_swap): Likewise.
10622 (expand_mem_thread_fence): Likewise.
10623 (expand_atomic_fetch_op): Likewise.
10625 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10627 * doc/generic.texi (OpenACC): Adjust URL.
10628 * doc/invoke.texi (C Dialect Options): Ditto.
10630 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10632 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10633 predicate for operand 1. Add (m,<S>) constraint.
10634 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10635 Prevent memory operand 1 with register operand 2.
10637 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10639 PR rtl-optimization/82024
10640 * combine.c (try_combine): If the combination result is a PARALLEL,
10641 and we only need to retain the SET in there that would be placed
10642 at I2, check that we can place that at I3 instead, before doing so.
10644 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10647 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10649 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10650 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10653 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10654 Jeff Law <law@redhat.com>
10656 * varasm.c (bss_initializer_p): Do not put constants into .bss
10657 (categorize_decl_for_section): Handle bss_initializer_p returning
10658 false when DECL_INITIAL is NULL.
10660 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10663 * config/s390/s390.c (s390_can_inline_p): New function.
10665 2017-09-01 Jeff Law <law@redhat.com>
10667 PR tree-optimization/82052
10668 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10669 Always initialize the returned slot after a hash table miss
10670 when INSERT is true.
10672 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10674 * config/s390/s390.md (mem_signal_fence): Remove.
10675 * doc/md.texi (mem_signal_fence): Remove.
10676 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10678 * target-insns.def (mem_signal_fence): Remove.
10680 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10683 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10686 * asan.c (create_odr_indicator): Strip name encoding from assembler
10687 name before appending it after __odr_asan_.
10689 2017-09-01 Martin Liska <mliska@suse.cz>
10691 PR tree-optimization/82059
10692 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10693 frequency only when an edge is redirected.
10695 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10697 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10698 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10699 (arc_conditional_register_usage): Remove ARC600 lp_count
10701 (arc_file_start): Emit Tag_ARC_CPU_variation.
10702 (arc_can_use_doloop_p): New conditions to use ZOLs.
10703 (hwloop_fail): New function.
10704 (hwloop_optimize): Likewise.
10705 (hwloop_pattern_reg): Likewise.
10706 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10707 (arc_reorg_loops): New function, calls reorg_loops.
10708 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10709 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10711 (arc_loop_hazard): Remove function, functionality moved into
10713 (arc_hazard): Remove arc_loop_hazard call.
10714 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10715 into hwloop_optimize.
10716 (arc_label_align): Remove ZOL handling.
10717 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10718 * config/arc/arc.md (doloop_begin): Remove pattern.
10719 (doloop_begin_i): Likewise.
10720 (doloop_end_i): Likewise.
10721 (doloop_fallback): Likewise.
10722 (doloop_fallback_m): Likewise.
10723 (doloop_end): Reimplement expand.
10724 (arc_lp): New pattern for LP instruction.
10725 (loop_end): New pattern.
10726 (loop_fail): Likewise.
10727 (decrement_and_branch_until_zero): Likewise.
10728 * config/arc/arc.opt (mlpc-width): New option.
10729 * doc/invoke.texi (mlpc-width): Document option.
10731 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10733 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10734 (arc_ccfsm_advance): Fix checking for delay slots.
10735 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10737 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10739 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10741 (movhi_insn): Update store instruction constraint which are saving
10742 6-bit short immediates.
10743 (movsi_insn): Consider also short scaled load operations.
10744 (zero_extendhisi2_i): Use Usd constraint instead of T.
10745 (extendhisi2_i): Add q constraint.
10746 (arc_clzsi2): Add type and length attributes.
10747 (arc_ctzsi2): Likewise.
10748 * config/arc/constraints.md (Usc): Update constraint, the
10749 assembler can parse two relocations for a single instruction.
10751 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10753 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10754 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10756 2017-08-31 Olivier Hainque <hainque@adacore.com>
10758 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10759 match as powerpc-wrs-vxworks*.
10761 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10763 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10764 register constraint for by-element operand.
10765 (aarch64_mls_elt_merge<mode>): Likewise.
10767 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10769 * config/arc/arc.c (arc_can_follow_jump): Check for short
10772 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10774 * config.gcc: Use g.opt for arc.
10775 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10776 functionality moved to ...
10777 (legitimate_scaled_address_p): New function, ...here.
10778 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10779 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10780 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10782 (arc_override_options): Handle G option.
10783 (arc_output_pic_addr_const): Correct function definition.
10784 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10785 (arc_decl_anon_ns_mem_p): Delete.
10786 (arc_in_small_data_p): Overhaul this function to take into
10787 consideration the value given via G option.
10788 (arc_rewrite_small_data_1): Renamed and corrected old
10789 arc_rewrite_small_data function.
10790 (arc_rewrite_small_data): New function.
10791 (small_data_pattern): Don't use pic_offset_table_rtx.
10792 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10793 * config/arc/simdext.md (movmisalignv2hi): Use
10794 prepare_move_operands function.
10796 (movmisalign*): Likewise.
10797 * doc/invoke.texi (ARC options): Document -G option.
10799 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10801 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10803 * config/arc/arc.c (arc_print_operand): Output scalled address for
10804 sdata whenever is possible.
10805 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10806 load/stores are available.
10807 (compact_sda_memory_operand): Check for the alignment required by
10808 code density instructions.
10809 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10811 * config/arc/constraints.md (Usd): Update constraint.
10812 (Us0): New constraint.
10813 (Usc): Update constraint.
10815 2017-08-31 Richard Biener <rguenther@suse.de>
10817 PR middle-end/82054
10818 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10819 function only once.
10821 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10823 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10824 Resize type_signature.
10826 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10827 Alan Hayward <alan.hayward@arm.com>
10828 David Sherwood <david.sherwood@arm.com>
10830 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10831 subregs whose inner modes can be stored in GPRs.
10832 (aarch64_classify_index): Likewise.
10834 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10835 Alan Hayward <alan.hayward@arm.com>
10836 David Sherwood <david.sherwood@arm.com>
10838 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10839 (V_INT_EQUIV): ...this.
10840 (v_cmp_result): Rename to...
10841 (v_int_equiv): ...this.
10842 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10843 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10844 (copysign<mode>3): Likewise.
10845 (aarch64_simd_bsl<mode>_internal): Likewise.
10846 (aarch64_simd_bsl<mode>): Likewise.
10847 (vec_cmp<mode><mode>): Likewise.
10848 (vcond<mode><mode>): Likewise.
10849 (vcond<v_cmp_mixed><mode>): Likewise.
10850 (vcondu<mode><v_cmp_mixed>): Likewise.
10851 (aarch64_cm<optab><mode>): Likewise.
10852 (aarch64_cmtst<mode>): Likewise.
10853 (aarch64_fac<optab><mode>): Likewise.
10854 (vec_perm_const<mode>): Likewise.
10855 (vcond_mask_<mode><v_cmp_result>): Rename to...
10856 (vcond_mask_<mode><v_int_equiv>): ...this.
10857 (vec_cmp<mode><v_cmp_result>): Rename to...
10858 (vec_cmp<mode><v_int_equiv>): ...this.
10860 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10861 Alan Hayward <alan.hayward@arm.com>
10862 David Sherwood <david.sherwood@arm.com>
10864 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10866 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10867 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10868 (UNSPEC_LD4_DREG): New unspecs.
10869 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10870 (aarch64_ld2<mode>_dreg_be): Replace with...
10871 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10873 (aarch64_ld3<mode>_dreg_le)
10874 (aarch64_ld3<mode>_dreg_be): Replace with...
10875 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10877 (aarch64_ld4<mode>_dreg_le)
10878 (aarch64_ld4<mode>_dreg_be): Replace with...
10879 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10882 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10884 PR tree-optimization/81987
10885 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10886 insert an initializer in a location not dominated by the stride
10889 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10891 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10892 on the entire header of the finally block in the fallthru case.
10894 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10896 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10898 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10900 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10901 rs6000_emit_move_from_cr and call renamed function.
10902 (rs6000_emit_prologue): Call renamed functions.
10903 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10904 movesi_from_cr, remove volatile CRs.
10906 2017-08-30 Jon Beniston <jon@beniston.com>
10907 Richard Biener <rguenther@suse.de>
10909 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10910 of VECTOR_MODE_P check.
10911 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10912 element vector types.
10914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10916 * df.h (df_read_modify_subreg_p): Remove in favor of...
10917 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10918 const_rtx instead of an rtx.
10919 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10920 * df-problems.c (df_word_lr_mark_ref): Likewise.
10921 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10922 (mark_pseudo_reg_dead): Likewise.
10923 (mark_ref_dead): Likewise.
10924 * reginfo.c (init_subregs_of_mode): Likewise.
10925 * sched-deps.c (sched_analyze_1): Likewise.
10926 * df-scan.c (df_def_record_1): Likewise.
10927 (df_uses_record): Likewise.
10928 (df_read_modify_subreg_p): Remove in favor of...
10929 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10930 const_rtx instead of an rtx.
10932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10933 Alan Hayward <alan.hayward@arm.com>
10934 David Sherwood <david.sherwood@arm.com>
10936 * rtl.h (partial_subreg_p): New function.
10937 * caller-save.c (save_call_clobbered_regs): Use it.
10938 * calls.c (expand_call): Likewise.
10939 * combine.c (combinable_i3pat): Likewise.
10940 (simplify_set): Likewise.
10941 (make_extraction): Likewise.
10942 (make_compound_operation_int): Likewise.
10943 (gen_lowpart_or_truncate): Likewise.
10944 (force_to_mode): Likewise.
10945 (make_field_assignment): Likewise.
10946 (reg_truncated_to_mode): Likewise.
10947 (record_truncated_value): Likewise.
10948 (move_deaths): Likewise.
10949 * cse.c (record_jump_cond): Likewise.
10950 (cse_insn): Likewise.
10951 * cselib.c (cselib_lookup_1): Likewise.
10952 * expmed.c (extract_bit_field_using_extv): Likewise.
10953 * function.c (assign_parm_setup_reg): Likewise.
10954 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10955 * ira-build.c (create_insn_allocnos): Likewise.
10956 * lra-coalesce.c (merge_pseudos): Likewise.
10957 * lra-constraints.c (match_reload): Likewise.
10958 (simplify_operand_subreg): Likewise.
10959 (curr_insn_transform): Likewise.
10960 * lra-lives.c (process_bb_lives): Likewise.
10961 * lra.c (new_insn_reg): Likewise.
10962 (lra_substitute_pseudo): Likewise.
10963 * regcprop.c (mode_change_ok): Likewise.
10964 (maybe_mode_change): Likewise.
10965 (copyprop_hardreg_forward_1): Likewise.
10966 * reload.c (push_reload): Likewise.
10967 (find_reloads): Likewise.
10968 (find_reloads_subreg_address): Likewise.
10969 * reload1.c (alter_reg): Likewise.
10970 (eliminate_regs_1): Likewise.
10971 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10973 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10975 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10976 back to if statements, including unpack.
10978 2017-08-30 Martin Liska <mliska@suse.cz>
10980 PR inline-asm/82001
10981 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10983 (func_checker::compare_asm_inputs_outputs): ... this function.
10984 (func_checker::compare_gimple_asm): Use the function to compare
10985 also ASM constrains.
10986 * ipa-icf-gimple.h: Rename the function.
10988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10989 Alan Hayward <alan.hayward@arm.com>
10990 David Sherwood <david.sherwood@arm.com>
10992 * coretypes.h (complex_mode): New type.
10993 * gdbhooks.py (build_pretty_printer): Handle it.
10994 * machmode.h (complex_mode): New class.
10995 (complex_mode::includes_p): New function.
10996 (is_complex_int_mode): Likewise.
10997 (is_complex_float_mode): Likewise.
10998 * genmodes.c (get_mode_class): Handle complex mode classes.
10999 * function.c (expand_function_end): Use is_complex_int_mode.
11001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11002 Alan Hayward <alan.hayward@arm.com>
11003 David Sherwood <david.sherwood@arm.com>
11005 * coretypes.h (scalar_mode_pod): New typedef.
11006 * gdbhooks.py (build_pretty_printer): Handle it.
11007 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11008 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11009 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11010 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11011 as_a <scalar_mode>.
11013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11014 Alan Hayward <alan.hayward@arm.com>
11015 David Sherwood <david.sherwood@arm.com>
11017 * machmode.h (mode_for_vector): Take a scalar_mode instead
11019 * stor-layout.c (mode_for_vector): Likewise.
11020 * explow.c (promote_mode): Use as_a <scalar_mode>.
11021 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11024 Alan Hayward <alan.hayward@arm.com>
11025 David Sherwood <david.sherwood@arm.com>
11027 * target.def (preferred_simd_mode): Take a scalar_mode
11028 instead of a machine_mode.
11029 * targhooks.h (default_preferred_simd_mode): Likewise.
11030 * targhooks.c (default_preferred_simd_mode): Likewise.
11031 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11032 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11033 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11034 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11035 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11036 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11037 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11038 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11040 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11041 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11042 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11043 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11044 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11045 * doc/tm.texi: Regenerate.
11046 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11050 Alan Hayward <alan.hayward@arm.com>
11051 David Sherwood <david.sherwood@arm.com>
11053 * target.def (scalar_mode_supported_p): Take a scalar_mode
11054 instead of a machine_mode.
11055 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11056 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11057 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11058 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11059 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11060 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11061 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11062 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11063 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11064 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11065 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11066 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11067 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11068 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11070 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11071 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11072 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11073 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11074 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11076 * doc/tm.texi: Regenerate.
11078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11079 Alan Hayward <alan.hayward@arm.com>
11080 David Sherwood <david.sherwood@arm.com>
11082 * coretypes.h (opt_scalar_mode): New typedef.
11083 * gdbhooks.py (build_pretty_printers): Handle it.
11084 * machmode.h (mode_iterator::get_2xwider): Add overload for
11086 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11088 * expr.c (convert_mode_scalar): Likewise.
11089 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11090 * optabs.c (expand_float): Likewise.
11091 (expand_fix): Likewise.
11092 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11095 Alan Hayward <alan.hayward@arm.com>
11096 David Sherwood <david.sherwood@arm.com>
11098 * optabs.c (expand_float): Explicitly check for scalars before
11099 using a branching expansion.
11100 (expand_fix): Likewise.
11102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11103 Alan Hayward <alan.hayward@arm.com>
11104 David Sherwood <david.sherwood@arm.com>
11106 * expr.c (convert_mode): Split scalar handling out into...
11107 (convert_mode_scalar): ...this new function. Treat the modes
11110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11111 Alan Hayward <alan.hayward@arm.com>
11112 David Sherwood <david.sherwood@arm.com>
11114 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11116 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11123 rather than a machine_mode.
11124 (fixed_from_string): Likewise.
11125 (fixed_convert): Likewise.
11126 (fixed_convert_from_int): Likewise.
11127 (fixed_convert_from_real): Likewise.
11128 (real_convert_from_fixed): Likewise.
11129 * fixed-value.c (fixed_from_double_int): Likewise.
11130 (fixed_from_string): Likewise.
11131 (fixed_convert): Likewise.
11132 (fixed_convert_from_int): Likewise.
11133 (fixed_convert_from_real): Likewise.
11134 (real_convert_from_fixed): Likewise.
11135 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11138 Alan Hayward <alan.hayward@arm.com>
11139 David Sherwood <david.sherwood@arm.com>
11141 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11142 of separate mode class checks. Do not allow vector modes here.
11143 (immed_wide_int_const): Use as_a <scalar_mode>.
11144 * explow.c (trunc_int_for_mode): Likewise.
11145 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11146 (wi::shwi): Likewise.
11147 (wi::min_value): Likewise.
11148 (wi::max_value): Likewise.
11149 * dwarf2out.c (loc_descriptor): Likewise.
11150 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11151 for CONST_WIDE_INT.
11153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11154 Alan Hayward <alan.hayward@arm.com>
11155 David Sherwood <david.sherwood@arm.com>
11157 * tree.h (SCALAR_TYPE_MODE): New macro.
11158 * expr.c (expand_expr_addr_expr_1): Use it.
11159 (expand_expr_real_2): Likewise.
11160 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11161 (fold_convert_const_fixed_from_int): Likewise.
11162 (fold_convert_const_fixed_from_real): Likewise.
11163 (native_encode_fixed): Likewise
11164 (native_encode_complex): Likewise
11165 (native_encode_vector): Likewise.
11166 (native_interpret_fixed): Likewise.
11167 (native_interpret_real): Likewise.
11168 (native_interpret_complex): Likewise.
11169 (native_interpret_vector): Likewise.
11170 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11171 (simd_clone_adjust_argument_types): Likewise.
11172 (simd_clone_init_simd_arrays): Likewise.
11173 (simd_clone_adjust): Likewise.
11174 * stor-layout.c (layout_type): Likewise.
11175 * tree.c (build_minus_one_cst): Likewise.
11176 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11177 * tree-inline.c (estimate_move_cost): Likewise.
11178 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11179 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11180 (vectorizable_reduction): Likewise.
11181 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11182 (vect_recog_mixed_size_cond_pattern): Likewise.
11183 (check_bool_pattern): Likewise.
11184 (adjust_bool_pattern): Likewise.
11185 (search_type_for_mask_1): Likewise.
11186 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11187 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11188 (vectorizable_load): Likewise.
11189 (vectorizable_store): Likewise.
11190 * ubsan.c (ubsan_encode_value): Likewise.
11191 * varasm.c (output_constant): Likewise.
11193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11194 Alan Hayward <alan.hayward@arm.com>
11195 David Sherwood <david.sherwood@arm.com>
11197 * coretypes.h (scalar_mode): New class.
11198 * machmode.h (scalar_mode): Likewise.
11199 (scalar_mode::includes_p): New function.
11200 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11201 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11202 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11203 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11204 * expmed.c (store_bit_field_1): Likewise.
11205 (extract_bit_field_1): Likewise.
11206 * expr.c (write_complex_part): Likewise.
11207 (read_complex_part): Likewise.
11208 (emit_move_complex_push): Likewise.
11209 (expand_expr_real_2): Likewise.
11210 * function.c (assign_parm_setup_reg): Likewise.
11211 (assign_parms_unsplit_complex): Likewise.
11212 * optabs.c (expand_binop): Likewise.
11213 * rtlanal.c (subreg_get_info): Likewise.
11214 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11215 * varasm.c (output_constant_pool_2): Likewise.
11217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11218 Alan Hayward <alan.hayward@arm.com>
11219 David Sherwood <david.sherwood@arm.com>
11221 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11223 (expmed_mult_highpart_optab): Likewise.
11224 (expmed_mult_highpart): Likewise.
11226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11227 Alan Hayward <alan.hayward@arm.com>
11228 David Sherwood <david.sherwood@arm.com>
11230 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11231 instead of a machine_mode.
11232 (builtin_memset_read_str): Likewise.
11233 * builtins.c (c_readstr): Likewise.
11234 (builtin_memcpy_read_str): Likewise.
11235 (builtin_strncpy_read_str): Likewise.
11236 (builtin_memset_read_str): Likewise.
11237 (builtin_memset_gen_str): Likewise.
11238 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11239 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11240 instead of a machine_mode.
11241 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11243 (make_extraction): Likewise.
11244 (try_widen_shift_mode): Take and return scalar_int_modes instead
11246 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11247 a scalar_int_mode instead of a machine_mode.
11248 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11249 (avr_addr_space_pointer_mode): Likewise.
11250 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11251 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11252 (msp430_unwind_word_mode): Likewise.
11253 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11254 (spu_addr_space_pointer_mode): Likewise.
11255 (spu_addr_space_address_mode): Likewise.
11256 (spu_libgcc_cmp_return_mode): Likewise.
11257 (spu_libgcc_shift_count_mode): Likewise.
11258 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11259 (rl78_addr_space_pointer_mode): Likewise.
11260 (fl78_unwind_word_mode): Likewise.
11261 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11263 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11264 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11265 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11266 (mips_valid_pointer_mode): Likewise.
11267 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11268 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11269 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11271 (ft32_addr_space_address_mode): Likewise.
11272 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11273 scalar_int_mode instead of a machine_mode.
11274 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11276 (m32c_addr_space_address_mode): Likewise.
11277 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11278 (rs6000_eh_return_filter_mode): Likewise.
11279 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11280 (rs6000_eh_return_filter_mode): Likewise.
11281 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11282 (s390_libgcc_shift_count_mode): Likewise.
11283 (s390_unwind_word_mode): Likewise.
11284 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11286 * target.def (mode_rep_extended): Likewise.
11287 (valid_pointer_mode): Likewise.
11288 (addr_space.valid_pointer_mode): Likewise.
11289 (eh_return_filter_mode): Return a scalar_int_mode rather than
11291 (libgcc_cmp_return_mode): Likewise.
11292 (libgcc_shift_count_mode): Likewise.
11293 (unwind_word_mode): Likewise.
11294 (addr_space.pointer_mode): Likewise.
11295 (addr_space.address_mode): Likewise.
11296 * doc/tm.texi: Regenerate.
11297 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11299 (do_jump): Use scalar_int_mode for local variables.
11300 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11301 rather than a machine_mode.
11302 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11303 (scompare_loc_descriptor_wide): Likewise.
11304 (scompare_loc_descriptor_narrow): Likewise.
11305 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11307 * except.c (sjlj_emit_dispatch_table): Likewise.
11308 (expand_builtin_eh_copy_values): Likewise.
11309 * explow.c (convert_memory_address_addr_space_1): Likewise.
11310 Take a scalar_int_mode rather than a machine_mode.
11311 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11312 than a machine_mode.
11313 (memory_address_addr_space): Use scalar_int_mode for local variables.
11314 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11315 rather than a machine_mode.
11316 * expmed.c (mask_rtx): Likewise.
11317 (init_expmed_one_conv): Likewise.
11318 (expand_mult_highpart_adjust): Likewise.
11319 (extract_high_half): Likewise.
11320 (expmed_mult_highpart_optab): Likewise.
11321 (expmed_mult_highpart): Likewise.
11322 (expand_smod_pow2): Likewise.
11323 (expand_sdiv_pow2): Likewise.
11324 (emit_store_flag_int): Likewise.
11325 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11327 (extract_low_bits): Likewise.
11328 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11330 * expr.c (pieces_addr::adjust): Likewise.
11331 (can_store_by_pieces): Likewise.
11332 (store_by_pieces): Likewise.
11333 (clear_by_pieces_1): Likewise.
11334 (expand_expr_addr_expr_1): Likewise.
11335 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11336 (expand_expr_real_1): Likewise.
11337 (try_casesi): Likewise.
11338 * final.c (shorten_branches): Likewise.
11339 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11340 type of "mode" to machine_mode.
11341 * internal-fn.c (expand_arith_overflow_result_store): Take a
11342 scalar_int_mode rather than a machine_mode.
11343 (expand_mul_overflow): Use scalar_int_mode for local variables.
11344 * loop-doloop.c (doloop_modify): Likewise.
11345 (doloop_optimize): Likewise.
11346 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11347 than a machine_mode.
11348 (expand_doubleword_shift_condmove): Likewise.
11349 (expand_doubleword_shift): Likewise.
11350 (expand_doubleword_clz): Likewise.
11351 (expand_doubleword_popcount): Likewise.
11352 (expand_doubleword_parity): Likewise.
11353 (expand_absneg_bit): Use scalar_int_mode for local variables.
11354 (prepare_float_lib_cmp): Likewise.
11355 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11356 rather than a machine_mode.
11357 (convert_memory_address_addr_space): Likewise.
11358 (get_mode_bounds): Likewise.
11359 (get_address_mode): Return a scalar_int_mode rather than a
11361 * rtlanal.c (get_address_mode): Likewise.
11362 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11363 than a machine_mode.
11364 * targhooks.c (default_mode_rep_extended): Likewise.
11365 (default_valid_pointer_mode): Likewise.
11366 (default_addr_space_valid_pointer_mode): Likewise.
11367 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11368 than a machine_mode.
11369 (default_libgcc_cmp_return_mode): Likewise.
11370 (default_libgcc_shift_count_mode): Likewise.
11371 (default_unwind_word_mode): Likewise.
11372 (default_addr_space_pointer_mode): Likewise.
11373 (default_addr_space_address_mode): Likewise.
11374 * targhooks.h (default_eh_return_filter_mode): Likewise.
11375 (default_libgcc_cmp_return_mode): Likewise.
11376 (default_libgcc_shift_count_mode): Likewise.
11377 (default_unwind_word_mode): Likewise.
11378 (default_addr_space_pointer_mode): Likewise.
11379 (default_addr_space_address_mode): Likewise.
11380 (default_mode_rep_extended): Take a scalar_int_mode rather than
11382 (default_valid_pointer_mode): Likewise.
11383 (default_addr_space_valid_pointer_mode): Likewise.
11384 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11386 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11387 rather than a machine_mode.
11388 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11389 for local variables.
11390 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11391 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11392 than a machine_mode.
11394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11395 Alan Hayward <alan.hayward@arm.com>
11396 David Sherwood <david.sherwood@arm.com>
11398 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11399 the mode argument to scalar_int_mode.
11400 (do_jump_by_parts_zero_rtx): Likewise.
11401 (do_jump_by_parts_equality_rtx): Likewise.
11402 (do_jump_by_parts_greater): Take a mode argument.
11403 (do_jump_by_parts_equality): Likewise.
11404 (do_jump_1): Update calls accordingly.
11406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11407 Alan Hayward <alan.hayward@arm.com>
11408 David Sherwood <david.sherwood@arm.com>
11410 * is-a.h (safe_dyn_cast): New function.
11411 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11412 (jump_table_for_label): Likewise.
11413 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11414 instead of an rtx_insn *.
11415 (shorten_branches): Use dyn_cast instead of LABEL_P and
11416 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11417 rtx_jump_table_data::get_data_mode.
11418 (final_scan_insn): Likewise.
11420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11421 Alan Hayward <alan.hayward@arm.com>
11422 David Sherwood <david.sherwood@arm.com>
11424 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11425 trying to combine a full-register integer set with a subreg
11428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11429 Alan Hayward <alan.hayward@arm.com>
11430 David Sherwood <david.sherwood@arm.com>
11432 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11433 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11440 when the two are known to be equal.
11442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11443 Alan Hayward <alan.hayward@arm.com>
11444 David Sherwood <david.sherwood@arm.com>
11446 * simplify-rtx.c (simplify_const_unary_operation): Use
11447 is_a <scalar_int_mode> instead of checking for a nonzero
11448 precision. Forcibly convert op_mode to a scalar_int_mode
11449 in that case. More clearly differentiate the operand and
11450 result modes and use the former when deciding what the value
11451 of a count-bits operation should be. Use is_int_mode instead
11452 of checking for a MODE_INT. Remove redundant check for whether
11453 this mode has a zero precision.
11455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11456 Alan Hayward <alan.hayward@arm.com>
11457 David Sherwood <david.sherwood@arm.com>
11459 * optabs.c (widen_leading): Change the type of the mode argument
11460 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11461 (widen_bswap): Likewise.
11462 (expand_parity): Likewise.
11463 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11464 (expand_ffs): Likewise.
11465 (epand_unop): Check for scalar integer modes before calling the
11468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11469 Alan Hayward <alan.hayward@arm.com>
11470 David Sherwood <david.sherwood@arm.com>
11472 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11474 (expand_expr_real_1): Update call accordingly.
11476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11481 value_mode arguments. Use scalar_int_mode internally.
11482 (store_bit_field_1): Rename the new integer mode from imode
11483 to op0_mode and use it instead of GET_MODE (op0). Update calls
11484 to store_split_bit_field, store_bit_field_using_insv and
11485 store_fixed_bit_field.
11486 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11487 Use scalar_int_mode internally. Use a bit count rather than a mode
11488 when calculating the largest bit size for get_best_mode.
11489 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11490 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11491 Remove assertion that OP0 has a scalar integer mode.
11492 (store_split_bit_field): Add op0_mode and value_mode arguments.
11493 Update calls to extract_fixed_bit_field.
11494 (extract_bit_field_using_extv): Add an op0_mode argument.
11495 Use scalar_int_mode internally.
11496 (extract_bit_field_1): Rename the new integer mode from imode to
11497 op0_mode and use it instead of GET_MODE (op0). Update calls to
11498 extract_split_bit_field, extract_bit_field_using_extv and
11499 extract_fixed_bit_field.
11500 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11501 to extract_split_bit_field and extract_fixed_bit_field_1.
11502 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11503 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11504 on the target mode.
11505 (extract_split_bit_field): Add an op0_mode argument. Update call
11506 to extract_fixed_bit_field.
11508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11509 Alan Hayward <alan.hayward@arm.com>
11510 David Sherwood <david.sherwood@arm.com>
11512 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11513 * explow.c (hard_function_value): Likewise.
11514 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11515 convert_to_mode call outside the loop.
11516 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11517 for the mode iterator. Require the mode specified by max_pieces
11519 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11521 (copy_blkmode_to_reg): Likewise.
11522 (set_storage_via_setmem): Likewise.
11523 * optabs.c (prepare_cmp_insn): Likewise.
11524 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11525 * stor-layout.c (finish_bitfield_representative): Likewise.
11527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11528 Alan Hayward <alan.hayward@arm.com>
11529 David Sherwood <david.sherwood@arm.com>
11531 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11532 * expr.c (convert_move): Use them.
11533 (convert_modes): Likewise.
11534 (store_expr_with_bounds): Likewise.
11536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11537 Alan Hayward <alan.hayward@arm.com>
11538 David Sherwood <david.sherwood@arm.com>
11540 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11541 parameter for the mode of "x". Remove the "known_x", "known_mode"
11542 and "known_ret" arguments. Change the type of the mode argument
11543 to scalar_int_mode.
11544 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11545 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11546 (reg_num_sign_bit_copies_for_combine): Likewise.
11547 * rtlanal.c (nonzero_bits1): Likewise.
11548 (num_sign_bit_copies1): Likewise.
11549 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11550 (reg_num_sign_bit_copies_general): Likewise.
11551 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11552 (reg_nonzero_bits_general): Likewise.
11554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11555 Alan Hayward <alan.hayward@arm.com>
11556 David Sherwood <david.sherwood@arm.com>
11558 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11559 than in subroutines. Return 1 for non-integer modes.
11560 (cached_num_sign_bit_copies): Change the type of the mode parameter
11561 to scalar_int_mode.
11562 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11563 classes. Handle CONST_INT_P first and then check whether X also
11564 has a scalar integer mode. Check the same thing for inner registers
11565 of a SUBREG and for values that are being extended or truncated.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11572 in subroutines. Return the mode mask for non-integer modes.
11573 (cached_nonzero_bits): Change the type of the mode parameter
11574 to scalar_int_mode.
11575 (nonzero_bits1): Likewise. Remove early exit for other mode
11576 classes. Handle CONST_INT_P first and then check whether X
11577 also has a scalar integer mode.
11579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11580 Alan Hayward <alan.hayward@arm.com>
11581 David Sherwood <david.sherwood@arm.com>
11583 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11584 Return a scalar_int_mode and assert that the size is greater than
11586 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11587 (op_by_pieces_d::op_by_pieces_d): Likewise.
11588 (op_by_pieces_d::run): Likewise.
11589 (can_store_by_pieces): Likewise.
11591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11592 Alan Hayward <alan.hayward@arm.com>
11593 David Sherwood <david.sherwood@arm.com>
11595 * combine.c (extract_left_shift): Add a mode argument and update
11597 (make_compound_operation_int): Change the type of the mode parameter
11598 to scalar_int_mode and update the call to extract_left_shift.
11600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11601 Alan Hayward <alan.hayward@arm.com>
11602 David Sherwood <david.sherwood@arm.com>
11604 * combine.c (simplify_and_const_int): Change the type of the mode
11605 parameter to scalar_int_mode.
11606 (simplify_and_const_int_1): Likewise. Update recursive call.
11608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11609 Alan Hayward <alan.hayward@arm.com>
11610 David Sherwood <david.sherwood@arm.com>
11612 * combine.c (simplify_compare_const): Check that the mode is a
11613 scalar_int_mode (rather than VOIDmode) before testing its
11615 (simplify_comparison): Move COMPARISON_P handling out of the
11616 loop and restrict the latter part of the loop to scalar_int_modes.
11617 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11618 and when considering SUBREG_REGs. Use is_int_mode instead of
11619 checking GET_MODE_CLASS against MODE_INT.
11621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11622 Alan Hayward <alan.hayward@arm.com>
11623 David Sherwood <david.sherwood@arm.com>
11625 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11626 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11627 shift_unit_mode and for modes involved in scalar shifts.
11629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11630 Alan Hayward <alan.hayward@arm.com>
11631 David Sherwood <david.sherwood@arm.com>
11633 * combine.c (force_int_to_mode): New function, split out from...
11634 (force_to_mode): ...here. Keep xmode up-to-date and use it
11635 instead of GET_MODE (x).
11637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11638 Alan Hayward <alan.hayward@arm.com>
11639 David Sherwood <david.sherwood@arm.com>
11641 * optabs-query.h (extraction_insn::struct_mode): Change type to
11642 opt_scalar_int_mode and update comment.
11643 (extraction_insn::field_mode): Change type to scalar_int_mode.
11644 (extraction_insn::pos_mode): Likewise.
11645 * combine.c (make_extraction): Update accordingly.
11646 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11647 (get_optab_extraction_insn): Likewise.
11648 * recog.c (simplify_while_replacing): Likewise.
11649 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11650 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11657 to a scalar_int_mode instead of a machine_mode.
11658 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11659 (get_best_mode): Return a boolean and use a pointer argument to store
11660 the selected mode. Replace the limit mode parameter with a bit limit.
11661 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11662 for the values returned by bit_field_mode_iterator::next_mode.
11663 (store_bit_field): Update call to get_best_mode.
11664 (store_fixed_bit_field): Likewise.
11665 (extract_fixed_bit_field): Likewise.
11666 * expr.c (optimize_bitfield_assignment_op): Likewise.
11667 * fold-const.c (optimize_bit_field_compare): Likewise.
11668 (fold_truth_andor_1): Likewise.
11669 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11670 Update for new type of m_mode.
11671 (get_best_mode): As above.
11673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11674 Alan Hayward <alan.hayward@arm.com>
11675 David Sherwood <david.sherwood@arm.com>
11677 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11678 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11679 (store_bit_field): Check is_a <scalar_int_mode> before calling
11680 strict_volatile_bitfield_p.
11681 (extract_bit_field): Likewise.
11683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11684 Alan Hayward <alan.hayward@arm.com>
11685 David Sherwood <david.sherwood@arm.com>
11687 * target.def (cstore_mode): Return a scalar_int_mode.
11688 * doc/tm.texi: Regenerate.
11689 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11690 * targhooks.h (default_cstore_mode): Likewise.
11691 * targhooks.c (default_cstore_mode): Likewise, using a forced
11693 * expmed.c (emit_cstore): Expect the target of the cstore to be
11696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11697 Alan Hayward <alan.hayward@arm.com>
11698 David Sherwood <david.sherwood@arm.com>
11700 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11702 (niter_desc): Likewise mode.
11703 (iv_analyze): Add a mode parameter.
11705 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11706 and change its type to scalar_int_mode.
11707 * loop-iv.c: Update commentary at head of file.
11708 (iv_constant): Pass the mode paraeter before the rtx it describes
11709 and change its type to scalar_int_mode. Remove VOIDmode handling.
11710 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11711 (iv_extend): Likewise.
11712 (shorten_into_mode): Likewise.
11713 (iv_add): Use scalar_int_mode.
11714 (iv_mult): Likewise.
11715 (iv_shift): Likewise.
11716 (canonicalize_iv_subregs): Likewise.
11717 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11718 it describes and change its mode to scalar_int_mode. Also change
11719 the type of the returned inner_mode to scalar_int_mode.
11720 (get_biv_step): Likewise, turning outer_mode from a pointer
11721 into a direct parameter. Update call to get_biv_step_1.
11722 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11723 iv_constant and get_biv_step.
11724 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11725 and change its type to scalar_int_mode. Don't initialise iv->mode
11726 to VOIDmode and remove later checks for its still being VOIDmode.
11727 Update calls to iv_analyze_op and iv_analyze_expr. Check
11728 is_a <scalar_int_mode> when changing the mode under consideration.
11729 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11730 Update call to iv_analyze_expr.
11731 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11732 inner register is not also a scalar_int_mode. Update call to
11734 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11735 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11736 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11737 separate mode class checks. Update calls to iv_analyze. Remove
11738 fix-up of VOIDmodes after iv_analyze_biv.
11739 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11740 don't have a scalar_int_mode. Update call to biv_p.
11742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11743 Alan Hayward <alan.hayward@arm.com>
11744 David Sherwood <david.sherwood@arm.com>
11746 * cfgexpand.c (convert_debug_memory_address): Use
11747 as_a <scalar_int_mode>.
11748 * combine.c (expand_compound_operation): Likewise.
11749 (make_extraction): Likewise.
11750 (change_zero_ext): Likewise.
11751 (simplify_comparison): Likewise.
11752 * cse.c (cse_insn): Likewise.
11753 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11754 (mem_loc_descriptor): Likewise.
11755 (loc_descriptor): Likewise.
11756 * expmed.c (init_expmed_one_mode): Likewise.
11757 (synth_mult): Likewise.
11758 (emit_store_flag_1): Likewise.
11759 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11760 of a comparison with size.
11761 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11762 (reduce_to_bit_field_precision): Likewise.
11763 * function.c (expand_function_end): Likewise.
11764 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11765 * loop-doloop.c (doloop_modify): Likewise.
11766 * optabs.c (expand_binop): Likewise.
11767 (expand_unop): Likewise.
11768 (expand_copysign_absneg): Likewise.
11769 (prepare_cmp_insn): Likewise.
11770 (maybe_legitimize_operand): Likewise.
11771 * recog.c (const_scalar_int_operand): Likewise.
11772 * rtlanal.c (get_address_mode): Likewise.
11773 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11774 (simplify_cond_clz_ctz): Likewise.
11775 * tree-nested.c (get_nl_goto_field): Likewise.
11776 * tree.c (build_vector_type_for_mode): Likewise.
11777 * var-tracking.c (use_narrower_mode): Likewise.
11779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11780 Alan Hayward <alan.hayward@arm.com>
11781 David Sherwood <david.sherwood@arm.com>
11783 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11784 * builtins.c (expand_builtin_signbit): Use it.
11785 * cfgexpand.c (expand_debug_expr): Likewise.
11786 * dojump.c (do_jump): Likewise.
11787 (do_compare_and_jump): Likewise.
11788 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11789 * expmed.c (make_tree): Likewise.
11790 * expr.c (expand_expr_real_2): Likewise.
11791 (expand_expr_real_1): Likewise.
11792 (try_casesi): Likewise.
11793 * fold-const-call.c (fold_const_call_ss): Likewise.
11794 * fold-const.c (unextend): Likewise.
11795 (extract_muldiv_1): Likewise.
11796 (fold_single_bit_test): Likewise.
11797 (native_encode_int): Likewise.
11798 (native_encode_string): Likewise.
11799 (native_interpret_int): Likewise.
11800 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11801 * internal-fn.c (expand_addsub_overflow): Likewise.
11802 (expand_neg_overflow): Likewise.
11803 (expand_mul_overflow): Likewise.
11804 (expand_arith_overflow): Likewise.
11805 * match.pd: Likewise.
11806 * stor-layout.c (layout_type): Likewise.
11807 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11808 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11809 * tree-ssanames.c (get_range_info): Likewise.
11810 * tree-switch-conversion.c (array_value_type) Likewise.
11811 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11812 (vect_recog_divmod_pattern): Likewise.
11813 (vect_recog_mixed_size_cond_pattern): Likewise.
11814 * tree-vrp.c (extract_range_basic): Likewise.
11815 (simplify_float_conversion_using_ranges): Likewise.
11816 * tree.c (int_fits_type_p): Likewise.
11817 * ubsan.c (instrument_bool_enum_load): Likewise.
11818 * varasm.c (mergeable_string_section): Likewise.
11819 (narrowing_initializer_constant_valid_p): Likewise.
11820 (output_constant): Likewise.
11822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11823 Alan Hayward <alan.hayward@arm.com>
11824 David Sherwood <david.sherwood@arm.com>
11826 * machmode.h (NARROWEST_INT_MODE): New macro.
11827 * expr.c (alignment_for_piecewise_move): Use it instead of
11828 GET_CLASS_NARROWEST_MODE (MODE_INT).
11829 (push_block): Likewise.
11830 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11832 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11835 Alan Hayward <alan.hayward@arm.com>
11836 David Sherwood <david.sherwood@arm.com>
11838 * postreload.c (move2add_valid_value_p): Change the type of the
11839 mode parameter to scalar_int_mode.
11840 (move2add_use_add2_insn): Add a mode parameter and use it instead
11842 (move2add_use_add3_insn): Likewise.
11843 (reload_cse_move2add): Update accordingly.
11845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11846 Alan Hayward <alan.hayward@arm.com>
11847 David Sherwood <david.sherwood@arm.com>
11849 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11851 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11852 * optabs.c (expand_unop): Likewise.
11854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11855 Alan Hayward <alan.hayward@arm.com>
11856 David Sherwood <david.sherwood@arm.com>
11858 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11859 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11860 (popcount_loc_descriptor): Likewise.
11861 (bswap_loc_descriptor): Likewise.
11862 (rotate_loc_descriptor): Likewise.
11863 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11864 calling the functions above.
11866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11867 Alan Hayward <alan.hayward@arm.com>
11868 David Sherwood <david.sherwood@arm.com>
11870 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11872 (try_combine): Likewise.
11873 (simplify_if_then_else): Likewise.
11874 * cse.c (cse_insn): Likewise.
11875 * dwarf2out.c (mem_loc_descriptor): Likewise.
11876 * emit-rtl.c (gen_lowpart_common): Likewise.
11877 * simplify-rtx.c (simplify_truncation): Likewise.
11878 (simplify_binary_operation_1): Likewise.
11879 (simplify_const_relational_operation): Likewise.
11880 (simplify_ternary_operation): Likewise.
11881 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11884 Alan Hayward <alan.hayward@arm.com>
11885 David Sherwood <david.sherwood@arm.com>
11887 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11888 * reload.c (push_reload): Likewise.
11889 (find_reloads): Likewise.
11891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11892 Alan Hayward <alan.hayward@arm.com>
11893 David Sherwood <david.sherwood@arm.com>
11895 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11896 (make_compound_operation_int): Likewise.
11897 (change_zero_ext): Likewise.
11898 * expr.c (convert_move): Likewise.
11899 (convert_modes): Likewise.
11900 * fwprop.c (forward_propagate_subreg): Likewise.
11901 * loop-iv.c (get_biv_step_1): Likewise.
11902 * optabs.c (widen_operand): Likewise.
11903 * postreload.c (move2add_valid_value_p): Likewise.
11904 * recog.c (simplify_while_replacing): Likewise.
11905 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11906 (simplify_binary_operation_1): Likewise. Remove redundant
11907 mode equality check.
11909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11910 Alan Hayward <alan.hayward@arm.com>
11911 David Sherwood <david.sherwood@arm.com>
11913 * combine.c (combine_simplify_rtx): Add checks for
11914 is_a <scalar_int_mode>.
11915 (simplify_if_then_else): Likewise.
11916 (make_field_assignment): Likewise.
11917 (simplify_comparison): Likewise.
11918 * ifcvt.c (noce_try_bitop): Likewise.
11919 * loop-invariant.c (canonicalize_address_mult): Likewise.
11920 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11923 Alan Hayward <alan.hayward@arm.com>
11924 David Sherwood <david.sherwood@arm.com>
11926 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11927 is_a <scalar_int_mode> instead of != BLKmode.
11929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11930 Alan Hayward <alan.hayward@arm.com>
11931 David Sherwood <david.sherwood@arm.com>
11933 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11934 instead of != VOIDmode.
11935 * combine.c (if_then_else_cond): Likewise.
11936 (change_zero_ext): Likewise.
11937 * dwarf2out.c (mem_loc_descriptor): Likewise.
11938 (loc_descriptor): Likewise.
11939 * rtlanal.c (canonicalize_condition): Likewise.
11940 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11943 Alan Hayward <alan.hayward@arm.com>
11944 David Sherwood <david.sherwood@arm.com>
11946 * simplify-rtx.c (simplify_binary_operation_1): Use
11947 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11950 Alan Hayward <alan.hayward@arm.com>
11951 David Sherwood <david.sherwood@arm.com>
11953 * wide-int.h (int_traits<unsigned char>) New class.
11954 (int_traits<unsigned short>) Likewise.
11955 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11956 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11958 * combine.c (set_nonzero_bits_and_sign_copies): Use
11959 is_a <scalar_int_mode>.
11960 (find_split_point): Likewise.
11961 (combine_simplify_rtx): Likewise.
11962 (simplify_logical): Likewise.
11963 (expand_compound_operation): Likewise.
11964 (expand_field_assignment): Likewise.
11965 (make_compound_operation): Likewise.
11966 (extended_count): Likewise.
11967 (change_zero_ext): Likewise.
11968 (simplify_comparison): Likewise.
11969 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11970 (ucompare_loc_descriptor): Likewise.
11971 (minmax_loc_descriptor): Likewise.
11972 (mem_loc_descriptor): Likewise.
11973 (loc_descriptor): Likewise.
11974 * expmed.c (init_expmed_one_mode): Likewise.
11975 * lra-constraints.c (lra_constraint_offset): Likewise.
11976 * optabs.c (prepare_libcall_arg): Likewise.
11977 * postreload.c (move2add_note_store): Likewise.
11978 * reload.c (operands_match_p): Likewise.
11979 * rtl.h (load_extend_op): Likewise.
11980 * rtlhooks.c (gen_lowpart_general): Likewise.
11981 * simplify-rtx.c (simplify_truncation): Likewise.
11982 (simplify_unary_operation_1): Likewise.
11983 (simplify_binary_operation_1): Likewise.
11984 (simplify_const_binary_operation): Likewise.
11985 (simplify_const_relational_operation): Likewise.
11986 (simplify_subreg): Likewise.
11987 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11988 * var-tracking.c (adjust_mems): Likewise.
11989 (prepare_call_arguments): Likewise.
11991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11992 Alan Hayward <alan.hayward@arm.com>
11993 David Sherwood <david.sherwood@arm.com>
11995 * machmode.h (is_int_mode): New fuction.
11996 * combine.c (find_split_point): Use it.
11997 (combine_simplify_rtx): Likewise.
11998 (simplify_if_then_else): Likewise.
11999 (simplify_set): Likewise.
12000 (simplify_shift_const_1): Likewise.
12001 (simplify_comparison): Likewise.
12002 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12003 * cse.c (notreg_cost): Likewise.
12004 (cse_insn): Likewise.
12005 * cselib.c (cselib_lookup_1): Likewise.
12006 * dojump.c (do_jump_1): Likewise.
12007 (do_compare_rtx_and_jump): Likewise.
12008 * dse.c (get_call_args): Likewise.
12009 * dwarf2out.c (rtl_for_decl_init): Likewise.
12010 (native_encode_initializer): Likewise.
12011 * expmed.c (emit_store_flag_1): Likewise.
12012 (emit_store_flag): Likewise.
12013 * expr.c (convert_modes): Likewise.
12014 (store_field): Likewise.
12015 (expand_expr_real_1): Likewise.
12016 * fold-const.c (fold_read_from_constant_string): Likewise.
12017 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12018 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12019 * optabs.c (expand_binop): Likewise.
12020 (expand_unop): Likewise.
12021 (expand_abs_nojump): Likewise.
12022 (expand_one_cmpl_abs_nojump): Likewise.
12023 * simplify-rtx.c (mode_signbit_p): Likewise.
12024 (val_signbit_p): Likewise.
12025 (val_signbit_known_set_p): Likewise.
12026 (val_signbit_known_clear_p): Likewise.
12027 (simplify_relational_operation_1): Likewise.
12028 * tree.c (vector_type_mode): Likewise.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * machmode.h (smallest_mode_for_size): Fix formatting.
12035 (smallest_int_mode_for_size): New function.
12036 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12037 instead of smallest_mode_for_size.
12038 * combine.c (make_extraction): Likewise.
12039 * config/arc/arc.c (arc_expand_movmem): Likewise.
12040 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12041 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12042 * config/s390/s390.c (s390_expand_insv): Likewise.
12043 * config/sparc/sparc.c (assign_int_registers): Likewise.
12044 * config/spu/spu.c (spu_function_value): Likewise.
12045 (spu_function_arg): Likewise.
12046 * coverage.c (get_gcov_type): Likewise.
12047 (get_gcov_unsigned_t): Likewise.
12048 * dse.c (find_shift_sequence): Likewise.
12049 * expmed.c (store_bit_field_1): Likewise.
12050 * expr.c (convert_move): Likewise.
12051 (store_field): Likewise.
12052 * internal-fn.c (expand_arith_overflow): Likewise.
12053 * optabs-query.c (get_best_extraction_insn): Likewise.
12054 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12055 * stor-layout.c (layout_type): Likewise.
12056 (initialize_sizetypes): Likewise.
12057 * targhooks.c (default_get_mask_mode): Likewise.
12058 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12061 Alan Hayward <alan.hayward@arm.com>
12062 David Sherwood <david.sherwood@arm.com>
12064 * machmode.h (opt_mode::else_blk): New function.
12065 (int_mode_for_mode): Declare.
12066 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12067 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12069 * cfgexpand.c (expand_debug_expr): Likewise.
12070 * combine.c (gen_lowpart_or_truncate): Likewise.
12071 (gen_lowpart_for_combine): Likewise.
12072 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12073 * config/avr/avr.c (avr_to_int_mode): Likewise.
12074 (avr_out_plus_1): Likewise.
12075 (avr_out_plus): Likewise.
12076 (avr_out_round): Likewise.
12077 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12078 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12079 (s390_expand_vcond): Likewise.
12080 * config/spu/spu.c (spu_split_immediate): Likewise.
12081 (spu_expand_mov): Likewise.
12082 * dse.c (get_stored_val): Likewise.
12083 * expmed.c (store_bit_field_1): Likewise.
12084 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12086 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12087 (extract_low_bits): Likewise.
12088 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12089 handling rather than repeating the check.
12090 (emit_group_store): Likewise.
12091 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12092 * optabs.c (expand_absneg_bit): Likewise.
12093 (expand_copysign_absneg): Likewise.
12094 (expand_copysign_bit): Likewise.
12095 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12096 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12097 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12098 * var-tracking.c (prepare_call_arguments): Likewise.
12099 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12100 int_mode_for_mode instead of mode_for_size.
12101 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12104 Alan Hayward <alan.hayward@arm.com>
12105 David Sherwood <david.sherwood@arm.com>
12107 * machmode.h (int_mode_for_size): New function.
12108 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12109 instead of mode_for_size.
12110 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12112 * combine.c (expand_field_assignment): Use int_mode_for_size
12113 instead of mode_for_size.
12114 (make_extraction): Likewise.
12115 (simplify_shift_const_1): Likewise.
12116 (simplify_comparison): Likewise.
12117 * dojump.c (do_jump): Likewise.
12118 * dwarf2out.c (mem_loc_descriptor): Likewise.
12119 * emit-rtl.c (init_derived_machine_modes): Likewise.
12120 * expmed.c (flip_storage_order): Likewise.
12121 (convert_extracted_bit_field): Likewise.
12122 * expr.c (copy_blkmode_from_reg): Likewise.
12123 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12124 * internal-fn.c (expand_mul_overflow): Likewise.
12125 * lower-subreg.c (simple_move): Likewise.
12126 * optabs-libfuncs.c (init_optabs): Likewise.
12127 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12128 * tree.c (vector_type_mode): Likewise.
12129 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12130 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12131 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12132 * tree-vect-stmts.c (vectorizable_load): Likewise.
12133 (vectorizable_store): Likewise.
12135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12136 Alan Hayward <alan.hayward@arm.com>
12137 David Sherwood <david.sherwood@arm.com>
12139 * coretypes.h (pod_mode): New type.
12140 (scalar_int_mode_pod): New typedef.
12141 * machmode.h (pod_mode): New class.
12142 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12143 * genmodes.c (emit_mode_int_n): Update accordingly.
12144 * lower-subreg.h (target_lower_subreg): Change type to
12145 scalar_int_mode_pod.
12146 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12147 scalar_int_mode_pod.
12149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12150 Alan Hayward <alan.hayward@arm.com>
12151 David Sherwood <david.sherwood@arm.com>
12153 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12154 machine_mode to scalar_int_mode.
12155 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12156 (rs6000_option_override_internal): Remove cast to int.
12157 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12158 machine_mode to scalar_int_mode.
12159 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12160 (rs6000_option_override_internal): Remove cast to int.
12161 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12162 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12164 * config/s390/s390.c (s390_expand_builtin): Likewise.
12165 * coretypes.h (scalar_int_mode): New type.
12166 (opt_scalar_int_mode): New typedef.
12167 * machmode.h (scalar_int_mode): New class.
12168 (scalar_int_mode::includes_p): New function.
12169 (byte_mode): Change type to scalar_int_mode.
12170 (word_mode): Likewise.
12171 (ptr_mode): Likewise.
12172 * emit-rtl.c (byte_mode): Likewise.
12173 (word_mode): Likewise.
12174 (ptr_mode): Likewise.
12175 (init_derived_machine_modes): Update accordingly.
12176 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12177 and MODE_PARTIAL_INT.
12178 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12179 opt_scalar_int_mode.
12181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12182 Alan Hayward <alan.hayward@arm.com>
12183 David Sherwood <david.sherwood@arm.com>
12185 * target.def (libgcc_floating_mode_supported_p): Take a
12187 * doc/tm.texi: Regenerate.
12188 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12190 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12191 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12195 Alan Hayward <alan.hayward@arm.com>
12196 David Sherwood <david.sherwood@arm.com>
12198 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12199 * doc/tm.texi: Regenerate.
12200 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12201 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12202 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12203 * targhooks.h (default_floatn_mode): Likewise.
12204 * targhooks.c (default_floatn_mode): Likewise.
12205 * tree.c (build_common_tree_nodes): Update accordingly.
12207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12208 Alan Hayward <alan.hayward@arm.com>
12209 David Sherwood <david.sherwood@arm.com>
12211 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12212 (mode_iterator::iterate_p): Likewise.
12213 (mode_iterator::get_wider): Likewise.
12214 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12217 Alan Hayward <alan.hayward@arm.com>
12218 David Sherwood <david.sherwood@arm.com>
12220 * coretypes.h (opt_scalar_float_mode): New typedef.
12221 * machmode.h (float_mode_for_size): New function.
12222 * emit-rtl.c (double_mode): Delete.
12223 (init_emit_once): Use float_mode_for_size.
12224 * stor-layout.c (layout_type): Likewise.
12225 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12228 Alan Hayward <alan.hayward@arm.com>
12229 David Sherwood <david.sherwood@arm.com>
12231 * output.h (assemble_real): Take a scalar_float_mode.
12232 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12233 * config/arm/arm.md (consttable_4): Likewise.
12234 (consttable_8): Likewise.
12235 (consttable_16): Likewise.
12236 * config/mips/mips.md (consttable_float): Likewise.
12237 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12238 * varasm.c (assemble_real): Take a scalar_float_mode.
12239 (output_constant_pool_2): Update accordingly.
12240 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12247 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12248 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12249 (native_encode_real): Likewise.
12250 (native_interpret_real): Likewise.
12251 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12252 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12255 Alan Hayward <alan.hayward@arm.com>
12256 David Sherwood <david.sherwood@arm.com>
12258 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12259 <scalar_float_mode>. Simplify.
12260 (gen_extend_conv_libfunc): Likewise.
12262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12263 Alan Hayward <alan.hayward@arm.com>
12264 David Sherwood <david.sherwood@arm.com>
12266 * coretypes.h (scalar_float_mode): New type.
12267 * machmode.h (mode_traits::from_int): Use machine_mode if
12268 USE_ENUM_MODES is defined.
12269 (is_a): New function.
12271 (dyn_cast): Likewise.
12272 (scalar_float_mode): New class.
12273 (scalar_float_mode::includes_p): New function.
12274 (is_float_mode): Likewise.
12275 * gdbhooks.py (MachineModePrinter): New class.
12276 (build_pretty_printer): Use it for scalar_float_mode.
12277 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12278 (format_helper::format_helper): Turn into a template.
12279 * genmodes.c (get_mode_class): New function.
12280 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12281 or machine_mode if none.
12282 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12283 as_a <scalar_float_mode>.
12284 * dwarf2out.c (mem_loc_descriptor): Likewise.
12285 (insert_float): Likewise.
12286 (add_const_value_attribute): Likewise.
12287 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12288 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12289 (expand_unop): Update accordingly.
12290 (expand_abs_nojump): Likewise.
12291 (expand_copysign_absneg): Take a scalar_float_mode.
12292 (expand_copysign_bit): Likewise.
12293 (expand_copysign): Update accordingly.
12295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12296 Alan Hayward <alan.hayward@arm.com>
12297 David Sherwood <david.sherwood@arm.com>
12299 * coretypes.h (opt_mode): New class.
12300 * machmode.h (opt_mode): Likewise.
12301 (opt_mode::else_void): New function.
12302 (opt_mode::require): Likewise.
12303 (opt_mode::exists): Likewise.
12304 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12305 (GET_MODE_2XWIDER_MODE): Likewise.
12306 (mode_iterator::get_wider): Update accordingly.
12307 (mode_iterator::get_2xwider): Likewise.
12308 (mode_iterator::get_known_wider): Likewise, turning into a template.
12309 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12310 forcing a wider mode to exist.
12311 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12312 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12313 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12314 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12315 * lower-subreg.c (init_lower_subreg): Likewise.
12316 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12317 on the final iteration.
12318 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12319 a wider mode exists before asking for a move pattern.
12320 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12321 forcing a wider mode to exist.
12322 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12323 returning false if no such mode exists.
12324 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12325 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12326 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12327 Avoid checking for a MODE_INT if we already know the mode is not a
12329 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12330 forcing a wider mode to exist.
12331 (expmed_mult_highpart_optab): Likewise.
12332 (expmed_mult_highpart): Likewise.
12333 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12335 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12336 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12337 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12338 * internal-fn.c (expand_mul_overflow): Update use of
12339 GET_MODE_2XWIDER_MODE.
12340 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12341 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12342 GET_MODE_WIDER_MODE.
12343 (convert_plusminus_to_widen): Likewise.
12344 * tree-switch-conversion.c (array_value_type): Likewise.
12345 * var-tracking.c (emit_note_insn_var_location): Likewise.
12346 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12347 Return false inside rather than outside the loop if no wider mode
12349 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12350 and GET_MODE_2XWIDER_MODE
12351 (can_compare_p): Use else_void.
12352 * gdbhooks.py (OptMachineModePrinter): New class.
12353 (build_pretty_printer): Use it for opt_mode.
12355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12356 Alan Hayward <alan.hayward@arm.com>
12357 David Sherwood <david.sherwood@arm.com>
12359 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12360 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12363 Alan Hayward <alan.hayward@arm.com>
12364 David Sherwood <david.sherwood@arm.com>
12366 * machmode.h (mode_traits): New structure.
12367 (get_narrowest_mode): New function.
12368 (mode_iterator::start): Likewise.
12369 (mode_iterator::iterate_p): Likewise.
12370 (mode_iterator::get_wider): Likewise.
12371 (mode_iterator::get_known_wider): Likewise.
12372 (mode_iterator::get_2xwider): Likewise.
12373 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12374 (FOR_EACH_MODE): Likewise.
12375 (FOR_EACH_MODE_FROM): Likewise.
12376 (FOR_EACH_MODE_UNTIL): Likewise.
12377 (FOR_EACH_WIDER_MODE): Likewise.
12378 (FOR_EACH_2XWIDER_MODE): Likewise.
12379 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12380 * combine.c (simplify_comparison): Likewise
12381 * config/i386/i386.c (type_natural_mode): Likewise.
12382 * cse.c (cse_insn): Likewise.
12383 * dse.c (find_shift_sequence): Likewise.
12384 * emit-rtl.c (init_derived_machine_modes): Likewise.
12385 (init_emit_once): Likewise.
12386 * explow.c (hard_function_value): Likewise.
12387 * expmed.c (extract_fixed_bit_field_1): Likewise.
12388 (extract_bit_field_1): Likewise.
12389 (expand_divmod): Likewise.
12390 (emit_store_flag_1): Likewise.
12391 * expr.c (init_expr_target): Likewise.
12392 (convert_move): Likewise.
12393 (alignment_for_piecewise_move): Likewise.
12394 (widest_int_mode_for_size): Likewise.
12395 (emit_block_move_via_movmem): Likewise.
12396 (copy_blkmode_to_reg): Likewise.
12397 (set_storage_via_setmem): Likewise.
12398 (compress_float_constant): Likewise.
12399 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12400 * optabs-query.c (get_best_extraction_insn): Likewise.
12401 * optabs.c (expand_binop): Likewise.
12402 (expand_twoval_unop): Likewise.
12403 (expand_twoval_binop): Likewise.
12404 (widen_leading): Likewise.
12405 (widen_bswap): Likewise.
12406 (expand_parity): Likewise.
12407 (expand_unop): Likewise.
12408 (prepare_cmp_insn): Likewise.
12409 (prepare_float_lib_cmp): Likewise.
12410 (expand_float): Likewise.
12411 (expand_fix): Likewise.
12412 (expand_sfix_optab): Likewise.
12413 * postreload.c (move2add_use_add2_insn): Likewise.
12414 * reg-stack.c (reg_to_stack): Likewise.
12415 * reginfo.c (choose_hard_reg_mode): Likewise.
12416 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12417 * stor-layout.c (mode_for_size): Likewise.
12418 (smallest_mode_for_size): Likewise.
12419 (mode_for_vector): Likewise.
12420 (finish_bitfield_representative): Likewise.
12421 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12422 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12423 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12424 * var-tracking.c (prepare_call_arguments): Likewise.
12426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12427 Alan Hayward <alan.hayward@arm.com>
12428 David Sherwood <david.sherwood@arm.com>
12430 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12431 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12432 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12433 * machmode.h (mode_size): Move earlier in file.
12434 (mode_precision): Likewise.
12435 (mode_inner): Likewise.
12436 (mode_nunits): Likewise.
12437 (mode_unit_size): Likewise.
12438 (unit_unit_precision): Likewise.
12439 (mode_wider): Likewise.
12440 (mode_2xwider): Likewise.
12441 (machine_mode): New class.
12442 (mode_to_bytes): New function.
12443 (mode_to_bits): Likewise.
12444 (mode_to_precision): Likewise.
12445 (mode_to_inner): Likewise.
12446 (mode_to_unit_size): Likewise.
12447 (mode_to_unit_precision): Likewise.
12448 (mode_to_nunits): Likewise.
12449 (GET_MODE_SIZE): Use mode_to_bytes.
12450 (GET_MODE_BITSIZE): Use mode_to_bits.
12451 (GET_MODE_PRECISION): Use mode_to_precision.
12452 (GET_MODE_INNER): Use mode_to_inner.
12453 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12454 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12455 (GET_MODE_NUNITS): Use mode_to_nunits.
12456 * system.h (ALWAYS_INLINE): New macro.
12457 * config/powerpcspe/powerpcspe-c.c
12458 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12459 int for arg1_mode and arg2_mode.
12461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12462 Alan Hayward <alan.hayward@arm.com>
12463 David Sherwood <david.sherwood@arm.com>
12465 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12466 Prefix mode names with E_ in case statements.
12467 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12468 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12469 (aarch64_split_simd_move): Likewise.
12470 (aarch64_gen_storewb_pair): Likewise.
12471 (aarch64_gen_loadwb_pair): Likewise.
12472 (aarch64_gen_store_pair): Likewise.
12473 (aarch64_gen_load_pair): Likewise.
12474 (aarch64_get_condition_code_1): Likewise.
12475 (aarch64_constant_pool_reload_icode): Likewise.
12476 (get_rsqrte_type): Likewise.
12477 (get_rsqrts_type): Likewise.
12478 (get_recpe_type): Likewise.
12479 (get_recps_type): Likewise.
12480 (aarch64_gimplify_va_arg_expr): Likewise.
12481 (aarch64_simd_container_mode): Likewise.
12482 (aarch64_emit_load_exclusive): Likewise.
12483 (aarch64_emit_store_exclusive): Likewise.
12484 (aarch64_expand_compare_and_swap): Likewise.
12485 (aarch64_gen_atomic_cas): Likewise.
12486 (aarch64_emit_bic): Likewise.
12487 (aarch64_emit_atomic_swap): Likewise.
12488 (aarch64_emit_atomic_load_op): Likewise.
12489 (aarch64_evpc_trn): Likewise.
12490 (aarch64_evpc_uzp): Likewise.
12491 (aarch64_evpc_zip): Likewise.
12492 (aarch64_evpc_ext): Likewise.
12493 (aarch64_evpc_rev): Likewise.
12494 (aarch64_evpc_dup): Likewise.
12495 (aarch64_gen_ccmp_first): Likewise.
12496 (aarch64_gen_ccmp_next): Likewise.
12497 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12498 (alpha_emit_xfloating_libcall): Likewise.
12499 (emit_insxl): Likewise.
12500 (alpha_arg_type): Likewise.
12501 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12502 (arc_preferred_simd_mode): Likewise.
12503 (arc_secondary_reload): Likewise.
12504 (get_arc_condition_code): Likewise.
12505 (arc_print_operand): Likewise.
12506 (arc_legitimate_constant_p): Likewise.
12507 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12508 * config/arc/arc.md (casesi_load): Likewise.
12509 (casesi_compact_jump): Likewise.
12510 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12511 (cc_use_register): Likewise.
12512 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12513 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12514 (arm_init_iwmmxt_builtins): Likewise.
12515 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12516 (neon_expand_vector_init): Likewise.
12517 (arm_attr_length_move_neon): Likewise.
12518 (maybe_get_arm_condition_code): Likewise.
12519 (arm_emit_vector_const): Likewise.
12520 (arm_preferred_simd_mode): Likewise.
12521 (arm_output_iwmmxt_tinsr): Likewise.
12522 (thumb1_output_casesi): Likewise.
12523 (thumb2_output_casesi): Likewise.
12524 (arm_emit_load_exclusive): Likewise.
12525 (arm_emit_store_exclusive): Likewise.
12526 (arm_expand_compare_and_swap): Likewise.
12527 (arm_evpc_neon_vuzp): Likewise.
12528 (arm_evpc_neon_vzip): Likewise.
12529 (arm_evpc_neon_vrev): Likewise.
12530 (arm_evpc_neon_vtrn): Likewise.
12531 (arm_evpc_neon_vext): Likewise.
12532 (arm_validize_comparison): Likewise.
12533 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12534 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12535 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12536 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12537 (c6x_preferred_simd_mode): Likewise.
12538 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12539 (epiphany_rtx_costs): Likewise.
12540 * config/epiphany/predicates.md (proper_comparison_operator):
12542 * config/frv/frv.c (condexec_memory_operand): Likewise.
12543 (frv_emit_move): Likewise.
12544 (output_move_single): Likewise.
12545 (output_condmove_single): Likewise.
12546 (frv_hard_regno_mode_ok): Likewise.
12547 (frv_matching_accg_mode): Likewise.
12548 * config/h8300/h8300.c (split_adds_subs): Likewise.
12549 (h8300_rtx_costs): Likewise.
12550 (h8300_print_operand): Likewise.
12551 (compute_mov_length): Likewise.
12552 (output_logical_op): Likewise.
12553 (compute_logical_op_length): Likewise.
12554 (compute_logical_op_cc): Likewise.
12555 (h8300_shift_needs_scratch_p): Likewise.
12556 (output_a_shift): Likewise.
12557 (compute_a_shift_length): Likewise.
12558 (compute_a_shift_cc): Likewise.
12559 (expand_a_rotate): Likewise.
12560 (output_a_rotate): Likewise.
12561 * config/i386/i386.c (classify_argument): Likewise.
12562 (function_arg_advance_32): Likewise.
12563 (function_arg_32): Likewise.
12564 (function_arg_64): Likewise.
12565 (function_value_64): Likewise.
12566 (ix86_gimplify_va_arg): Likewise.
12567 (ix86_legitimate_constant_p): Likewise.
12568 (put_condition_code): Likewise.
12569 (split_double_mode): Likewise.
12570 (ix86_avx256_split_vector_move_misalign): Likewise.
12571 (ix86_expand_vector_logical_operator): Likewise.
12572 (ix86_split_idivmod): Likewise.
12573 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12574 (ix86_build_const_vector): Likewise.
12575 (ix86_build_signbit_mask): Likewise.
12576 (ix86_match_ccmode): Likewise.
12577 (ix86_cc_modes_compatible): Likewise.
12578 (ix86_expand_branch): Likewise.
12579 (ix86_expand_sse_cmp): Likewise.
12580 (ix86_expand_sse_movcc): Likewise.
12581 (ix86_expand_int_sse_cmp): Likewise.
12582 (ix86_expand_vec_perm_vpermi2): Likewise.
12583 (ix86_expand_vec_perm): Likewise.
12584 (ix86_expand_sse_unpack): Likewise.
12585 (ix86_expand_int_addcc): Likewise.
12586 (ix86_split_to_parts): Likewise.
12587 (ix86_vectorize_builtin_gather): Likewise.
12588 (ix86_vectorize_builtin_scatter): Likewise.
12589 (avx_vpermilp_parallel): Likewise.
12590 (inline_memory_move_cost): Likewise.
12591 (ix86_tieable_integer_mode_p): Likewise.
12592 (x86_maybe_negate_const_int): Likewise.
12593 (ix86_expand_vector_init_duplicate): Likewise.
12594 (ix86_expand_vector_init_one_nonzero): Likewise.
12595 (ix86_expand_vector_init_one_var): Likewise.
12596 (ix86_expand_vector_init_concat): Likewise.
12597 (ix86_expand_vector_init_interleave): Likewise.
12598 (ix86_expand_vector_init_general): Likewise.
12599 (ix86_expand_vector_set): Likewise.
12600 (ix86_expand_vector_extract): Likewise.
12601 (emit_reduc_half): Likewise.
12602 (ix86_emit_i387_round): Likewise.
12603 (ix86_mangle_type): Likewise.
12604 (ix86_expand_round_sse4): Likewise.
12605 (expand_vec_perm_blend): Likewise.
12606 (canonicalize_vector_int_perm): Likewise.
12607 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12608 (expand_vec_perm_1): Likewise.
12609 (expand_vec_perm_interleave3): Likewise.
12610 (expand_vec_perm_even_odd_pack): Likewise.
12611 (expand_vec_perm_even_odd_1): Likewise.
12612 (expand_vec_perm_broadcast_1): Likewise.
12613 (ix86_vectorize_vec_perm_const_ok): Likewise.
12614 (ix86_expand_vecop_qihi): Likewise.
12615 (ix86_expand_mul_widen_hilo): Likewise.
12616 (ix86_expand_sse2_abs): Likewise.
12617 (ix86_expand_pextr): Likewise.
12618 (ix86_expand_pinsr): Likewise.
12619 (ix86_preferred_simd_mode): Likewise.
12620 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12621 * config/i386/sse.md (*andnot<mode>3): Likewise.
12622 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12623 (*<code><mode>3): Likewise.
12624 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12625 (ia64_expand_atomic_op): Likewise.
12626 (ia64_arg_type): Likewise.
12627 (ia64_mode_to_int): Likewise.
12628 (ia64_scalar_mode_supported_p): Likewise.
12629 (ia64_vector_mode_supported_p): Likewise.
12630 (expand_vec_perm_broadcast): Likewise.
12631 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12632 (iq2000_function_arg_advance): Likewise.
12633 (iq2000_function_arg): Likewise.
12634 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12635 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12636 (m68k_libcall_value): Likewise.
12637 (m68k_function_value): Likewise.
12638 (sched_attr_op_type): Likewise.
12639 * config/mcore/mcore.c (mcore_output_move): Likewise.
12640 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12642 (microblaze_function_arg): Likewise.
12643 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12644 (mips_print_operand): Likewise.
12645 (mips_mode_ok_for_mov_fmt_p): Likewise.
12646 (mips_vector_mode_supported_p): Likewise.
12647 (mips_preferred_simd_mode): Likewise.
12648 (mips_expand_vpc_loongson_even_odd): Likewise.
12649 (mips_expand_vec_unpack): Likewise.
12650 (mips_expand_vi_broadcast): Likewise.
12651 (mips_expand_vector_init): Likewise.
12652 (mips_expand_vec_reduc): Likewise.
12653 (mips_expand_msa_cmp): Likewise.
12654 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12655 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12656 (cc_flags_for_mode): Likewise.
12657 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12658 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12659 (nds32_output_casesi_pc_relative): Likewise.
12660 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12661 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12662 (nvptx_gen_unpack): Likewise.
12663 (nvptx_gen_pack): Likewise.
12664 (nvptx_gen_shuffle): Likewise.
12665 (nvptx_gen_wcast): Likewise.
12666 (nvptx_preferred_simd_mode): Likewise.
12667 * config/pa/pa.c (pa_secondary_reload): Likewise.
12668 * config/pa/predicates.md (base14_operand): Likewise.
12669 * config/powerpcspe/powerpcspe-c.c
12670 (altivec_resolve_overloaded_builtin): Likewise.
12671 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12673 (rs6000_preferred_simd_mode): Likewise.
12674 (output_vec_const_move): Likewise.
12675 (rs6000_expand_vector_extract): Likewise.
12676 (rs6000_split_vec_extract_var): Likewise.
12677 (reg_offset_addressing_ok_p): Likewise.
12678 (rs6000_legitimate_offset_address_p): Likewise.
12679 (rs6000_legitimize_address): Likewise.
12680 (rs6000_emit_set_const): Likewise.
12681 (rs6000_const_vec): Likewise.
12682 (rs6000_emit_move): Likewise.
12683 (spe_build_register_parallel): Likewise.
12684 (rs6000_darwin64_record_arg_recurse): Likewise.
12685 (swap_selector_for_mode): Likewise.
12686 (spe_init_builtins): Likewise.
12687 (paired_init_builtins): Likewise.
12688 (altivec_init_builtins): Likewise.
12689 (do_load_for_compare): Likewise.
12690 (rs6000_generate_compare): Likewise.
12691 (rs6000_expand_float128_convert): Likewise.
12692 (emit_load_locked): Likewise.
12693 (emit_store_conditional): Likewise.
12694 (rs6000_output_function_epilogue): Likewise.
12695 (rs6000_handle_altivec_attribute): Likewise.
12696 (rs6000_function_value): Likewise.
12697 (emit_fusion_gpr_load): Likewise.
12698 (emit_fusion_p9_load): Likewise.
12699 (emit_fusion_p9_store): Likewise.
12700 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12701 (fusion_gpr_mem_load): Likewise.
12702 (fusion_addis_mem_combo_load): Likewise.
12703 (fusion_addis_mem_combo_store): Likewise.
12704 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12705 (fusion_gpr_mem_load): Likewise.
12706 (fusion_addis_mem_combo_load): Likewise.
12707 (fusion_addis_mem_combo_store): Likewise.
12708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12710 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12711 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12712 (rs6000_preferred_simd_mode): Likewise.
12713 (output_vec_const_move): Likewise.
12714 (rs6000_expand_vector_extract): Likewise.
12715 (rs6000_split_vec_extract_var): Likewise.
12716 (reg_offset_addressing_ok_p): Likewise.
12717 (rs6000_legitimate_offset_address_p): Likewise.
12718 (rs6000_legitimize_address): Likewise.
12719 (rs6000_emit_set_const): Likewise.
12720 (rs6000_const_vec): Likewise.
12721 (rs6000_emit_move): Likewise.
12722 (rs6000_darwin64_record_arg_recurse): Likewise.
12723 (swap_selector_for_mode): Likewise.
12724 (paired_init_builtins): Likewise.
12725 (altivec_init_builtins): Likewise.
12726 (rs6000_expand_float128_convert): Likewise.
12727 (emit_load_locked): Likewise.
12728 (emit_store_conditional): Likewise.
12729 (rs6000_output_function_epilogue): Likewise.
12730 (rs6000_handle_altivec_attribute): Likewise.
12731 (rs6000_function_value): Likewise.
12732 (emit_fusion_gpr_load): Likewise.
12733 (emit_fusion_p9_load): Likewise.
12734 (emit_fusion_p9_store): Likewise.
12735 * config/rx/rx.c (rx_gen_move_template): Likewise.
12736 (flags_from_mode): Likewise.
12737 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12738 (s390_slb_comparison): Likewise.
12739 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12740 (s390_vector_mode_supported_p): Likewise.
12741 (s390_cc_modes_compatible): Likewise.
12742 (s390_match_ccmode_set): Likewise.
12743 (s390_canonicalize_comparison): Likewise.
12744 (s390_emit_compare_and_swap): Likewise.
12745 (s390_branch_condition_mask): Likewise.
12746 (s390_rtx_costs): Likewise.
12747 (s390_secondary_reload): Likewise.
12748 (__SECONDARY_RELOAD_CASE): Likewise.
12749 (s390_expand_cs): Likewise.
12750 (s390_preferred_simd_mode): Likewise.
12751 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12752 * config/sh/sh.c (sh_print_operand): Likewise.
12753 (dump_table): Likewise.
12754 (sh_secondary_reload): Likewise.
12755 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12756 * config/sh/sh.md (casesi_worker_1): Likewise.
12757 (casesi_worker_2): Likewise.
12758 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12759 (fcc_comparison_operator): Likewise.
12760 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12761 (emit_soft_tfmode_cvt): Likewise.
12762 (sparc_preferred_simd_mode): Likewise.
12763 (output_cbranch): Likewise.
12764 (sparc_print_operand): Likewise.
12765 (sparc_expand_vec_perm_bmask): Likewise.
12766 (vector_init_bshuffle): Likewise.
12767 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12768 (spu_vector_mode_supported_p): Likewise.
12769 (spu_expand_insv): Likewise.
12770 (spu_emit_branch_or_set): Likewise.
12771 (spu_handle_vector_attribute): Likewise.
12772 (spu_builtin_splats): Likewise.
12773 (spu_builtin_extract): Likewise.
12774 (spu_builtin_promote): Likewise.
12775 (spu_expand_sign_extend): Likewise.
12776 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12777 (tilegx_simd_int): Likewise.
12778 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12779 (tilepro_simd_int): Likewise.
12780 * config/v850/v850.c (const_double_split): Likewise.
12781 (v850_print_operand): Likewise.
12782 (ep_memory_offset): Likewise.
12783 * config/vax/vax.c (vax_rtx_costs): Likewise.
12784 (vax_output_int_move): Likewise.
12785 (vax_output_int_add): Likewise.
12786 (vax_output_int_subtract): Likewise.
12787 * config/visium/predicates.md (visium_branch_operator): Likewise.
12788 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12789 (visium_print_operand_address): Likewise.
12790 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12791 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12792 (xtensa_expand_conditional_branch): Likewise.
12793 (xtensa_copy_incoming_a7): Likewise.
12794 (xtensa_output_literal): Likewise.
12795 * dfp.c (decimal_real_maxval): Likewise.
12796 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12799 Alan Hayward <alan.hayward@arm.com>
12800 David Sherwood <david.sherwood@arm.com>
12802 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12803 (mode_nunits_inline): Likewise.
12804 (mode_inner_inline): Likewise.
12805 (mode_unit_size_inline): Likewise.
12806 (mode_unit_precision_inline): Likewise.
12807 (emit_insn_modes_h): Likewise. Also emit a #define of the
12809 (emit_mode_wider): Add an E_ prefix to mode names.
12810 (emit_mode_complex): Likewise.
12811 (emit_mode_inner): Likewise.
12812 (emit_mode_adjustments): Likewise.
12813 (emit_mode_int_n): Likewise.
12814 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12815 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12816 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12817 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12818 (CRC32_BUILTIN, ENTRY): Likewise.
12819 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12820 (aarch64_pop_regs): Likewise.
12821 (aarch64_process_components): Likewise.
12822 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12823 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12824 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12825 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12826 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12827 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12829 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12830 (pcmpistr): Likewise.
12831 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12832 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12833 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12835 * config/rl78/rl78.c (mduc_regs): Likewise.
12836 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12837 (htm_expand_builtin): Likewise.
12838 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12839 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12840 * config/xtensa/xtensa.c (print_operand): Likewise.
12841 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12842 (NUM_MODE_VECTOR_INT): Likewise.
12843 * genoutput.c (null_operand): Likewise.
12844 (output_operand_data): Likewise.
12845 * genrecog.c (print_parameter_value): Likewise.
12846 * lra.c (debug_operand_data): Likewise.
12848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12850 * dwarf2out.c (scompare_loc_descriptor_wide)
12851 (scompare_loc_descriptor_narrow): New functions, split out from...
12852 (scompare_loc_descriptor): ...here.
12853 * expmed.c (emit_store_flag_int): New function, split out from...
12854 (emit_store_flag): ...here.
12856 2017-08-30 Richard Biener <rguenther@suse.de>
12858 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12859 (dwarf2out_early_finish): Move setting of AT_pubnames from
12860 early debug output to early finish.
12862 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12864 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12865 and -mdata-region to the assembler.
12867 2017-08-30 Richard Biener <rguenther@suse.de>
12869 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12871 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12872 (dwarf2out_early_global_decl): Only generate a DIE for the
12873 abstract origin if it doesn't already exist or is a declaration DIE.
12874 (resolve_addr): Do not add the linkage name twice when
12875 generating a stub DIE for the DW_TAG_GNU_call_site target.
12877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12879 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12880 Use machine_mode rather than int for arg1_mode.
12882 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12885 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12886 that the second argument of the built-in functions to unpack
12887 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12888 switch statement instead a lot of if statements.
12889 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12890 Allow 64-bit values to be in Altivec registers as well as
12891 traditional floating point registers.
12892 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12894 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12896 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12897 MAX_REGS_PER_ADDRESS == 1.
12899 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12901 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12902 * config/i386/i386.c (ix86_option_override_internal): Simplify
12903 setting of opts->x_flag_entry.
12905 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12906 Jakub Jelinek <jakub@redhat.com>
12907 Richard Biener <rguenther@suse.de>
12909 PR tree-optimization/81503
12910 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12911 folded constant fits in the target type; reorder tests for clarity.
12913 2017-08-29 Martin Liska <mliska@suse.cz>
12915 * passes.def: Include pass_lower_switch.
12916 * stmt.c (dump_case_nodes): Remove and move to
12917 tree-switch-conversion.
12918 (case_values_threshold): Likewise.
12919 (expand_switch_as_decision_tree_p): Likewise.
12920 (emit_case_decision_tree): Likewise.
12921 (expand_case): Likewise.
12922 (balance_case_nodes): Likewise.
12923 (node_has_low_bound): Likewise.
12924 (node_has_high_bound): Likewise.
12925 (node_is_bounded): Likewise.
12926 (emit_case_nodes): Likewise.
12927 (struct simple_case_node): New struct.
12928 (add_case_node): Remove.
12929 (emit_case_dispatch_table): Use vector instead of case_list.
12930 (reset_out_edges_aux): Remove.
12931 (compute_cases_per_edge): Likewise.
12932 (expand_case): Build list of simple_case_node.
12933 (expand_sjlj_dispatch_table): Use it.
12934 * tree-switch-conversion.c (struct case_node): Moved from
12935 stmt.c and adjusted.
12936 (emit_case_nodes): Likewise.
12937 (node_has_low_bound): Likewise.
12938 (node_has_high_bound): Likewise.
12939 (node_is_bounded): Likewise.
12940 (case_values_threshold): Likewise.
12941 (reset_out_edges_aux): Likewise.
12942 (compute_cases_per_edge): Likewise.
12943 (add_case_node): Likewise.
12944 (dump_case_nodes): Likewise.
12945 (balance_case_nodes): Likewise.
12946 (expand_switch_as_decision_tree_p): Likewise.
12947 (emit_jump): Likewise.
12948 (emit_case_decision_tree): Likewise.
12949 (try_switch_expansion): Likewise.
12950 (do_jump_if_equal): Likewise.
12951 (emit_cmp_and_jump_insns): Likewise.
12952 (fix_phi_operands_for_edge): New function.
12953 (record_phi_operand_mapping): Likewise.
12954 (class pass_lower_switch): New pass.
12955 (pass_lower_switch::execute): New function.
12956 (make_pass_lower_switch): Likewise.
12957 (conditional_probability):
12958 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12959 * tree-pass.h: Add make_pass_lower_switch.
12961 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12964 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12967 2017-08-29 Richard Biener <rguenther@suse.de>
12969 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12970 we do not add a DW_AT_inline attribute twice.
12971 (gen_subprogram_die): Remove code setting DW_AT_inline on
12972 DECL_ABSTRACT_P nodes.
12974 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12976 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12977 calls to internal functions.
12978 (gimplify_modify_expr): Likewise.
12979 * tree-call-cdce.c (use_internal_fn): Likewise.
12980 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12981 (convert_to_divmod): Set the nothrow flag.
12982 * tree-if-conv.c (predicate_mem_writes): Likewise.
12983 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12984 (vectorizable_call): Likewise.
12985 (vectorizable_store): Likewise.
12986 (vectorizable_load): Likewise.
12987 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12988 (vect_recog_mask_conversion_pattern): Likewise.
12990 2017-08-29 Martin Liska <mliska@suse.cz>
12993 * gcc.c (driver_handle_option): Add new argument.
12994 * opts-common.c (handle_option): Pass
12995 target_option_override_hook.
12996 * opts-global.c (lang_handle_option): Add new option.
12997 (set_default_handlers): Add new argument.
12998 (decode_options): Likewise.
12999 * opts.c (target_handle_option): Likewise.
13000 (common_handle_option): Call target_option_override_hook.
13001 * opts.h (struct cl_option_handler_func): Add hook for
13002 target option override.
13003 (struct cl_option_handlers): Likewise.
13004 (set_default_handlers): Add new argument.
13005 (decode_options): Likewise.
13006 (common_handle_option): Likewise.
13007 (target_handle_option): Likewise.
13008 * toplev.c (toplev::main): Pass targetm.target_option.override
13011 2017-08-29 Richard Biener <rguenther@suse.de>
13012 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13014 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13015 life to the active subtree.
13017 2017-08-28 Jeff Law <law@redhat.com>
13019 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13020 derive_equivalences.
13021 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13022 Code moved into....
13023 (edge_info::derive_equivalences): New private member function
13025 * tree-ssa-dom.c (class edge_info): Changed from a struct
13026 to a class. Add ctor/dtor, methods and data members.
13027 (edge_info::edge_info): Renamed from allocate_edge_info.
13028 Initialize additional members.
13029 (edge_info::~edge_info): New.
13030 (free_dom_edge_info): Delete the edge info.
13031 (record_edge_info): Use new class & associated member functions.
13032 Tighten forms for testing for edge equivalences.
13033 (record_temporary_equivalences): Iterate over the simple
13034 equivalences rather than assuming there's only one per edge.
13035 (cprop_into_successor_phis): Iterate over the simple
13036 equivalences rather than assuming there's only one per edge.
13037 (optimize_stmt): Use operand_equal_p rather than pointer
13038 equality for mini-DSE code.
13040 2017-08-28 Nathan Sidwell <nathan@acm.org>
13042 * gcc.c (execute): Fold SIGPIPE handling into switch
13043 statement. Adjust internal error message.
13045 2017-08-28 Richard Biener <rguenther@suse.de>
13048 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13049 Do nothing for removed DIEs.
13051 2017-08-28 Richard Biener <rguenther@suse.de>
13053 PR tree-optimization/81977
13054 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13057 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13060 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13061 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13062 using targetm.gen_mem_thread_fence.
13064 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13067 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13068 predicate to register_operand. Reorder operands.
13069 (*btr<mode>): Ditto.
13070 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13071 (*btr<mode>_mask): Ditto.
13073 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13075 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13076 * config/rs6000/xmmintrin.h: New file.
13077 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13079 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13082 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13083 parameter and_insn and return it.
13084 (recombine_lvx_pattern): Insert a copy to ensure availability of
13085 the base register of the copied masking operation at the point of
13086 the instruction replacement.
13087 (recombine_stvx_pattern): Likewise.
13089 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13091 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13092 undocumented switches.
13093 (-mpower9-dform-vector): Likewise.
13094 (-mpower9-dform): Likewise.
13095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13096 comments to delete references to -mpower9-dform* switches.
13097 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13098 Delete reference to -mpower9-dform* switches, test for
13099 -mpower9-vector instead.
13100 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13101 (OTHER_P9_VECTOR_MASKS): Likewise.
13102 (POWERPC_MASKS): Likewise.
13103 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13104 tests against -mpower9-dform* to -mpower9-vector. Delete code
13105 that checked for -mpower9-dform* consistancy with other options.
13106 Add test for -mpower9-misc to enable other power9 switches.
13107 (rs6000_init_hard_regno_mode_ok): Likewise.
13108 (rs6000_option_override_internal): Likewise.
13109 (rs6000_emit_prologue): Likewise.
13110 (rs6000_emit_epilogue): Likewise.
13111 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13112 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13113 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13115 (emit_fusion_p9_store): Likewise.
13116 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13117 resetting these macros if the assembler does not support ISA 3.0
13119 (TARGET_P9_DFORM_VECTOR): Likewise.
13120 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13121 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13123 2017-08-25 Alan Modra <amodra@gmail.com>
13125 PR rtl-optimization/81747
13126 * cse.c (cse_extended_basic_block): Don't attempt to record
13127 equivalences for degenerate conditional branches that branch
13128 to their fall-through.
13130 2017-08-24 Martin Sebor <msebor@redhat.com>
13132 PR middle-end/81908
13133 * gimple-fold.c (size_must_be_zero_p): New function.
13134 (gimple_fold_builtin_memory_op): Call it.
13136 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13138 * config/rs6000/mm_malloc.h: New file.
13140 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13142 PR tree-optimization/81913
13143 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13144 analysis when either IVs in condition can wrap.
13146 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13148 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13149 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13151 2017-08-24 Richard Biener <rguenther@suse.de>
13154 * targhooks.c (default_target_can_inline_p): Properly
13155 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13156 is present and always compare.
13157 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13158 infer -mfpmath=sse from TARGET_SSE_P.
13159 (ix86_can_inline_p): Properly use target_option_default_node when
13160 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13162 2017-08-24 Richard Biener <rguenther@suse.de>
13165 * dwarf2out.c (output_die): Handle flag_generate_offload like
13167 (output_comp_unit): Likewise.
13168 (gen_array_type_die): Likewise.
13169 (dwarf2out_early_finish): Likewise.
13170 (note_variable_value_in_expr): Likewise.
13171 (dwarf2out_finish): Likewise. Adjust assert.
13172 * cgraphunit.c (symbol_table::compile): Move setting of
13173 flag_generate_offload earlier ...
13174 (symbol_table::finalize_compilation_unit): ... here, before
13175 early debug finalization.
13177 2017-08-24 Richard Biener <rguenther@suse.de>
13179 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13180 and ipa-fnsummary.h.
13181 (ix86_can_inline_p): When ix86_fpmath flags do not match
13182 check whether the callee uses FP math at all.
13184 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13186 PR middle-end/81931
13187 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13188 instead of TYPE_PRECISION.
13190 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13191 Alan Hayward <alan.hayward@arm.com>
13192 David Sherwood <david.sherwood@arm.com>
13194 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13196 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13198 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13200 (ix86_frame::stack_realign_allocate): New field.
13201 (struct machine_frame_state): Modify comments.
13202 (machine_frame_state::sp_realigned_fp_end): New field.
13203 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13204 layout calculation.
13205 (sp_valid_at): Add assertion to assure no attempt to access invalid
13206 offset of a realigned stack.
13207 (fp_valid_at): Likewise.
13208 (choose_baseaddr): Modify comments.
13209 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13210 ix86_expand_prologue.
13211 (ix86_expand_prologue): Modify stack realignment and allocation.
13212 (ix86_expand_epilogue): Modify comments.
13213 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13214 avx2_runtime, avx512f, and avx512f_runtime.
13216 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13218 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13219 (mstackrealign): Do not init to -1.
13220 * config/i386/i386.c (ix86_option_override_internal):
13221 Check opts_set, not opts when setting default value of
13222 opts->x_ix86_force_align_arg_pointer.
13224 2017-08-23 Richard Biener <rguenther@suse.de>
13226 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13227 lang_hooks.decl_printable_name.
13228 * print-rtl-function.c (print_rtx_function): Likewise.
13229 * tree-pretty-print.c (dump_function_header): Likewise.
13231 2017-08-23 Richard Biener <rguenther@suse.de>
13234 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13235 -g0 at compile-time.
13237 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13239 PR middle-end/19706
13240 * doc/sourcebuild.texi (Other hardware attributes):
13243 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13245 PR middle-end/19706
13246 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13247 Fix single-use check.
13249 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13251 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13253 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13255 * doc/install.texi: Modify to add more details on running selected
13258 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13260 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13261 is combined with -mabi=ms.
13262 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13265 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13267 PR tree-optimization/81488
13268 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13269 and cached_basis fields.
13270 (MAX_SPREAD): New constant.
13271 (alloc_cand_and_find_basis): Initialize new fields.
13272 (clear_visited): New function.
13273 (create_phi_basis_1): Rename from create_phi_basis, set visited
13274 and cached_basis fields.
13275 (create_phi_basis): New wrapper function.
13276 (phi_add_costs_1): Rename from phi_add_costs, add spread
13277 parameter, set visited field, short-circuit when limits reached.
13278 (phi_add_costs): New wrapper function.
13279 (record_phi_increments_1): Rename from record_phi_increments, set
13281 (record_phi_increments): New wrapper function.
13282 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13283 (phi_incr_cost): New wrapper function.
13284 (all_phi_incrs_profitable_1): Rename from
13285 all_phi_incrs_profitable, set visited field.
13286 (all_phi_incrs_profitable): New wrapper function.
13288 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13289 Alan Hayward <alan.hayward@arm.com>
13290 David Sherwood <david.sherwood@arm.com>
13292 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13293 that takes the outer and inner modes.
13294 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13295 comparison as the canonical test for a paradoxical subreg.
13296 * combine.c (simplify_set): Use paradoxical_subreg_p.
13297 (make_extraction): Likewise.
13298 (force_to_mode): Likewise.
13299 (rtx_equal_for_field_assignment_p): Likewise.
13300 (gen_lowpart_for_combine): Likewise.
13301 (simplify_comparison): Likewise.
13302 * cse.c (equiv_constant): Likewise.
13303 * expmed.c (store_bit_field_1): Likewise.
13304 * final.c (alter_subreg): Likewise.
13305 * fwprop.c (propagate_rtx): Likewise.
13306 (forward_propagate_subreg): Likewise.
13307 * ira-conflicts.c (ira_build_conflicts): Likewise.
13308 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13309 * lra-constraints.c (curr_insn_transform): Likewise.
13310 (split_reg): Likewise.
13311 * lra-eliminations.c (move_plus_up): Likewise.
13312 (lra_eliminate_regs_1): Likewise.
13313 * recog.c (general_operand): Likewise.
13314 * ree.c (combine_reaching_defs): Likewise.
13315 * reload.c (push_reload): Likewise.
13316 (find_reloads): Likewise.
13317 * reload1.c (elimination_effects): Likewise.
13318 (compute_reload_subreg_offset): Likewise.
13319 (choose_reload_regs): Likewise.
13320 * rtlanal.c (subreg_lsb_1): Likewise.
13321 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13322 (simplify_subreg): Likewise.
13323 * var-tracking.c (track_loc_p): Likewise.
13324 * emit-rtl.c (byte_lowpart_offset): Likewise.
13325 (paradoxical_subreg_p): Delete out-of-line definition.
13327 2017-08-22 Jeff Law <law@redhat.com>
13329 PR tree-optimization/81741
13330 PR tree-optimization/71947
13331 * tree-ssa-dom.c: Include tree-inline.h.
13332 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13333 equivalences if one is more expensive to compute than the other.
13334 * tree-ssa-scopedtables.h (class const_or_copies): Make
13335 record_const_or_copy_raw method private.
13336 (class avail_exprs_stack): New method simplify_binary_operation.
13337 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13338 avail_exprs_stack::simplify_binary_operation as needed.
13339 (avail_exprs_stack::simplify_binary_operation): New function.
13341 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13343 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13344 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13345 (DOT_SYMBOLS): Likewise.
13346 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13347 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13348 (RS6000_ABI_NAME): Likewise.
13349 (TARGET_CMODEL): Likewise.
13350 (TOC_SECTION_ASM_OP): Likewise.
13351 (SET_CMODEL): New macro.
13352 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13354 2017-08-22 Richard Biener <rguenther@suse.de>
13356 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13357 to allow for free-lang-data replacements similar to verify_type_variant.
13359 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13360 Alan Hayward <alan.hayward@arm.com>
13361 David Sherwood <david.sherwood@arm.com>
13363 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13364 VOIDmode for the LABEL_REF.
13366 2017-08-22 Richard Biener <rguenther@suse.de>
13368 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13370 2017-08-22 Richard Biener <rguenther@suse.de>
13372 * common.opt (feliminate-dwarf2-dups): Ignore.
13373 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13374 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13375 same_die_p_wrap, compute_section_prefix,
13376 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13377 (comdat_symbol_id, comdat_symbol_number): Likewise.
13378 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13380 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13381 (output_die): Mark unreachable path unreachable.
13382 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13383 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13384 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13385 (dwarf2out_early_finish): Likewise.
13387 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13389 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13391 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13394 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13395 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13396 (avr_attribute_table) <io, io_low, address>: Initialize
13397 .decl_required with true.
13399 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13401 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13402 undocumented debugging options.
13403 (-mvsx-scalar-double): Likewise.
13404 (-mallow-df-permute): Likewise.
13405 (-mvectorize-builtins): Likewise.
13406 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13407 (rs6000_builtin_vectorized_function): Likewise.
13408 (rs6000_builtin_md_vectorized_function): Likewise.
13409 (rs6000_opt_vars): Likewise.
13411 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13414 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13415 (*btrq_imm): Rename from *btrq.
13416 (*btcq_imm): Rename from *btcq.
13417 (btsc): New code attribute.
13418 (*<btsc><mode>): New insn pattern.
13419 (*btr<mode>): Ditto.
13420 (*<btsc><mode>_mask): New insn_and_split pattern.
13421 (*btr<mode>_mask): Ditto.
13423 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13424 Alan Hayward <alan.hayward@arm.com>
13425 David Sherwood <david.sherwood@arm.com>
13427 * function.c (pad_below): Simplify padding calculation.
13429 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13430 Alan Hayward <alan.hayward@arm.com>
13431 David Sherwood <david.sherwood@arm.com>
13433 * target.def (function_prologue): Remove frame size argument.
13434 (function_epilogue): Likewise.
13435 * doc/tm.texi: Regenerate.
13436 * final.c (final_start_function): Update call to function_prologue.
13437 (final_end_function): Update call to function_epilogue.
13438 (default_function_pro_epilogue): Remove frame size argument.
13439 * output.h (default_function_pro_epilogue): Likewise.
13440 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13441 (arm_output_function_prologue): Likewise.
13442 * config/frv/frv.c (frv_function_prologue): Likewise.
13443 (frv_function_epilogue): Likewise.
13444 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13445 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13446 (ia64_output_function_epilogue): Likewise.
13447 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13448 (m32r_output_function_epilogue): Likewise.
13449 * config/microblaze/microblaze.c (microblaze_function_prologue)
13450 (microblaze_function_epilogue): Likewise.
13451 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13452 (mips_output_function_epilogue): Likewise.
13453 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13454 (mmix_target_asm_function_epilogue): Likewise.
13455 * config/msp430/msp430.c (msp430_start_function): Likewise.
13456 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13457 (nds32_asm_function_epilogue): Likewise.
13458 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13459 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13460 (pa_output_function_epilogue): Likewise.
13461 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13462 (rs6000_output_function_epilogue): Likewise.
13463 * config/rl78/rl78.c (rl78_start_function): Likewise.
13464 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13465 (rs6000_output_function_epilogue): Likewise.
13466 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13467 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13468 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13469 (sparc_asm_function_epilogue): Likewise.
13471 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13473 * tree.h (type_has_mode_precision_p): New function.
13474 * convert.c (convert_to_integer_1): Use it.
13475 * expr.c (expand_expr_real_2): Likewise.
13476 (expand_expr_real_1): Likewise.
13477 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13478 * match.pd: Likewise.
13479 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13480 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13481 * tree-tailcall.c (process_assignment): Likewise.
13482 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13483 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13484 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13485 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13486 (vectorizable_assignment): Likewise.
13487 (vectorizable_shift): Likewise.
13488 (vectorizable_operation): Likewise.
13489 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13491 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13493 * match.pd: Add pow (C, x) simplification.
13495 2017-08-21 Richard Biener <rguenther@suse.de>
13497 PR tree-optimization/81900
13498 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13499 for blocks with abnormal predecessors.
13500 (compute_antic): Do not set visited flag prematurely.
13502 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13505 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13507 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13509 * stor-layout.h (vector_type_mode): Move to...
13510 * tree.h (vector_type_mode): ...here.
13511 * stor-layout.c (vector_type_mode): Move to...
13512 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13514 2017-08-21 Richard Biener <rguenther@suse.de>
13516 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13517 register_external_die hooks.
13518 (debug_false_tree_charstarstar_uhwistar): Declare.
13519 (debug_nothing_tree_charstar_uhwi): Likewise.
13520 * debug.c (do_nothing_debug_hooks): Adjust.
13521 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13522 (debug_nothing_tree_charstar_uhwi): Likewise.
13523 * dbxout.c (dbx_debug_hooks): Adjust.
13524 (xcoff_debug_hooks): Likewise.
13525 * sdbout.c (sdb_debug_hooks): Likewise.
13526 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13527 * dwarf2out.c (macinfo_label_base): New global.
13528 (dwarf2out_register_external_die): New function for the
13529 register_external_die hook.
13530 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13531 (dwarf2_debug_hooks): Use them.
13532 (dwarf2_lineno_debug_hooks): Adjust.
13533 (struct die_struct): Add with_offset flag.
13534 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13535 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13536 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13537 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13538 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13539 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13540 defining section names for the early LTO debug variants.
13541 (reset_indirect_string): New helper.
13542 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13543 (print_dw_val): Add support for offsetted symbol references.
13544 (get_ultimate_context): Split out from is_cxx.
13545 (is_cxx): Use get_ultimate_context.
13546 (is_fortran): Add decl overload.
13547 (compute_comp_unit_symbol): Split out worker from
13548 compute_section_prefix.
13549 (compute_section_prefix): Call compute_comp_unit_symbol and
13550 set comdat_type_p here.
13551 (output_die): Skip DIE symbol output for the LTO added one.
13552 Handle DIE symbol references with offset.
13553 (output_comp_unit): Guard section name mangling properly.
13554 For LTO debug sections emit a symbol at the section beginning
13555 which we use to refer to its DIEs.
13556 (add_abstract_origin_attribute): For DIEs registered via
13557 dwarf2out_register_external_die directly refer to the early
13558 DIE rather than indirectly through the shadow one we created.
13559 Remove obsolete call to dwarf2out_abstract_function for
13560 non-function/block origins.
13561 (gen_array_type_die): When generating early LTO debug do
13562 not emit DW_AT_string_length.
13563 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13564 late when in LTO. As suggested place a gcc_unreachable for
13565 the DECL_ABSTRACT_P case.
13566 (gen_subprogram_die): Avoid another specification DIE
13567 for early built declarations/definitions for the late LTO case.
13568 (gen_variable_die): Add type references for late duplicated VLA dies
13570 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13571 we have the abstract instance already.
13572 (process_scope_var): Adjust decl DIE contexts in LTO which
13573 first puts them in limbo.
13574 (gen_decl_die): Do not generate type DIEs late apart from
13575 types for VLAs or for decls we do not yet have a DIE. Do not
13576 call dwarf2out_abstract_function late.
13577 (dwarf2out_early_global_decl): Make sure to create DIEs
13578 for abstract instances of a decl first.
13579 (dwarf2out_late_global_decl): Adjust comment.
13580 (output_macinfo_op): With multiple macro sections use
13581 macinfo_label_base to distinguish labels.
13582 (output_macinfo): Likewise. Update macinfo_label_base.
13583 Pass in the line info label.
13584 (note_variable_value_in_expr): When generating LTO resolve
13585 all variable values here by generating DIEs as needed.
13586 (init_sections_and_labels): Add early LTO debug flag parameter
13587 and generate different sections and names if set. Add generation
13588 counter for the labels so we can have multiple of them.
13589 (reset_dies): Helper to allow DIEs to be output multiple times.
13590 (dwarf2out_finish): When outputting DIEs to the fat part of an
13591 LTO object first reset DIEs.
13592 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13593 (modified_type_die): Check for decl_ultimate_origin being self
13595 (gen_type_die_with_usage): Likewise.
13596 (gen_typedef_die): Allow decl_ultimate_origin being self.
13597 (set_decl_abstract_flags): Remove.
13598 (set_block_abstract_flags): Likewise.
13599 (dwarf2out_abstract_function): Treat the early generated DIEs
13600 as the abstract copy and only add DW_AT_inline and
13601 DW_AT_artificial here and call set_decl_origin_self.
13602 If the DIE has an abstract origin don't do anything.
13603 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13604 if we have none yet (Go fails to build one, PR78628).
13605 (variably_modified_type_p): Prevent endless recursion for Ada
13606 cyclic pointer types.
13607 * lto-streamer-in.c: Include debug.h.
13608 (dref_queue): New global.
13609 (lto_read_tree_1): Stream in DIE references.
13610 (lto_input_tree): Register DIE references.
13611 (input_function): Stream DECL_DEBUG_ARGS.
13612 * lto-streamer-out.c: Include debug.h.
13613 (lto_write_tree_1): Output DIE references.
13614 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13615 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13616 (output_function): Stream DECL_DEBUG_ARGS.
13617 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13618 Stream DECL_ABSTRACT_ORIGIN.
13619 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13620 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13621 DECL_CONTEXT for file-scope decls.
13622 * lto-streamer.h (struct dref_entry): Declare.
13623 (dref_queue): Likewise.
13624 * cfgexpand.c (pass_expand::execute): Do not call the
13625 outlining_inline_function hook here.
13626 * lto-wrapper.c (debug_obj): New global.
13627 (tool_cleanup): Unlink it if required.
13628 (debug_objcopy): New function.
13629 (run_gcc): Handle early debug sections in the IL files by
13630 extracting them to separate files, partially linkin them and
13631 feeding the result back as result to the linker.
13632 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13633 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13634 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13635 sections into a separate segment.
13636 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13638 (darwin_asm_dwarf_section): Likewise.
13639 (darwin_asm_output_dwarf_offset): Likewise.
13640 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13642 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13643 Alan Hayward <alan.hayward@arm.com>
13644 David Sherwood <david.sherwood@arm.com>
13646 * read-md.h (md_reader::record_potential_iterator_use): Replace
13647 pointer argument with an rtx and an index.
13648 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13649 (apply_mode_iterator): Likewise.
13650 (apply_code_iterator): Likewise.
13651 (apply_int_iterator): Likewise.
13652 (apply_subst_iterator): Likewise.
13653 (record_iterator_use): Likewise.
13654 (record_attribute_use): Likewise.
13655 (md_reader::record_potential_iterator_use): Likewise. Update calls
13656 to record_iterator_use and apply_iterator.
13657 (iterator_use): Replace ptr with x and index.
13658 (attribute_use): Likewise.
13659 (apply_attribute_uses): Update calls to apply_iterator.
13660 (apply_iterators): Likewise. Update initialization of iterator_use.
13661 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13662 and record_potential_iterator_use.
13663 (rtx_reader::read_rtx_operand): Likewise.
13665 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13666 Alan Hayward <alan.hayward@arm.com>
13667 David Sherwood <david.sherwood@arm.com>
13669 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13672 2017-08-21 Richard Biener <rguenther@suse.de>
13674 PR middle-end/81884
13675 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13676 at struct end conservatively when comparing common bases.
13678 2017-08-21 Richard Biener <rguenther@suse.de>
13680 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13681 (mem_ref_in_stmt): Remove.
13682 (determine_max_movement): Use ref index to get at the reference.
13683 (invariantness_dom_walker::before_dom_children): Deal with
13684 lim data already initialized.
13685 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13687 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13689 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13690 (TARGET_ISA_ROUND): Ditto.
13691 (TARGET_ROUND): Ditto.
13692 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13693 * config/i386/i386.md: Ditto.
13694 * config/i386/sse.md: Ditto.
13695 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13696 with OPTION_MASK_ISA_SSE4_1.
13698 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13701 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13702 __builtin_ia32_lzcnt_u16.
13704 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13707 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13708 (rs6000_set_current_function): Rewrite function to use it.
13710 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13713 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13714 and TYPE_WARN_IF_NOT_ALIGN.
13715 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13716 (handle_warn_if_not_align): New.
13717 (place_union_field): Call handle_warn_if_not_align.
13718 (place_field): Call handle_warn_if_not_align.
13719 Copy TYPE_WARN_IF_NOT_ALIGN.
13720 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13721 (layout_type): Likewise.
13722 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13724 (tree_decl_common): Add warn_if_not_align.
13725 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13726 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13727 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13728 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13729 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13730 * doc/extend.texi: Document warn_if_not_aligned attribute.
13731 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13733 2017-08-17 Martin Liska <mliska@suse.cz>
13736 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13737 (get_data_dependence): Use it as pointer type.
13738 (distribute_loop): Likewise.
13740 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13742 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13743 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13744 (p8_vmrgow_<mode>_direct): New define_insn.
13745 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13746 handle endianness for vmrgew and vmrgow permute patterns.
13748 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13750 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13751 * config/rs6000/rs6000-cpus.def: Remove comment.
13752 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13753 (POWERPC_MASKS): Likewise.
13754 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13755 use of TARGET_VSX_TIMODE.
13756 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13757 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13758 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13759 (rs6000_option_override_internal): Remove dead code.
13760 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13761 (rs6000_legitimize_reload_address): Likewise.
13762 (rs6000_legitimate_address_p): Likewise.
13763 (rs6000_opt_masks): Delete "vsx-timode".
13764 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13765 from function comment.
13766 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13767 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13768 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13770 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13771 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13772 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13773 (VSX_TI): Likewise.
13775 (define_peephole2): Likewise.
13777 2017-08-17 Martin Sebor <msebor@redhat.com>
13780 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13781 past the end of an array.
13782 (test_pp_format): Add test cases.
13784 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13786 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13787 missing ECF_NOTHROW flags.
13789 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13792 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13793 operands residing in integer registers.
13794 (*vsx_le_perm_load_<mode>): Likewise.
13795 (*vsx_le_perm_store_<mode>): Likewise.
13796 (define_peephole2): Add peepholes to optimize the above.
13798 2017-08-17 Marek Polacek <polacek@redhat.com>
13800 PR middle-end/81814
13801 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13802 to mimic what shorten_compare did. Change the return type to bool.
13803 (fold_cond_expr_with_comparison): Update call to
13804 operand_equal_for_comparison_p.
13805 (fold_ternary_loc): Likewise.
13807 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13809 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13811 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13813 2017-08-17 Richard Biener <rguenther@suse.de>
13815 * tree-ssa-structalias.c (solve_graph): When propagating
13816 to successors update the graphs succ edges and avoid duplicate work.
13818 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13821 * config/i386/i386.c (ix86_option_override_internal): Save target
13822 specific options after ix86_stack_protector_guard_reg was changed.
13824 2017-08-17 Richard Biener <rguenther@suse.de>
13826 PR tree-optimization/81827
13827 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13828 (new_var_info): Initialize it conservatively.
13829 (get_call_vi): Mark register vars.
13830 (new_scalar_tmp_constraint_exp): Likewise.
13831 (handle_rhs_call): Likewise.
13832 (handle_const_call): Likewise.
13833 (create_function_info_for): Likewise.
13834 (solve_constraints): Sort varinfos to separate register from
13835 non-register vars to pack points-to solution bitmaps during
13838 2017-08-17 Marek Polacek <polacek@redhat.com>
13840 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13842 2017-08-17 Richard Biener <rguenther@suse.de>
13844 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13845 to true when overflow is undefined and we saturated the result.
13847 2017-08-17 Alan Modra <amodra@gmail.com>
13850 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13851 Don't use store multiple if only one reg needs saving.
13852 (interesting_frame_related_regno): New function.
13853 (rs6000_frame_related): Don't emit frame info for regs that
13855 (rs6000_emit_epilogue): Likewise.
13857 2017-08-16 Nathan Sidwell <nathan@acm.org>
13859 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13860 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13861 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13862 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13863 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13864 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13865 (verify_type): Adjust for TYPE_BINFO move.
13866 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13867 process TYPE_BINFO directly.
13868 (hash_tree): Likewise.
13869 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13871 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13874 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13876 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13878 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13881 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13882 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13883 Add ix86_binary_operator_ok to insn constraint.
13884 (*iordi_1_bts): Ditto.
13885 (*xordi_1_btc): Ditto.
13886 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13887 Update corresponding peephole2 pattern.
13891 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13893 PR tree-optimization/81832
13894 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13895 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13897 2017-08-16 Marek Polacek <polacek@redhat.com>
13900 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13901 regarding size_zero_node.
13903 2017-08-16 Martin Liska <mliska@suse.cz>
13906 * config.gcc: Respect previously set extra_objs in case
13909 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13911 PR tree-optimization/81835
13912 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13913 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13914 not depend on the phi.
13916 2017-08-16 Alan Modra <amodra@gmail.com>
13918 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13921 2017-08-16 Alan Modra <amodra@gmail.com>
13923 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13924 (save_reg_p): ..into this. Update all callers.
13925 (first_reg_to_save): Simplify.
13927 2017-08-16 Alan Modra <amodra@gmail.com>
13929 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13932 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13936 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13937 constant count if that count is less than 32.
13939 2017-08-15 Nathan Sidwell <nathan@acm.org>
13941 * gcc.c (execute): Emit friendlier message if inferior is killed
13942 by an external cause.
13944 2017-08-15 Richard Biener <rguenther@suse.de>
13946 PR tree-optimization/81790
13947 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13948 CONSTRUCTORs from simplifying and VN.
13950 2017-08-14 Martin Sebor <msebor@redhat.com>
13952 * builtin-attrs.def: Add comments.
13954 2017-08-14 Martin Sebor <msebor@redhat.com>
13957 * doc/extend.texi (attribute nonstring): Document new attribute.
13959 2017-08-14 Martin Sebor <msebor@redhat.com>
13962 * tree-diagnostic.c (default_tree_printer): Handle %G.
13963 * gimple-pretty-print.h (percent_G_format): Declare new function.
13964 * gimple-pretty-print.c (percent_G_format): Define.
13965 * tree-pretty-print.c (percent_K_format): Add argument.
13967 2017-08-14 Martin Sebor <msebor@redhat.com>
13969 PR translation/79998
13970 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13971 Remove a stray space.
13973 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13976 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13977 (*iordi_1_bts): Ditto.
13978 (*xordi_1_btc): Ditto.
13980 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13983 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13984 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13986 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13987 quoted strings, and make more translator-friendly.
13988 (darwin_rs6000_override_options): Likewise.
13989 (rs6000_option_override_internal): Likewise.
13990 (rs6000_return_in_memory): Fix overlong line.
13991 (init_cmulative_args): Use quoted strings, and make more
13992 translator-friendly.
13993 (rs6000_pass_by_reference): Fix overlong line.
13994 (def_builtin): Use quoted strings.
13995 (altivec_expand_predicate_builtin): Use quoted strings, and make
13996 more translator-friendly.
13997 (htm_expand_builtin): Use quoted strings.
13998 (cpu_expand_builtin): Use quoted strings, and make more
13999 translator-friendly.
14000 (altivec_expand_builtin): Likewise.
14001 (paired_expand_predicate_builtin): Likewise.
14002 (rs6000_invalid_builtin): Likewise.
14003 (builtin_function_type): Use quoted strings.
14004 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14005 more translator-friendly.
14006 (rs6000_trampoline_init): Likewise.
14007 (rs6000_handle_altivec_attribute): Likewise.
14008 (rs6000_inner_target_options): Use quoted strings.
14009 (rs6000_disable_incompatible_switches): Likewise.
14010 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14011 strings, and make more translator-friendly.
14012 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14014 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14016 PR tree-optimization/81799
14017 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14018 cond_expr to simple gimple operand.
14020 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14022 PR middle-end/46932
14023 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14025 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14029 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14030 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14031 TARGET_GASISR_PROLOGUES.
14032 * config/avr/avr.c (avr_option_override): Same.
14033 (avr_pass_pre_proep::execute): Same.
14035 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14038 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14039 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14041 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14043 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14044 define_insn_and_split. Split to a memory load from 0 in
14045 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14046 using PTR mode iterator.
14047 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14048 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14049 (*add_tp_<mode>): Redefine as define_insn_and_split.
14050 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14051 address space. Merge with *add_tp_x32 using PTR mode iterator.
14052 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14053 Split to an add with a memory load from 0 in
14054 DEFAULT_TLS_SEG_REG address space.
14056 2017-08-12 Andrew Pinski <apinski@cavium.com>
14058 * config/aarch64/aarch64-option-extensions.def (rdma):
14059 Fix feature string to what Linux prints out in /proc/cpuinfo.
14061 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14064 * dwarf2out.c (modified_type_die): For C typedef types that have
14065 an ultimate origin, process the ultimate origin instead of the
14067 (gen_typedef_die): Assert that input DECLs have no ultimate
14069 (gen_type_die_with_usage): For typedef variants that have an
14070 ultimate origin, just call gen_decl_die on the original DECL.
14071 (process_scope_var): Avoid creating DIEs for local typedefs and
14072 concrete static variables.
14074 2017-08-12 Alan Modra <amodra@gmail.com>
14078 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14079 match gnu-user.h startfile.
14080 (ENDFILE_LINUX_SPEC): Similarly.
14082 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14085 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14087 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14089 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14090 * config/aarch64/aarch64.md (mov<mode>): Change.
14091 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14092 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14093 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14095 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14097 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14098 for storage order barriers.
14100 2017-08-11 Martin Liska <mliska@suse.cz>
14103 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14104 variables of void type.
14106 2017-08-11 Martin Liska <mliska@suse.cz>
14108 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14109 TARGET_SUPPORTS_ALIASES.
14110 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14111 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14112 (optimize_weakref): Likewise.
14113 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14114 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14115 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14117 2017-08-11 Martin Liska <mliska@suse.cz>
14120 * config/i386/i386.c (make_resolver_func): Do complete
14121 refactoring of the function.
14123 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14126 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14127 * config/i386/i386.c (ix86_stack_protect_guard): Use
14128 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14129 * doc/invoke.texi (x86 Options): Document
14130 -mstack-protector-guard-symbol= option.
14132 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14134 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14135 * config/i386/i386.c (ix86_split_stack_guard): New function.
14136 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14137 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14138 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14139 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14140 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14141 (split_stack_space_check): Call ix86_split_stack_guard.
14143 2017-08-10 Martin Sebor <msebor@redhat.com>
14145 * print-tree.c (print_node): Print location using the established
14147 Replace spaces with colons.
14148 (debug_raw, debug): Ditto.
14150 2017-08-10 Martin Sebor <msebor@redhat.com>
14153 * pretty-print.c (pp_format): Correct the handling of %s precision.
14155 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14158 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14160 (ix86_finalize_stack_frame_flags): This. Also clear
14161 frame_pointer_needed if -fno-omit-frame-pointer is used without
14163 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14164 with ix86_finalize_stack_frame_flags.
14165 (ix86_expand_epilogue): Likewise.
14166 (ix86_expand_split_stack_prologue): Likewise.
14167 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14169 2017-08-10 Martin Liska <mliska@suse.cz>
14172 * c-attribs.c (handle_target_attribute):
14173 Report warning for an empty string argument of target attribute.
14175 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14178 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14180 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14181 or DECL_NONLOCAL labels.
14182 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14183 or DECL_NONLOCAL labels here.
14185 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14187 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14188 to indicate when early gimple folding has been disabled.
14189 (rs6000_gimple_fold_builtin): Add debug content.
14190 (rs6000_invalid_builtin): Fix whitespace.
14191 (rs6000_expand_builtin): Fix whitespace.
14192 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14194 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14197 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14198 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14200 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14202 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14203 (qdf24xx): Likewise.
14204 * config/aarch64/aarch64.md: Include falkor.md.
14205 * config/aarch64/falkor.md: New.
14207 2017-08-09 Marek Polacek <polacek@redhat.com>
14210 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14211 * diagnostic.c (emit_diagnostic): Add a comment.
14212 (emit_diagnostic_valist): New function.
14214 2017-08-09 Marek Polacek <polacek@redhat.com>
14217 * input.c (make_location): New overload.
14218 * input.h (make_location): Declare.
14220 2017-08-08 Alan Modra <amodra@gmail.com>
14221 H.J. Lu <hongjiu.lu@intel.com>
14224 * gcc.c (NO_PIE_SPEC): Delete.
14225 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14227 (LINK_PIE_SPEC): ..to here.
14228 (LINK_COMMAND_SPEC): Support -no-pie.
14229 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14230 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14232 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14233 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14234 (ENDFILE_CRTEND_SPEC): Similarly.
14236 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14239 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14240 (mstack-protector-guard-offset=): Ditto.
14241 * config/i386/i386.c (ix86_option_override): Handle
14242 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14244 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14245 ix86_stack_protect_guard_offset variables.
14246 (TARGET_STACK_PROTECT_GUARD): Always define.
14247 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14248 and -mstack-protector-guard-offset= options.
14250 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14252 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14253 boundary case for the last candidate.
14255 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14257 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14258 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14260 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14262 PR middle-end/19706
14263 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14264 * config/aarch64/aarch64-builtins.c
14265 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14266 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14267 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14269 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14270 Andrew Pinski <pinskia@gmail.com>
14272 PR middle-end/19706
14273 * internal-fn.def (XORSIGN): New.
14274 * optabs.def (xorsign_optab): New.
14275 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14276 (convert_expand_mult_copysign): New.
14277 (pass_optimize_widening_mul::execute): Call
14278 convert_expand_mult_copysign.
14280 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14282 PR tree-optimization/81354
14283 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14284 Insert on edges rather than explicitly creating landing pads.
14285 (analyze_candidates_and_replace): Commit edge inserts.
14287 2017-08-08 Richard Biener <rguenther@suse.de>
14289 PR middle-end/81719
14290 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14291 (expand_simple_operations): Also look through ADDR_EXPRs with
14292 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14294 2017-08-08 Richard Biener <rguenther@suse.de>
14296 PR tree-optimization/81723
14297 * tree-vect-slp.c (struct bst_traits): New hash traits.
14298 (bst_fail): New global.
14299 (vect_build_slp_tree_2): New worker, split out from ...
14300 (vect_build_slp_tree): ... this now wrapping it with using
14301 bst_fail set to cache SLP tree build fails. Properly handle
14303 (vect_analyze_slp_instance): Allocate and free bst_fail.
14305 2017-08-08 Martin Liska <mliska@suse.cz>
14308 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14309 LABEL_DECLs that can be from a different function.
14311 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14313 PR tree-optimization/81744
14314 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14315 loop's number of iterations.
14317 2017-08-08 Martin Liska <mliska@suse.cz>
14319 * asan.c: Include header files.
14320 * attribs.c (build_decl_attribute_variant): New function moved
14322 (build_type_attribute_qual_variant): Likewise.
14323 (cmp_attrib_identifiers): Likewise.
14324 (simple_cst_list_equal): Likewise.
14325 (omp_declare_simd_clauses_equal): Likewise.
14326 (attribute_value_equal): Likewise.
14327 (comp_type_attributes): Likewise.
14328 (build_type_attribute_variant): Likewise.
14329 (lookup_ident_attribute): Likewise.
14330 (remove_attribute): Likewise.
14331 (merge_attributes): Likewise.
14332 (merge_type_attributes): Likewise.
14333 (merge_decl_attributes): Likewise.
14334 (merge_dllimport_decl_attributes): Likewise.
14335 (handle_dll_attribute): Likewise.
14336 (attribute_list_equal): Likewise.
14337 (attribute_list_contained): Likewise.
14338 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14339 (lookup_attribute_by_prefix): Likewise.
14340 * bb-reorder.c: Include header files.
14341 * builtins.c: Likewise.
14342 * calls.c: Likewise.
14343 * cfgexpand.c: Likewise.
14344 * cgraph.c: Likewise.
14345 * cgraphunit.c: Likewise.
14346 * convert.c: Likewise.
14347 * dwarf2out.c: Likewise.
14348 * final.c: Likewise.
14349 * fold-const.c: Likewise.
14350 * function.c: Likewise.
14351 * gimple-expr.c: Likewise.
14352 * gimple-fold.c: Likewise.
14353 * gimple-pretty-print.c: Likewise.
14354 * gimple.c: Likewise.
14355 * gimplify.c: Likewise.
14356 * hsa-common.c: Likewise.
14357 * hsa-gen.c: Likewise.
14358 * internal-fn.c: Likewise.
14359 * ipa-chkp.c: Likewise.
14360 * ipa-cp.c: Likewise.
14361 * ipa-devirt.c: Likewise.
14362 * ipa-fnsummary.c: Likewise.
14363 * ipa-inline.c: Likewise.
14364 * ipa-visibility.c: Likewise.
14366 * lto-cgraph.c: Likewise.
14367 * omp-expand.c: Likewise.
14368 * omp-general.c: Likewise.
14369 * omp-low.c: Likewise.
14370 * omp-offload.c: Likewise.
14371 * omp-simd-clone.c: Likewise.
14372 * opts-global.c: Likewise.
14373 * passes.c: Likewise.
14374 * predict.c: Likewise.
14375 * sancov.c: Likewise.
14376 * sanopt.c: Likewise.
14377 * symtab.c: Likewise.
14378 * toplev.c: Likewise.
14379 * trans-mem.c: Likewise.
14380 * tree-chkp.c: Likewise.
14381 * tree-eh.c: Likewise.
14382 * tree-into-ssa.c: Likewise.
14383 * tree-object-size.c: Likewise.
14384 * tree-parloops.c: Likewise.
14385 * tree-profile.c: Likewise.
14386 * tree-ssa-ccp.c: Likewise.
14387 * tree-ssa-live.c: Likewise.
14388 * tree-ssa-loop.c: Likewise.
14389 * tree-ssa-sccvn.c: Likewise.
14390 * tree-ssa-structalias.c: Likewise.
14391 * tree-ssa.c: Likewise.
14392 * tree-streamer-in.c: Likewise.
14393 * tree-vectorizer.c: Likewise.
14394 * tree-vrp.c: Likewise.
14395 * tsan.c: Likewise.
14396 * ubsan.c: Likewise.
14397 * varasm.c: Likewise.
14398 * varpool.c: Likewise.
14399 * tree.c: Remove functions moved to attribs.[ch].
14400 * tree.h: Likewise.
14401 * config/aarch64/aarch64.c: Add attrs.h header file.
14402 * config/alpha/alpha.c: Likewise.
14403 * config/arc/arc.c: Likewise.
14404 * config/arm/arm.c: Likewise.
14405 * config/avr/avr.c: Likewise.
14406 * config/bfin/bfin.c: Likewise.
14407 * config/c6x/c6x.c: Likewise.
14408 * config/cr16/cr16.c: Likewise.
14409 * config/cris/cris.c: Likewise.
14410 * config/darwin.c: Likewise.
14411 * config/epiphany/epiphany.c: Likewise.
14412 * config/fr30/fr30.c: Likewise.
14413 * config/frv/frv.c: Likewise.
14414 * config/ft32/ft32.c: Likewise.
14415 * config/h8300/h8300.c: Likewise.
14416 * config/i386/winnt.c: Likewise.
14417 * config/ia64/ia64.c: Likewise.
14418 * config/iq2000/iq2000.c: Likewise.
14419 * config/lm32/lm32.c: Likewise.
14420 * config/m32c/m32c.c: Likewise.
14421 * config/m32r/m32r.c: Likewise.
14422 * config/m68k/m68k.c: Likewise.
14423 * config/mcore/mcore.c: Likewise.
14424 * config/microblaze/microblaze.c: Likewise.
14425 * config/mips/mips.c: Likewise.
14426 * config/mmix/mmix.c: Likewise.
14427 * config/mn10300/mn10300.c: Likewise.
14428 * config/moxie/moxie.c: Likewise.
14429 * config/msp430/msp430.c: Likewise.
14430 * config/nds32/nds32-isr.c: Likewise.
14431 * config/nds32/nds32.c: Likewise.
14432 * config/nios2/nios2.c: Likewise.
14433 * config/nvptx/nvptx.c: Likewise.
14434 * config/pa/pa.c: Likewise.
14435 * config/pdp11/pdp11.c: Likewise.
14436 * config/powerpcspe/powerpcspe.c: Likewise.
14437 * config/riscv/riscv.c: Likewise.
14438 * config/rl78/rl78.c: Likewise.
14439 * config/rx/rx.c: Likewise.
14440 * config/s390/s390.c: Likewise.
14441 * config/sh/sh.c: Likewise.
14442 * config/sol2.c: Likewise.
14443 * config/sparc/sparc.c: Likewise.
14444 * config/spu/spu.c: Likewise.
14445 * config/stormy16/stormy16.c: Likewise.
14446 * config/tilegx/tilegx.c: Likewise.
14447 * config/tilepro/tilepro.c: Likewise.
14448 * config/v850/v850.c: Likewise.
14449 * config/vax/vax.c: Likewise.
14450 * config/visium/visium.c: Likewise.
14451 * config/xtensa/xtensa.c: Likewise.
14453 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14456 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14457 constraints since the -mupper-regs-* switches have been
14459 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14460 into a vector from a double word element that was extracted from
14461 another vector, and eliminate extra XXPERMDI instructions.
14462 (vsx_concat_<mode>_2): Likewise.
14463 (vsx_concat_<mode>_3): Likewise.
14464 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14465 concat to allow optimizing inserts from previous extracts.
14467 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14469 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14470 memory reference to a SSP offset in TLS address space.
14471 (ix86_print_operand) <case '@'>: Remove.
14472 (ix86_print_operand_punct_valid_p): Remove '@' code.
14473 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14474 UNSPEC_SP_TLS_TEST.
14475 (stack_tls_protect_set_<mode>): Remove.
14476 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14477 (stack_tls_protect_test_<mode>): Remove.
14478 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14480 2017-08-07 Olivier Hainque <hainque@adacore.com>
14483 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14485 2017-08-07 Douglas Rupp <rupp@adacore.com>
14487 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14488 variable was referenced as multidir in command.
14490 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14493 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14496 2017-08-07 Martin Liska <mliska@suse.cz>
14498 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14499 * config/rl78/rl78.c: Add include of attribs.h.
14500 * config/sh/sh.c: Likewise.
14501 * config/v850/v850.c: Likewise.
14503 2017-08-07 Tom de Vries <tom@codesourcery.com>
14505 PR middle-end/78266
14506 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14508 2017-08-07 Martin Liska <mliska@suse.cz>
14510 * config/mips/mips.c: Include attribs.h.
14512 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14515 * doc/invoke.texi: Document change in behvaior for -Ofast for
14518 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14520 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14522 (aarch64_pop_regs): Likewise.
14523 (aarch64_gen_load_pair): Likewise.
14524 (aarch64_save_callee_saves): Likewise.
14525 (aarch64_restore_callee_saves): Likewise.
14527 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14529 * config/i386/i386.c: Revert the last change.
14531 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14534 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14536 (ix86_finalize_stack_frame_flags): This. Also clear
14537 frame_pointer_needed if -fno-omit-frame-pointer is used without
14539 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14540 with ix86_finalize_stack_frame_flags.
14541 (ix86_expand_epilogue): Likewise.
14542 (ix86_expand_split_stack_prologue): Likewise.
14544 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14547 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14548 to P_AES for Westmere.
14550 2017-08-07 Jonathan Yong <10walls@gmail.com>
14552 * config/i386/mingw.opt (fset-stack-executable): Removed.
14553 * config/i386/cygming.opt (fset-stack-executable): Moved
14555 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14557 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14559 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14561 2017-08-07 Marek Polacek <polacek@redhat.com>
14563 PR middle-end/81737
14564 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14566 2017-08-07 Martin Liska <mliska@suse.cz>
14568 * attribs.h (canonicalize_attr_name): New function.
14569 (cmp_attribs): Move from c-format.c and adjusted.
14570 (is_attribute_p): Moved from tree.h.
14571 * tree-inline.c: Add new includes.
14572 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14573 (private_is_attribute_p): Remove.
14574 (private_lookup_attribute): Likewise.
14575 (private_lookup_attribute_by_prefix): Simplify.
14576 (remove_attribute): Use is_attribute_p.
14577 * tree.h: Remove removed declarations.
14579 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14581 PR middle-end/81698
14582 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14583 instead of computing it in the function. Formatting fix.
14584 (expand_case): Don't rely on default_edge being the first edge,
14585 clear it if removing it, pass default_edge to
14586 emit_case_dispatch_table.
14587 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14590 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14592 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14593 insn in the function, emit NOP after the insn.
14595 2017-08-06 Tom de Vries <tom@codesourcery.com>
14597 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14600 2017-08-06 Tom de Vries <tom@codesourcery.com>
14602 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14605 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14607 PR tree-optimization/57371
14608 * match.pd: New pattern.
14610 2017-08-04 Marek Polacek <polacek@redhat.com>
14612 PR middle-end/81695
14613 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14614 perform the computation in offset_int.
14616 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14618 PR tree-optimization/81136
14619 * tree-vectorizer.h: Include tree-hash-traits.h.
14620 (vec_base_alignments): New typedef.
14621 (vec_info): Add a base_alignments field.
14622 (vect_record_base_alignments): Declare.
14623 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14625 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14626 (create_data_ref): Add an is_conditional_in_stmt argument.
14627 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14628 the is_conditional_in_stmt field.
14629 (data_ref_loc): Add an is_conditional_in_stmt field.
14630 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14631 (find_data_references_in_stmt): Update call to create_data_ref.
14632 (graphite_find_data_references_in_stmt): Likewise.
14633 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14634 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14635 (vect_record_base_alignment): New function.
14636 (vect_record_base_alignments): Likewise.
14637 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14638 for nested statements even if we fail to compute a misalignment.
14639 Use pooled base alignments for unconditional references.
14640 (vect_find_same_alignment_drs): Compare base addresses instead
14642 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14643 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14645 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14647 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14648 Add an explicit name for the enum. Use auto_vec for slp_instances
14649 and grouped_stores.
14650 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14652 (_bb_vec_info): Add a constructor and destructor.
14653 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14654 (destroy_loop_vec_info): Delete.
14655 (vect_destroy_datarefs): Likewise.
14656 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14657 (vec_info::vec_info): New function.
14658 (vec_info::~vec_info): Likewise.
14659 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14660 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14661 destroy_loop_vec_info.
14662 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14663 (_loop_vec_info::_loop_vec_info): ...this.
14664 (destroy_loop_vec_info): Replace with...
14665 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14666 the stmt_vec_infos. Leave handling of vec_info information to its
14667 destructor. Remove explicit vector releases.
14668 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14669 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14670 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14671 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14672 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14673 (destroy_bb_vec_info): Replace with...
14674 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14675 information to its destructor.
14676 (vect_slp_analyze_bb_1): Use new and delete instead of
14677 new_bb_vec_info and destroy_bb_vec_info.
14678 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14681 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14683 * tree-data-ref.h (subscript): Add access_fn field.
14684 (data_dependence_relation): Add could_be_independent_p.
14685 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14686 (same_access_functions): Move to tree-data-ref.c.
14687 * tree-data-ref.c (ref_contains_union_access_p): New function.
14688 (access_fn_component_p): Likewise.
14689 (access_fn_components_comparable_p): Likewise.
14690 (dr_analyze_indices): Add a reference to access_fn_component_p.
14691 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14693 (constant_access_functions): Likewise.
14694 (add_other_self_distances): Likewise.
14695 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14696 (initialize_data_dependence_relation): Use XCNEW and remove
14697 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14698 of access functions that have the same type. Allow the
14699 subsequence to end with different bases in some circumstances.
14700 Record the chosen access functions in SUB_ACCESS_FN.
14701 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14702 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14703 (subscript_dependence_tester_1): Likewise dra and drb.
14704 (build_classic_dist_vector): Update calls accordingly.
14705 (subscript_dependence_tester): Likewise.
14706 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14707 DDR_COULD_BE_INDEPENDENT_P.
14708 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14709 comp_alias_ddrs instead of may_alias_ddrs.
14710 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14712 (vect_analyze_data_ref_dependence): Use it if
14713 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14714 distance vectors if that fails.
14715 (dependence_distance_ge_vf): New function.
14716 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14717 LOOP_VINFO_MAY_ALIAS_DDRS.
14719 2017-08-04 Richard Biener <rguenther@suse.de>
14721 PR middle-end/81705
14722 * fold-const.c (fold_binary_loc): Properly restrict
14723 minus_var0 && minus_var1 case when associating undefined overflow
14726 2017-08-04 Tom de Vries <tom@codesourcery.com>
14728 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14730 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14732 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14733 Don't start diagnostic messages with a capital letter.
14734 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14736 (rs6000_invalid_builtin): Likewise.
14737 (rs6000_trampoline_init): Likewise.
14739 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14742 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14743 after setting changeable df flags.
14745 2017-08-03 Richard Biener <rguenther@suse.de>
14747 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14748 up if the use is in USE - X.
14750 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14752 * toplev.c (dumpfile.h): New include.
14753 (internal_error_reentered): New static function. Use it...
14754 (internal_error_function): ...here to handle reentered internal_error.
14756 2017-08-03 Richard Biener <rguenther@suse.de>
14758 PR middle-end/81148
14759 * fold-const.c (split_tree): Add minus_var and minus_con
14760 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14761 here but always use minus_*.
14762 (associate_trees): Assert we never associate with MINUS_EXPR
14763 and NULL first operand. Do not recurse for PLUS_EXPR operands
14764 when associating as MINUS_EXPR either.
14765 (fold_binary_loc): Track minus_var and minus_con.
14767 2017-08-03 Tom de Vries <tom@codesourcery.com>
14770 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14771 ACCEL_COMPILER, apply finish_options on
14772 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14774 2017-08-03 Tom de Vries <tom@codesourcery.com>
14777 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14778 function_entry_patch_area_size > 0.
14780 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14783 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14784 instead of 10??LU, perform unit multiplication in wide_int,
14785 don't change alloc_object_size_limit if the limit is larger
14788 PR tree-optimization/81655
14789 PR tree-optimization/81588
14790 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14791 the case when ranges[i].low and high are 1 for unsigned type with
14794 PR middle-end/81052
14795 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14796 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14798 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14800 * tree-vrp.h: Add include guard.
14802 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14805 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14806 (ud2): New insn pattern.
14807 * config/i386/i386.c (ix86_expand_epilogue):
14808 For naked functions, generate ud2 instead of trap insn.
14810 2017-08-02 Marek Polacek <polacek@redhat.com>
14813 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14815 2017-08-02 Tom de Vries <tom@codesourcery.com>
14816 Cesar Philippidis <cesar@codesourcery.com>
14818 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14819 Add missing edge probabilities.
14821 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14823 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14824 Correct endianness.
14826 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14828 PR middle-end/79499
14829 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14830 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14831 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14832 prologue_seq sequences - if any.
14834 2017-08-02 Richard Biener <rguenther@suse.de>
14836 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14837 via vectors if supported, integer extracts via punning if supported
14838 or otherwise vector extracts.
14840 2017-08-02 Richard Biener <rguenther@suse.de>
14842 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14844 (bitmap_insert_into_set): ... this.
14846 2017-08-02 Richard Biener <rguenther@suse.de>
14848 PR tree-optimization/81633
14850 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14852 PR tree-optimization/71752
14853 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14855 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14857 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14858 (machine_function::call_ms2sysv_pad_out): Remove field.
14859 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14860 (ix86_compute_frame_layout): Likewise.
14862 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14865 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14866 attribute with interrupt attribute.
14868 2017-08-01 Andrew Pinski <apinski@cavium.com>
14870 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14871 BIT_INSERT_EXPR's operand 1
14872 to see if the types precision matches.
14874 2017-08-01 Martin Liska <mliska@suse.cz>
14876 PR middle-end/70140
14877 * builtins.c (expand_builtin_memcpy_args): Remove.
14878 (expand_builtin_memcpy): Call newly added function
14879 expand_builtin_memory_copy_args.
14880 (expand_builtin_memcpy_with_bounds): Likewise.
14881 (expand_builtin_mempcpy): Remove last argument.
14882 (expand_builtin_mempcpy_with_bounds): Likewise.
14883 (expand_builtin_memory_copy_args): New function created from
14884 expand_builtin_mempcpy_args with small modifications.
14885 (expand_builtin_mempcpy_args): Remove.
14886 (expand_builtin_stpcpy): Remove unused argument.
14887 (expand_builtin): Likewise.
14888 (expand_builtin_with_bounds): Likewise.
14890 2017-08-01 Martin Liska <mliska@suse.cz>
14893 Make mempcpy more optimal (PR middle-end/70140).
14895 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14899 __builtin_vec_cmpne verify both arguments are compatible vectors
14900 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14901 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14902 move computation of aligned to after checking the argument types.
14906 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14907 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14910 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14911 Doug Rupp <rupp@adacore.com>
14912 Olivier Hainque <hainque@adacore.com>
14914 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14915 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14917 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14918 for TARGET_OS_CPP_BUILTIN.
14919 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14920 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14922 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14923 passing required abi options to the assembler for EABI configurations.
14924 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14925 of .text.hot and .text.unlikely sections for kernel modules when
14926 using ARM style exceptions.
14927 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14928 options. Add EXTRA_CC1_SPEC.
14929 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14931 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14933 (ARM_TARGET2_DWARF_FORMAT): Define.
14934 * config/arm/t-vxworks: Adjust multilib control to removal of the
14935 Diab command line options.
14937 2017-08-01 Martin Liska <mliska@suse.cz>
14939 PR gcov-profile/81561
14940 * gcov.c (unblock): Make unblocking safe as we need to preserve
14941 index correspondence of blocks and block_lists.
14943 2017-08-01 Richard Biener <rguenther@suse.de>
14945 PR tree-optimization/81181
14946 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14947 (compute_antic): ... end of iteration here.
14949 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14951 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14952 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14953 (ftree-slp-vectorize): Likewise.
14954 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14955 can no longer be set independent of flag_tree_loop_vectorize.
14956 * omp-general.c (emp_max_vf): Likewise.
14957 * opts.c (enable_fdo_optimizations): Remove references to
14958 flag_tree_vectorize, these are now implicit.
14959 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14960 and leave it for the options machinery.
14962 2017-08-01 Martin Liska <mliska@suse.cz>
14964 PR middle-end/70140
14965 * builtins.c (expand_builtin_memcpy_args): Remove.
14966 (expand_builtin_memcpy): Call newly added function
14967 expand_builtin_memory_copy_args.
14968 (expand_builtin_memcpy_with_bounds): Likewise.
14969 (expand_builtin_mempcpy): Remove last argument.
14970 (expand_builtin_mempcpy_with_bounds): Likewise.
14971 (expand_builtin_memory_copy_args): New function created from
14972 expand_builtin_mempcpy_args with small modifications.
14973 (expand_builtin_mempcpy_args): Remove.
14974 (expand_builtin_stpcpy): Remove unused argument.
14975 (expand_builtin): Likewise.
14976 (expand_builtin_with_bounds): Likewise.
14978 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14981 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14982 print "ds:" only for immediates in generic address space.
14984 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14987 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14988 (ix86_function_ok_for_sibcall): Return false for naked functions.
14990 2017-08-01 Richard Biener <rguenther@suse.de>
14992 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14993 (compute_antic): Seed worklist with exit block predecessors.
14994 * cfganal.c (dfs_find_deadend): For a cycle return the source
14995 of the edge closing it.
14997 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14999 * config/aarch64/aarch64.c
15000 (aarch64_can_const_movi_rtx_p): Move 0 check.
15002 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15004 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15005 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15007 * match.pd: Ditto in address comparison pattern.
15009 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15011 PR tree-optimization/81627
15012 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15013 closed ssa form for store-store chain.
15015 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15017 PR tree-optimization/81620
15018 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15019 for store-store chain.
15021 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15023 PR tree-optimization/81588
15024 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15025 ranges[i].in_p, invert comparison code ccode. For >/>=,
15026 swap rhs1 and rhs2 and comparison code unconditionally,
15027 for </<= don't do that. Don't swap rhs1/rhs2 again if
15028 ranges[i].in_p, instead invert comparison code ccode if
15029 opcode or oe->rank is BIT_IOR_EXPR.
15032 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15033 a direct optab to conversion optab.
15034 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15035 with GET_MODE_INNER as last argument instead of optab_handler.
15036 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15037 vector extraction if possible and optab is available.
15038 * expr.c (store_constructor): Use convert_optab_handler instead
15039 of optab_handler. Use vector initialization from smaller
15040 vectors if possible and optab is available.
15041 * tree-vect-stmts.c (vectorizable_load): Likewise.
15042 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15043 now have two modes.
15044 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15045 of vec_init from half-sized vectors with the same element mode.
15046 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15047 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15048 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15049 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15050 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15051 after mode in gen_vec_extract* calls.
15052 (vec_extract<mode>): Renamed to ...
15053 (vec_extract<mode><ssescalarmodelower>): ... this.
15054 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15055 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15056 element mode after mode in gen_vec_init* calls.
15057 (VEC_INIT_HALF_MODE): New mode iterator.
15058 (vec_init<mode>): Renamed to ...
15059 (vec_init<mode><ssescalarmodelower>): ... this.
15060 (vec_init<mode><ssehalfvecmodelower>): New expander.
15061 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15062 (vec_extractv2sfsf): ... this.
15063 (vec_initv2sf): Renamed to ...
15064 (vec_initv2sfsf): ... this.
15065 (vec_extractv2si): Renamed to ...
15066 (vec_extractv2sisi): ... this.
15067 (vec_initv2si): Renamed to ...
15068 (vec_initv2sisi): ... this.
15069 (vec_extractv4hi): Renamed to ...
15070 (vec_extractv4hihi): ... this.
15071 (vec_initv4hi): Renamed to ...
15072 (vec_initv4hihi): ... this.
15073 (vec_extractv8qi): Renamed to ...
15074 (vec_extractv8qiqi): ... this.
15075 (vec_initv8qi): Renamed to ...
15076 (vec_initv8qiqi): ... this.
15077 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15078 (vec_init<mode>): Renamed to ...
15079 (vec_init<mode><VEC_base_l>): ... this.
15080 (vec_extract<mode>): Renamed to ...
15081 (vec_extract<mode><VEC_base_l>): ... this.
15082 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15083 (vec_initv2sfsf): ... this.
15084 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15085 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15086 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15087 element mode after mode in gen_vec_init* calls.
15088 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15089 (vec_init<mode><Vel>): ... this.
15090 (vec_extract<mode>): Renamed to ...
15091 (vec_extract<mode><Vel>): ... this.
15092 * config/aarch64/iterators.md (Vel): New mode attribute.
15093 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15094 Add element mode after mode in gen_vec_extract* calls.
15095 * config/s390/vector.md (non_vec_l): New mode attribute.
15096 (vec_extract<mode>): Renamed to ...
15097 (vec_extract<mode><non_vec_l>): ... this.
15098 (vec_init<mode>): Renamed to ...
15099 (vec_init<mode><non_vec_l>): ... this.
15100 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15101 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15103 * config/arm/iterators.md (V_elem_l): New mode attribute.
15104 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15105 (vec_extract<mode><V_elem_l>): ... this.
15106 (vec_extractv2di): Renamed to ...
15107 (vec_extractv2didi): ... this.
15108 (vec_init<mode>): Renamed to ...
15109 (vec_init<mode><V_elem_l>): ... this.
15110 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15111 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15112 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15113 Add element mode after gen_vec_extract* calls.
15114 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15115 (vec_init<mode><unitmode>): ... this.
15116 (vec_extract<mode>): Renamed to ...
15117 (vec_extract<mode><unitmode>): ... this.
15118 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15119 (vec_init<mode><unitmode>): ... this.
15120 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15121 (vec_initv2sfsf): ... this.
15122 (vec_extractv2sf): Renamed to ...
15123 (vec_extractv2sfsf): ... this.
15124 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15125 Add element mode after gen_vec_extract* calls.
15126 * config/mips/mips.md (unitmode): New mode iterator.
15127 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15128 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15129 * config/spu/spu.md (inner_l): New mode attribute.
15130 (vec_init<mode>): Renamed to ...
15131 (vec_init<mode><inner_l>): ... this.
15132 (vec_extract<mode>): Renamed to ...
15133 (vec_extract<mode><inner_l>): ... this.
15134 * config/sparc/sparc.md (veltmode): New mode iterator.
15135 (vec_init<VMALL:mode>): Renamed to ...
15136 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15137 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15138 (vec_initv2sisi): ... this.
15139 (vec_initv2sf): Renamed to ...
15140 (vec_initv2sfsf): ... this.
15141 (vec_extractv2sf): Renamed to ...
15142 (vec_extractv2sfsf): ... this.
15143 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15144 (vec_init<mode>): Renamed to ...
15145 (vec_init<mode><VEC_base_l>): ... this.
15146 (vec_extract<mode>): Renamed to ...
15147 (vec_extract<mode><VEC_base_l>): ... this.
15148 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15149 (vec_initv2sfsf): ... this.
15150 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15151 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15152 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15153 gen_vec_init* calls.
15155 2017-08-01 Richard Biener <rguenther@suse.de>
15157 PR tree-optimization/81297
15158 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15159 TREE_OVERFLOW from INTEGER_CSTs.
15161 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15163 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15165 2017-07-31 Carl Love <cel@us.ibm.com>
15167 * config/rs6000/rs6000-c: Add support for built-in functions
15168 vector signed char vec_xl_be (signed long long, signed char *);
15169 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15170 vector signed int vec_xl_be (signed long long, signed int *);
15171 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15172 vector signed long long vec_xl_be (signed long long, signed long long *);
15173 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15174 vector signed short vec_xl_be (signed long long, signed short *);
15175 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15176 vector double vec_xl_be (signed long long, double *);
15177 vector float vec_xl_be (signed long long, float *);
15178 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15179 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15180 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15182 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15183 (altivec_expand_builtin): Add switch statement to call
15184 altivec_expand_xl_be for each builtin.
15185 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15186 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15187 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15188 __builtin_vsx_le_be_v16qi.
15189 * doc/extend.texi: Update the built-in documentation file for the
15190 new built-in functions.
15192 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15195 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15197 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15199 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15201 * config.gcc: Add z14.
15202 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15203 CPU model numbers for z13s and z14.
15204 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15206 * config/s390/s390-opts.h (enum processor_type): Rename
15207 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15208 * config/s390/s390.c (processor_table): Add field for CPU name to
15209 be passed to Binutils.
15210 (s390_asm_output_machine_for_arch): Use the new field in
15211 processor_table for Binutils.
15212 (s390_expand_builtin): Replace arch12 with z14.
15213 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15214 (s390_get_sched_attrmask): Likewise.
15215 (s390_get_unit_mask): Likewise.
15216 * config/s390/s390.opt: Add z14 to processor_type enum.
15218 2017-07-31 Martin Jambor <mjambor@suse.cz>
15221 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15222 regardless of optimization level.
15224 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15225 Martin Liska <mliska@suse.cz>
15227 * predict.def: Remove old comment and adjust probability.
15228 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15229 PREDICT statements.
15231 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15234 * config/i386/i386.c (ix86_function_naked): New function.
15235 (ix86_can_use_return_insn_p): Return false for naked functions.
15236 (ix86_expand_prologue): Skip prologue for naked functions.
15237 (ix86_expand_epilogue): Skip epilogue for naked functions
15238 and emit trap instruction.
15239 (ix86_warn_func_return): New function.
15240 (ix86_attribute_table): Add "naked" attribute specification.
15241 (TARGET_WARN_FUNC_RETURN): Define.
15242 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15244 2017-07-31 Martin Liska <mliska@suse.cz>
15246 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15247 (dump_gimple_bb_header): Always dump BB info.
15248 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15250 2017-07-31 Martin Liska <mliska@suse.cz>
15253 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15254 also with current_function_decl non-null equality.
15256 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15259 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15260 change type to the element type, instead add eltype variable and
15261 use it where we are interested in the element type.
15263 PR tree-optimization/81603
15264 * ipa-polymorphic-call.c
15265 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15266 offset arithmetic in offset_int, bail out if the resulting bit offset
15267 doesn't fit into shwi.
15269 2017-07-31 Martin Liska <mliska@suse.cz>
15271 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15272 (gimplify_save_expr): Fix comment.
15274 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15277 * config/i386/i386.c (ix86_function_arg): Update arguments for
15279 (ix86_compute_frame_layout): Set the initial stack offset to
15280 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15281 INCOMING_FRAME_SP_OFFSET.
15282 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15283 stack before exception handler returns.
15284 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15285 the 'ERROR_CODE' for exception handler.
15287 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15289 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15290 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15291 (ASM_OUTPUT_REG_POP): Ditto.
15292 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15293 instead of asm_fprintf to output pure string.
15295 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15297 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15298 to imported_module_or_decl hook.
15299 (debug_nothing_tree_tree_tree_bool): Remove.
15300 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15301 * debug.c (do_nothing_debug_hooks): Use
15302 debug_nothing_tree_tree_tree_bool_bool instead of
15303 debug_nothing_tree_tree_tree_bool.
15304 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15305 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15306 * sdbout.c (sdb_debug_hooks): Likewise.
15307 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15308 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15310 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15311 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15312 attribute, don't add anything.
15314 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15316 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15317 (fold_build2_stat_loc): Likewise.
15318 (fold_build3_stat_loc): Likewise.
15319 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15320 (fold_build1_loc): Remove macro.
15321 (fold_build2_loc): Likewise.
15322 (fold_build3_loc): Likewise.
15324 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15326 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15327 (gimple_build_debug_bind_source_stat): Likewise.
15328 * gimple.h (gimple_build_debug_bind): Remove macro.
15329 (gimple_build_debug_bind_source): Likewise.
15331 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15333 * bitmap.c (bitmap_alloc): Adjust.
15334 (bitmap_gc_alloc): Likewise.
15335 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15337 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15339 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15340 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15341 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15342 (bitmap_gc_alloc_stat): Likewise.
15343 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15345 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15347 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15348 * rtl.h (shallow_copy_rtx): Remove macro.
15350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15352 * emit-rtl.c (gen_raw_REG): Adjust.
15353 * gengenrtl.c (gendef): Likewise.
15354 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15355 * rtl.h (rtx_alloc): Remove macro.
15357 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15359 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15360 (build_tree_list_stat): Likewise.
15361 * tree.h (build_tree_list): Remove macro.
15362 (build_tree_list_vec): Likewise.
15364 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15366 * tree.c (make_vector_stat): Remove _stat from name.
15367 (build_vector_stat): Likewise.
15368 * tree.h (make_vector_stat): Remove macro.
15369 (build_vector_stat): Likewise.
15371 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15373 * tree.h (build_var_debug_value): Remove prototype.
15375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15377 * tree.c (tree_cons_stat): Remove _stat from name.
15378 * tree.h (tree_cons): Remove macro.
15380 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15382 * tree.c (build_vl_exp_stat): Remove _stat from name.
15383 * tree.h (build_vl_exp): Remove macro.
15385 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15387 * tree.c (build_decl_stat): Remove _stat from name.
15388 * tree.h (build_decl): Remove macro.
15390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15392 * gimple.c (gimple_build_with_ops_stat): Adjust.
15393 (gimple_alloc_stat): Remove _stat from name.
15394 * gimple.h (gimple_alloc): Remove macro.
15396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15398 * tree.c (make_tree_vec_stat): Remove _stat from name.
15399 (grow_tree_vec_stat): Likewise.
15400 * tree.h (make_tree_vec_stat): Adjust prototype.
15401 (grow_tree_vec_stat): Likewise.
15402 (make_tree_vec): Remove macro.
15403 (grow_tree_vec): Likewise.
15405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15407 * fold-const.c (fold_build1_stat_loc): Adjust.
15408 (fold_build2_stat_loc): Likewise.
15409 (fold_build3_stat_loc): Likewise.
15410 * tree.c (build0_stat): Remove _stat from name.
15411 (build1_stat): Likewise.
15412 (build2_stat): Likewise.
15413 (build3_stat): Likewise.
15414 (build4_stat): Likewise.
15415 (build5_stat): Likewise.
15416 * tree.h (build1_loc): Remove macro, and rename _stat function
15418 (build2_loc): Likewise.
15419 (build3_loc): Likewise.
15420 (build4_loc): Likewise.
15421 (build5_loc): Likewise.
15423 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15425 * tree.c (make_int_cst_stat): Remove _stat from name.
15426 * tree.h (make_int_cst_stat): Adjust prototype.
15427 (make_int_cst): Remove macro.
15429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15431 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15432 * tree.h (make_tree_binfo_stat): Adjust prototype.
15433 (make_tree_binfo): Remove.
15435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15437 * tree.c (copy_node_stat): Rename to copy_node.
15438 (build_distinct_type_copy): Adjust.
15439 * tree.h (copy_node_stat): Adjust prototype.
15440 (copy_node): Remove macro.
15442 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15444 * tree.c (make_node_stat): rename to make_node.
15445 (build_tree_list_stat): Adjust.
15446 (build0_stat): Likewise.
15447 (build2_stat): Likewise.
15448 (build3_stat): Likewise.
15449 (build4_stat): Likewise.
15450 (build5_stat): Likewise.
15451 (build_decl_stat): Likewise.
15452 * tree.h (make_node_stat): Adjust prototype.
15453 (make_node): remove macro.
15455 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15457 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15458 (PPC_FEATURE2_SCV): Likewise.
15459 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15461 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15463 * config/aarch64/aarch64.c
15464 (aarch64_internal_mov_immediate): Add new special pattern.
15465 * config/aarch64/aarch64.md (*movdi_aarch64):
15466 Add reg/32bit const mov case.
15468 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15469 Richard Sandiford <richard.sandiford@linaro.org>
15471 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15472 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15473 Add integer and movi cases.
15474 (movi-split-hf-df-sf split, fp16): New.
15475 (enabled): Added TARGET_FP_F16INST.
15476 * config/aarch64/iterators.md (GPF_HF): New.
15477 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15479 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15481 * config/aarch64/aarch64.c
15482 (aarch64_simd_container_mode): Add prototype.
15483 (aarch64_expand_mov_immediate): Add HI support.
15484 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15485 (aarch64_can_const_movi_rtx_p): New.
15486 (aarch64_preferred_reload_class):
15487 Remove restrictions of using FP registers for certain SIMD operations.
15488 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15489 (aarch64_valid_floating_const): Add integer move validation.
15490 (aarch64_simd_imm_scalar_p): Remove.
15491 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15492 (aarch64_legitimate_constant_p): Expand list of supported cases.
15493 * config/aarch64/aarch64-protos.h
15494 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15495 (aarch64_reinterpret_float_as_int): New.
15496 (aarch64_simd_imm_scalar_p): Remove.
15497 * config/aarch64/constraints.md (Uvi): New.
15498 (Dd): Split into Ds and new Dd.
15499 * config/aarch64/aarch64.md (*movsi_aarch64):
15501 (*movdi_aarch64): Add SIMD mov case.
15503 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15505 * tree-predcom.c: (struct chain): Handle store-store chain in which
15506 stores for elimination only store loop invariant values.
15507 (execute_pred_commoning_chain): Ditto.
15508 (prepare_initializers_chain_store_elim): Ditto.
15509 (prepare_finalizers): Ditto.
15510 (is_inv_store_elimination_chain): New function.
15511 (initialize_root_vars_store_elim_1): New function.
15513 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15515 * tree-predcom.c: Revise general description of the pass.
15516 (enum chain_type): New enum type for store elimination.
15517 (struct chain): New field supporting store elimination.
15518 (struct component): Ditto.
15519 (dump_chain): Dump store-stores chain.
15520 (release_chain): Release resources.
15521 (split_data_refs_to_components): Compute and create component
15522 contains only stores for elimination.
15523 (get_chain_last_ref_at): New function.
15524 (make_invariant_chain): Initialization.
15525 (make_rooted_chain): Specify chain type in parameter and record it.
15526 (add_looparound_copies): Skip for store-stores chain.
15527 (determine_roots_comp): Compute type of chain and pass it to
15529 (initialize_root_vars_store_elim_2): New function.
15530 (finalize_eliminated_stores): New function.
15531 (remove_stmt): Handle store for elimination.
15532 (execute_pred_commoning_chain): Execute predictive commoning on
15533 store-store chains.
15534 (determine_unroll_factor): Skip unroll for store-stores chain.
15535 (prepare_initializers_chain_store_elim): New function.
15536 (prepare_initializers_chain): Hanlde store-store chain.
15537 (prepare_finalizers_chain, prepare_finalizers): New function.
15538 (tree_predictive_commoning_loop): Return integer value indicating
15539 if loop is unrolled or lcssa form is corrupted.
15540 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15542 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15544 * tree-predcom.c (initialize_root): Delete.
15545 (execute_pred_commoning_chain): Initialize root vars and replace
15546 reference of non-combined chain directly, rather than call above
15549 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15551 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15552 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15554 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15556 * tree-predcom.c (struct chain): New field init_seq.
15557 (release_chain): Release init_seq.
15558 (prepare_initializers_chain): Record intialization stmts in above
15560 (insert_init_seqs): New function.
15561 (tree_predictive_commoning_loop): Call insert_init_seqs.
15563 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15565 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15567 2017-07-28 Richard Biener <rguenther@suse.de>
15569 * match.pd: Remove superfluous :c.
15570 * genmatch.c (simplify::id): Add member.
15571 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15573 (current_id): New global.
15574 (dt_node::parent): Move from ...
15575 (dt_operand::parent): ... here. Add for_id member.
15576 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15577 (decision_tree::find_node): Relax order requirement when
15578 merging DT_TRUE nodes to ones inbetween the current simplify
15579 and the one we try to merge with. Add diagnostic whenever
15580 we need to enforce pattern order by not merging.
15581 (decision_tree::insert): Set current_id.
15582 (decision_tree::print_node): Dump parent node and for_id.
15583 (parser::last_id): Add member.
15584 (parser::push_simplify): Assign unique id.
15585 (parser::parser): Initialize last_id.
15587 2017-07-28 Martin Liska <mliska@suse.cz>
15590 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15591 gimple_build_debug_bind.
15593 2017-07-28 Richard Biener <rguenther@suse.de>
15595 PR tree-optimization/81502
15596 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15598 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15600 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15601 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15602 for BIT_FIELD_REF args.
15603 * fold-const.c (make_bit_field_ref): Likewise.
15604 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15606 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15609 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15610 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15611 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15612 Or it into SANITIZER_UNDEFINED.
15613 * ubsan.c: Include gimple-fold.h and varasm.h.
15614 (ubsan_expand_ptr_ifn): New function.
15615 (instrument_pointer_overflow): New function.
15616 (maybe_instrument_pointer_overflow): New function.
15617 (instrument_object_size): Formatting fix.
15618 (pass_ubsan::execute): Call instrument_pointer_overflow
15619 and maybe_instrument_pointer_overflow.
15620 * internal-fn.c (expand_UBSAN_PTR): New function.
15621 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15622 * sanitizer.def (__ubsan_handle_pointer_overflow,
15623 __ubsan_handle_pointer_overflow_abort): New builtins.
15624 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15625 * internal-fn.def (UBSAN_PTR): New internal function.
15626 * opts.c (sanitizer_opts): Add pointer-overflow.
15627 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15628 * fold-const.c (build_range_check): Compute pointer range check in
15629 integral type if pointer arithmetics would be needed. Formatting
15632 2017-07-28 Martin Liska <mliska@suse.cz>
15635 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15636 parameters that are of a variable-length.
15638 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15640 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15642 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15643 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15644 (CRT_CALL_STATIC_FUNCTION): Likewise.
15645 (ASM_DEFAULT_SPEC): New define.
15646 (ASM_SPEC32): Likewise.
15647 (ASM_SPEC64): Likewise.
15648 (ASM_SPEC_COMMON): Likewise.
15649 (ASM_SPEC): Likewise.
15650 (INVALID_64BIT): Likewise.
15651 (LINK_OS_DEFAULT_SPEC): Likewise.
15652 (LINK_OS_SPEC32): Likewise.
15653 (LINK_OS_SPEC64): Likewise.
15654 (POWERPC_LINUX): Likewise.
15655 (PTRDIFF_TYPE): Likewise.
15656 (RESTORE_FP_PREFIX): Likewise.
15657 (RESTORE_FP_SUFFIX): Likewise.
15658 (SAVE_FP_PREFIX): Likewise.
15659 (SAVE_FP_SUFFIX): Likewise.
15660 (SIZE_TYPE): Likewise.
15661 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15662 (TARGET_64BIT): Likewise.
15663 (TARGET_64BIT): Likewise.
15664 (TARGET_AIX): Likewise.
15665 (WCHAR_TYPE_SIZE): Likewise.
15666 (WCHAR_TYPE): Undefine.
15667 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15668 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15669 (CPP_OS_RTEMS_SPEC): Delete.
15670 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15671 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15673 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15675 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15677 PR tree-optimization/81578
15678 * tree-parloops.c (build_new_reduction): Bail out if
15679 reduction_code isn't one of the standard OpenMP reductions.
15680 Move the details printing after that decision.
15682 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15684 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15685 related to reload_in_progress.
15686 (splat_input_operand): Likewise.
15687 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15689 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15691 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15692 (TARGET_INSTANTIATE_DECLS): Likewise.
15693 (legitimate_indexed_address_p): Delete reload_in_progress code.
15694 (rs6000_debug_legitimate_address_p): Likewise.
15695 (rs6000_eliminate_indexed_memrefs): Likewise.
15696 (rs6000_emit_le_vsx_store): Likewise.
15697 (rs6000_emit_move_si_sf_subreg): Likewise.
15698 (rs6000_emit_move): Likewise.
15699 (register_to_reg_type): Likewise.
15700 (rs6000_pre_atomic_barrier): Likewise.
15701 (rs6000_machopic_legitimize_pic_address): Likewise.
15702 (rs6000_allocate_stack_temp): Likewise.
15703 (rs6000_address_for_fpconvert): Likewise.
15704 (rs6000_address_for_altivec): Likewise.
15705 (rs6000_secondary_memory_needed_rtx): Delete function.
15706 (rs6000_check_sdmode): Likewise.
15707 (rs6000_alloc_sdmode_stack_slot): Likewise.
15708 (rs6000_instantiate_decls): Likewise.
15709 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15710 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15711 Delete reload_in_progress.
15712 (*vec_reload_and_plus_<mptrsize>): Likewise.
15713 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15714 (vsx_div_v2di): Likewise.
15715 (vsx_udiv_v2di): Likewise.
15717 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15719 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15720 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15721 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15722 (rs6000_debug_reg_global): Delete print of LRA status.
15723 (rs6000_option_override_internal): Delete dead LRA related code.
15724 (rs6000_lra_p): Delete function.
15725 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15727 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15729 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15730 * config/riscv/rtems.h: New file.
15732 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15733 Sudakshina Das <sudi.das@arm.com>
15735 * config/aarch64/aarch64.md
15736 (define_split for and<mode>3nr_compare): Move
15737 non aarch64_logical_operand to a register.
15738 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15739 register immediate operand to a register.
15740 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15742 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15744 PR middle-end/81564
15745 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15747 2017-07-27 Richard Biener <rguenther@suse.de>
15749 PR tree-optimization/81573
15750 PR tree-optimization/81494
15751 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15752 multi defuse cycle case.
15754 2017-07-27 Richard Biener <rguenther@suse.de>
15756 PR tree-optimization/81571
15757 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15760 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15762 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15763 earlier and only if MASK_FPU is set. Adjust formatting.
15765 2017-07-27 Martin Liska <mliska@suse.cz>
15767 * opt-functions.awk: Add validation of value of Init.
15768 * optc-gen.awk: Pass new argument.
15770 2017-07-27 Martin Liska <mliska@suse.cz>
15772 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15773 Fix wrong condition.
15775 2017-07-27 Martin Liska <mliska@suse.cz>
15777 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15778 BBs and edges seen by autoFDO.
15780 2017-07-27 Richard Biener <rguenther@suse.de>
15782 PR tree-optimization/81502
15783 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15784 with incompatible but same sized type.
15785 (execute_update_addresses_taken): Likewise.
15787 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15789 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15790 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15792 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15795 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15796 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15797 Change s_operand to memory_operand.
15799 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15801 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15802 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15803 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15804 Emit instructions rather than returning an expression. Handle TFmode
15805 and KFmode by casting to TImode.
15806 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15807 (rs6000_emit_le_vsx_store): Likewise.
15808 * config/rs6000/vsx.md (VSX_TI): New iterator.
15809 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15810 (*vsx_le_undo_permute_<mode>): Likewise.
15811 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15812 emit the split sequence.
15813 (*vsx_le_perm_store_<mode>): Likewise.
15815 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15817 PR tree-optimization/81555
15818 PR tree-optimization/81556
15819 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15820 if true, force CHANGED for the recursive invocation.
15821 (reassociate_bb): Remember original length of ops array, pass
15822 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15824 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15825 attributes for noipa attribute. For naked attribute use
15826 lookup_attribute first before lookup_attribute_spec.
15827 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15829 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15830 for functions with noipa attribute.
15831 (cgraph_externally_visible_p): Return true for functions with noipa
15833 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15834 for functions with noipa attribute.
15835 * doc/extend.texi: Document noipa function attribute.
15836 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15837 also for functions with noipa attribute.
15838 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15840 2017-07-26 Andrew Pinski <apinski@cavium.com>
15842 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15843 vec_unalign_load_cost and vec_unalign_store_cost.
15845 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15847 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15848 -mvsx-small-integer option.
15849 (ISA_3_0_MASKS_IEEE): Likewise.
15850 (OTHER_VSX_VECTOR_MASKS): Likewise.
15851 (POWERPC_MASKS): Likewise.
15852 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15853 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15854 code, only testing for DImode being allowed in non-VSX floating
15856 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15857 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15859 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15860 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15861 TARGET_P8_VECTOR test.
15862 (rs6000_secondary_reload_simple_move): Likewise.
15863 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15864 since TARGET_P9_VECTOR was already tested.
15865 (rs6000_opt_masks): Remove -mvsx-small-integer.
15866 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15867 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15869 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15870 test for TARGET_VEXTRACTUB was used, and that uses
15872 (p9 extract splitter): Likewise.
15873 (vsx_extract_<mode>_di_p9): Likewise.
15874 (vsx_extract_<mode>_store_p9): Likewise.
15875 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15876 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15877 the elimination of TARGET_VSX_SMALL_INTEGER.
15878 (vsx_extract_<mode>_p8): Likewise.
15879 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15880 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15881 (vsx_set_<mode>_p9): Likewise.
15882 (vsx_set_v4sf_p9): Likewise.
15883 (vsx_set_v4sf_p9_zero): Likewise.
15884 (vsx_insert_extract_v4sf_p9): Likewise.
15885 (vsx_insert_extract_v4sf_p9_2): Likewise.
15886 * config/rs6000/rs6000.md (sign extend splitter): Change
15887 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15888 (floatsi<mode>2_lfiwax_mem): Likewise.
15889 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15890 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15891 since a test for TARGET_P9_VECTOR was used.
15892 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15893 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15894 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15895 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15896 TARGET_P8_VECTOR test.
15897 (fix_trunc<mode>si2_stfiwx): Likewise.
15898 (fix_trunc<mode>si2_internal): Likewise.
15899 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15900 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15902 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15903 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15904 TARGET_P8_VECTOR test.
15905 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15906 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15907 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15909 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15910 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15911 since a test for TARGET_P9_VECTOR was used.
15912 (splitter for loading small constants): Likewise.
15914 2017-07-26 Andrew Pinski <apinski@cavium.com>
15916 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15919 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15922 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15923 (fp_valid_at): Likewise.
15925 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15927 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15928 (qdf24xx_addrcost_table): Likewise.
15929 (cortexa57_tunings): Update to use generic_branch_cost.
15930 (cortexa72_tunings): Likewise.
15931 (cortexa73_tunings): Likewise.
15932 (qdf24xx_tunings): Likewise.
15934 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15936 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15937 (thunderx2t99_branch_cost): Likewise.
15938 (cortexa35_tunings): Update to use generic_branch_cost.
15939 (cortexa53_tunings): Likewise.
15940 (cortexa57_tunings): Likewise.
15941 (cortexa72_tunings): Likewise.
15942 (cortexa73_tunings): Likewise.
15943 (thunderx2t99_tunings): Likewise.
15945 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15947 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15948 (sparc_option_override): Honour MASK_FSMULD.
15949 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15950 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15951 * config/sparc/sparc.opt (mfsmuld): New option.
15952 * doc/invoke.texi (mfsmuld): Document option.
15954 2017-07-26 Marek Polacek <polacek@redhat.com>
15956 PR middle-end/70992
15957 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15959 2017-07-26 Richard Biener <rguenther@suse.de>
15961 * gimple-match-head.c (do_valueize): Return OP if valueize
15963 (get_def): New helper to get at the def stmt of a SSA name
15964 if valueize allows.
15965 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15966 do_valueize to get at the def stmt.
15967 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15969 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15971 PR middle-end/46932
15972 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15974 2017-07-26 Martin Liska <mliska@suse.cz>
15977 * function.c (expand_function_start): Make expansion of
15978 nonlocal_goto_save_area after parm_birth_insn.
15980 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15982 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15983 from all CPU target flags enable members.
15985 2017-07-26 Richard Biener <rguenther@suse.de>
15987 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15988 (decision_tree::gen): Make 'type' const.
15989 (write_predicate): Likewise.
15991 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15993 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15994 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15995 (rs6000_option_override_internal): Likewise.
15996 (rs6000_expand_vector_set): Likewise.
15997 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15998 (TARGET_UPPER_REGS_SF): Likewise.
15999 (TARGET_UPPER_REGS_DI): Likewise.
16000 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16001 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16002 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16003 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16004 (Splitters for DI constants in Altivec registers): Likewise.
16005 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16006 (vsx_set_v4sf_p9): Likewise.
16007 (vsx_set_v4sf_p9_zero): Likewise.
16008 (vsx_insert_extract_v4sf_p9): Likewise.
16009 (vsx_insert_extract_v4sf_p9_2): Likewise.
16011 2017-07-25 Carl Love <cel@us.ibm.com>
16013 * doc/extend.texi: Update the built-in documentation file for the
16014 existing built-in functions
16015 vector signed char vec_cnttz (vector signed char);
16016 vector unsigned char vec_cnttz (vector unsigned char);
16017 vector signed short vec_cnttz (vector signed short);
16018 vector unsigned short vec_cnttz (vector unsigned short);
16019 vector signed int vec_cnttz (vector signed int);
16020 vector unsigned int vec_cnttz (vector unsigned int);
16021 vector signed long long vec_cnttz (vector signed long long);
16022 vector unsigned long long vec_cnttz (vector unsigned long long);
16024 2017-07-25 Andrew Pinski <apinski@cavium.com>
16026 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16027 accesses where the use is for the first operand of a BIT_INSERT.
16029 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16032 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16033 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16035 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16037 * config/i386/gstabs.h: Delete.
16038 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16040 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16042 * config/i386/i386.c (ix86_decompose_address): Do not check for
16043 register RTX when looking at index_reg or base_reg.
16044 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16046 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16048 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16049 to update EH info here.
16051 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16053 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16055 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16057 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16059 2017-07-25 Torsten Duwe <duwe@suse.de>
16061 * common.opt: Introduce -fpatchable-function-entry
16062 command line option, and its variables function_entry_patch_area_size
16063 and function_entry_patch_area_start.
16064 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16065 including a two-value parser.
16066 * target.def (print_patchable_function_entry): New target hook.
16067 * targhooks.h (default_print_patchable_function_entry): New function.
16068 * targhooks.c (default_print_patchable_function_entry): Likewise.
16069 * toplev.c (process_options): Switch off IPA-RA if
16070 patchable function entries are being generated.
16071 * varasm.c (assemble_start_function): Look at the
16072 patchable-function-entry command line switch and current
16073 function attributes and maybe generate NOP instructions by
16074 calling the print_patchable_function_entry hook.
16075 * doc/extend.texi: Document patchable_function_entry attribute.
16076 * doc/invoke.texi: Document -fpatchable_function_entry
16077 command line option.
16078 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16080 * doc/tm.texi: Re-generate.
16082 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16085 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16086 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16088 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16090 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16092 2017-07-25 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/81455
16095 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16096 not walk in cycles when looking for guards.
16098 2017-07-25 Richard Biener <rguenther@suse.de>
16100 PR tree-optimization/81529
16101 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16102 when optimizing backedge uses.
16104 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16106 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16108 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16109 to dl_section_ref. On AIX, append an expression to subtract
16110 the size of the section length to dl_section_ref.
16112 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16114 * configure.ac: If any of the config.* scripts fail, exit 1.
16115 * configure: Regenerate.
16117 2017-07-25 Richard Biener <rguenther@suse.de>
16119 PR middle-end/81546
16120 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16121 of immediate uses, be more verbose on errors.
16123 2017-07-25 Richard Biener <rguenther@suse.de>
16125 PR tree-optimization/81510
16126 * tree-vect-loop.c (vect_is_simple_reduction): When the
16127 reduction stmt is not inside the loop bail out.
16129 2017-07-25 Richard Biener <rguenther@suse.de>
16131 PR tree-optimization/81303
16132 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16133 profitability check against LOOP_VINFO_NITERSM1.
16135 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16137 * domwalk.c (cmp_bb_postorder): Simplify.
16138 (sort_bbs_postorder): New function. Use it...
16139 (dom_walker::walk): ...here to optimize common cases.
16141 2017-07-25 Martin Liska <mliska@suse.cz>
16144 * ipa-visibility.c (function_and_variable_visibility): Make the
16145 redirection just on target that supports aliasing.
16146 Fix GNU coding style.
16148 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16151 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16153 * config/arm/rtems.h: Likewise.
16154 * config/bfin/rtems.h: Likewise.
16155 * config/i386/rtemself.h: Likewise.
16156 * config/lm32/rtems.h: Likewise.
16157 * config/m32c/rtems.h: Likewise.
16158 * config/m68k/rtemself.h: Likewise.
16159 * config/microblaze/rtems.h: Likewise.
16160 * config/mips/rtems.h: Likewise.
16161 * config/moxie/rtems.h: Likewise.
16162 * config/nios2/rtems.h: Likewise.
16163 * config/powerpcspe/rtems.h: Likewise.
16164 * config/rs6000/rtems.h: Likewise.
16165 * config/rtems.h: Likewise.
16166 * config/sh/rtems.h: Likewise.
16167 * config/sh/rtemself.h: Likewise.
16168 * config/sparc/rtemself.h: Likewise.
16170 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16173 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16174 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16175 * tree-ssa-structalias.c (alias_get_name): Same.
16177 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16180 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16181 instructions if no du chain is found.
16183 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16185 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16187 2017-07-25 Richard Biener <rguenther@suse.de>
16189 PR middle-end/81505
16190 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16193 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16195 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16196 upper-regs options.
16197 (ISA_2_7_MASKS_SERVER): Likewise.
16198 (ISA_3_0_MASKS_IEEE): Likewise.
16199 (OTHER_P8_VECTOR_MASKS): Likewise.
16200 (OTHER_VSX_VECTOR_MASKS): Likewise.
16201 (POWERPC_MASKS): Likewise.
16202 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16203 duplicate list of options.
16204 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16205 explicit -mupper-regs options.
16206 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16207 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16208 alias for -mupper-regs-df.
16209 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16210 (rs6000_init_hard_regno_mode_ok): Likewise.
16211 (rs6000_option_override_internal): Likewise.
16212 (rs6000_opt_masks): Likewise.
16213 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16214 options in terms of whether -mvsx or -mpower8-vector was used.
16215 (TARGET_UPPER_REGS_DI): Likewise.
16216 (TARGET_UPPER_REGS_SF): Likewise.
16217 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16218 -mupper-regs-* options.
16220 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16222 * passes.c (emergency_dump_function): Print some empty lines and a
16223 header before the RTL dump.
16225 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16227 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16229 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16232 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16233 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16235 2017-07-24 Carl Love <cel@us.ibm.com>
16237 * config/rs6000/rs6000-c.c: Add support for built-in functions
16238 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16239 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16240 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16241 vec_extract_fp_from_shortl): Add defines for the two builtins.
16242 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16243 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16245 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16246 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16247 * doc/extend.texi: Update the built-in documentation file for the
16248 new built-in function.
16250 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16253 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16255 * doc/generic.texi: Likewise.
16256 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16257 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16259 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16261 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16262 (aarch64_mls_elt_merge<mode>): Likewise.
16264 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16266 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16267 having __cxa_atexit.
16269 2017-07-23 Michael Collison <michael.collison@arm.com>
16271 * config/arm/arm.c (arm_option_override): Deprecate
16272 use of -mstructure-size-boundary.
16273 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16274 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16276 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16279 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16280 Reduce cost estimate for direct moves.
16282 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16285 * config/i386/i386.c (ix86_option_override_internal): Disable
16286 BMI, BMI2 and TBM instructions for -m16.
16288 2017-07-21 Carl Love <cel@us.ibm.com>
16290 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16291 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16292 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16293 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16294 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16295 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16296 VMULOSW): New enum "unspec" values.
16297 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16298 altivec_vmulosw): New patterns.
16299 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16300 VMULOSW): Add definitions.
16302 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16304 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16305 (qdf24xx): Likewise.
16306 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16307 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16308 (AARCH64_FL_V8_1): Renumber.
16309 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16310 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16311 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16312 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16313 rdma to feature modifiers list.
16315 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16317 PR middle-end/56727
16318 * ipa-visibility (function_and_variable_visibility): Convert
16319 recursive PLT call to direct call if appropriate.
16321 2017-07-21 Andrew Pinski <apinski@cavium.com>
16323 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16324 operand 1 to see if the types precision matches.
16325 * fold-const.c (operand_equal_p): Likewise.
16327 2017-07-21 Richard Biener <rguenther@suse.de>
16329 PR tree-optimization/81303
16330 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16331 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16332 (vect_peeling_hash_get_lowest_cost): Adjust.
16333 (vect_enhance_data_refs_alignment): Likewise. Use
16334 vect_get_peeling_costs_all_drs to compute the penalty for no
16335 peeling to match up costs.
16337 2017-07-21 Richard Biener <rguenther@suse.de>
16339 PR tree-optimization/81500
16340 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16341 we didn't identify a reduction path.
16343 2017-07-21 Tom de Vries <tom@codesourcery.com>
16344 Cesar Philippidis <cesar@codesourcery.com>
16346 PR gcov-profile/81442
16347 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16350 2017-07-21 Tom de Vries <tom@codesourcery.com>
16353 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16355 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16356 nvptx_override_options_after_change.
16358 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16360 * dwarf2out.c (output_file_names): Avoid double testing for
16361 dwarf_version >= 5.
16363 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16365 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16367 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16369 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16371 (try_crossjump_to_edge): Do not punt on partitioned functions.
16373 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16375 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16376 Put all BBs reachable only via paths crossing cold region to cold
16378 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16380 2016-07-21 Richard Biener <rguenther@suse.de>
16382 PR tree-optimization/81303
16383 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16384 into account prologue and epilogue iterations when raising
16385 min_profitable_iters to sth at least covering one vector iteration.
16387 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16389 * config/arm/arm.c (arm_test_cpu_arch_dat):
16392 2017-07-20 Nathan Sidwell <nathan@acm.org>
16394 Remove TYPE_METHODS.
16395 * tree.h (TYPE_METHODS): Delete.
16396 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16397 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16398 (dbxout_type_methods): Scan TYPE_FIELDS.
16399 (dbxout_type): Don't check TYPE_METHODS here.
16400 * function.c (use_register_for_decl): Always ignore register for
16401 class types when not optimizing.
16402 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16403 * tree.c (free_lang_data_in_type): Stitch out member functions and
16404 templates from TYPE_FIELDS.
16405 (build_distinct_type_copy, verify_type_variant,
16406 verify_type): Member fns are on TYPE_FIELDS.
16407 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16408 * tree-pretty-print.c (dump_generic_node): Likewise.
16410 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16413 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16414 V2TImode and V4TImode.
16415 (ix86_expand_vector_extract): Likewise.
16416 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16417 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16418 (ssescalarmode): Handle V4TImode and V2TImode.
16419 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16420 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16421 (VEXTRACTI128_MODE): New mode iterator.
16422 (splitter for *vec_extractv?ti first element): New.
16423 (VEC_INIT_MODE): New mode iterator.
16424 (vec_init<mode>): Consolidate 3 expanders into one using
16425 VEC_INIT_MODE mode iterator.
16427 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16429 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16430 non_spilled_static_chain_regno_p.
16432 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16434 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16436 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16438 * bb-reorder.c (connect_traces): Allow copying of blocks within
16441 2017-07-20 Richard Biener <rguenther@suse.de>
16443 * gimple.h (gimple_phi_result): Add gphi * overload.
16444 (gimple_phi_result_ptr): Likewise.
16445 (gimple_phi_arg): Likewise. Adjust index assert to only
16446 allow actual argument accesses rather than all slots available
16448 (gimple_phi_arg_def): Add gphi * overload.
16449 * tree-phinodes.c (make_phi_node): Initialize only actual
16451 (resize_phi_node): Clear memory not covered by old node,
16452 do not initialize excess argument slots.
16453 (reserve_phi_args_for_new_edge): Initialize new argument slot
16456 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16458 PR tree-optimization/81388
16460 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16462 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16463 by removing computation of may_be_zero.
16465 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16466 Tom de Vries <tom@codesourcery.com>
16468 PR middle-end/81030
16469 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16470 when gimple level profile disagrees with what RTL expander did.
16472 2017-07-20 Richard Biener <rguenther@suse.de>
16474 PR tree-optimization/61171
16475 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16476 (vect_analyze_stmt): Add slp instance parameter.
16477 (vectorizable_reduction): Likewise.
16478 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16479 (vect_is_simple_reduction): Deal with chains not detected
16480 as SLP reduction chain, specifically not properly associated
16481 chains containing a mix of plus/minus.
16482 (get_reduction_op): Remove.
16483 (get_initial_defs_for_reduction): Simplify, pass in whether
16484 this is a reduction chain, pass in the SLP node for the PHIs.
16485 (vect_create_epilog_for_reduction): Get the SLP instance as
16487 (vectorizable_reduction): Get the SLP instance as arg.
16488 During analysis remember the SLP node with the PHIs in the
16489 instance. Simplify getting at the vectorized reduction PHIs.
16490 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16491 through SLP instance.
16492 (vect_slp_analyze_operations): Likewise.
16493 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16494 (vect_transform_stmt): Likewise.
16496 2017-07-20 Tom de Vries <tom@codesourcery.com>
16498 PR tree-optimization/81489
16499 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16500 read of phi arg location to before loop that modifies phi.
16502 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16504 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16507 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16509 PR middle-end/81331
16510 * except.c (execute): Fix ordering issue.
16512 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16514 PR rtl-optimization/81423
16515 * combine.c (make_compound_operation_int): Don't try to optimize
16516 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16518 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16520 PR rtl-optimization/81423
16521 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16522 with a constant that is -1 in the truncated to mode.
16524 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16526 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16527 (determine_unlikely_bbs): ... here.
16528 * predict.h (propagate_unlikely_bbs_forward): Declare.
16529 * cfgexpand.c (pass_expand::execute): Use it.
16530 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16532 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16533 propagate_unlikely_bbs_forward.
16535 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16537 PR middle-end/81331
16538 * except.c (maybe_add_nop_after_section_switch): New function.
16541 2017-07-19 Tom de Vries <tom@codesourcery.com>
16543 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16545 2017-07-19 Tom de Vries <tom@codesourcery.com>
16547 * gimple.h (gimple_phi_arg): Make assert more strict.
16549 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16551 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16552 * config/rs6000/mmintrin.h: New file.
16553 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16555 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16557 PR tree-optimization/81346
16558 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16560 2017-07-19 Tom de Vries <tom@codesourcery.com>
16562 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16564 2017-07-19 Tom de Vries <tom@codesourcery.com>
16566 * config/nvptx/nvptx-modes.def: Add V2DImode.
16567 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16568 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16569 (nvptx_output_mov_insn): Handle lack of mov.b128.
16570 (nvptx_print_operand): Handle 'H' and 'L' codes.
16571 (nvptx_vector_mode_supported): Allow V2DImode.
16572 (nvptx_preferred_simd_mode): New function.
16573 (nvptx_data_alignment): New function.
16574 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16575 nvptx_preferred_simd_mode.
16576 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16578 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16580 2017-07-19 Tom de Vries <tom@codesourcery.com>
16582 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16583 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16584 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16585 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16586 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16587 (mov<VECIM>_insn): New define_insn.
16588 (define_expand "mov<VECIM>): New define_expand.
16590 2017-07-19 Tom de Vries <tom@codesourcery.com>
16592 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16594 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16596 PR tree-optimization/81346
16597 * fold-const.h (fold_div_compare, range_check_type): Declare.
16598 * fold-const.c (range_check_type): New function.
16599 (build_range_check): Use range_check_type.
16600 (fold_div_compare): No longer static, rewritten into
16601 a match.pd helper function.
16602 (fold_comparison): Don't call fold_div_compare here.
16603 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16604 as helper function.
16606 2017-07-19 Nathan Sidwell <nathan@acm.org>
16608 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16609 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16610 * tree.c (find_decls_types_r, verify_type): Use
16611 TYPE_{MIN,MAX}_VALUE_RAW.
16612 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16613 (hash_tree): Likewise.
16614 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16616 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16619 2017-07-18 Tom de Vries <tom@codesourcery.com>
16621 PR middle-end/81464
16622 * omp-expand.c (expand_omp_for_static_chunk): Handle
16623 equal-argument loop exit phi.
16625 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16628 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16629 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16630 operand 2 predicate.
16631 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16632 operand 2 predicate.
16633 (ror,rol -> rorx splitters): Use const_int_operand as
16634 operand 2 predicate.
16636 2017-06-18 Richard Biener <rguenther@suse.de>
16638 PR tree-optimization/81410
16639 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16640 the gap in the ! slp_perm SLP case after each group.
16642 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16644 PR middle-end/81463
16645 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16648 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16650 PR middle-end/81462
16651 * predict.c (set_even_probabilities): Cleanup; do not affect
16652 probabilities that are already known.
16653 (combine_predictions_for_bb): Call even when count is set.
16655 2017-07-18 Nathan Sidwell <nathan@acm.org>
16657 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16660 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16663 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16664 optimization for loop niter analysis.
16666 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16669 * config/avr/avr.c (avr_optimize_casesi): Don't use
16670 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16672 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16674 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16675 body_cost_vec from _vect_peel_extended_info.
16676 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16677 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16680 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16682 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16684 2017-07-18 Richard Biener <rguenther@suse.de>
16686 PR tree-optimization/80620
16687 PR tree-optimization/81403
16688 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16689 info when re-using a VN table entry.
16691 2017-07-18 Richard Biener <rguenther@suse.de>
16693 PR tree-optimization/81418
16694 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16695 vectype_in. Verify that with lane-reducing reduction operations
16696 we have a single def-use cycle.
16698 2017-07-17 Carl Love <cel@us.ibm.com>
16700 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16702 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16703 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16704 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16705 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16706 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16707 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16708 VMULOSW): New enum "unspec" values.
16709 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16710 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16711 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16712 altivec_vmulosw): New patterns.
16713 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16714 VMULOSW): Add definitions.
16715 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16717 * config/alpha/alpha.c: Include predict.h.
16719 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16721 * tree-vrp.c (compare_assert_loc): Fix comparison function
16722 to return predictable results.
16724 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16726 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16728 (subdi3): Likewise.
16729 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16730 * doc/invoke.texi (mexpand-adddi): Update text.
16732 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16734 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16735 that also clobbers the CC register. The old expand code is moved
16737 (*arc_clzsi2): ... here.
16738 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16739 the CC register. The old expand code is moved to ...
16740 (arc_ctzsi2): ... here.
16742 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16744 * config/arc/arc.opt (mindexed-loads): Use initial value
16745 TARGET_INDEXED_LOADS_DEFAULT.
16746 (mauto-modify-reg): Use initial value
16747 TARGET_AUTO_MODIFY_REG_DEFAULT.
16748 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16749 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16750 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16751 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16753 2017-07-17 Martin Liska <mliska@suse.cz>
16756 * opts.c (finish_options): Do not allow -fgnu-tm
16757 w/ -fsanitize={kernel-,}address. Say sorry.
16759 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16762 * tree-loop-distribution.c (classify_partition): Only assert on
16763 numer of iterations.
16764 (merge_dep_scc_partitions): Delete prameter. Update function call.
16765 (distribute_loop): Remove code handling loop with unknown niters.
16766 (pass_loop_distribution::execute): Skip loop with unknown niters.
16768 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16771 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16772 function sort_partitions_by_post_order.
16774 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16776 PR tree-optimization/81374
16777 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16778 the max index of basic blocks, rather than number of basic blocks.
16780 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16782 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16784 (arc_legitimate_pic_operand_p): Likewise.
16785 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16787 (arc_needs_pcl_p): Likewise.
16788 (arc_legitimate_pc_offset_p): Likewise.
16789 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16790 function is also used in constrains.md.
16791 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16792 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16793 PLUS. Only return true/false in known cases, otherwise assert.
16794 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16795 is already called in arc_legitimate_constant_p.
16796 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16798 (LEGITIMATE_PIC_OPERAND_P): Use
16799 arc_raw_symbolic_reference_mentioned_p function.
16800 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16805 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16806 Andrew Burgess <andrew.burgess@embecosm.com>
16808 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16809 (arc_return_address_register): New function.
16810 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16811 (arc_handle_fndecl_attribute): Add naked attribute.
16812 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16813 (TARGET_WARN_FUNC_RETURN): Likewise.
16814 (arc_allocate_stack_slots_for_args): New function.
16815 (arc_warn_func_return): Likewise.
16816 (machine_function): Change type fn_type.
16817 (arc_compute_function_type): Consider new naked function type,
16818 change function return type.
16819 (arc_must_save_register): Adapt to handle new
16820 arc_compute_function_type's return type.
16821 (arc_expand_prologue): Likewise.
16822 (arc_expand_epilogue): Likewise.
16823 (arc_return_address_regs): Delete.
16824 (arc_return_address_register): New function.
16825 (arc_epilogue_uses): Use above function.
16826 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16827 (arc_function_type): Change encoding, add naked type.
16828 (ARC_INTERRUPT_P): Change to handle the new encoding.
16829 (ARC_FAST_INTERRUPT_P): Likewise.
16830 (ARC_NORMAL_P): Define.
16831 (ARC_NAKED_P): Likewise.
16832 (arc_compute_function_type): Delete prototype.
16833 * config/arc/arc.md (in_ret_delay_slot): Use
16834 arc_return_address_register function.
16835 (simple_return): Likewise.
16836 (p_return_i): Likewise.
16838 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16840 PR tree-optimization/81428
16841 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16842 can't be built for those types.
16844 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16846 Remove stuff dead since r239246.
16848 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16849 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16850 (avr_inform_devices): Remove dead stuff.
16852 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16854 * config/arm/arm_neon.h: Fix softp typo.
16856 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16858 PR tree-optimization/81365
16859 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16860 aggregate moves onto bb predecessor edges, make sure there are no
16861 loads that could alias the lhs in between the start of bb and the
16864 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16867 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16868 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16869 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16871 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16873 PR tree-optimization/81396
16874 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16875 (init_symbolic_number): Initialize it to 1.
16876 (perform_symbolic_merge): Add n_ops from both operands into the new
16878 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16879 without base_addr as useless if they need more than one operation.
16880 (bswap_replace): Handle !bswap case for NULL base_addr.
16882 2017-07-17 Tom de Vries <tom@codesourcery.com>
16885 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16888 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16890 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16891 conditional builtin define __FIX_LEON3FT_B2BST.
16893 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16895 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16896 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16899 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16901 PR rtl-optimization/81424
16902 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16903 to remove potential trapping from operands if -fnon-call-exceptions.
16905 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16907 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16908 profile_proability for scalling.
16909 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16911 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16913 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16915 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16917 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16918 fixpoint arithmetics.
16920 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16922 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16923 fixpoint arithmetics.
16925 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16927 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16928 fixpoint arithmetics.
16930 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16932 * profile-count.h (profile_probability::from_reg_br_prob_note,
16933 profile_probability::to_reg_br_prob_note): New functions.
16934 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16935 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16936 * predict.c (probability_reliable_p): Update.
16937 (edge_probability_reliable_p): Update.
16938 (br_prob_note_reliable_p): Update.
16939 (invert_br_probabilities): Update.
16940 (add_reg_br_prob_note): New function.
16941 (combine_predictions_for_insn): Update.
16942 * asan.c (asan_clear_shadow): Update.
16943 * cfgbuild.c (compute_outgoing_frequencies): Update.
16944 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16945 (update_br_prob_note): Update.
16946 (rtl_verify_edges): Update.
16947 (purge_dead_edges): Update.
16948 (fixup_reorder_chain): Update.
16949 * emit-rtl.c (try_split): Update.
16950 * ifcvt.c (cond_exec_process_insns): Update.
16951 (cond_exec_process_if_block): Update.
16952 (dead_or_predicable): Update.
16953 * internal-fn.c (expand_addsub_overflow): Update.
16954 (expand_neg_overflow): Update.
16955 (expand_mul_overflow): Update.
16956 * loop-doloop.c (doloop_modify): Update.
16957 * loop-unroll.c (compare_and_jump_seq): Update.
16958 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16959 * predict.h: Update.
16960 * reorg.c (mostly_true_jump): Update.
16962 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16963 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16964 * config/arc/arc.c: (emit_unlikely_jump): Update.
16965 * config/arm/arm.c: (emit_unlikely_jump): Update.
16966 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16967 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16968 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16969 (ix86_print_operand): Update.
16970 (ix86_split_fp_branch): Update.
16971 (predict_jump): Update.
16972 * config/ia64/ia64.c (ia64_print_operand): Update.
16973 * config/mmix/mmix.c (mmix_print_operand): Update.
16974 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16975 (rs6000_expand_split_stack_prologue): Update.
16976 * config/rs6000/rs6000.c: Update.
16977 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16978 (s390_expand_vec_movstr): Update.
16979 (s390_expand_cs_tdsi): Update.
16980 (s390_expand_split_stack_prologue): Update.
16981 * config/sh/sh.c (sh_print_operand): Update.
16982 (expand_cbranchsi4): Update.
16983 (expand_cbranchdi4): Update.
16984 * config/sparc/sparc.c (output_v9branch): Update.
16985 * config/spu/spu.c (get_branch_target): Update.
16986 (ea_load_store_inline): Update.
16987 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16988 * config/tilepro/tilepro.c: Update.
16990 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16992 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16993 (gimplify_save_expr): Likewise.
16995 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16997 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16999 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17001 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17003 * timevar.def (TV_IPA_FNSUMMARY): Define.
17005 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17007 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17008 to back store errata sensitive sequence from being generated.
17009 (sqrtdf2_fix): Likewise.
17011 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17013 * tree-ssa-threadupdate.c (compute_path_counts,
17014 update_joiner_offpath_counts): Use profile_probability.
17016 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17019 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17021 * config/arm/arm-c.c (arm_cpu_builtins): Define
17022 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17024 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17027 array entries to represent __ieee128 versions of the
17028 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17029 scalar_extract_sig, and scalar_insert_exp built-in functions.
17030 (altivec_resolve_overloaded_builtin): Add special case handling
17031 for the __builtin_scalar_insert_exp function, as represented by
17032 the P9V_BUILTIN_VEC_VSIEDP constant.
17033 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17034 exponent support for __ieee128 argument.
17035 (VSESQP): Add scalar extract signature support for __ieee128
17037 (VSTDCNQP): Add scalar test negative support for __ieee128
17039 (VSIEQP): Add scalar insert exponent support for __int128 argument
17040 with __ieee128 result.
17041 (VSIEQPF): Add scalar insert exponent support for __ieee128
17042 argument with __ieee128 result.
17043 (VSTDCQP): Add scalar test data class support for __ieee128
17045 (VSTDCNQP): Add overload support for scalar test negative with
17046 __ieee128 argument.
17047 (VSTDCQP): Add overload support for scalar test data class
17048 __ieee128 argument.
17049 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17050 UNSPEC_VSX_SXSIGDP.
17051 (UNSPEC_VSX_SIEXPQP): New constant.
17052 (xsxexpqp): New insn for VSX scalar extract exponent quad
17054 (xsxsigqp): New insn for VSX scalar extract significand quad
17056 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17057 precision with floating point argument.
17058 (xststdcqp): New expand for VSX scalar test data class quad
17060 (xststdcnegqp): New expand for VSX scalar test negative quad
17062 (xststdcqp): New insn to match expansions for VSX scalar test data
17063 class quad precision and VSX scalar test negative quad precision.
17064 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17065 special case operand checking to enforce that second operand of
17066 VSX scalar test data class with quad precision argument is a 7-bit
17068 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17069 prototypes and descriptions of __ieee128 versions of
17070 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17071 scalar_test_data_class, and scalar_test_neg built-in functions.
17073 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17075 PR tree-optimization/81162
17076 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17077 replace a negate with an add.
17079 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17081 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17083 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17085 * config/arm/arm-c.c (arm_cpu_builtins): Define
17086 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17088 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17090 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17091 (armv8-r): Set ARM Cortex-R52 as default CPU.
17092 * config/arm/arm-tables.opt: Regenerate.
17093 * config/arm/arm-tune.md: Regenerate.
17094 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17096 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17097 extension for -mcpu=cortex-r52.
17099 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17101 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17102 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17103 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17104 (fp-armv8): Define it as FP_ARMv8 only.
17105 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17106 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17108 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17109 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17110 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17111 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17112 than TARGET_FPU_ARMV8.
17113 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17114 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17115 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17117 * config/arm/neon.md (neon_vrint): Likewise.
17118 (neon_vcvt): Likewise.
17119 (neon_<fmaxmin_op><mode>): Likewise.
17120 (<fmaxmin><mode>3): Likewise.
17121 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17122 * config/arm/predicates.md (arm_cond_move_operator): Check against
17123 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17125 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17127 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17128 to top of function.
17130 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17132 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17133 loop in comment with memset.
17135 2017-07-14 Martin Liska <mliska@suse.cz>
17137 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17138 * dwarf2out.c (is_java): Remove the function.
17139 (output_pubname): Remove usage of the function.
17140 (lower_bound_default): Remove usage of DW_LANG_Java.
17141 (gen_compile_unit_die): Likewise.
17142 * gcc.c: Remove compiler defaults for .java and .zip files.
17143 * gimple-expr.c (remove_suffix): Change as there's no longer
17144 extension than 4-letter one.
17145 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17146 (gimplify_save_expr): Likewise.
17147 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17148 as it's possible even for other languages than Java.
17149 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17150 * lto-opts.c (lto_write_options): Remove reference to Java.
17151 * opts.c (strip_off_ending): Update file extension handling.
17152 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17153 * tree-eh.c (lower_resx): Likewise.
17154 * tree.c (free_lang_data_in_type): Remove dead code.
17155 (find_decls_types_r): Likewise.
17156 (build_common_builtin_nodes): Remove Java from a comment.
17157 (verify_type): Remove dead code.
17158 * varasm.c (assemble_external): Remove Java from a comment.
17160 2017-07-14 Martin Liska <mliska@suse.cz>
17162 * opts.c (finish_options): Add quotes.
17163 (common_handle_option): Likewise.
17165 2017-07-14 Martin Liska <mliska@suse.cz>
17167 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17168 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17169 Remove N_SO_PASCAL.
17170 * dwarf2out.c (lower_bound_default): Do not handle
17172 (gen_compile_unit_die): Likewise.
17173 * gcc.c: Remove default extension binding for GNU Pascal.
17174 * stmt.c: Remove Pascal language from a comment.
17175 * xcoffout.c: Likewise.
17177 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17180 * diagnostic-show-locus.c (fixit_cmp): New function.
17181 (layout::layout): Sort m_fixit_hints.
17182 (column_range::column_range): Assert that the values are valid.
17183 (struct char_span): New struct.
17184 (correction::overwrite): New method.
17185 (struct source_line): New struct.
17186 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17187 calls in terms of classes source_line and char_span, and
17188 correction::overwrite.
17189 (selftest::test_overlapped_fixit_printing_2): New function.
17190 (selftest::diagnostic_show_locus_c_tests): Call it.
17192 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17194 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17195 early if there is no lhs.
17197 2017-07-13 Martin Liska <mliska@suse.cz>
17199 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17200 (gen_reference_type_die): Likewise.
17201 * stor-layout.c: Remove Pascal-related comment.
17203 2017-07-13 Martin Liska <mliska@suse.cz>
17205 * opts.c (finish_options): Add quotes to error messages.
17206 (parse_sanitizer_options): Likewise.
17208 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17210 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17212 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17214 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17216 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17218 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17220 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17222 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17225 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17226 provides the hardware capability bits, define the macro
17227 __BUILTIN_CPU_SUPPORTS__.
17228 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17229 if GLIBC does not provide the hardware capability bits. Add a
17230 gcc_unreachable call if the built-in cpu function is neither
17231 __builtin_cpu_is nor __builtin_cpu_supports.
17232 (rs6000_get_function_versions_dispatcher): Change the warning
17233 that an old GLIBC is used which does not export the capability
17234 bits to be an error.
17235 * doc/extend.texi (target_clones attribute): Document the
17236 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17237 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17238 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17239 the macros defined by GCC if the newer GLIBC is available.
17241 2017-07-12 Jeff Law <law@redhat.com>
17243 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17244 remaining includes slightly.
17245 * config/riscv/riscv-builtins.c: Include profile-count.h.
17247 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17250 * config/avr/avr.c (avr_set_current_function): In diagnostic
17251 messages: Quote keywords and (parts of) identifiers.
17252 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17255 2017-07-12 Carl Love <cel@us.ibm.com>
17257 * config/rs6000/rs6000-c.c: Add support for built-in functions
17258 vector bool char vec_revb (vector bool char);
17259 vector bool short vec_revb (vector short char);
17260 vector bool int vec_revb (vector bool int);
17261 vector bool long long vec_revb (vector bool long long);
17262 * doc/extend.texi: Update the built-in documentation file for the
17263 new built-in functions.
17265 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17267 * config/s390/s390.md: Remove movcc splitter.
17269 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17271 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17272 load/store on condition.
17274 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17277 * config/avr/avr.c (avr_encode_section_info)
17278 [progmem && !TREE_READONLY]: Error if progmem object needs
17281 2017-07-11 Michael Collison <michael.collison@arm.com>
17283 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17286 2017-07-11 Carl Love <cel@us.ibm.com>
17288 * config/rs6000/rs6000-c.c: Add support for builtins
17289 vector unsigned int vec_parity_lsbb (vector signed int);
17290 vector unsigned int vec_parity_lsbb (vector unsigned int);
17291 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17292 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17293 vector unsigned long long vec_parity_lsbb (vector signed long long);
17294 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17295 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17296 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17297 * doc/extend.texi: Update the built-in documentation file for the
17298 new built-in functions.
17300 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17302 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17303 (layout::m_primary_loc): New field.
17304 (layout::layout): Initialize new field. Move location filtering
17305 logic from here to...
17306 (layout::maybe_add_location_range): ...this new method. Add
17307 support for filtering to just the lines already specified by other
17309 (layout::will_show_line_p): New method.
17310 (gcc_rich_location::add_location_if_nearby): New method.
17311 (selftest::test_add_location_if_nearby): New test function.
17312 (selftest::diagnostic_show_locus_c_tests): Call it.
17313 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17316 2017-07-11 Tom de Vries <tom@codesourcery.com>
17318 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17319 (bb_first_real_insn): New function.
17320 (nvptx_single): Add extra initialization of broadcasted condition
17323 2017-07-11 Nathan Sidwell <nathan@acm.org>
17325 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17327 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17329 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17330 Binutils doc as TEXI will mess them up.
17331 * doc/invoke.texi (AVR Options): Same here.
17333 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17335 * config/sparc/sparc.opt (mfix-ut700): New option.
17336 (mfix-gr712rc): Likewise.
17337 (sparc_fix_b2bst): New variable.
17338 * doc/invoke.texi (SPARC options): Document them.
17339 (ARM options): Fix warnings.
17340 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17341 instructions to prevent sequences that can trigger the store-store
17342 errata for certain LEON3FT processors.
17343 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17344 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17345 * config/sparc/sparc.md (fix_b2bst): New attribute.
17346 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17348 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17351 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17353 (*rsqrtsf2_sse): Ditto.
17355 (div<mode>3): Macroize insn from divdf3 and divsf3
17356 using MODEF mode iterator.
17358 2017-07-10 Martin Sebor <msebor@redhat.com>
17360 PR tree-optimization/80397
17361 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17362 instead of testing for equality to INTEGER_TYPE.
17364 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17366 * config.gcc: Remove uclibc from arc target spec.
17368 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17370 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17372 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17375 * lto-wrapper.c (remove_option): New function.
17376 (merge_and_complain): Merge PIC/PIE options more realistically.
17378 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17380 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17384 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17385 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17386 * config.in: Regenerate.
17387 * configure: Regenerate.
17388 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17389 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17390 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17391 (TARGET_GASISR_PROLOGUES): ...target mask.
17392 * common/config/avr/avr-common.c
17393 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17394 Set -mgas-isr-prologues.
17395 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17396 INSERT_PASS_BEFORE for it.
17397 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17398 * config/avr/avr.c (avr_option_override)
17399 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17400 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17401 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17402 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17403 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17405 (make_avr_pass_pre_proep): New function.
17406 (emit_push_sfr) <treg>: Add argument to function and use it
17407 instead of TMP_REG.
17408 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17409 and set machine->gasisr.yes.
17410 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17411 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17412 __gcc_isr.n_pushed to .L__stack_usage.
17413 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17414 (avr_asm_final_postscan_insn): ...this new static function.
17415 * config/avr/avr.h (machine_function)
17416 <is_no_gccisr, use_L__stack_usage>: New fields.
17417 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17418 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17419 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17420 (gasisr, *gasisr): New expander and insn.
17421 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17422 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17423 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17425 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17427 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17430 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17432 Move jump-tables out of .text again.
17435 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17436 (ASM_OUTPUT_ADDR_VEC): New function.
17437 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17438 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17439 INSN_ADDRESSes as asm comment.
17440 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17441 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17442 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17443 * config/avr/avr.md (*tablejump): Adjust comment.
17444 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17445 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17447 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17448 (avr_output_addr_vec): New proto.
17449 (avr_log_t) <insn_addresses>: New field.
17451 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17454 * config/i386/i386.c (ix86_function_arg_advance): Set
17455 outgoing_args_on_stack to true if there are outgoing arguments
17457 (ix86_function_arg): Likewise.
17458 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17459 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17460 * config/i386/i386.h (machine_function): Add
17461 outgoing_args_on_stack.
17463 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17465 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17466 supporting pthreds.
17467 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17469 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17471 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17472 (REAL_H): Remove $(MACHMODE_H).
17473 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17475 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17476 $(MACHMODE_H) and double-int.h.
17477 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17479 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17482 2017-07-07 Andrew Pinski <apinski@cavium.com>
17484 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17485 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17487 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17489 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17490 Add warning if GCC was not configured to link against a GLIBC that
17491 exports the hardware capability bits.
17492 (make_resolver_func): Make resolver function private and not a
17493 COMDAT function. Create the name with clone_function_name instead
17494 of make_unique_name.
17497 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17498 correct operand in doing the split.
17500 2017-07-07 Carl Love <cel@us.ibm.com>
17502 * config/rs6000/rs6000-c: Add support for built-in function
17503 vector unsigned short vec_pack_to_short_fp32 (vector float,
17505 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17506 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17507 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17508 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17509 (convert_4f32_8i16): Add define_expand.
17510 * doc/extend.texi: Update the built-in documentation file for the
17511 new built-in function.
17513 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17515 * config/sparc/m8.md: New file.
17516 * config/sparc/sparc.md: Include m8.md.
17518 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17520 * config/sparc/sparc.opt: New option -mvis4b.
17521 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17522 (sparc_option_override): Handle VIS4B.
17523 (enum sparc_builtins): Define
17524 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17525 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17526 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17527 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17528 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17529 (check_constant_argument): New function.
17530 (sparc_vis_init_builtins): Define builtins
17531 __builtin_vis_dictunpack{8,16,32},
17532 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17533 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17534 __builtin_vis_fpcmpde{8,16,32}shl and
17535 __builtin_vis_fpcmpur{8,16,32}shl.
17536 (sparc_expand_builtin): Check that the constant operands to
17537 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17538 constant and in range.
17539 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17541 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17542 (SPARC_IMM5_P): Likewise.
17543 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17544 (enabled): Handle vis4b.
17545 (UNSPEC_DICTUNPACK): New unspec.
17546 (UNSPEC_FPCMPSHL): Likewise.
17547 (UNSPEC_FPUCMPSHL): Likewise.
17548 (UNSPEC_FPCMPDESHL): Likewise.
17549 (UNSPEC_FPCMPURSHL): Likewise.
17550 (cpu_feature): New CPU feature `vis4b'.
17551 (dictunpack{8,16,32}): New insns.
17552 (FPCSMODE): New mode iterator.
17553 (fpcscond): New code iterator.
17554 (fpcsucond): Likewise.
17555 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17556 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17557 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17558 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17559 * config/sparc/constraints.md: Define constraints `q' for unsigned
17560 2-bit integer constants and `t' for unsigned 5-bit integer
17562 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17564 (imm5_operand_dictunpack16): Likewise.
17565 (imm5_operand_dictunpack32): Likewise.
17566 (imm2_operand): Likewise.
17567 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17568 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17569 ditunpack* and fpcmp*shl builtins.
17571 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17573 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17574 * config.in: Add HAVE_AS_SPARC6 define.
17575 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17577 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17579 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17580 (CPP_CPU_SPEC): Handle m8.
17581 (ASM_CPU_SPEC): Likewise.
17582 * config/sparc/sparc-opts.h (enum processor_type): Add
17584 * config/sparc/sparc.c (m8_costs): New struct.
17585 (sparc_option_override): Handle TARGET_CPU_m8.
17586 (sparc32_initialize_trampoline): Likewise.
17587 (sparc64_initialize_trampoline): Likewise.
17588 (sparc_issue_rate): Likewise.
17589 (sparc_register_move_cost): Likewise.
17590 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17591 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17592 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17593 (CPP_CPU_SPEC): Handle M8.
17594 (ASM_CPU_SPEC): Likewise.
17595 (AS_M8_FLAG): Define.
17596 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17597 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17598 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17600 * configure: Regenerate.
17601 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17604 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17606 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17608 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17609 ("*movdi_insn_sp32"): Do not set v3pipe.
17610 ("*movsi_insn"): Likewise.
17611 ("*movdi_insn_sp64"): Likewise.
17612 ("*movsf_insn"): Likewise.
17613 ("*movdf_insn_sp32"): Likewise.
17614 ("*movdf_insn_sp64"): Likewise.
17615 ("*zero_extendsidi2_insn_sp64"): Likewise.
17616 ("*sign_extendsidi2_insn"): Likewise.
17617 ("*mov<VM32:mode>_insn"): Likewise.
17618 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17619 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17620 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17621 ("<vlop:code><VL:mode>3"): Likewise.
17622 ("*not_<vlop:code><VL:mode>3"): Likewise.
17623 ("*nand<VL:mode>_vis"): Likewise.
17624 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17625 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17626 ("one_cmpl<VL:mode>2"): Likewise.
17627 ("faligndata<VM64:mode>_vis"): Likewise.
17628 ("alignaddrsi_vis"): Likewise.
17629 ("alignaddrdi_vis"): Likweise.
17630 ("alignaddrlsi_vis"): Likewise.
17631 ("alignaddrldi_vis"): Likewise.
17632 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17633 ("bmaskdi_vis"): Likewise.
17634 ("bmasksi_vis"): Likewise.
17635 ("bshuffle<VM64:mode>_vis"): Likewise.
17636 ("cmask8<P:mode>_vis"): Likewise.
17637 ("cmask16<P:mode>_vis"): Likewise.
17638 ("cmask32<P:mode>_vis"): Likewise.
17639 ("pdistn<P:mode>_vis"): Likewise.
17640 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17642 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17644 * config/sparc/sparc.md ("subtype"): New insn attribute.
17645 ("*wrgsr_sp64"): Set insn subtype.
17646 ("*rdgsr_sp64"): Likewise.
17647 ("alignaddrsi_vis"): Likewise.
17648 ("alignaddrdi_vis"): Likewise.
17649 ("alignaddrlsi_vis"): Likewise.
17650 ("alignaddrldi_vis"): Likewise.
17651 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17652 ("fexpand_vis"): Likewise.
17653 ("fpmerge_vis"): Likewise.
17654 ("faligndata<VM64:mode>_vis"): Likewise.
17655 ("bshuffle<VM64:mode>_vis"): Likewise.
17656 ("cmask8<P:mode>_vis"): Likewise.
17657 ("cmask16<P:mode>_vis"): Likewise.
17658 ("cmask32<P:mode>_vis"): Likewise.
17659 ("fchksm16_vis"): Likewise.
17660 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17661 ("fmean16_vis"): Likewise.
17662 ("fp<plusminus_insn>64_vis"): Likewise.
17663 ("<plusminus_insn>v8qi3"): Likewise.
17664 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17665 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17666 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17667 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17668 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17669 ("*movqi_insn"): Likewise.
17670 ("*movhi_insn"): Likewise.
17671 ("*movsi_insn"): Likewise.
17672 ("movsi_pic_gotdata_op"): Likewise.
17673 ("*movdi_insn_sp32"): Likewise.
17674 ("*movdi_insn_sp64"): Likewise.
17675 ("movdi_pic_gotdata_op"): Likewise.
17676 ("*movsf_insn"): Likewise.
17677 ("*movdf_insn_sp32"): Likewise.
17678 ("*movdf_insn_sp64"): Likewise.
17679 ("*zero_extendhisi2_insn"): Likewise.
17680 ("*zero_extendqihi2_insn"): Likewise.
17681 ("*zero_extendqisi2_insn"): Likewise.
17682 ("*zero_extendqidi2_insn"): Likewise.
17683 ("*zero_extendhidi2_insn"): Likewise.
17684 ("*zero_extendsidi2_insn_sp64"): Likewise.
17685 ("ldfsr"): Likewise.
17686 ("prefetch_64"): Likewise.
17687 ("prefetch_32"): Likewise.
17688 ("tie_ld32"): Likewise.
17689 ("tie_ld64"): Likewise.
17690 ("*tldo_ldub_sp32"): Likewise.
17691 ("*tldo_ldub1_sp32"): Likewise.
17692 ("*tldo_ldub2_sp32"): Likewise.
17693 ("*tldo_ldub_sp64"): Likewise.
17694 ("*tldo_ldub1_sp64"): Likewise.
17695 ("*tldo_ldub2_sp64"): Likewise.
17696 ("*tldo_ldub3_sp64"): Likewise.
17697 ("*tldo_lduh_sp32"): Likewise.
17698 ("*tldo_lduh1_sp32"): Likewise.
17699 ("*tldo_lduh_sp64"): Likewise.
17700 ("*tldo_lduh1_sp64"): Likewise.
17701 ("*tldo_lduh2_sp64"): Likewise.
17702 ("*tldo_lduw_sp32"): Likewise.
17703 ("*tldo_lduw_sp64"): Likewise.
17704 ("*tldo_lduw1_sp64"): Likewise.
17705 ("*tldo_ldx_sp64"): Likewise.
17706 ("*mov<VM32:mode>_insn"): Likewise.
17707 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17708 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17710 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17712 * config/sparc/sparc.md ("type"): New insn type viscmp.
17713 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17715 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17716 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17717 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17718 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17720 ("n7_vis_logical_11cycle"): Likewise.
17721 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17722 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17723 * config/sparc/niagara.md ("niag_vis"): Likewise.
17724 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17725 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17727 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17729 * config/sparc/sparc.md: New instruction type `bmask'.
17730 (bmaskdi_vis): Use the `bmask' type.
17731 (bmasksi_vis): Likewise.
17732 * config/sparc/ultra3.md (us3_array): Likewise.
17733 * config/sparc/niagara7.md (n7_array): Likewise.
17734 * config/sparc/niagara4.md (n4_array): Likewise.
17735 * config/sparc/niagara2.md (niag2_vis): Likewise.
17736 (niag3_vis): Likewise.
17737 * config/sparc/niagara.md (niag_vis): Likewise.
17739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17741 * ipa-comdats.c: Remove optimize check from gate.
17742 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17743 for functions not optimized.
17744 (ipa_fn_summary_read): Skip optimize check.
17745 (ipa_fn_summary_write): Likewise.
17746 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17748 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17750 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17751 (check_callers): Give up on caller which is not optimized.
17752 (inline_small_functions): Likewise.
17753 (ipa_inline): Do not give up when not optimizing.
17754 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17755 away unoptimizes cdtors.
17756 (whole_program_function_and_variable_visibility): Do
17757 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17758 * ipa.c (process_references): Do not check optimize.
17759 (symbol_table::remove_unreachable_nodes): Update optimize check.
17760 (set_writeonly_bit): Update optimize check.
17761 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17762 (pass_ipa_single_use::gate): Remove.
17764 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17766 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17767 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17768 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17769 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17770 permute_load, permute_store, adjust_extract, adjust_splat,
17771 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17772 replace_swap_with_copy, dump_swap_insn_table,
17773 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17774 recombine_lvx_pattern, recombine_stvx_pattern,
17775 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17776 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17777 to file rs6000-p8swap.c.
17778 * config/rs6000/rs6000-p8swap.c: New file.
17779 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17780 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17781 and rs6000*-*-* targets.
17783 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17785 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17787 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17789 * lto-wrapper.c (merge_and_complain): Do not merge
17790 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17791 fsigned_zeros, ftrapping_math, fwrapv.
17792 (append_compiler_options): Do not track these options.
17793 (append_linker_options): Likewie
17795 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17797 * cgraphunit.c (cgraph_node::finalize_function): When
17798 !flag_toplevel_reorde set no_reorder flag.
17799 (varpool_node::finalize_decl): Likewise.
17800 (symbol_table::compile): Drop no toplevel reorder path.
17802 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17804 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17805 edges; zero probability is not better than uninitialized.
17807 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17809 * asan.h (asan_sanitize_allocas_p): Declare.
17810 * asan.c (asan_sanitize_allocas_p): New function.
17811 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17812 (handle_builtin_alloca): Likewise.
17813 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17814 if !asan_sanitize_allocas_p.
17815 * params.def (asan-instrument-allocas): Add new option.
17816 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17817 * opts.c (common_handle_option): Disable allocas sanitization for
17820 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17822 * asan.c: Include gimple-fold.h.
17823 (get_last_alloca_addr): New function.
17824 (handle_builtin_stackrestore): Likewise.
17825 (handle_builtin_alloca): Likewise.
17826 (asan_emit_allocas_unpoison): Likewise.
17827 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17828 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17829 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17830 (instrument_builtin_call): Pass gimple iterator to
17831 get_mem_refs_of_builtin_call.
17832 (last_alloca_addr): New global.
17833 * asan.h (asan_emit_allocas_unpoison): Declare.
17834 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17835 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17836 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17837 if function calls alloca.
17838 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17839 * gimple-fold.h (replace_call_with_value): Declare.
17840 * internal-fn.c: Include asan.h.
17841 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17842 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17844 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17846 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17847 (C_SELFTEST_FLAGS): New.
17848 (CPP_SELFTEST_FLAGS): New.
17849 (SELFTEST_DEPS): New, from deps of s-selftest.
17850 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17851 (CPP_SELFTEST_DEPS): New.
17852 (selftest): Add dependency on s-selftest-c++.
17853 (s-selftest): Rename to...
17854 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17855 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17856 than SELFTEST_FLAGS.
17857 (selftest-gdb): Rename to...
17858 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17860 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17861 (selftest-valgrind): Rename to...
17862 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17864 (selftest-valgrind): Reintroduce as an alias for
17865 selftest-c-valgrind.
17866 (s-selftest-c++): New.
17867 (selftest-c++-gdb): New.
17868 (selftest-c++-valgrind): New.
17870 2017-07-06 Olivier Hainque <hainque@adacore.com>
17872 * gcc.c (process_command): When deciding if undefined variables
17873 should be ignored when processing specs, accept "gcc -v" as well.
17875 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17877 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17878 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17880 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17882 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17883 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17884 * config/arm/arm-tables.opt: Regenerate.
17885 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17887 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17889 2017-07-06 Carl Love <cel@us.ibm.com>
17891 * ChangeLog: Clean up from mid air collision
17893 2017-07-06 Carl Love <cel@us.ibm.com>
17895 * config/rs6000/rs6000-c.c: Add support for built-in functions
17896 vector signed int vec_subc (vector signed int, vector signed int);
17897 vector signed __int128 vec_subc (vector signed __int128,
17898 vector signed __int128);
17899 vector unsigned __int128 vec_subc (vector unsigned __int128,
17900 vector unsigned __int128);
17901 vector signed int vec_sube (vector signed int, vector signed int,
17902 vector signed int);
17903 vector unsigned int vec_sube (vector unsigned int,
17904 vector unsigned int,
17905 vector unsigned int);
17906 vector signed __int128 vec_sube (vector signed __int128,
17907 vector signed __int128,
17908 vector signed__int128);
17909 vector unsigned __int128 vec_sube (vector unsigned __int128,
17910 vector unsigned __int128,
17911 vector unsigned __int128);
17912 vector signed int vec_subec (vector signed int, vector signed int,
17913 vector signed int);
17914 vector unsigned int vec_subec (vector unsigned int,
17915 vector unsigned int,
17916 vector unsigned int);
17917 vector signed __int128 vec_subec (vector signed __int128,
17918 vector signed __int128,
17919 vector signed__int128);
17920 vector unsigned __int128 vec_subec (vector unsigned __int128,
17921 vector unsigned __int128,
17922 vector unsigned __int128);
17923 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17924 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17925 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17926 BU_ALTIVEC_OVERLOAD_X definitions.
17927 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17928 * doc/extend.texi: Update the built-in documentation file for the new
17929 built-in functions.
17931 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17934 * diagnostic-show-locus.c (layout::layout): Use start and finish
17935 spelling location for the start and finish of each range.
17936 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17937 Add unused aspect param.
17938 * input.c (expand_location_1): Add "aspect" param, and use it
17939 to access the correct part of the location.
17940 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17942 (expand_location_to_spelling_point): Likewise.
17943 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17944 param, and pass it to expand_location_1.
17946 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17948 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17949 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17950 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17951 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17952 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17953 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17954 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17955 _mm_maskz_getmant_ss): New intrinsics.
17956 (__builtin_ia32_getexpss128_mask): Changed to ...
17957 __builtin_ia32_getexpss128_round ... this.
17958 (__builtin_ia32_getexpsd128_mask): Changed to ...
17959 __builtin_ia32_getexpsd128_round ... this.
17960 * config/i386/i386-builtin-types.def
17961 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17962 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17963 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17964 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17965 __builtin_ia32_getmantss_mask_round): New builtins.
17966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17967 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17968 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17969 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17970 * config/i386/sse.md
17971 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17972 avx512f_sgetexp<mode><mask_scalar_name>
17973 <round_saeonly_scalar_name> ... this.
17974 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17975 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17976 vgetexp<ssescalarmodesuffix>
17977 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17978 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17979 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17980 avx512f_vgetmant<mode><mask_scalar_name>
17981 <round_saeonly_scalar_name> ... this.
17982 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17983 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17984 vgetmant<ssescalarmodesuffix>
17985 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17986 %0<mask_scalar_operand4>, %1, %2
17987 <round_saeonly_scalar_mask_op4>, %3} ... this.
17988 * config/i386/subst.md (mask_scalar_operand4,
17989 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17990 round_saeonly_scalar_nimm_predicate): New subst attributes.
17992 2017-07-06 Julia Koval <julia.koval@intel.com>
17994 * config/i386/i386.c (ix86_erase_embedded_rounding):
17995 Remove code for old rounding pattern.
17997 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17999 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18001 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18003 * doc/sourcebuild.texi (Test Directives, Variants of
18004 dg-require-support): Add documentation for dg-require-stack-check.
18006 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18008 * config/i386/subst.md (mask_scalar, round_scalar,
18009 round_saeonly_scalar): New meta-templates.
18010 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18011 round_scalar_mask_operand3, round_scalar_mask_op3,
18012 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18013 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18014 round_saeonly_scalar_constraint,
18015 round_saeonly_scalar_prefix): New subst attribute.
18016 * config/i386/sse.md
18017 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18018 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18019 <round_scalar_name> ... this.
18020 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18021 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18022 <round_scalar_name> ... this.
18023 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18024 <sse>_vm<code><mode>3<mask_scalar_name>
18025 <round_saeonly_scalar_name> ... this.
18026 (v<plusminus_mnemonic><ssescalarmodesuffix>
18027 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18028 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18029 v<plusminus_mnemonic><ssescalarmodesuffix>
18030 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18031 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18032 (v<multdiv_mnemonic><ssescalarmodesuffix>
18033 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18034 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18035 v<multdiv_mnemonic><ssescalarmodesuffix>
18036 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18037 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18038 (v<maxmin_float><ssescalarmodesuffix>
18039 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18040 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18041 v<maxmin_float><ssescalarmodesuffix>
18042 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18043 %0<mask_scalar_operand3>, %1, %<iptr>2
18044 <round_saeonly_scalar_mask_op3>} ... this.
18046 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18048 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18049 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18051 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18052 Alan Hayward <alan.hayward@arm.com>
18053 David Sherwood <david.sherwood@arm.com>
18055 * combine.c (simplify_if_then_else): Remove "enum" before
18057 * compare-elim.c (can_eliminate_compare): Likewise.
18058 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18060 (aarch64_lookup_simd_builtin_type): Likewise.
18061 (aarch64_simd_builtin_type): Likewise.
18062 (aarch64_init_simd_builtin_types): Likewise.
18063 (aarch64_simd_expand_args): Likewise.
18064 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18066 (aarch64_reverse_mask): Likewise.
18067 (aarch64_simd_emit_reg_reg_move): Likewise.
18068 (aarch64_gen_adjusted_ldpstp): Likewise.
18069 (aarch64_ccmp_mode_to_code): Likewise.
18070 (aarch64_operands_ok_for_ldpstp): Likewise.
18071 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18072 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18074 (aarch64_min_divisions_for_recip_mul): Likewise.
18075 (aarch64_reassociation_width): Likewise.
18076 (aarch64_get_condition_code_1): Likewise.
18077 (aarch64_simd_emit_reg_reg_move): Likewise.
18078 (aarch64_simd_attr_length_rglist): Likewise.
18079 (aarch64_reverse_mask): Likewise.
18080 (aarch64_operands_ok_for_ldpstp): Likewise.
18081 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18082 (aarch64_gen_adjusted_ldpstp): Likewise.
18083 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18085 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18086 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18087 (arm_lookup_simd_builtin_type): Likewise.
18088 (arm_simd_builtin_type): Likewise.
18089 (arm_init_simd_builtin_types): Likewise.
18090 (arm_expand_builtin_args): Likewise.
18091 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18092 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18093 (ft32_setup_incoming_varargs): Likewise.
18094 (ft32_function_arg): Likewise.
18095 (ft32_function_arg_advance): Likewise.
18096 (ft32_pass_by_reference): Likewise.
18097 (ft32_arg_partial_bytes): Likewise.
18098 (ft32_valid_pointer_mode): Likewise.
18099 (ft32_addr_space_pointer_mode): Likewise.
18100 (ft32_addr_space_legitimate_address_p): Likewise.
18101 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18103 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18104 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18105 (iamcu_alignment): Likewise.
18106 (canonicalize_vector_int_perm): Likewise.
18107 (ix86_noce_conversion_profitable_p): Likewise.
18108 (ix86_mpx_bound_mode): Likewise.
18109 (ix86_operands_ok_for_move_multiple): Likewise.
18110 * config/microblaze/microblaze-protos.h
18111 (microblaze_expand_conditional_branch_reg): Likewise.
18112 * config/microblaze/microblaze.c
18113 (microblaze_expand_conditional_branch_reg): Likewise.
18114 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18116 (rs6000_reassociation_width): Likewise.
18117 (rs6000_invalid_binary_op): Likewise.
18118 (fusion_p9_p): Likewise.
18119 (emit_fusion_p9_load): Likewise.
18120 (emit_fusion_p9_store): Likewise.
18121 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18123 (riscv_hard_regno_mode_ok_p): Likewise.
18124 (riscv_address_insns): Likewise.
18125 (riscv_split_symbol): Likewise.
18126 (riscv_legitimize_move): Likewise.
18127 (riscv_function_value): Likewise.
18128 (riscv_hard_regno_nregs): Likewise.
18129 (riscv_expand_builtin): Likewise.
18130 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18131 (riscv_build_integer): Likewise.
18132 (riscv_split_integer): Likewise.
18133 (riscv_legitimate_constant_p): Likewise.
18134 (riscv_cannot_force_const_mem): Likewise.
18135 (riscv_regno_mode_ok_for_base_p): Likewise.
18136 (riscv_valid_base_register_p): Likewise.
18137 (riscv_valid_offset_p): Likewise.
18138 (riscv_valid_lo_sum_p): Likewise.
18139 (riscv_classify_address): Likewise.
18140 (riscv_legitimate_address_p): Likewise.
18141 (riscv_address_insns): Likewise.
18142 (riscv_load_store_insns): Likewise.
18143 (riscv_force_binary): Likewise.
18144 (riscv_split_symbol): Likewise.
18145 (riscv_force_address): Likewise.
18146 (riscv_legitimize_address): Likewise.
18147 (riscv_move_integer): Likewise.
18148 (riscv_legitimize_const_move): Likewise.
18149 (riscv_legitimize_move): Likewise.
18150 (riscv_address_cost): Likewise.
18151 (riscv_subword): Likewise.
18152 (riscv_output_move): Likewise.
18153 (riscv_canonicalize_int_order_test): Likewise.
18154 (riscv_emit_int_order_test): Likewise.
18155 (riscv_function_arg_boundary): Likewise.
18156 (riscv_pass_mode_in_fpr_p): Likewise.
18157 (riscv_pass_fpr_single): Likewise.
18158 (riscv_pass_fpr_pair): Likewise.
18159 (riscv_get_arg_info): Likewise.
18160 (riscv_function_arg): Likewise.
18161 (riscv_function_arg_advance): Likewise.
18162 (riscv_arg_partial_bytes): Likewise.
18163 (riscv_function_value): Likewise.
18164 (riscv_pass_by_reference): Likewise.
18165 (riscv_setup_incoming_varargs): Likewise.
18166 (riscv_print_operand): Likewise.
18167 (riscv_elf_select_rtx_section): Likewise.
18168 (riscv_save_restore_reg): Likewise.
18169 (riscv_for_each_saved_reg): Likewise.
18170 (riscv_register_move_cost): Likewise.
18171 (riscv_hard_regno_mode_ok_p): Likewise.
18172 (riscv_hard_regno_nregs): Likewise.
18173 (riscv_class_max_nregs): Likewise.
18174 (riscv_memory_move_cost): Likewise.
18175 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18176 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18177 (rl78_addr_space_address_mode): Likewise.
18178 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18180 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18181 (rs6000_reassociation_width): Likewise.
18182 (rs6000_invalid_binary_op): Likewise.
18183 (fusion_p9_p): Likewise.
18184 (emit_fusion_p9_load): Likewise.
18185 (emit_fusion_p9_store): Likewise.
18186 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18187 (ok_for_simple_move_operands): Likewise.
18188 (ok_for_simple_move_strict_operands): Likewise.
18189 (ok_for_simple_arith_logic_operands): Likewise.
18190 (visium_legitimize_reload_address): Likewise.
18191 (visium_select_cc_mode): Likewise.
18192 (output_cbranch): Likewise.
18193 (visium_split_double_move): Likewise.
18194 (visium_expand_copysign): Likewise.
18195 (visium_expand_int_cstore): Likewise.
18196 (visium_expand_fp_cstore): Likewise.
18197 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18198 (visium_function_arg): Likewise.
18199 (visium_function_arg_advance): Likewise.
18200 (visium_libcall_value): Likewise.
18201 (visium_setup_incoming_varargs): Likewise.
18202 (visium_legitimate_constant_p): Likewise.
18203 (visium_legitimate_address_p): Likewise.
18204 (visium_legitimize_address): Likewise.
18205 (visium_secondary_reload): Likewise.
18206 (visium_register_move_cost): Likewise.
18207 (visium_memory_move_cost): Likewise.
18208 (prepare_move_operands): Likewise.
18209 (ok_for_simple_move_operands): Likewise.
18210 (ok_for_simple_move_strict_operands): Likewise.
18211 (ok_for_simple_arith_logic_operands): Likewise.
18212 (visium_function_value_1): Likewise.
18213 (rtx_ok_for_offset_p): Likewise.
18214 (visium_legitimize_reload_address): Likewise.
18215 (visium_split_double_move): Likewise.
18216 (visium_expand_copysign): Likewise.
18217 (visium_expand_int_cstore): Likewise.
18218 (visium_expand_fp_cstore): Likewise.
18219 (visium_split_cstore): Likewise.
18220 (visium_select_cc_mode): Likewise.
18221 (visium_split_cbranch): Likewise.
18222 (output_cbranch): Likewise.
18223 (visium_print_operand_address): Likewise.
18224 * expmed.c (flip_storage_order): Likewise.
18225 * expmed.h (emit_cstore): Likewise.
18226 (flip_storage_order): Likewise.
18227 * genrecog.c (validate_pattern): Likewise.
18228 * hsa-gen.c (gen_hsa_addr): Likewise.
18229 * internal-fn.c (expand_arith_overflow): Likewise.
18230 * ira-color.c (allocno_copy_cost_saving): Likewise.
18231 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18232 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18233 (process_invariant_for_inheritance): Likewise.
18234 * lra-eliminations.c (move_plus_up): Likewise.
18235 * omp-low.c (lower_oacc_reductions): Likewise.
18236 * simplify-rtx.c (simplify_subreg): Likewise.
18237 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18238 (TARGET_CHKP_BOUND_MODE): Likewise..
18239 * targhooks.c (default_chkp_bound_mode): Likewise.
18240 (default_setup_incoming_vararg_bounds): Likewise.
18241 * targhooks.h (default_chkp_bound_mode): Likewise.
18242 (default_setup_incoming_vararg_bounds): Likewise.
18243 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18244 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18245 (have_whole_vector_shift): Likewise.
18246 * tree-vect-stmts.c (vectorizable_load): Likewise.
18247 * doc/tm.texi: Regenerate.
18249 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18251 Graceful degrade if Binutils PR21472 is not available.
18254 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18255 .rodata in flash test fails.
18256 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18257 * confgure: Regenerate.
18258 * config.in: Regenerate.
18259 * config/avr/avr.c (avr_asm_named_section)
18260 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18261 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18262 (avr_asm_init_sections): Same.
18264 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18266 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18267 (fma<VH:mode>4_intrinsic): Likewise.
18268 (*fmsub<VCVTF:mode>4): Likewise.
18269 (*fmsub<VH:mode>4_intrinsic): Likewise.
18271 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18274 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18275 Don't depend on "optimize > 0".
18276 (out_movhi_r_mr, out_movqi_mr_r): Same.
18277 (out_movhi_mr_r, out_movqi_r_mr): Same.
18278 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18279 io_address_operand on "optimize > 0".
18281 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18283 * tree-loop-distribution.c: Add general explanantion on the pass.
18284 (generate_loops_for_partition): Mark distributed loop.
18285 (pg_add_dependence_edges): New parameter. Handle alias data
18286 dependence specially and record it in the parameter if asked.
18287 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18288 (init_partition_graph_vertices, add_partition_graph_edge): New.
18289 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18290 (free_partition_graph_vdata, build_partition_graph): New.
18291 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18292 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18293 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18294 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18295 (version_for_distribution_p, finalize_partitions): New.
18296 (distribute_loop): Handle alias data dependence specially. Factor
18297 out loop fusion code as functions and call these functions.
18299 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18301 * tree-loop-distribution.c (classify_partition): New parameter and
18302 better handle reduction statement.
18303 (rdg_build_partitions): Revise comment.
18304 (distribute_loop): Compute statements in all partitions and pass it
18305 to classify_partition.
18307 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18309 * tree-loop-distribution.c (enum partition_type): New.
18310 (struct partition): New field type.
18311 (partition_merge_into): Add parameter. Update partition type.
18312 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18313 (build_rdg_partition_for_vertex): Compute partition type.
18314 (rdg_build_partitions): Dump partition type.
18315 (distribute_loop): Update calls to partition_merge_into.
18317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18319 * tree-loop-distribution.c (struct ddr_hasher): New.
18320 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18322 (classify_partition): Call get_data_dependence.
18323 (pg_add_dependence_edges): Ditto.
18324 (distribute_loop): Release data dependence hash table.
18326 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18328 * tree-loop-distribution.c (ref_base_address): Delete.
18329 (similar_memory_accesses): Rename ...
18330 (share_memory_accesses): ... to this. Check if partitions access
18331 the same memory reference.
18332 (distribute_loop): Call share_memory_accesses.
18334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18336 * tree-loop-distribution.c (struct partition): New field recording
18337 its data reference.
18338 (partition_alloc, partition_free): Init and release data refs.
18339 (partition_merge_into): Merge data refs.
18340 (build_rdg_partition_for_vertex): Collect data refs for partition.
18341 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18343 (distribute_loop): Remve data refs from vertice data of partition
18346 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18348 * tree-loop-distribution.c (params.h): Include header file.
18349 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18350 (datarefs_vec): New global var.
18351 (create_rdg_vertices): Use datarefs_vec directly.
18352 (free_rdg): Don't free data references.
18353 (build_rdg): Update use. Don't free data references.
18354 (distribute_loop): Compute global variable for data references.
18355 Bail out if there are too many data references.
18357 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18359 * tree-loop-distribution.c (loop_nest): New global var.
18360 (build_rdg): Use loop directly, rather than loop nest.
18361 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18363 (distribute_loop): Compute global variable loop nest. Update use.
18365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18367 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18368 (partition_merge_into): New parameter. Dump reason for fusion.
18369 (distribute_loop): Update use of partition_merge_into.
18371 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18373 * tree-loop-distribution.c (bb_top_order_index): New.
18374 (bb_top_order_index_size, bb_top_order_cmp): New.
18375 (stmts_from_loop): Use topological order.
18376 (pass_loop_distribution::execute): Compute and release topological
18377 order for basic blocks.
18379 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18381 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18384 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18386 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18387 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18388 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18389 * internal-fn.def (LOOP_DIST_ALIAS): New.
18390 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18391 (fold_loop_internal_call): ... this.
18392 (vect_loop_dist_alias_call): New function.
18393 (set_uid_loop_bbs): Call fold_loop_internal_call.
18394 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18397 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18400 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18401 Require dead FLAGS_REG at the beginning of a peephole.
18403 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18406 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18407 arguments in the call to __builtin_ia32_sbb_u32.
18408 (_subborrow_u64): Swap _X and _Y arguments in the call to
18409 __builtin_ia32_sbb_u64.
18411 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18414 * tree-vrp.c (compare_assert_loc): Turn into a function template
18415 with stable template parameter. Only test if a->e is NULL,
18416 !a->e == !b->e has been verified already. Use e == NULL or
18417 e != NULL instead of e or ! e tests. If stable is true, don't use
18418 iterative_hash_expr, on the other side allow a or b or both NULL
18419 and sort the NULLs last.
18420 (process_assert_insertions): Sort using compare_assert_loc<false>
18421 instead of compare_assert_loc, later sort using
18422 compare_assert_loc<true> before calling process_assert_insertions_for
18423 in a loop. Use break instead of continue once seen NULL pointer.
18425 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18427 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18428 Cortex-R7 and Cortex-R8 processors.
18430 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18432 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18433 uninitialized while src is not.
18435 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18437 * common/config/arm/arm-common.c: Adjust include path for
18439 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18440 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18441 (arm-cpu-data.h): Likewise.
18442 (arm-cpu-cdata.h): Likewise.
18443 * config/arm/arm-cpu.h: Delete.
18444 * config/arm/arm-cpu-cdata.h: Delete.
18445 * config/arm/arm-cpu-data.h: Delete.
18447 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18449 * config/arm/arm-cpus.in (cortex-a55): New.
18450 (cortex-a75): Likewise.
18451 (cortex-a75.cortex-a55): Likewise.
18452 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18454 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18455 * config/arm/arm-cpu-cdata.h: Regenerate.
18456 * config/arm/arm-cpu-data.h: Regenerate.
18457 * config/arm/arm-cpu.h: Regenerate.
18458 * config/arm/arm-tables.opt: Regenerate.
18459 * config/arm/arm-tune.md: Regenerate.
18461 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18463 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18465 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18467 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18470 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18472 PR tree-optimization/81292
18473 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18474 full_string_p, also call adjust_related_strinfos if the adjustment
18475 is simple, otherwise invalidate related strinfos.
18477 2017-07-04 Martin Liska <mliska@suse.cz>
18480 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18481 newly created variable as DECL_IGNORED_P.
18483 2017-07-04 Martin Liska <mliska@suse.cz>
18486 * ipa-inline.c (inline_small_functions):
18487 Use xstrdup_for_dump.
18489 2017-07-04 Tom de Vries <tom@codesourcery.com>
18491 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18493 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18496 * config/darwin.c (darwin_function_switched_text_sections):
18499 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18501 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18503 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18505 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18507 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18509 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18510 min_profitable_iters, and th as inclusive lower bounds.
18511 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18512 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18513 for min_profitable_iters and min_profitable_estimate.
18514 (vect_transform_loop): Treat th as an inclusive lower bound.
18515 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18517 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18520 * config/darwin.c (darwin_function_switched_text_sections):
18521 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18522 in two pieces, and suppress the use of buf.
18524 2017-07-03 Nathan Sidwell <nathan@acm.org>
18526 * hash-table.h (hash_table_mod1): Fix indentation.
18528 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18530 PR middle-end/81290
18531 * predict.c (force_edge_cold): Be more careful about propagation
18533 * profile-count.h (profile_probability::guessed,
18534 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18536 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18538 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18540 * doc/invoke.texi (rcpc architecture extension): Document it.
18542 2017-07-03 Richard Biener <rguenther@suse.de>
18544 PR tree-optimization/60510
18545 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18546 the scalar reduction PHI and use it.
18547 (vectorizable_reduction): Properly guard the single_defuse_cycle
18548 path for non-SLP reduction chains where we cannot use it.
18549 Rework reduc_def/index and vector type deduction. Rework
18550 vector operand gathering during reduction op code-gen.
18551 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18552 chains dissolve the chain and leave it to non-SLP reduction
18555 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18557 * tree-data-ref.h (dr_alignment): Declare.
18558 * tree-data-ref.c (dr_alignment): New function.
18559 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18560 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18562 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18564 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18566 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18567 and base_misalignment fields.
18568 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18569 * tree-data-ref.c: Include builtins.h.
18570 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18571 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18572 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18573 * tree-vect-data-refs.c: Include tree-cfg.h.
18574 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18575 fields instead of calculating an alignment here.
18576 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18577 innermost_loop_behavior fields.
18579 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18581 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18583 (DR_STEP_ALIGNMENT): New macro.
18584 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18585 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18586 (create_data_ref): Print it.
18587 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18588 to tell whether the step preserves vector (mis)alignment.
18589 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18590 Move the check for an integer step and generalise to all INTEGER_CST.
18591 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18592 Print the outer step alignment.
18594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18596 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18597 with offset_alignment.
18598 (DR_ALIGNED_TO): Delete.
18599 (DR_OFFSET_ALIGNMENT): New macro.
18600 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18601 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18602 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18603 (create_data_ref): Likewise.
18604 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18605 (vect_analyze_data_refs): Likewise.
18606 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18607 creating dummy innermost behavior.
18609 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18611 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18612 with a "innermost_loop_behavior *" and refeence tree.
18613 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18614 (create_data_ref): Update call accordingly.
18615 * tree-predcom.c (find_looparound_phi): Likewise.
18617 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18619 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18620 fields with dr_wrt_vec_loop.
18621 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18622 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18623 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18624 (vect_dr_behavior): New function.
18625 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18626 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18627 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18628 track whether the step preserves the misalignment.
18629 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18630 Use vect_dr_behavior.
18631 (vect_setup_realignment): Update call accordingly.
18632 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18633 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18634 call to vect_create_addr_base_for_vector_ref.
18635 (vect_create_cond_for_align_checks): Likewise.
18636 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18637 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18638 (vect_recog_mask_conversion_pattern): Likewise.
18639 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18640 (new_stmt_vec_info): Remove redundant zeroing.
18642 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18644 * common/config/arm/arm-common.c (arm_be8_option): New function.
18645 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18646 (ISA_ARMv6): Add isa_bit_be8.
18647 * config/arm/arm.h (arm_be8_option): Add prototype.
18648 (BE8_SPEC_FUNCTION): New define.
18649 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18650 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18651 (mlittle-endian): Similarly.
18652 (mbe8, mbe32): New options.
18653 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18654 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18656 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18658 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18660 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18662 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18663 (cleanup_tree_cfg_bb): Use it.
18664 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18666 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18668 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18671 * loop-doloop.c (add_test): Update profile.
18673 2017-07-03 Martin Liska <mliska@suse.cz>
18676 * sanopt.c (rewrite_usage_of_param): New function.
18677 (sanitize_rewrite_addressable_params): Likewise.
18678 (pass_sanopt::execute): Call rewrite_usage_of_param.
18680 2017-07-03 Richard Biener <rguenther@suse.de>
18682 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18683 back to using VIEW_CONVERT_EXPR.
18685 2017-07-03 Martin Liska <mliska@suse.cz>
18688 * doc/extend.texi: Document when a resolver function is
18689 generated for target_clones.
18691 2017-07-03 Martin Liska <mliska@suse.cz>
18693 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18694 and shadow memory of auto variables which are subject of
18695 use-after-scope sanitization.
18696 (asan_expand_mark_ifn): Add do set only when is_poison.
18698 2016-07-03 Richard Biener <rguenther@suse.de>
18700 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18702 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18703 (vectorizable_reduction): Perform reduction PHI creation when
18704 visiting a reduction PHI and adjust and simplify code generation
18705 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18706 (vect_transform_loop): Visit reduction PHIs.
18707 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18708 defs into the SLP tree.
18709 (vect_build_slp_tree): Reduction defs terminate the recursion.
18710 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18712 (vect_get_vec_defs_for_stmt_copy): Export.
18713 (vect_get_vec_defs): Likewise.
18714 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18716 (vect_get_vec_defs_for_stmt_copy): Declare.
18717 (vect_get_vec_defs): Likewise.
18719 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18721 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18722 parameter with a "loop" parameter and use it instead of the
18723 loop containing DR_STMT. Don't check simple_iv when doing
18724 BB analysis. Describe the two analysis modes in the comment.
18726 2017-07-03 Tom de Vries <tom@codesourcery.com>
18728 PR tree-optimization/69468
18729 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18730 (find_same_succ_bb): Handle ignore_edge_flags.
18732 2017-07-03 Tom de Vries <tom@codesourcery.com>
18734 PR tree-optimization/81192
18735 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18737 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18739 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18741 2017-07-03 Tom de Vries <tom@codesourcery.com>
18743 PR tree-optimization/81192
18744 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18745 BB_SAME_SUCC (bb) == NULL.
18747 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18749 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18752 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18754 * dumpfile.c: Include profile-count.h
18755 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18757 (insert_cond_bb): Update profile.
18758 * tree-cfg.h (insert_cond_bb): Update prototype.
18759 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18760 * tree-dump.c: Do not include tree-cfg.
18762 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18764 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18766 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18768 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18771 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18773 * tree-complex.c (expand_complex_div_wide): update profile.
18775 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18776 Alan Hayward <alan.hayward@arm.com>
18777 David Sherwood <david.sherwood@arm.com>
18779 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18780 (CORETYPES_H): New define.
18781 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18782 (insn-modes-inline.h, s-modes-inline-h): New rules.
18783 (generated_files): Add insn-modes-inline.h.
18784 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18785 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18786 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18787 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18788 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18789 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18790 (build/gencodes.o, build/genconditions.o): Likewise.
18791 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18792 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18793 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18794 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18795 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18796 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18797 * coretypes.h: Include everything up to real.h for generators.
18798 Include insn-modes.h first. Include wide-int-print.h after
18799 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18800 * machmode.h: Don't include insn-modes.h here.
18801 * function-tests.c: Remove includes of signop.h, machmode.h,
18802 double-int.h and wide-int.h.
18804 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18806 * optc-save-gen.awk: Likewise.
18807 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18808 * godump.c: Remove include of wide-int-print.h.
18809 * pretty-print.h: Likewise.
18810 * wide-int-print.cc: Likewise.
18811 * wide-int.cc: Likewise.
18812 * hash-map-tests.c: Remove include of signop.h.
18813 * hash-set-tests.c: Likewise.
18814 * rtl-tests.c: Likewise.
18815 * mkconfig.sh: Remove include of machmode.h.
18816 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18818 (emit_insn_modes_inline_h): ...this new function. Emit the code
18819 into an insn-modes-inline.h header file, adding appropriate
18820 include guards and end comments.
18821 (emit_insn_modes_c_header): Remove include of machmode.h.
18822 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18824 (main): Handle -i flag and call emit_insn_modes_inline_h when
18827 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18829 * tree-ssa-strlen.c (strinfo): Rename the length field to
18830 nonzero_chars. Add a full_string_p field.
18831 (compare_nonzero_chars, zero_length_string_p): New functions.
18832 (get_addr_stridx): Add an offset_out parameter.
18833 Use compare_nonzero_chars.
18834 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18835 (new_strinfo): Update after above changes to strinfo.
18836 (set_endptr_and_length): Set full_string_p.
18837 (get_string_length): Update after above changes to strinfo.
18838 (unshare_strinfo): Update call to new_strinfo.
18839 (maybe_invalidate): Likewise.
18840 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18841 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18842 as a uhwi instead of an shwi. Update after above changes to
18843 strinfo and new_strinfo.
18844 (zero_length_string): Assert that chainsi contains full strings.
18845 Use zero_length_string_p. Update call to new_strinfo.
18846 (adjust_related_strinfos): Update after above changes to strinfo.
18847 Copy full_string_p from origsi.
18848 (adjust_last_stmt): Use zero_length_string_p.
18849 (handle_builtin_strlen): Update after above changes to strinfo and
18850 new_strinfo. Install the lhs as the string length if the previous
18851 entry didn't describe a full string.
18852 (handle_builtin_strchr): Update after above changes to strinfo
18854 (handle_builtin_strcpy): Likewise.
18855 (handle_builtin_strcat): Likewise.
18856 (handle_builtin_malloc): Likewise.
18857 (handle_pointer_plus): Likewise.
18858 (handle_builtin_memcpy): Likewise. Track nonzero characters
18859 that aren't necessarily followed by a nul terminator.
18860 (handle_char_store): Likewise.
18862 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18864 PR tree-optimization/80769
18865 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18866 for malloc and calloc. Document the new invariant that all related
18867 strinfos have delayed lengths or none do.
18868 (verify_related_strinfos): Move earlier in file.
18869 (set_endptr_and_length): New function, split out from...
18870 (get_string_length): ...here. Also set the lengths of related
18872 (zero_length_string): Assert that chainsi has known (rather than
18874 (adjust_related_strinfos): Likewise.
18876 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18878 PR tree-optimization/81136
18879 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18880 assert that two references with the same misalignment have the same
18881 compile-time misalignment if those compile-time misalignments
18884 2017-07-01 Andi Kleen <ak@linux.intel.com>
18886 * print-tree.c (print_node): Print all attributes.
18888 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18890 * cfg.c (scale_bbs_frequencies): New function.
18891 * cfg.h (scale_bbs_frequencies): Declare it.
18892 * cfgloopanal.c (single_likely_exit): Cleanup.
18893 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18895 (scale_loop_profile): Likewise.
18896 (loop_version): Likewise.
18897 (create_empty_loop_on_edge): Update.
18898 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18899 scale_loop_frequencies, scale_loop_profile, loopify,
18900 loop_version): Update prototypes.
18901 * modulo-sched.c (sms_schedule): Update.
18902 * predict.c (unlikely_executed_edge_p): Also check probability.
18903 (probably_never_executed_edge_p): Fix typo.
18904 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18905 * tree-parloops.c (gen_parallel_loop): Update.
18906 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18907 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18908 * tree-ssa-loop-split.c (split_loop): Update.
18909 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18910 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18911 (vect_loop_versioning): Update.
18912 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18914 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18916 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18918 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18920 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18921 to keep profile consistent.
18923 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18925 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18926 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18927 * profile-count.h (max_safe_multiplier): Make unsigned.
18928 (profile_count::guessed_zero): New.
18930 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18932 * bb-reorder.c (fix_up_crossing_landing_pad,
18933 fix_crossing_conditional_branches): Use make_single_succ_edge
18934 to keep profile consistent.
18936 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18938 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18941 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18944 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18945 the right scopes, make sure cond_jump isn't preserved between multiple
18946 iterations. Search for fallthru edge whenever there are 3+ edges and
18947 use find_fallthru_edge for it.
18949 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18951 Patch by Alexander Monakov <amonakov@ispras.ru>
18952 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18953 probabilities consistently.
18955 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18957 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18958 emit_cmp_and_jump_insns.
18960 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18963 * tree-inline.c (expand_call_inline): Combine profile statuses.
18965 2017-06-30 Andrew Pinski <apinski@cavium.com>
18967 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18968 fold_stmt returned true.
18970 2017-06-30 Nathan Sidwell <nathan@acm.org>
18972 * ggc.h (empty_string): Delete.
18973 * cfgexpand.c (expand_asm_stmt): Use plain "".
18974 * optabs.c (expand_asm_memory_barrier): Likewise.
18975 * stringpool.c (empty_string): Delete.
18976 (digit_vector, digit_string): Delete.
18977 (ggc_alloc_string): Use plain "", don't optimize single digit
18978 strings. Use ggc_alloc_atomic.
18980 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18982 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18983 comparison set and one other set, use the cost of the non-comparison
18986 2017-06-30 Nathan Sidwell <nathan@acm.org>
18988 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18991 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18993 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18994 loops. Remove now unneeded calls to gimple_switch_set_label() that
18995 just set removed labels to NULL_TREE.
18997 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18999 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19000 (set_range_info): ...here. Only call set_range_info_raw if domain
19002 (set_nonzero_bits): Call set_range_info_raw.
19003 * tree-ssanames.h (set_range_info_raw): New.
19005 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19008 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19009 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19010 of nonimmediate_operand and <store_mask_constraint> instead of m
19011 for the input operand. For V8FI iterator, always split if input
19012 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19013 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19014 <store_mask_predicate> instead of register_operand and
19015 <store_mask_constraint> instead of v for the input operand. Make
19016 sure both operands aren't MEMs for if not <mask_applied>.
19018 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19020 * lto-wrapper.c (copy_file) Close both file descriptors before
19023 2017-06-30 Martin Liska <mliska@suse.cz>
19026 * multiple_target.c (create_dispatcher_calls): Make ifunc
19027 also for function that don't have calls or are not referenced.
19029 2017-06-30 Richard Biener <rguenther@suse.de>
19031 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19032 analyze the first scalar stmt. Move vector type computation
19033 for the BB case here from ...
19034 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19035 live operation processing in the SLP case properly.
19037 2017-06-30 Richard Biener <rguenther@suse.de>
19039 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19041 2017-06-30 Martin Liska <mliska@suse.cz>
19044 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19045 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19047 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19049 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19052 2017-06-30 Richard Biener <rguenther@suse.de>
19054 PR tree-optimization/81249
19055 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19056 condition reduction result to original scalar type.
19058 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19060 * profile-count.h (enum profile_quality): Fix typos and whitespace
19063 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19065 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19066 type for branch probabilities.
19068 2017-06-29 Julian Brown <julian@codesourcery.com>
19069 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19071 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19072 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19073 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19074 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19076 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19078 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19079 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19080 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19081 CC usage from generic code to here.
19082 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19083 CC usage into the target macros.
19085 2017-06-29 Maya Rashish <coypu@sdf.org>
19087 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19090 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19092 * arm/arm-builtins.c: Include profile-count.h
19093 * except.c (sjlj_emit_function_enter): Use
19094 profile_probability::unlikely.
19096 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19098 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19099 and tocrel_offset be pointer args rather than implicitly using
19101 (legitimate_constant_pool_address_p, rs6000_emit_move,
19102 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19103 tocrel_offset and use in toc_relative_expr_p call.
19104 (print_operand, print_operand_address): Use static tocrel_base_oac
19105 and tocrel_offset_oac.
19106 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19109 2017-06-29 Maya Rashish <coypu@sdf.org>
19111 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19113 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19115 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19116 objects, take into account only the alignment of 'op0' and 'mode1' if
19119 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19121 * ccmp.c (ccmp_tree_comparison_p): New function.
19122 (ccmp_candidate_p): Update to use above function.
19123 (get_compare_parts): New function.
19124 (expand_ccmp_next): Update to use new functions.
19125 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19127 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19128 take mode as argument.
19129 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19130 * expr.c (expand_expr_real_1): Pass mode as argument.
19132 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19134 * combine.c (combine_instructions): Print insns to dump_file, together
19137 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19139 * asan.c (asan_emit_stack_protection): Update.
19140 (create_cond_insert_point): Update.
19141 * auto-profile.c (afdo_propagate_circuit): Update.
19142 * basic-block.h (struct edge_def): Turn probability to
19143 profile_probability.
19144 (EDGE_FREQUENCY): Update.
19145 * bb-reorder.c (find_traces_1_round): Update.
19146 (better_edge_p): Update.
19147 (sanitize_hot_paths): Update.
19148 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19149 (make_single_succ_edge): Update.
19150 (check_bb_profile): Update.
19151 (dump_edge_info): Update.
19152 (update_bb_profile_for_threading): Update.
19153 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19154 probabilitycount to 0.
19155 * cfgbuild.c (compute_outgoing_frequencies): Update.
19156 * cfgcleanup.c (try_forward_edges): Update.
19157 (outgoing_edges_match): Update.
19158 (try_crossjump_to_edge): Update.
19159 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19160 (expand_gimple_tailcall): Update.
19161 (construct_init_block): Use make_single_succ_edge.
19162 (construct_exit_block): Use make_single_succ_edge.
19163 * cfghooks.c (verify_flow_info): Update.
19164 (redirect_edge_succ_nodup): Update.
19165 (split_edge): Update.
19166 (account_profile_record): Update.
19167 * cfgloopanal.c (single_likely_exit): Update.
19168 * cfgloopmanip.c (scale_loop_profile): Update.
19169 (set_zero_probability): Remove.
19170 (duplicate_loop_to_header_edge): Update.
19171 * cfgloopmanip.h (loop_version): Update prototype.
19172 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19173 (force_nonfallthru_and_redirect): Update.
19174 (update_br_prob_note): Update.
19175 (rtl_verify_edges): Update.
19176 (purge_dead_edges): Update.
19177 (rtl_lv_add_condition_to_bb): Update.
19178 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19179 * cgraphunit.c (init_lowered_empty_function): Update.
19180 (cgraph_node::expand_thunk): Update.
19181 * cilk-common.c: Include profile-count.h
19182 * dojump.c (inv): Remove.
19183 (jumpifnot): Update.
19184 (jumpifnot_1): Update.
19185 (do_jump_1): Update.
19187 (do_jump_by_parts_greater_rtx): Update.
19188 (do_compare_rtx_and_jump): Update.
19189 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19190 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19191 * dwarf2cfi.c: Include profile-count.h
19192 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19193 (sjlj_emit_dispatch_table): Likewise.
19194 * explow.c: Include profile-count.h
19195 * expmed.c (emit_store_flag_force): Update.
19196 (do_cmp_and_jump): Update.
19197 * expr.c (compare_by_pieces_d::generate): Update.
19198 (compare_by_pieces_d::finish_mode): Update.
19199 (emit_block_move_via_loop): Update.
19200 (store_expr_with_bounds): Update.
19201 (store_constructor): Update.
19202 (expand_expr_real_2): Update.
19203 (expand_expr_real_1): Update.
19204 * expr.h (try_casesi, try_tablejump): Update prototypes.
19205 * gimple-pretty-print.c (dump_probability): Update.
19206 (dump_profile): New.
19207 (dump_gimple_label): Update.
19208 (dump_gimple_bb_header): Update.
19209 * graph.c (draw_cfg_node_succ_edges): Update.
19210 * hsa-gen.c (convert_switch_statements): Update.
19211 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19212 (find_if_case_1): Update.
19213 (find_if_case_2): Update.
19214 * internal-fn.c (expand_arith_overflow_result_store): Update.
19215 (expand_addsub_overflow): Update.
19216 (expand_neg_overflow): Update.
19217 (expand_mul_overflow): Update.
19218 (expand_vector_ubsan_overflow): Update.
19219 * ipa-cp.c (good_cloning_opportunity_p): Update.
19220 * ipa-split.c (split_function): Use make_single_succ_edge.
19221 * ipa-utils.c (ipa_merge_profiles): Update.
19222 * loop-doloop.c (add_test): Update.
19223 (doloop_modify): Update.
19224 * loop-unroll.c (compare_and_jump_seq): Update.
19225 (unroll_loop_runtime_iterations): Update.
19226 * lra-constraints.c (lra_inheritance): Update.
19227 * lto-streamer-in.c (input_cfg): Update.
19228 * lto-streamer-out.c (output_cfg): Update.
19229 * mcf.c (adjust_cfg_counts): Update.
19230 * modulo-sched.c (sms_schedule): Update.
19231 * omp-expand.c (expand_omp_for_init_counts): Update.
19232 (extract_omp_for_update_vars): Update.
19233 (expand_omp_ordered_sink): Update.
19234 (expand_omp_for_ordered_loops): Update.
19235 (expand_omp_for_generic): Update.
19236 (expand_omp_for_static_nochunk): Update.
19237 (expand_omp_for_static_chunk): Update.
19238 (expand_cilk_for): Update.
19239 (expand_omp_simd): Update.
19240 (expand_omp_taskloop_for_outer): Update.
19241 (expand_omp_taskloop_for_inner): Update.
19242 * omp-simd-clone.c (simd_clone_adjust): Update.
19243 * optabs.c (expand_doubleword_shift): Update.
19244 (expand_abs): Update.
19245 (emit_cmp_and_jump_insn_1): Update.
19246 (expand_compare_and_swap_loop): Update.
19247 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19248 * predict.c (predictable_edge_p): Update.
19249 (edge_probability_reliable_p): Update.
19250 (set_even_probabilities): Update.
19251 (combine_predictions_for_insn): Update.
19252 (combine_predictions_for_bb): Update.
19253 (propagate_freq): Update.
19254 (estimate_bb_frequencies): Update.
19255 (force_edge_cold): Update.
19256 * profile-count.c (profile_count::dump): Add missing space into dump.
19257 (profile_count::debug): Add newline.
19258 (profile_count::differs_from_p): Explicitly convert to unsigned.
19259 (profile_count::stream_in): Update.
19260 (profile_probability::dump): New member function.
19261 (profile_probability::debug): New member function.
19262 (profile_probability::differs_from_p): New member function.
19263 (profile_probability::differs_lot_from_p): New member function.
19264 (profile_probability::stream_in): New member function.
19265 (profile_probability::stream_out): New member function.
19266 * profile-count.h (profile_count_quality): Rename to ...
19267 (profile_quality): ... this one.
19268 (profile_probability): New.
19269 (profile_count): Update.
19270 * profile.c (compute_branch_probabilities): Update.
19271 * recog.c (peep2_attempt): Update.
19272 * sched-ebb.c (schedule_ebbs): Update.
19273 * sched-rgn.c (find_single_block_region): Update.
19274 (compute_dom_prob_ps): Update.
19275 (schedule_region): Update.
19276 * sel-sched-ir.c (compute_succs_info): Update.
19277 * stmt.c (struct case_node): Update.
19278 (do_jump_if_equal): Update.
19279 (get_outgoing_edge_probs): Update.
19280 (conditional_probability): Update.
19281 (emit_case_dispatch_table): Update.
19282 (expand_case): Update.
19283 (expand_sjlj_dispatch_table): Update.
19284 (emit_case_nodes): Update.
19285 * targhooks.c: Update.
19286 * tracer.c (better_p): Update.
19287 (find_best_successor): Update.
19288 * trans-mem.c (expand_transaction): Update.
19289 * tree-call-cdce.c: Update.
19290 * tree-cfg.c (gimple_split_edge): Upate.
19291 (move_sese_region_to_fn): Upate.
19292 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19293 * tree-eh.c (lower_resx): Upate.
19294 (cleanup_empty_eh_move_lp): Upate.
19295 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19296 * tree-inline.c (copy_edges_for_bb): Update.
19297 (copy_cfg_body): Update.
19298 * tree-parloops.c (gen_parallel_loop): Update.
19299 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19300 (gimple_gen_time_profiler): Update.
19301 * tree-ssa-dce.c (remove_dead_stmt): Update.
19302 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19303 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19304 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19305 (unloop_loops): Update.
19306 (try_peel_loop): Update.
19307 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19308 * tree-ssa-loop-split.c (connect_loops): Update.
19309 (split_loop): Update.
19310 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19311 (hoist_guard): Update.
19312 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19313 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19314 (value_replacement): Update.
19315 * tree-ssa-reassoc.c (branch_fixup): Update.
19316 * tree-ssa-tail-merge.c (replace_block_by): Update.
19317 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19318 (create_edge_and_update_destination_phis): Update.
19319 (compute_path_counts): Update.
19320 (recompute_probabilities): Update.
19321 (update_joiner_offpath_counts): Update.
19322 (freqs_to_counts_path): Update.
19323 (duplicate_thread_path): Update.
19324 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19325 (struct switch_conv_info): Update.
19326 (gen_inbound_check): Update.
19327 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19328 (vect_do_peeling): Update.
19329 (vect_loop_versioning): Update.
19330 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19331 (optimize_mask_stores): Update.
19332 * ubsan.c (ubsan_expand_null_ifn): Update.
19333 * value-prof.c (gimple_divmod_fixed_value): Update.
19334 (gimple_divmod_fixed_value_transform): Update.
19335 (gimple_mod_pow2): Update.
19336 (gimple_mod_pow2_value_transform): Update.
19337 (gimple_mod_subtract): Update.
19338 (gimple_mod_subtract_transform): Update.
19339 (gimple_ic): Update.
19340 (gimple_stringop_fixed_value): Update.
19341 (gimple_stringops_transform): Update.
19342 * value-prof.h: Update.
19344 2017-06-29 Carl Love <cel@us.ibm.com>
19346 * config/rs6000/rs6000-c.c: Add support for built-in functions
19347 vector signed int vec_signed (vector float);
19348 vector signed long long vec_signed (vector double);
19349 vector signed int vec_signed2 (vector double, vector double);
19350 vector signed int vec_signede (vector double);
19351 vector signed int vec_signedo (vector double);
19352 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19353 instruction generator.
19354 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19355 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19356 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19358 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19359 vunsignede_v2df): Add define_expands.
19360 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19361 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19362 VEC_UNSIGNEDO): Add definitions.
19363 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19364 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19365 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19366 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19367 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19368 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19369 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19370 vec_unsignede, vec_unsignedo): Add builtin defines.
19371 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19373 * doc/extend.texi: Update the built-in documentation file for the
19374 new built-in functions.
19376 2017-06-29 Richard Biener <rguenther@suse.de>
19378 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19379 reduction chains to LOOP_VINFO_REDUCTIONS.
19380 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19381 SLP reductions after processing reduction chains.
19383 2017-06-29 Nathan Sidwell <nathan@acm.org>
19385 * builtins.c (fold_builtin_FUNCTION): Use
19386 lang_hooks.decl_printable_name.
19388 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19390 PR middle-end/81194
19391 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19392 with only one label.
19393 * stmt.c (expand_case): Assert NCASES is greater than one.
19395 2017-06-29 Richard Biener <rguenther@suse.de>
19397 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19399 (group_case_labels): Likewise.
19400 (find_taken_edge): Push sanity checking on val to workers...
19401 (find_taken_edge_cond_expr): ... here
19402 (find_taken_edge_switch_expr): ... and here, handle cases
19403 with just a default label.
19404 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19405 (group_case_labels): Likewise.
19406 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19407 group_case_labels does anything cleanup the CFG again.
19409 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19411 PR tree-optimization/81196
19412 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19413 exit condition comparing two IVs.
19415 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19417 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19418 profile to the dummy entry at the end of the list of architectures.
19419 * config/arm/arm-cpu-cdata.h: Regenerated.
19421 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19422 Michael Collison <michael.collison@arm.com>
19425 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19427 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19428 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19429 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19430 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19431 of shift when the shift amount is masked with constant equal to
19432 the size of the mode.
19433 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19436 2017-06-29 Martin Liska <mliska@suse.cz>
19438 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19440 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19443 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19446 (edge_growth_cache_entry::edge_growth_cache_entry): New
19448 (reset_edge_growth_cache): Update to use constructor.
19450 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19452 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19453 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19454 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19456 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19458 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19459 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19461 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19463 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19464 (*-linux-uclibc*): Add t-uclibc tmake_file.
19465 * config/t-musl: New.
19466 * config/t-uclibc: New.
19468 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19470 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19472 (gen_comm_data): Emit architectural setting of arch_prof.
19473 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19475 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19476 (armv8-m.base, armv8-m.main): Likewise.
19477 * arm-protos.h (arm_build_target): Add profile field.
19478 (arch_option): Likewise.
19479 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19481 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19482 arm_active_target.profile.
19484 2017-06-28 Richard Biener <rguenther@suse.de>
19486 PR middle-end/81227
19487 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19488 TYPE_OVERFLOW_WRAPS.
19489 * match.pd (negate_expr_p): Likewise.
19490 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19491 fold_build2, not fold_binary.
19493 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19495 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19496 Convert memory address to Pmode.
19497 (aarch64_print_operand): Assert MEM operands are always Pmode.
19499 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19502 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19503 Remove redundant if.
19504 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19505 * config/arm/aarch-common-protos.h
19506 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19507 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19509 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19512 * multiple_target.c (create_dispatcher_calls): Set the default
19513 clone to be static, not public.
19515 2017-06-28 Richard Biener <rguenther@suse.de>
19517 * tree-vect-loop.c (vectorizable_reduction): Move special
19518 cond reduction IV var creation ...
19519 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19520 parameter. Use STMT_VINFO_VECTYPE.
19521 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19524 2017-06-28 Martin Liska <mliska@suse.cz>
19527 * ipa-visibility.c (non_local_p): Handle visibility.
19529 2017-06-28 Martin Liska <mliska@suse.cz>
19532 * common.opt: Add IntegerRange to various options.
19533 * opt-functions.awk (integer_range_info): New function.
19534 * optc-gen.awk: Add integer_range_info to cl_options struct.
19535 * opts-common.c (decode_cmdline_option): Handle
19536 CL_ERR_INT_RANGE_ARG.
19537 (cmdline_handle_error): Likewise.
19538 * opts.c (print_filtered_help): Show valid interval in
19539 when --help is provided.
19540 * opts.h (struct cl_option): Add range_min and range_max fields.
19541 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19543 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19545 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19546 (x * C EQ/NE y * C): New transformation.
19548 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19550 * genmultilib (combination_space): Accept '+' in option names.
19552 2017-06-28 Martin Liska <mliska@suse.cz>
19555 * asan.c (instrument_derefs): Bail out inner references
19556 that are hard register variables.
19558 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19561 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19562 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19564 2017-06-28 Richard Biener <rguenther@suse.de>
19566 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19567 (vect_get_slp_defs): Adjust.
19568 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19570 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19572 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19573 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19574 (vectorizable_reduction): Adjust.
19575 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19577 (vect_get_slp_defs): Likewise.
19578 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19579 (vectorizable_bswap): Adjust.
19580 (vectorizable_call): Likewise.
19581 (vectorizable_conversion): Likewise.
19582 (vectorizable_assignment): Likewise.
19583 (vectorizable_shift): Likewise.
19584 (vectorizable_operation): Likewise.
19585 (vectorizable_store): Likewise.
19586 (vectorizable_condition): Likewise.
19587 (vectorizable_comparison): Likewise.
19589 2017-06-28 Michael Collison <michael.collison@arm.com>
19592 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19594 (arm_gen_movmemqi): Removed unused variable 'i'.
19595 Convert 'for' loop into 'while' loop.
19596 (arm_expand_prologue): Remove last unnecessary set of insn.
19597 (thumb_pop): Remove unused variable 'pushed_words'.
19598 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19600 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19602 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19603 * config/s390/s390-protos.h: Add prototype of
19604 s390_rel_address_ok_p.
19605 * config/s390/s390.c (s390_got_symbol): New function.
19606 (s390_rel_address_ok_p): New function.
19607 (legitimize_pic_address): Use s390_rel_address_ok_p.
19608 (s390_load_got): Use s390_got_symbol.
19609 (s390_option_override): Issue error if
19610 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19611 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19613 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19615 2017-06-27 Andrew Pinski <apinski@cavium.com>
19617 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19618 (X * copysign (1.0, X)): New pattern.
19619 (X * copysign (1.0, -X)): New pattern.
19620 (copysign (-1.0, CST)): New pattern.
19622 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19624 * genmultilib (combination_space): Remove variable.
19625 Validate reuse rules against regular expression for any sequence
19626 of multilib options in any order.
19628 2017-06-27 Michael Collison <michael.collison@arm.com>
19630 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19631 call aarch64_split_simd_combine.
19632 * (aarch64_combine_internal<mode>): Delete pattern.
19633 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19634 Allow register and subreg operands.
19636 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19638 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19639 specific need, just fallback on defaults.
19640 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19642 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19643 Olivier Hainque <hainque@adacore.com>
19645 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19647 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19648 targets. Pick a default if no particular attempt applied.
19649 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19652 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19654 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19655 (x86_64-wrs-vxworks7): Likewise.
19657 2017-06-27 Marek Polacek <polacek@redhat.com>
19660 * ubsan.c (instrument_null): Check get_base_address's result for null.
19662 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19664 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19666 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19668 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19669 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19670 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19671 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19672 New function types.
19673 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19674 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19675 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19676 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19677 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19678 BUILT_IN_FEUPDATEENV): New builtins.
19679 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19680 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19681 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19682 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19684 (builtin_structptr_types): Adjust size.
19685 * tree.c (builtin_structptr_types): Add four entries.
19687 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19688 Olivier Hainque <hainque@adacore.com>
19690 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19691 (TLS_SYM): New local macro, forcing reference to __tls__ on
19692 link command lines for VxWorks 7 RTPs, triggering initialization
19694 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19695 OS features TLS support, true for RTPs on VxWorks 7.
19696 * config/vxworks.c (vxworks_override_options): Setup emutls
19699 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19701 * predict.c (test_prediction_value_range): Use -1U instead of -1
19702 to avoid narrowing conversion warning.
19703 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19704 to avoid narrowing conversion warning.
19705 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19707 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19709 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19711 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19712 64bit configurations.
19713 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19714 (SIZE_TYPE): Likewise.
19715 * config/vxworks.c (vxworks_emutls_var_fields): Use
19716 long_unsigned_type_node instead of unsigned_type_node as the offset
19717 field type, which is "pointer" mode in emutls.c.
19719 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19722 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19724 PR middle-end/81207
19725 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19726 gimple_vuse copying separately from gimple_vdef copying.
19728 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19730 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19733 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19734 Olivier Hainque <hainque@adacore.com>
19736 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19737 for all vxworks7 targets.
19738 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19739 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19740 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19741 variations for VX6/VX7 and 32/64bits later on in ...
19742 (VXWORKS_LIB_SPEC): Leverage new macros.
19743 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19744 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19746 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19748 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19749 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19751 2017-06-26 Carl Love <cel@us.ibm.com>
19753 * config/rs6000/rs6000-c.c: Add support for built-in functions
19754 vector bool char vec_reve (vector bool char);
19755 vector signed char vec_reve (vector signed char);
19756 vector unsigned char vec_reve (vector unsigned char);
19757 vector bool int vec_reve (vector bool int);
19758 vector signed int vec_reve (vector signed int);
19759 vector unsigned int vec_reve (vector unsigned int);
19760 vector bool long long vec_reve (vector bool long long);
19761 vector signed long long vec_reve (vector signed long long);
19762 vector unsigned long long vec_reve (vector unsigned long long);
19763 vector bool short vec_reve (vector bool short);
19764 vector signed short vec_reve (vector signed short);
19765 vector double vec_reve (vector double);
19766 vector float vec_reve (vector float);
19767 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19768 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19769 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19770 (altivec_vreve): New pattern.
19771 * config/rs6000/altivec.h (vec_reve): New define.
19772 * doc/extend.texi (vec_rev): Update the built-in documentation file
19773 for the new built-in functions.
19775 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19777 PR tree-optimization/71815
19778 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19780 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19782 (slsr_process_phi): Likewise.
19783 (replace_uncond_cands_and_profitable_phis): Don't replace a
19784 multiply candidate with a stride of 1 (copy or cast).
19785 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19787 (lowest_cost_path): Likewise.
19788 (total_savings): Likewise.
19790 2017-06-26 Richard Biener <rguenther@suse.de>
19793 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19794 Use def_builtin_pure for all gather builtins.
19796 2017-06-26 Richard Biener <rguenther@suse.de>
19798 PR tree-optimization/81203
19799 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19800 non-dominating BBs.
19802 2017-06-26 Marek Polacek <polacek@redhat.com>
19805 * doc/invoke.texi: Document -Wmultistatement-macros.
19807 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19809 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19810 arm_neon_ok_no_float_abi effective target.
19812 2017-06-26 Richard Biener <rguenther@suse.de>
19814 PR tree-optimization/80928
19815 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19816 (copy_bbs): Set BB_DUPLICATED flag early.
19817 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19819 (execute_on_shrinking_pred): Likewise.
19820 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19821 BB_DUPLICATED blocks.
19822 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19823 iterate over all PHIs considering removal of *gsi.
19825 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19827 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19830 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19832 * config/rs6000/rs6000-string.c: (expand_block_clear,
19833 do_load_for_compare, select_block_compare_mode,
19834 compute_current_alignment, expand_block_compare,
19835 expand_strncmp_align_check, expand_strn_compare,
19836 expand_block_move, rs6000_output_load_multiple)
19837 Move functions related to string/block move/compare
19838 to a separate file.
19839 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19840 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19841 for this function which is now used in two files.
19842 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19843 * config.gcc: Add rs6000-string.o to extra_objs for
19844 targets powerpc*-*-* and rs6000*-*-*.
19846 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19849 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19850 32-bit, since indexed is not valid for DImode.
19851 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19852 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19853 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19854 (define_peephole2 for Altivec d-form store): Likewise.
19857 * multiple_target.c (create_dispatcher_calls): Only create the
19858 dispatcher call if the function is the default clone of a
19859 versioned function.
19861 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19863 PR middle-end/80902
19864 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19865 a call, force the call to not be a tail call.
19867 2017-06-23 Jeff Law <law@redhat.com>
19869 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19872 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19874 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19875 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19876 handling for early expansion of vector shifts (sl,sr,sra,rl).
19877 (builtin_function_type): Add vector shift right instructions
19878 to the unsigned argument list.
19880 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19882 rtl-optimizatoin/79286
19883 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19884 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19885 trap. PIC register plus a const unspec without offset can never trap.
19887 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19889 * tree.h (builtin_structptr_type): New type.
19890 (builtin_structptr_types): Declare new array.
19891 * tree.c (builtin_structptr_types): New array.
19892 (free_lang_data, build_common_tree_nodes): Use it.
19894 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19897 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19900 2017-06-22 Matt Turner <mattst88@gmail.com>
19902 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19903 Lake models to skylake case. Assume skylake for unknown
19904 models with clflushopt.
19906 2017-06-22 Jeff Law <law@redhat.com>
19908 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19909 frame sizes that do not satisfy aarch64_uimm12_shift.
19911 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19913 * profile-count.h (apply_probability,
19914 apply_scale, probability_in): Fix checks for zero.
19916 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19918 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19919 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19921 2016-06-22 Richard Biener <rguenther@suse.de>
19923 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19924 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19925 REDUC_MAX_EXPR support.
19926 (vectorizable_reduction): Likewise.
19927 (vect_create_epilog_for_reduction): Likewise.
19929 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19931 * match.pd (A / (1 << B) -> A >> B): New.
19932 * generic-match-head.c: Include optabs-tree.h.
19933 * gimple-match-head.c: Likewise.
19934 * optabs-tree.h (target_supports_op_p): New.
19935 * optabs-tree.c (target_supports_op_p): New.
19937 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19939 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19940 $gcc_cv_ld --help output.
19941 (gcc_cv_ld_demangle): Likewise.
19942 (gcc_cv_ld_eh_frame_hdr): Likewise.
19943 (gcc_cv_ld_pie): Likewise.
19944 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19945 (gcc_cv_ld_buildid): Likewise.
19946 (gcc_cv_ld_sysroot): Likewise.
19947 (ld_bndplt_support): Likewise.
19948 (ld_pushpopstate_support): Likewise.
19949 * configure: Regenerate.
19950 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19952 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19955 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19956 operands indexes to avoid gap between operands and match_dups.
19958 2017-06-21 Andrew Pinski <apinski@cavium.com>
19960 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19961 Increment Arith_shift and Arith_shift_reg by 1.
19962 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19964 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19965 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19966 (aarch64_strip_extend): Add new argument and test for it.
19967 (aarch64_cheap_mult_shift_p): New function.
19968 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19969 add a cost if it is true.
19970 Update calls to aarch64_strip_extend.
19971 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19973 2017-06-21 Andrew Pinski <apinski@cavium.com>
19975 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19977 (thunderxt88): Likewise.
19978 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19979 (thunderx_prefetch_tune): New variable.
19980 (thunderx2t99_prefetch_tune): Update for the correct values.
19981 (thunderxt88_tunings): New variable.
19982 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19983 generic_prefetch_tune.
19984 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19986 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19988 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19989 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19990 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19991 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19992 (aarch64_atomic_cas<mode>, GPI): Likewise.
19994 2017-06-21 Martin Liska <mliska@suse.cz>
19996 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19997 statements on cold and hot labels.
19998 * predict.c (tree_estimate_probability_bb): Remove the
19999 prediction from this place.
20001 2017-06-21 Martin Liska <mliska@suse.cz>
20003 PR tree-optimization/79489
20004 * gimplify.c (maybe_add_early_return_predict_stmt): New
20006 (gimplify_return_expr): Call the function.
20007 * predict.c (tree_estimate_probability_bb): Remove handling
20009 * predict.def: Update comment about early return predictor.
20010 * gimple-predict.h (is_gimple_predict): New function.
20011 * predict.def: Change default value of early return to 66.
20012 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20014 * passes.def: Put pass_strip_predict_hints to the beginning of
20017 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20019 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20020 FUNCTION_DECL declarations.
20021 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20023 (dwaf2out_decl): Likewise.
20024 * godump.c (go_early_global_decl): Skip call to the real debug hook
20025 for FUNCTION_DECL declarations.
20026 * passes.c (rest_of_decl_compilation): Skip call to the
20027 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20028 -fdump-go-spec is passed.
20030 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20032 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20033 Reorder for compactness.
20034 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20035 (def_builtin_pure, def_builtin_pure2): New functions.
20036 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20038 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20040 * match.pd (nop_convert): New predicate.
20041 ((A +- CST1) +- CST2): Allow some NOP conversions.
20043 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20046 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20047 with ctors/dtors if GOVD_SHARED is set.
20049 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20051 * config/aarch64/aarch64.md (movti_aarch64):
20052 Emit mov rather than orr.
20053 (movtf_aarch64): Likewise.
20054 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20055 Emit mov rather than orr.
20057 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20059 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20060 Swap alternatives, make integer dup more expensive.
20062 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20064 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20065 Return true for non-tls symbols.
20067 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20069 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20070 (cortex-a75): Likewise.
20071 (cortex-a75.cortex-a55): Likewise.
20072 * config/aarch64/aarch64-tune.md: Regenerate.
20073 * doc/invoke.texi (-mtune): Document new values for -mtune.
20075 2017-06-21 Tom de Vries <tom@codesourcery.com>
20077 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20078 stack_size feature.
20079 (Effective-Target Keywords, Other attributes): Suggest using
20080 dg-add-options stack_size feature to get stack limit in stack_size
20081 effective target documentation.
20083 2017-06-21 Julian Brown <julian@codesourcery.com>
20084 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20086 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20087 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20088 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20090 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20091 attribute type list for neon_multiply.
20092 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20093 attribute type list for neon_multiply.
20094 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20095 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20096 attribute type list for neon_multiply.
20097 * config/arm/types.md (crypto_pmull): Add.
20098 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20099 attribute type list.
20101 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20103 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20106 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20108 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20109 to make sure not to dereference a NULL cost_classes_ptr pointer.
20111 2017-06-20 Carl Love <cel@us.ibm.com>
20113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20114 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20115 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20116 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20117 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20118 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20119 VMULOSW): New enum "unspec" values.
20120 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20121 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20122 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20123 altivec_vmulosw): New patterns.
20124 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20125 VMULOSW): Add definitions.
20127 2017-06-20 Julia Koval <julia.koval@intel.com>
20129 * config/i386/i386.c: Fix rounding expand for new pattern.
20130 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20132 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20134 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20135 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20137 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20139 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20142 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20144 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20145 architecture, then by implementer ID.
20146 * config/aarch64/aarch64-tune.md: Regenerate.
20148 2017-06-20 Richard Biener <rguenther@suse.de>
20150 PR middle-end/81097
20151 * fold-const.c (split_tree): Fold to type before negating.
20153 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20155 * diagnostic-show-locus.c
20156 (selftest::test_fixit_deletion_affecting_newline): New function.
20157 (selftest::diagnostic_show_locus_c_tests): Call it.
20159 2017-06-20 Andreas Schwab <schwab@suse.de>
20162 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20165 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20167 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20168 __ARM_FEATURE_COPROC according to support.
20170 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20172 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20173 Rewritten to avoid overflow for > 32-bit pointers.
20176 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20177 by removing enum keyword.
20178 (ubsan_type_descriptor): Likewise. Formatting fix.
20181 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20182 splitter): Require TARGET_SSE2 in the condition.
20184 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20187 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20188 for doing vector set of SFmode on ISA 3.0.
20189 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20190 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20192 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20193 SFmode value into a V4SF variable that was extracted from another
20194 V4SF variable without converting the element to double precision
20195 and back to single precision vector format.
20196 (vsx_insert_extract_v4sf_p9_2): Likewise.
20198 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20200 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20201 in UWHI to avoid undefined overflow.
20204 * ubsan.h (enum ubsan_encode_value_phase): New.
20205 (ubsan_encode_value): Change second argument to
20206 enum ubsan_encode_value_phase with default value of
20207 UBSAN_ENCODE_VALUE_GENERIC.
20208 * ubsan.c (ubsan_encode_value): Change second argument to
20209 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20210 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20211 create_tmp_var_raw instead of create_tmp_var and use a
20213 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20214 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20215 ubsan_encode_value callers.
20218 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20219 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20220 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20222 2017-06-19 Richard Biener <rguenther@suse.de>
20224 PR middle-end/81118
20225 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20226 estimates if we changed anything.
20228 2017-06-19 Richard Biener <rguenther@suse.de>
20230 PR tree-optimization/80887
20231 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20232 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20233 simplified lookups, then reset mprts_hook.
20234 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20236 (try_to_simplify): Likewise.
20238 2017-06-19 Martin Liska <mliska@suse.cz>
20241 * gimplify.c (gimplify_switch_expr):
20242 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20244 2017-06-19 Martin Liska <mliska@suse.cz>
20246 * doc/install.texi: Document that PGO runs in 4 stages.
20248 2017-06-19 Martin Liska <mliska@suse.cz>
20251 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20252 to dispatcher function name.
20253 * multiple_target.c (replace_function_decl): New function.
20254 (create_dispatcher_calls): Redirect both edges and references.
20256 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20258 * profile-count.c (profile_count::dump): Dump quality.
20259 (profile_count::differs_from_p): Update for unsigned val.
20260 * profile-count.h (profile_count_quality): New enum.
20261 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20263 2017-06-19 Richard Biener <rguenther@suse.de>
20265 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20266 struct function as arg.
20267 (estimate_numbers_of_iterations): Export overload with loop arg.
20268 (free_numbers_of_iterations_estimates_loop): Use an overload of
20269 free_numbers_of_iterations_estimates instead.
20270 * tree-cfg.c (remove_bb): Adjust.
20271 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20272 * tree-parloops.c (gen_parallel_loop): Likewise.
20273 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20275 (tree_unroll_loops_completely): Likewise.
20276 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20277 Use an overload instead and export.
20278 (estimated_loop_iterations): Adjust.
20279 (max_loop_iterations): Likewise.
20280 (likely_max_loop_iterations): Likewise.
20281 (estimate_numbers_of_iterations): Take struct function as arg
20283 (loop_exits_before_overflow): Adjust.
20284 (free_numbers_of_iterations_estimates_loop): Use an overload.
20285 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20286 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20288 2017-06-19 Richard Biener <rguenther@suse.de>
20291 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20292 RANGE_EXPR conservatively.
20294 2017-06-16 Carl Love <cel@us.ibm.com>
20296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20297 definitions for vec_float, vec_float2, vec_floato,
20298 vec_floate built-ins.
20299 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20300 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20302 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20303 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20304 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20305 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20306 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20307 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20308 vec_floato): Add builtin defines.
20309 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20310 Update the built-in documentation file for the new built-in
20313 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20315 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20316 (mthumb): Mark as the negative of -marm.
20318 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20320 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20322 (ARM Options, -mtune): Document that this accepts the same
20323 extension options as -mcpu.
20324 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20328 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20329 permitted extensions.
20331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20333 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20334 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20335 (armv8-m.main): Add option +nodsp.
20336 * config/arm/arm-cpu-cdata.h: Regenerated.
20338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20340 * config/arm/t-fuchsia: New file.
20341 * config.gcc (arm*-*-fuchsia*): Use it.
20343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20345 * config/arm/t-symbian: Rewrite for new option infrastructure.
20347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20349 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20350 (MULTILIB_REQUIRED): Likewise.
20352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20354 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20355 (MULTILIB_RESUE): Likewise.
20356 (MULTILIB_MATCHES): Likewise.
20357 (MULTLIB_REQUIRED): Likewise.
20359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20361 * config/arm/t-rtems: Rewrite for new option framework.
20363 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20365 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20366 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20367 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20368 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20369 * config/arm/t-multilib: ... here.
20370 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20371 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20372 armv7-a and armv8*-a when A-profile libraries have not been built.
20373 * config/arm/t-rmprofile: Rewrite.
20375 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20377 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20378 with a backslash. Remove the backslash after substituting unescaped
20380 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20384 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20385 the driver to pass through the default setting of -mfloat-abi.
20386 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20388 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20389 (all_feat_combs): New rule.
20390 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20392 * config/arm/t-aprofile: Rewrite.
20394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20396 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20397 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20398 fpu is not specified by the user/command-line.
20399 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20400 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20401 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20402 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20403 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20404 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20408 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20409 * config/arm/t-arm-elf: Rewritten.
20411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20413 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20414 have some floating-point instructions.
20415 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20416 (TARGET_MAYBE_HARD_FLOAT): New macro.
20417 * config/arm/arm-builtins.c (arm_init_builtins): Use
20418 TARGET_MAYBE_HARD_FLOAT.
20419 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20423 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20424 (configargs.h): Include it.
20425 (arm_print_hint_for_fpu_option): New function.
20426 (arm_parse_fpu_option): New function.
20427 (candidate_extension): New class.
20428 (arm_canon_for_multilib): New function.
20429 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20430 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20431 (ARCH_CANONICAL_SPECS): New macro.
20432 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20436 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20437 are set after handling multilib fragments. Set target_cpu_default2
20440 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20442 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20444 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20445 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20447 (check_arch): Likewise.
20448 * config/arm/arm.c (arm_configure_build_target): Handle
20449 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20450 options in the default.
20452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20454 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20455 when an option is an alias of another.
20456 * config/arm/parsecpu.awk (optalias): New parser token.
20457 (gen_comm_data): Mark non-alias options as such. Emit entries
20458 for extension aliases.
20459 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20460 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20461 (armv6kz, armv6zk, armv6t2): Likewise.
20462 (armv7): Make vfpv3-d16 an alias.
20463 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20465 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20466 Sort in canonical order.
20467 (armv8-a): Sort in canonical order.
20468 (armv8.1-a, armv8.2-a): Likewise.
20469 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20471 (cortex-a9): Sort in canonical order.
20472 * config/arm/arm.c (selftests.h): Include it.
20473 (arm_test_cpu_arch_data): New function.
20474 (arm_run_self_tests): New function.
20475 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20476 (targetm): Move declaration to the end of the file.
20477 * arm-cpu-cdata.h: Regenerated.
20479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20481 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20482 call to target_mode_check describing the type of option passed.
20483 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20484 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20485 arm_parse_cpu_option_name to match parameters against list of
20487 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20488 arm_arch_core_flags data structure.
20489 * config/arm/arm-cpu_cdata.h: Regenerated.
20491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20493 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20495 (arm_print_hint_for_cpu_option): Likewise.
20496 (arm_print_hint_for_arch_option): Likewise.
20497 (arm_parse_cpu_option_name): Likewise.
20498 (arm_parse_arch_option_name): Likewise.
20499 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20500 of entries in the all_fpus list.
20501 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20502 (arm_parse_cpu_option_name): Declare.
20503 (arm_parse_arch_option_name): Declare.
20504 (arm_parse_option_features): Declare.
20505 (arm_intialize_isa): Declare.
20506 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20508 (gen_comm_data): ... here. Make definitions non-static.
20509 * config/arm/arm-cpu-data.h: Regenerated.
20510 * config/arm/arm-cpu-cdata.h: Regenerated.
20512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20514 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20515 (cpu_arch_extension): New structure.
20516 (cpu_arch_option, arch_option, cpu_option): New structures.
20517 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20518 architecture types.
20519 (gen_data): Generate new format data tables.
20520 * config/arm/arm.c (cpu_tune): New structure.
20521 (cpu_option, processors): Delete.
20522 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20523 (arm_print_hint_for_cpu_option): ... this and ...
20524 (arm_print_hint_for_arch_option): ... this.
20525 (arm_parse_arch_cpu_name): Delete. Replace with ...
20526 (arm_parse_cpu_option_name): ... this and ...
20527 (arm_parse_arch_option_name): ... this.
20528 (arm_unrecognized_feature): Change type of target parameter to
20530 (arm_parse_arch_cpu_features): Delete. Replace with ...
20531 (arm_parse_option_features): ... this.
20532 (arm_configure_build_target): Rework to use new configuration data
20534 (arm_print_tune_info): Rework for new configuration data tables.
20535 * config/arm/arm-cpu-data.h: Regenerated.
20536 * config/arm/arm-cpu.h: Regenerated.
20538 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20540 * Makefile.in (OBJS): Move sbitmap.o from here ...
20541 (OBJS-libcommon): ... to here.
20543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20545 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20546 (ISA_ALL_CRYPTO): New macro.
20547 (ISA_ALL_SIMD): New macro
20548 (ISA_ALL_FP): New macro.
20549 * config/arm/arm.c (fpu_bitlist): Update initializer.
20550 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20552 (arm9e): Add fpu. Add option for nofp
20553 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20554 (arm926ej-s, arm1026ej-s): Likewise.
20555 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20556 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20557 neon-fp16, neon-vfpv4, nofp and nosimd.
20558 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20559 (cortex-a8): Add fpu. Add option for nofp.
20560 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20561 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20562 (cortex-r4f): Add fpu.
20563 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20564 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20566 (cortex-r8): Likewise.
20567 (cortex-m4): Add fpu. Add option for nofp.
20568 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20569 (cortex-a17.cortex-a7): Likewise.
20570 (cortex-a32): Add fpu. Add options for crypto and nofp.
20571 (cortex-a35, cortex-a53): Likewise.
20572 (cortex-a57): Add fpu. Add option for crypto.
20573 (cortex-a72, cortex-a73): Likewise.
20574 (exynos-m1): Likewise.
20575 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20576 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20577 (cortex-m33): Add fpu. Add option for nofp.
20578 * config/arm/arm-cpu-cdata.h: Regenerated
20579 * config/arm/arm-cpu-data.h: Regenerated.
20581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20583 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20584 (armv5te, armv5tej): Likewise.
20585 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20586 (armv7): Add options fp and vfpv3-d16.
20587 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20588 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20590 (armv7ve): Likewise.
20591 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20592 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20593 (armv8-a): Add nocrypto option.
20594 (armv8.1-a, armv8.2-a): Likewise.
20595 (armv8-m.main): add options fp, fp.dp and nofp.
20597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20599 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20601 (armv8-a+crc): Delete.
20602 (armv8.1-a): Add options simd, crypto and nofp.
20603 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20604 (armv8.2-a+fp16): Delete.
20605 (armv8-m.main): Add option dsp.
20606 (armv8-m.main+dsp): Delete.
20607 (cortex-a8): Add fpu. Add nofp option.
20608 (cortex-a9): Add fpu. Add nofp and nosimd options.
20609 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20610 link to main cpu and architecture data structures.
20611 (gen_comm_data): Only put isa attributes from the main architecture
20613 (option): New statement for architecture and CPU entries.
20614 * arm.c (struct cpu_option): New structure.
20615 (struct processors): Add entry for options.
20616 (arm_unrecognized_feature): New function.
20617 (arm_parse_arch_cpu_name): Ignore any characters after the first
20619 (arm_parse_arch_cpu_feature): New function.
20620 (arm_configure_build_target): Separate out any CPU and architecture
20621 features and parse separately. Don't error out if -mfpu=auto is
20622 used with only an architecture string.
20623 (arm_print_asm_arch_directives): New function.
20624 (arm_file_start): Call it.
20625 * config/arm/arm-cpu-cdata.h: Regenerated.
20626 * config/arm/arm-cpu-data.h: Likewise.
20627 * config/arm/arm-tables.opt: Likewise.
20629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20631 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20632 assembler when it is not -mfpu=auto.
20634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20636 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20637 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20638 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20639 (ASM_CPU_SPEC): Rewrite.
20640 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20641 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20642 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20643 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20644 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20645 copied string is NUL-terminated. Also strip any characters prefixed
20647 (arm_rewrite_selected_arch): New function.
20648 (arm_rewrite_march): New function.
20650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20652 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20653 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20654 (march, mcpu, mtune): Convert to string-based options.
20655 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20656 (arm_parse_arch_cpu_name): New function.
20657 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20658 identify selected architecture or CPU.
20659 (arm_option_save): New function.
20660 (TARGET_OPTION_SAVE): Redefine.
20661 (arm_option_restore): Restore string options.
20662 (arm_option_print): Print string options.
20664 2017-06-16 Martin Sebor <msebor@redhat.com>
20666 PR tree-optimization/80933
20667 PR tree-optimization/80934
20668 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20669 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20670 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20671 (gimple_fold_builtin): Call them.
20673 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20675 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20676 as unlikely; update profile.
20678 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20680 * predict.c (force_edge_cold): Handle declaring edges impossible
20683 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20685 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20687 (try_unroll_loop_completely): Fix reporting.
20689 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20691 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20693 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20696 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20697 if given a non-constant argument for an intrinsic which requires a
20700 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20702 * profile.c (compare_freqs): New function.
20703 (branch_prob): Sort edge list.
20704 (find_spanning_tree): Assume that the list is priority sorted.
20706 2017-06-16 Richard Biener <rguenther@suse.de>
20708 PR tree-optimization/81090
20709 * passes.def (pass_record_bounds): Remove.
20710 * tree-pass.h (make_pass_record_bounds): Likewise.
20711 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20712 make_pass_record_bounds): Likewise.
20713 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20714 not free niter estimates at the beginning but at the end.
20715 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20717 2017-06-16 Richard Biener <rguenther@suse.de>
20719 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20720 initializer to workaround ICE in host GCC 4.8.
20722 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20724 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20726 (clone_inlined_nodes): Update.
20728 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20730 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20731 prefetch settings, and enable prefetching by default at -O3.
20733 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20735 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20736 Set flag_prefetch_loop_arrays according to tuning data.
20738 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20740 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20741 New tune structure.
20742 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20743 [Unrelated to main purpose of the patch] Place the pointer field last
20744 to enable type checking errors when tune structure are wrongly merged.
20745 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20746 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20747 (thunderx2t99_prefetch_tune): New tune constants.
20748 (tune_params *_tunings): Update all tunings (no functional change).
20749 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20750 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20751 from tunings structures.
20753 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20756 * ubsan.c (instrument_null): Add T argument, use it instead
20757 of computing it based on IS_LHS.
20758 (instrument_object_size): Likewise.
20759 (pass_ubsan::execute): Adjust instrument_null and
20760 instrument_object_size callers to pass gimple_get_lhs or
20761 gimple_assign_rhs1 result to it. Use instrument_null instead of
20762 calling get_base_address and instrument_mem_ref. Handle
20763 aggregate call arguments for object-size sanitization.
20765 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20767 PR tree-optimization/81089
20768 * tree-vrp.c (is_masked_range_test): Validate operands of
20771 2017-06-15 Martin Sebor <msebor@redhat.com>
20774 * dumpfile.c (dump_register): Avoid calling memset to initialize
20775 a class with a default ctor.
20776 * gcc.c (struct compiler): Remove const qualification.
20777 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20778 * hash-table.h: Ditto.
20779 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20781 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20782 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20784 * params.h (struct param_info): Make struct members non-const.
20785 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20786 with default initialization.
20787 * vec.h (vec_copy_construct, vec_default_construct): New helper
20789 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20790 with vec_copy_construct.
20791 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20792 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20793 * doc/invoke.texi (-Wclass-memaccess): Document.
20795 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20797 * emit-rtl.h (is_leaf): Update comment about local
20798 register allocator.
20800 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20803 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20804 for a variable to have a section before checking if the section has a
20806 Set section to.persistent if persistent attribute is set.
20807 Warn if .persistent attribute is used on an automatic variable.
20809 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20811 PR rtl-optimization/80474
20812 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20814 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20816 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20819 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20821 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20822 (casesi): Emit bounds checking as RTL.
20823 (casesi_internal_mips16_<mode>): Remove bounds checking.
20825 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20827 * config/xtensa/xtensa.c (xtensa_option_override): Append
20828 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20829 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20830 xtensa_doloop_hooks): Define unconditionally.
20831 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20833 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20834 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20835 for it in xtensa_option_override.
20836 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20837 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20839 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20841 * doc/cppopts.texi: Document '-' special value to -MF.
20843 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20845 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20846 (cortex_a53_fconst): Likewise.
20847 (cortex_a53_fpmul): Likewise.
20848 (cortex_a53_f_load_64): Likewise.
20849 (cortex_a53_f_load_many): Likewise.
20850 (cortex_a53_advsimd_alu): Likewise.
20851 (cortex_a53_advsimd_alu_q): Likewise.
20852 (cortex_a53_advsimd_mul): Likewise.
20853 (cortex_a53_advsimd_mul_q): Likewise.
20854 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20855 Add missing fmul, r2f_cvt and fconst cases.
20857 2017-06-14 Richard Biener <rguenther@suse.de>
20859 PR middle-end/81088
20860 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20862 (fold_binary_loc): When associating do not treat pre-existing
20863 TREE_OVERFLOW on literal constants as a reason to allow
20864 TREE_OVERFLOW on associated literal constants.
20866 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20868 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20869 (MASK_FEATURES): New macro.
20870 * config/sparc/sparc.c (sparc_option_override): Remove the special
20871 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20873 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20875 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20876 a division of 0 if non-call exceptions are enabled.
20878 2017-06-14 Andrew Pinski <apinski@cavium.com>
20879 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20882 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20883 Improve vector initialization code gen for only variable case.
20885 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20887 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20889 2017-06-14 Richard Biener <rguenther@suse.de>
20891 PR tree-optimization/81083
20892 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20895 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20897 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20898 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20899 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20900 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20901 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20902 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20904 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20906 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20907 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20909 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20911 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20913 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20915 * config/rs6000/t-rtems: Don't handle SPE.
20917 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20919 * config/rs6000/t-linux: Don't handle SPE.
20921 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20923 * config/rs6000/eabispe.h: Delete file.
20925 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20927 * config/rs6000/t-spe: Delete file.
20929 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20931 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20932 (rs6000_legitimate_offset_address_p): Return false for anything in
20933 V2SImode or V2SFmode.
20935 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20937 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20938 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20939 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20941 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20942 (rs6000_legitimate_offset_address_p): Ditto.
20943 (rs6000_emit_move): Ditto.
20944 (rs6000_init_builtins): Remove V4HI_type_node.
20946 2017-06-13 Martin Liska <mliska@suse.cz>
20949 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20950 (gate_asan): Likewise.
20951 * asan.h (asan_no_sanitize_address_p): Remove the function.
20952 (sanitize_flags_p): New function.
20953 * builtins.def: Fix coding style.
20954 * common.opt: Use renamed enum value.
20955 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20956 * doc/extend.texi: Document no_sanitize attribute.
20957 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20958 to SANITIZE_UNDEFINED_NONDEFAULT.
20959 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20960 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20961 Use sanitize_flags_p.
20962 * gimplify.c (gimplify_function_tree): Likewise.
20963 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20964 * opts.c (parse_no_sanitize_attribute): New function.
20965 (common_handle_option): Use renamed enum value.
20966 * opts.h (parse_no_sanitize_attribute): Declare.
20967 * tree.c (sanitize_flags_p): New function.
20968 * tree.h: Declared here.
20969 * tsan.c: Use sanitize_flags_p.
20970 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20971 (instrument_mem_ref): Likewise.
20972 (instrument_bool_enum_load): Likewise.
20973 (do_ubsan_in_current_function): Remove the function.
20974 (pass_ubsan::execute): Use sanitize_flags_p.
20975 * ubsan.h: Remove do_ubsan_in_current_function
20976 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20977 (dump_function_to_file): Use it here.
20979 2017-06-13 Martin Jambor <mjambor@suse.cz>
20981 PR tree-optimization/80803
20982 PR tree-optimization/81063
20983 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20984 (propagate_subaccesses_across_link): Enqueue subtree whenever
20985 necessary instead of relying on the caller.
20987 2017-06-13 Martin Jambor <mjambor@suse.cz>
20989 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20990 that have a first_link.
20991 (sort_and_splice_var_accesses): Do not check first_link before
20993 (subtree_mark_written_and_enqueue): Likewise.
20994 (propagate_all_subaccesses): Likewise and do not stop at first
20995 parent with a first_link.
20997 2017-06-13 Martin Jambor <mjambor@suse.cz>
20999 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21002 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21004 * match.pd: New pattern.
21006 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21008 * tree-vrp.c (is_masked_range_test): New function.
21009 (register_edge_assert_for): Determine ranges for
21012 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21014 PR tree-optimization/67328
21015 * fold-const.c (maskable_range_p): New function.
21016 (build_range_check): Generate bittests if possible.
21018 2017-06-13 Martin Liska <mliska@suse.cz>
21020 * gimple-pretty-print.c (dump_probability): Add new argument.
21021 (dump_edge_probability): Dump both probability and count.
21022 (dump_gimple_label): Likewise.
21023 (dump_gimple_bb_header): Likewise.
21025 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21028 * config/avr/avr-devices.c: Fix indentation.
21029 * config/avr/gen-avr-mmcu-specs.c: Dito.
21031 2017-06-13 Richard Biener <rguenther@suse.de>
21033 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21034 instead get vector type from stmt_info.
21035 (vectorizable_reduction): Adjust. Remove dead code.
21037 2017-06-13 Richard Biener <rguenther@suse.de>
21039 PR middle-end/81065
21040 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21041 case of C * (x * C2 + C3).
21042 (fold_addr_of_array_ref_difference): Properly fold index difference.
21044 2017-06-12 David S. Miller <davem@davemloft.net>
21047 * config/sparc/sparc.md (return expander): Emit frame blockage if
21048 function uses alloca.
21050 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21052 * combine.c (make_field_assignment): Check len rather than the mode
21053 precision when calling force_to_mode.
21055 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21057 Support multilibs and devices that see flash in RAM address range.
21060 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21061 (avr_mcu_t) <flash_pm_offset>: New field.
21062 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21063 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21064 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21065 (AVR_TINY_PM_OFFSET): Remove macro.
21066 * config/avr/avr.opt (-mshort-calls): New option.
21067 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21068 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21069 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21070 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21071 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21072 instead of avr_arch->have_jmp_call.
21073 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21074 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21075 avr_arch->flash_pm_offset to define.
21076 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21077 new field flash_pm_offset. Add entry for avrxmega3.
21078 (avr_texinfo): Add entry for avrxmega3.
21079 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21080 attiny212, attiny214,
21081 attiny412, attiny414, attiny416, attiny417,
21082 attiny814, attiny816, attiny817,
21083 attiny1614, attiny1616, attiny1617,
21084 attiny3214, attiny3216, attiny3217.
21085 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21086 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21087 (avr_print_operand_address) [AVR_TINY]: Same.
21088 (avr_asm_init_sections) <readonly_data_section>: Only patch
21089 callback if avr_arch->flash_pm_offset = 0.
21090 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21091 for rodata if avr_arch->flash_pm_offset != 0.
21092 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21093 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21094 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21095 (m_options): Append opt_rcall.
21096 (m_dirnames): Append dir_rcall.
21097 * config/avr/t-multilib: Regenerate.
21099 * configure.ac [target=avr]: Check whether avrxmega3 default
21100 linker description file works as needed.
21101 * configure: Regenerate.
21102 * doc/avr-mmcu.texi: Regenerate.
21103 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21104 <__AVR_ARCH__>: Document avrxmega3 and 103.
21105 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21106 <__AVR_SHORT_CALLS__>: Document it.
21107 <__AVR_PM_BASE_ADDRESS__>: Document it.
21108 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21109 (AVR Variable Attributes) <progmem>: Document this is
21110 not needed for avrxmega3.
21111 (AVR Named Address Spaces) <__flash>: Dito.
21113 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21115 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21117 2017-06-12 Doug Rupp <rupp@adacore.com>
21119 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21120 Append vxworks-stdint.h to the tm_file list.
21121 * config/vxworks-stdint.h: New file.
21123 2017-06-12 Martin Liska <mliska@suse.cz>
21125 PR tree-optimization/81041
21126 * tree-profile.c (gimple_gen_ic_func_profiler):
21127 Create an extra BB in profile-generate
21128 (gimple_gen_time_profiler): Likewise.
21130 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21132 PR tree-optimization/81003
21133 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21134 (update_range_test): Use it instead of force_gimple_operand_gsi.
21136 2017-06-12 Richard Biener <rguenther@suse.de>
21138 PR tree-optimization/81053
21139 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21140 with backedge value not defined in loop. Simplify def stmt
21143 2017-06-11 Tom de Vries <tom@codesourcery.com>
21146 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21148 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21149 nvptx_cannot_force_const_mem.
21151 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21153 * opts.c (finish_options): Move test for flag_split_stack after
21154 it has been initialized.
21156 2017-06-11 Jason Merrill <jason@redhat.com>
21158 * tree.h (id_equal): New.
21159 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21160 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21161 instead of strcmp of IDENTIFIER_POINTER.
21163 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21165 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21166 (mark_all_inlined_calls_cdtor): Fix formating.
21167 (inline_transform): Rescale profile before inlining.
21169 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21171 * cgraph.h (cgraph_edge::clone): Update prototype.
21172 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21173 (cgraph_node::create_clone): Update.
21174 (cgraph_node::create_version_clone): Update.
21175 * tree-inline.c (copy_bb): Update.
21176 (expand_call_inline): Update.
21178 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21180 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21181 factored out from ...
21182 (rs6000_emit_prologue): ... here.
21184 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21186 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21187 factored out from ...
21188 (rs6000_emit_prologue): ... here.
21190 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21192 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21194 (handle_missing_profiles): Fix computation of tp_first_run.
21195 (counts_to_freqs): Do not touch freqs when count is 0.
21197 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21199 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21202 2017-06-10 Tom de Vries <tom@codesourcery.com>
21204 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21205 attributes): Document signal effective target.
21207 2017-06-10 Tom de Vries <tom@codesourcery.com>
21209 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21210 Document effective target stack_size.
21212 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21214 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21215 to the edit_context if they can be auto-applied.
21217 2017-06-9 Ian Lance Taylor <iant@golang.org>
21219 * opts.c (finish_options): If -fsplit-stack, disable implicit
21220 -forder-blocks-and-partition.
21221 * doc/invoke.texi (Optimize Options): Document that when using
21222 -fsplit-stack -forder-blocks-and-partition is not implicitly
21225 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21227 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21228 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21229 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21230 * builtins.def (abort, trap, unreachable): Declare cold.
21231 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21232 * tree-core.h (ECF_COLD): New.
21233 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21234 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21236 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21238 * predict.c (unlikely_executed_stmt_p): Cleanup.
21240 2017-06-09 Richard Biener <rguenther@suse.de>
21242 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21243 model if the ref is always written to.
21245 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21247 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21249 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21251 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21254 2017-06-09 Tom de Vries <tom@codesourcery.com>
21257 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21258 "target cannot support label values" when encountering LABEL_REF.
21260 2017-06-09 Martin Liska <mliska@suse.cz>
21262 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21263 (gimple_gen_ic_func_profiler): Emit direct comparison
21264 of __gcov_indirect_call_callee with NULL.
21265 (gimple_gen_time_profiler): Change probability from
21266 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21268 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21270 * profile.c (edge_gcov_counts): Turn to pointer.
21271 (compute_branch_probabilities, compute_branch_probabilities): Update.
21272 (branch_prob): Do not clear edge_gcov_count.
21273 * profile.h (edge_gcov_counts): Turn to pointer.
21274 (edge_gcov_count): Update.
21276 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21278 * gimple.h (gimple_check_failed): Mark cold.
21280 2017-06-09 Richard Biener <rguenther@suse.de>
21282 PR tree-optimization/66623
21283 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21284 refactor check_reduction into two parts, properly computing
21285 whether we have to check reduction validity for outer loop
21288 2017-06-09 Richard Biener <rguenther@suse.de>
21290 PR tree-optimization/79483
21291 * graphite-scop-detection.c (order): New global.
21292 (get_order): Compute bb to order mapping that satisfies code
21293 generation constraints.
21294 (cmp_pbbs): New helper.
21295 (build_scops): Start domwalk at entry block, sort generated
21298 2017-06-09 Richard Biener <rguenther@suse.de>
21300 PR middle-end/81007
21301 * ipa-polymorphic-call.c
21302 (ipa_polymorphic_call_context::restrict_to_inner_class):
21303 Skip FIELD_DECLs with error_mark_node type.
21304 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21307 2017-06-09 Martin Liska <mliska@suse.cz>
21309 * predict.c (struct branch_predictor): New struct.
21310 (test_prediction_value_range): New test.
21311 (predict_c_tests): New function.
21312 * selftest-run-tests.c (selftest::run_tests): Run the function.
21313 * selftest.h: Declare new tests.
21315 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21318 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21319 gen_add3_insn did not fail.
21320 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21321 r0, construct that number in a temporary reg and add that reg to r0.
21322 If asked to put the result in r0 as well, fail.
21324 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21326 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21327 for early expansion of vec_eqv.
21329 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21331 PR middle-end/81005
21332 * ubsan.c (instrument_null): Avoid pointless code temporary.
21333 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21335 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21339 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21342 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21345 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21347 * predict.c (unlikely_executed_edge_p): Move ahead.
21348 (probably_never_executed_edge_p): Use it.
21350 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21352 PR middle-end/79988
21353 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21354 gimple_call_builtin_p call.
21356 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21358 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21359 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21360 rtl_check_failed_type2, rtl_check_failed_code1,
21361 rtl_check_failed_code2, rtl_check_failed_code_mode,
21362 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21363 rtvec_check_failed_bounds, rtl_check_failed_flag,
21364 _fatal_insn_not_found, _fatal_insn): Likewise.
21365 * tree.h (tree_contains_struct_check_failed,
21366 tree_check_failed, tree_not_check_failed,
21367 tree_class_check_failed, tree_range_check_failed,
21368 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21369 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21370 tree_operand_check_failed, omp_clause_check_failed,
21371 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21374 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21376 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21377 flag_branch_probabilities.
21378 * ipa-inline.c (edge_badness): Likewise.
21379 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21380 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21381 * predict.c (maybe_hot_frequency_p): Likewise.
21382 (probably_never_executed): Likewise.
21383 * sched-ebb.c (schedule_ebbs): Likewise.
21384 * sched-rgn.c (find_single_block_region): Likewise.
21385 * tracer.c (tail_duplicate): Likewise.
21387 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21389 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21390 longer requires x_flag_profile_use.
21392 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21394 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21395 instead of flag_reorder_blocks_and_partition.
21396 * dbxout.c (dbxout_function_end): Likewise.
21397 * dwarf2out.c (gen_subprogram_die): Likewise.
21398 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21399 * hw-doloop.c (reorg_loops): Likewise.
21400 * varasm.c (assemble_start_function,
21401 assemble_end_function): Likewise.
21402 (decide_function_section): Do not check for
21403 flag_reorder_blocks_and_partition.
21405 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21407 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21409 (chkp_get_hard_register_fake_addr_expr): Ditto.
21410 (chkp_build_addr_expr): Add check for hard reg case.
21411 (chkp_parse_array_and_component_ref): Ditto.
21412 (chkp_find_bounds_1): Ditto.
21413 (chkp_process_stmt): Don't generate bounds store for
21416 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21418 * predict.c (maybe_hot_bb_p): Do not check profile status.
21419 (maybe_hot_edge_p): Likewise.
21420 (probably_never_executed): Check for zero counts even if profile
21422 (unlikely_executed_edge_p): New function.
21423 (unlikely_executed_stmt_p): New function.
21424 (unlikely_executed_bb_p): New function.
21425 (set_even_probabilities): Use unlikely predicates.
21426 (combine_predictions_for_bb): Likewise.
21427 (predict_paths_for_bb): Likewise.
21428 (predict_paths_leading_to_edge): Likewise.
21429 (determine_unlikely_bbs): New function.
21430 (estimate_bb_frequencies): Use it.
21431 (compute_function_frequency): Use zero counts even if profile is
21433 * profile-count.h: Fix typo.
21435 2017-08-08 Julia Koval <julia.koval@intel.com>
21437 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21438 _mm512_mask_cvtsepi16_storeu_epi8,
21439 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21440 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21441 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21442 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21443 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21444 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21445 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21446 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21447 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21448 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21449 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21450 __builtin_ia32_pmovuswb256mem_mask,
21451 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21452 __builtin_ia32_pmovwb512mem_mask): New builtins.
21454 2017-08-08 Julia Koval <julia.koval@intel.com>
21456 PR target/73350,80862
21457 * config/i386/subst.md (round): Fix round pattern.
21458 * config/i386/i386.c (ix86_erase_embedded_rounding):
21459 Fix erasing rounding for the fixed pattern.
21461 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21463 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21465 2017-06-08 Martin Liska <mliska@suse.cz>
21467 PR gcov-profile/80911
21468 * gcov.c (block_info::block_info): New constructor.
21470 2017-06-07 Carl Love <cel@us.ibm.com>
21472 * config/rs6000/rs6000-c: The return type of the following
21473 built-in functions was implemented as int not long long. Fix sign
21474 of return value for the unsigned version of vec_mulo and vec_mule.
21475 vector unsigned long long vec_bperm (vector unsigned long long,
21476 vector unsigned char)
21477 vector signed long long vec_mule (vector signed int,
21479 vector unsigned long long vec_mule (vector unsigned int,
21480 vector unsigned int)
21481 vector signed long long vec_mulo (vector signed int,
21483 vector unsigned long long vec_mulo (vector unsigned int,
21484 vector unsigned int)
21485 * doc/extend.texi: Fix the documentation for the built-in
21488 2017-06-07 Carl Love <cel@us.ibm.com>
21491 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21494 2017-06-07 Carl Love <cel@us.ibm.com>
21496 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21497 support, Generate doublehv for signed int/float for BE case only.
21499 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21501 * doc/invoke.texi (mcx16): Rewrite.
21503 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21505 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21506 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21507 *mov<mode>_softfloat, and an anonymous splitter): Use
21508 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21510 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21512 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21514 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21515 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21516 (rs6000_debug_reg_global): Adjust.
21517 (rs6000_init_hard_regno_mode_ok): Adjust.
21518 (rs6000_dbx_register_number): Adjust.
21519 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21520 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21521 Remove SPE_ACC and SPEFSCR.
21522 (REG_ALLOC_ORDER): Ditto.
21523 (FRAME_POINTER_REGNUM): Change to 111.
21524 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21525 (REG_CLASS_NAMES): Ditto.
21526 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21527 (REGISTER_NAMES): Ditto.
21528 (ADDITIONAL_REG_NAMES): Ditto.
21529 (rs6000_reg_names): Ditto.
21530 * config/rs6000/rs6000.md: Renumber some register number
21533 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21535 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21537 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21538 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21540 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21541 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21542 Delete the SPE high registers.
21543 (REG_ALLOC_ORDER): Ditto.
21544 (enum reg_class): Remove SPE_HIGH_REGS.
21545 (REG_CLASS_NAMES): Ditto.
21546 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21547 (REGISTER_NAMES): Ditto.
21548 (rs6000_reg_names): Ditto.
21549 * doc/tm.texi.in: Remove SPE as example.
21550 * doc/tm.texi: Regenerate.
21552 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21554 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21555 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21556 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21557 * config/rs6000/rs6000.md (type): Remove "brinc".
21559 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21562 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21563 * config/rs6000/linuxspe.h: Delete file.
21564 * config/rs6000/rs6000.md: Don't include spe.md.
21565 * config/rs6000/spe.h: Delete file.
21566 * config/rs6000/spe.md: Delete file.
21567 * config/rs6000/t-rs6000: Remove spe.md.
21569 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21571 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21572 (reg_or_none500mem_operand): Delete.
21573 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21574 instead of reg_or_none500mem_operand.
21576 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21579 handling of SPE flags.
21580 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21582 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21584 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21586 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21587 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21588 paired_divv2sf3): Similar.
21589 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21590 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21591 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21593 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21594 Rename the paired_* instruction patterns.
21595 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21597 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21598 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21599 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21600 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21601 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21602 PAIRED_VECTOR_MODE.
21603 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21604 (spe_func_has_64bit_regs_p): Delete.
21605 (spe_expand_predicate_builtin): Delete.
21606 (spe_expand_evsel_builtin): Delete.
21607 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21608 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21609 (invalid_e500_subreg): Delete.
21610 (rs6000_legitimize_address): Always force_reg op2 as well, for
21611 paired single memory accesses.
21612 (rs6000_member_type_forces_blk): Delete.
21613 (rs6000_spe_function_arg): Delete.
21614 (rs6000_expand_unop_builtin): Delete SPE handling.
21615 (rs6000_expand_binop_builtin): Ditto.
21616 (spe_expand_stv_builtin): Delete.
21617 (bdesc_2arg_spe): Delete.
21618 (spe_expand_builtin): Delete.
21619 (spe_expand_predicate_builtin): Delete.
21620 (spe_expand_evsel_builtin): Delete.
21621 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21622 (spe_init_builtins): Delete.
21623 (spe_func_has_64bit_regs_p): Delete.
21624 (savres_routine_name): Delete "info" parameter. Adjust callers.
21625 (rs6000_emit_stack_reset): Ditto.
21626 (rs6000_dwarf_register_span): Delete.
21627 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21628 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21629 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21631 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21633 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21634 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21635 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21636 mulv2sf3, divv2sf3): Delete expanders.
21638 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21640 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21642 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21644 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21645 * config/rs6000/rs6000.c: Ditto.
21647 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21649 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21650 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21651 comparison_operator.
21653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21655 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21656 * config/rs6000/rs6000.opt: Ditto.
21657 * config/rs6000/t-rtems: Ditto.
21659 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21661 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21662 TARGET_E500_SINGLE by 0, simplify.
21663 * config/rs6000/rs6000.c: Ditto.
21664 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21665 (spe_build_register_parallel): Delete.
21666 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21667 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21668 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21669 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21670 (E500_CONVERT): Delete.
21671 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21673 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21676 * config/rs6000/dfp.md: Ditto.
21677 (negdd2, *negdd2_fpr): Merge.
21678 (absdd2, *absdd2_fpr): Merge.
21679 (negtd2, *negtd2_fpr): Merge.
21680 (abstd2, *abstd2_fpr): Merge.
21681 * config/rs6000/e500.h: Delete file.
21682 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21683 TARGET_FPRS by 1 and simplify.
21684 * config/rs6000/rs6000-c.c: Ditto.
21685 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21686 TARGET_DF_SPE by 0.
21687 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21689 * config/rs6000/rs6000.md: Ditto.
21690 (floatdidf2, *floatdidf2_fpr): Merge.
21691 (move_from_CR_gt_bit): Delete.
21692 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21693 (E500_CR_IOR_COMPARE): Delete.
21694 (All patterns that require !TARGET_FPRS): Delete.
21695 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21697 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21699 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21701 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21703 * graphds.c (add_edge): Intitialize edge's attached data.
21704 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21705 pointer parameter. Call pointed function on each edge during
21706 graph traversing. Skip traversing the edge when the function
21708 (graphds_dfs, graphds_scc): Ditto.
21709 (for_each_edge): New parameter. Pass the new parameter to callback
21711 * graphds.h (skip_edge_callback): New function pointer type.
21712 (graphds_dfs, graphds_scc): New function pointer parameter.
21713 (graphds_edge_callback, for_each_edge): New parameter.
21715 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21717 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21718 out code checking if runtime alias check is possible to below ...
21719 Call the new function.
21720 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21721 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21723 2017-06-07 Marek Polacek <polacek@redhat.com>
21726 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21727 TYPE_OVERFLOW_WRAPS check.
21729 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21731 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21732 if versioning is required.
21733 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21734 peeling with the check for versioning.
21736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21738 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21739 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21740 Set true to new parameter if new ssa variable is defined.
21741 (vect_gen_vector_loop_niters): Refactor. Set range information
21742 for the new vector loop bound variable.
21743 (vect_do_peeling): Ditto.
21745 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21747 * tree-affine.c (ssa.h): Include header file.
21748 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21749 has wrapping overflow behavior.
21751 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21753 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21755 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21757 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21758 (tree_to_aff_combination): ... here.
21760 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21762 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21763 reg_pressure model function.
21764 (ivopts_global_cost_for_size): Delete.
21765 (determine_set_costs, iv_ca_recount_cost): Call new model function
21766 ivopts_estimate_reg_pressure.
21768 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21770 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21771 expensive than udiv. Remove floating point cases from mod.
21773 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21775 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21776 Increase idiv cost.
21778 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21780 * config/aarch64/aarch64.md
21781 (copysignsf3): Fix mask generation.
21783 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21785 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21787 (class dump_manager): Add register_dumps method.
21788 * dumpfile.c: Include langhooks.h.
21789 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21790 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21791 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21792 (dump_manager::dump_register): Start with 512 entries instead of 32.
21793 (dump_manager::register_dumps): New method.
21794 * toplev.c (general_init): Instead of invoking register_dumps
21795 langhook, invoke register_dumps method on the dump manager.
21796 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21799 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21801 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21802 Say that # requires an associated define_split to exist, and that
21803 the define_split must be suitable for use after register allocation.
21805 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21807 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21808 (compute_outgoing_frequencies): Also initialize zero counts.
21809 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21810 around loops; preserve more of profile when nothing changes.
21812 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21814 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21816 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21817 * config/arm/arm-cpu-cdata.h: Regenerate.
21818 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21819 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21820 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21821 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21822 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21824 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21826 * config/arm/t-rmprofile: Likewise.
21827 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21829 2017-06-06 David S. Miller <davem@davemloft.net>
21832 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21833 blockage if function uses alloca.
21835 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21837 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21838 New "uid" fields to hold pretty-print IDs of group and ref.
21839 Memory references are now identified as <group_id>:<ref_id>
21840 instead of using [random] addresses.
21841 (dump_mem_details): Simplify, no functional change.
21842 (dump_mem_ref): Simplify and make output more concise.
21843 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21844 (find_or_create_group): Initialize group uid.
21845 (record_ref): Initialize ref uid. Improve debug output.
21846 (prune_group_by_reuse, should_issue_prefetch_p,)
21847 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21848 (mark_nontemporal_store, determine_loop_nest_reuse):
21849 Improve debug output.
21851 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21853 * dbgcnt.def (prefetch): New debug counter.
21854 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21855 (schedule_prefetches): Stop issueing prefetches if debug counter
21858 2017-06-06 Tom de Vries <tom@codesourcery.com>
21860 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21861 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21863 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21865 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21866 Use aarch64_reg_or_zero predicate for operand 4.
21867 (aarch64_compare_and_swap<mode> define_insn_and_split):
21868 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21869 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21871 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21873 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21874 (arm_compute_save_core_reg_mask): This.
21875 (thumb1_compute_save_reg_mask): Rename into ...
21876 (thumb1_compute_save_core_reg_mask): This.
21877 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21878 (arm_compute_frame_layout): Likewise.
21880 2017-06-06 Richard Biener <rguenther@suse.de>
21882 PR tree-optimization/80974
21883 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21884 keep or clear leaders SSA info.
21886 2017-06-06 Tom de Vries <tom@codesourcery.com>
21888 * config/nvptx/nvptx.c (split_mode_p): New function.
21889 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21891 2017-06-06 Tom de Vries <tom@codesourcery.com>
21893 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21895 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21898 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21901 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21903 * shrink-wrap.c (handle_simple_exit): Update profile.
21904 (try_shrink_wrapping): Upate profile.
21906 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21908 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21909 (tree_guess_outgoing_edge_probabilities): New.
21910 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21911 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21913 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21915 * ipa-split.c (split_function): Initialize return bb profile.
21917 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21919 * profile.c (compute_branch_probabilities): Also initialize
21920 EXIT_BLOCK profile.
21922 2017-06-06 Richard Biener <rguenther@suse.de>
21924 PR tree-optimization/80928
21925 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21926 (vect_analyze_loop_operations): Properly guard analysis for
21928 (vect_transform_loop): Likewise.
21929 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21930 (vect_model_induction_cost): Do not cost for pure SLP.
21931 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21932 of induction in inner loop vectorization.
21933 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21934 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21935 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21937 (vect_analyze_slp_cost_1): Cost induction.
21938 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21939 (vect_get_slp_vect_defs): Likewise.
21940 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21941 (vect_transform_stmt): Handle SLP reductions.
21942 * tree-vectorizer.h (vectorizable_induction): Adjust.
21944 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21946 * config/rs6000/rs6000.c (make_resolver_func): Update
21947 init_lowered_empty_function call.
21949 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21951 * doc/invoke.texi: Document the -fprofile-abs-path option.
21952 * common.opt (fprofile-abs-path): New option.
21953 * gcov-io.h (gcov_write_filename): Declare.
21954 * gcov-io.c (gcov_write_filename): New function.
21955 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21956 * profile.c (output_location): Likewise.
21958 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21960 * shring-wrap.c: Revert accidental commit.
21962 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21964 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21966 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21968 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21970 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21971 profile in callgraph edge.
21972 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21973 (apply_scale): Likewise.
21974 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21975 Also scale profile when inlining function with zero profile.
21976 (initialize_cfun): Update exit block profile even when it is zero.
21977 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21978 when profile is read.
21980 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21982 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21983 (CLONE_*): New constants to define the processors we can generate
21984 code for with the target_clone attribute.
21985 (rs6000_clone_map): New array to identify which clone processors
21986 the current program is running on.
21987 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21988 target_clone attribute.
21989 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21990 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21991 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21992 (cpu_expand_builtin): Add support for target_clone attribute.
21993 (rs6000_valid_attribute_p): Allow "default" attribute.
21994 (get_decl_name): New debug function to simplify printing the
21995 current function name in debugging statements.
21996 (rs6000_clone_priority): New functions to support the target_clone
21997 attribute, and be able to generate code to switch between ISA 2.05
21998 through ISA 3.0 (power6 through power9).
21999 (rs6000_compare_version_priority): Likewise.
22000 (rs6000_get_function_versions_dispatcher): Likewise.
22001 (make_resolver_func): Likewise.
22002 (add_condition_to_bb): Likewise.
22003 (dispatch_function_versions): Likewise.
22004 (rs6000_generate_version_dispatcher_body): Likewise.
22005 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22006 (fusion_gpr_load_p): Fix a spacing issue.
22007 * doc/extend.texi (Common Function Attributes): Document that the
22008 PowerPC supports the target_clone attribute.
22010 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22012 * config/arm/arm.h: explain F symbol found in description of ARM
22013 register allocation in its legend.
22015 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22017 * config/mips/frame-header-opt.c: Include profile-count.h.
22018 * config/riscv/riscv.c: Include profile-count.h
22020 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22022 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22024 (sm_set_flag_if_changed): Add bbs field.
22025 (execute_sm_if_changed_flag_set): Pass BBS.
22026 (execute_sm): Update.
22028 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22030 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22033 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22035 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22036 (peephole2): New peephole2 to emit the above.
22037 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22039 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22041 * config/aarch64/aarch64.c (define_peephole2 above
22042 *sub_<shift>_<mode>): New peephole.
22044 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22046 * config/i386/i386.c (make_resolver_func): Update.
22047 * Makefile.in: Add profile-count.h and profile-count.o
22048 * auto-profile.c (afdo_indirect_call): Update to new API.
22049 (afdo_set_bb_count): Update.
22050 (afdo_propagate_edge): Update.
22051 (afdo_propagate_circuit): Update.
22052 (afdo_calculate_branch_prob): Update.
22053 (afdo_annotate_cfg): Update.
22054 * basic-block.h: Include profile-count.h
22055 (struct edge_def): Turn count to profile_count.
22056 (struct basic_block_def): Likewie.
22057 (REG_BR_PROB_BASE): Move to profile-count.h
22058 (RDIV): Move to profile-count.h
22059 * bb-reorder.c (max_entry_count): Turn to profile_count.
22060 (find_traces): Update.
22061 (rotate_loop):Update.
22062 (connect_traces):Update.
22063 (sanitize_hot_paths):Update.
22064 * bt-load.c (migrate_btr_defs): Update.
22065 * cfg.c (RDIV): Remove.
22066 (init_flow): Use alloc_block.
22067 (alloc_block): Uninitialize count.
22068 (unchecked_make_edge): Uninitialize count.
22069 (check_bb_profile): Update.
22070 (dump_edge_info): Update.
22071 (dump_bb_info): Update.
22072 (update_bb_profile_for_threading): Update.
22073 (scale_bbs_frequencies_int): Update.
22074 (scale_bbs_frequencies_gcov_type): Update.
22075 (scale_bbs_frequencies_profile_count): New.
22076 * cfg.h (update_bb_profile_for_threading): Update.
22077 (scale_bbs_frequencies_profile_count): Declare.
22078 * cfgbuild.c (compute_outgoing_frequencies): Update.
22079 (find_many_sub_basic_blocks): Update.
22080 * cfgcleanup.c (try_forward_edges): Update.
22081 (try_crossjump_to_edge): Update.
22082 * cfgexpand.c (expand_gimple_tailcall): Update.
22083 (construct_exit_block): Update.
22084 * cfghooks.c (verify_flow_info): Update.
22085 (dump_bb_for_graph): Update.
22086 (split_edge): Update.
22087 (make_forwarder_block): Update.
22088 (duplicate_block): Update.
22089 (account_profile_record): Update.
22090 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22091 (get_estimated_loop_iterations): Update.
22092 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22093 (single_likely_exit): Update.
22094 * cfgloopmanip.c (scale_loop_profile): Update.
22096 (set_zero_probability): Update.
22097 (lv_adjust_loop_entry_edge): Update.
22098 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22099 (purge_dead_edges): Update.
22100 (rtl_account_profile_record): Update.
22101 * cgraph.c (cgraph_node::create): Uninitialize count.
22102 (symbol_table::create_edge): Uninitialize count.
22103 (cgraph_update_edges_for_call_stmt_node): Update.
22104 (cgraph_edge::dump_edge_flags): Update.
22105 (cgraph_node::dump): Update.
22106 (cgraph_edge::maybe_hot_p): Update.
22107 * cgraph.h: Include profile-count.h
22108 (create_clone), create_edge, create_indirect_edge): Update.
22109 (cgraph_node): Turn count to profile_count.
22110 (cgraph_edge0: Likewise.
22111 (make_speculative, clone): Update.
22112 (create_edge): Update.
22113 (init_lowered_empty_function): Update.
22114 * cgraphclones.c (cgraph_edge::clone): Update.
22115 (duplicate_thunk_for_node): Update.
22116 (cgraph_node::create_clone): Update.
22117 * cgraphunit.c (cgraph_node::analyze): Update.
22118 (cgraph_node::expand_thunk): Update.
22119 * final.c (dump_basic_block_info): Update.
22120 * gimple-streamer-in.c (input_bb): Update.
22121 * gimple-streamer-out.c (output_bb): Update.
22122 * graphite.c (print_global_statistics): Update.
22123 (print_graphite_scop_statistics): Update.
22124 * hsa-brig.c: Include basic-block.h.
22125 * hsa-dump.c: Include basic-block.h.
22126 * hsa-gen.c (T sum_slice): Update.
22127 (convert_switch_statements):Update.
22128 * hsa-regalloc.c: Include basic-block.h.
22129 * ipa-chkp.c (chkp_produce_thunks): Update.
22130 * ipa-cp.c (struct caller_statistics): Update.
22131 (init_caller_stats): Update.
22132 (gather_caller_stats): Update.
22133 (ipcp_cloning_candidate_p): Update.
22134 (good_cloning_opportunity_p): Update.
22135 (get_info_about_necessary_edges): Update.
22136 (dump_profile_updates): Update.
22137 (update_profiling_info): Update.
22138 (update_specialized_profile): Update.
22139 (perhaps_add_new_callers): Update.
22140 (decide_about_value): Update.
22141 (ipa_cp_c_finalize): Update.
22142 * ipa-devirt.c (struct odr_type_warn_count): Update.
22143 (struct decl_warn_count): Update.
22144 (struct final_warning_record): Update.
22145 (possible_polymorphic_call_targets): Update.
22146 (ipa_devirt): Update.
22147 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22148 * ipa-icf.c (sem_function::merge): Update.
22149 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22150 * ipa-inline.c (compute_uninlined_call_time): Update.
22151 (compute_inlined_call_time): Update.
22152 (want_inline_small_function_p): Update.
22153 (want_inline_self_recursive_call_p): Update.
22154 (edge_badness): Update.
22155 (lookup_recursive_calls): Update.
22156 (recursive_inlining): Update.
22157 (inline_small_functions): Update.
22158 (dump_overall_stats): Update.
22159 (dump_inline_stats): Update.
22160 * ipa-profile.c (ipa_profile_generate_summary): Update.
22161 (ipa_propagate_frequency): Update.
22162 (ipa_profile): Update.
22163 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22164 * ipa-utils.c (ipa_merge_profiles): Update.
22165 * loop-doloop.c (doloop_modify): Update.
22166 * loop-unroll.c (report_unroll): Update.
22167 (unroll_loop_runtime_iterations): Update.
22168 * lto-cgraph.c (lto_output_edge): Update.
22169 (lto_output_node): Update.
22170 (input_node): Update.
22171 (input_edge): Update.
22172 (merge_profile_summaries): Update.
22173 * lto-streamer-in.c (input_cfg): Update.
22174 * lto-streamer-out.c (output_cfg): Update.
22175 * mcf.c (create_fixup_graph): Update.
22176 (adjust_cfg_counts): Update.
22177 (sum_edge_counts): Update.
22178 * modulo-sched.c (sms_schedule): Update.
22179 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22180 * predict.c (maybe_hot_count_p): Update.
22181 (probably_never_executed): Update.
22182 (dump_prediction): Update.
22183 (combine_predictions_for_bb): Update.
22184 (propagate_freq): Update.
22185 (handle_missing_profiles): Update.
22186 (counts_to_freqs): Update.
22187 (rebuild_frequencies): Update.
22188 (force_edge_cold): Update.
22189 * predict.h: Include profile-count.h
22190 (maybe_hot_count_p, counts_to_freqs): UPdate.
22191 * print-rtl-function.c: Do not include cfg.h
22192 * print-rtl.c: Include basic-block.h
22193 * profile-count.c: New file.
22194 * profile-count.h: New file.
22195 * profile.c (is_edge_inconsistent): Update.
22196 (correct_negative_edge_counts): Update.
22197 (is_inconsistent): Update.
22198 (set_bb_counts): Update.
22199 (read_profile_edge_counts): Update.
22200 (compute_frequency_overlap): Update.
22201 (compute_branch_probabilities): Update; Initialize and deinitialize
22203 (branch_prob): Update.
22204 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22205 (edge_gcov_count): New.
22206 (bb_gcov_count): New.
22207 * shrink-wrap.c (try_shrink_wrapping): Update.
22208 * tracer.c (better_p): Update.
22209 * trans-mem.c (expand_transaction): Update.
22210 (ipa_tm_insert_irr_call): Update.
22211 (ipa_tm_insert_gettmclone_call): Update.
22212 * tree-call-cdce.c: Update.
22213 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22214 (gimple_duplicate_sese_tail): Update.
22215 (gimple_account_profile_record): Update.
22216 (execute_fixup_cfg): Update.
22217 * tree-inline.c (copy_bb): Update.
22218 (copy_edges_for_bb): Update.
22219 (initialize_cfun): Update.
22220 (freqs_to_counts): Update.
22221 (copy_cfg_body): Update.
22222 (expand_call_inline): Update.
22223 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22224 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22225 (try_unroll_loop_completely): Update.
22226 (try_peel_loop): Update.
22227 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22228 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22229 * tree-ssa-loop-split.c (connect_loops): Update.
22230 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22231 * tree-ssa-reassoc.c (branch_fixup): Update.
22232 * tree-ssa-tail-merge.c (replace_block_by): Update.
22233 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22234 (compute_path_counts): Update.
22235 (update_profile): Update.
22236 (recompute_probabilities): Update.
22237 (update_joiner_offpath_counts): Update.
22238 (estimated_freqs_path): Update.
22239 (freqs_to_counts_path): Update.
22240 (clear_counts_path): Update.
22241 (ssa_fix_duplicate_block_edges): Update.
22242 (duplicate_thread_path): Update.
22243 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22244 (struct switch_conv_info): Update.
22245 * tree-tailcall.c (decrease_profile): Update.
22246 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22247 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22248 * value-prof.c (check_counter): Update.
22249 (gimple_divmod_fixed_value): Update.
22250 (gimple_mod_pow2): Update.
22251 (gimple_mod_subtract): Update.
22252 (gimple_ic_transform): Update.
22253 (gimple_stringop_fixed_value): Update.
22254 * value-prof.h (gimple_ic): Update.
22256 2017-06-02 Carl Love <cel@us.ibm.com>
22258 * config/rs6000/rs6000-c: Add support for built-in functions
22259 vector double vec_doublee (vector signed int);
22260 vector double vec_doublee (vector unsigned int);
22261 vector double vec_doublee (vector float);
22262 vector double vec_doubleh (vector signed int);
22263 vector double vec_doubleh (vector unsigned int);
22264 vector double vec_doubleh (vector float);
22265 vector double vec_doublel (vector signed int);
22266 vector double vec_doublel (vector unsigned int);
22267 vector double vec_doublel (vector float);
22268 vector double vec_doubleo (vector signed int);
22269 vector double vec_doubleo (vector unsigned int);
22270 vector double vec_doubleo (vector float);.
22271 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22272 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22274 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22275 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22276 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22278 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22279 vec_doublel, vec_doubleh.
22280 * doc/extend.texi: Update the built-in documentation file for the
22281 new built-in functions.
22283 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22286 * ipa-inline-analysis.c (free_growth_caches): Set
22287 edge_removal_hook_holder to NULL after removing it.
22289 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22291 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22292 comparision with zero.
22294 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22296 for early expansion of vec_min and vec_max builtins.
22297 (builtin_function_type): Add min/max unsigned variants to those
22298 identified as having unsigned arguments.
22300 2017-06-02 Olivier Hainque <hainque@adacore.com>
22302 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22304 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22306 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22307 Use VALL_F16 iterator rather than VALL.
22309 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22311 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22312 Emit CBNZ inside loop when doing a strong exchange and comparing
22313 against zero. Generate the CC flags after the loop.
22315 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22317 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22318 (dl_section_ref): New.
22319 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22320 On AIX, append an expression to subtract the size of the
22321 section length to dl_section_ref.
22323 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22325 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22326 for early expansion of vector absolute builtins.
22328 2017-06-02 Richard Biener <rguenther@suse.de>
22330 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22331 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22333 2017-06-02 Richard Biener <rguenther@suse.de>
22335 PR tree-optimization/80948
22336 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22337 stmt order as well.
22339 2017-06-02 Richard Biener <rguenther@suse.de>
22341 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22343 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22344 for inductions as relevant.
22346 2017-06-02 Richard Biener <rguenther@suse.de>
22348 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22349 (vectorizable_induction): ... this. Remove dead code.
22351 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22353 * builtins. (expand_builtin_alloca): Remove second parameter and
22354 infer its value from the first parameter instead.
22355 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22357 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22359 PR rtl-optimization/80903
22360 * loop-doloop.c (add_test): Unshare sequence.
22362 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22364 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22366 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22368 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22370 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22371 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22372 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22373 (xlogue_layout::get_stub_rtx): Make static.
22374 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22375 (xlogue_layout::compute_stub_managed_regs): Rename to...
22376 (xlogue_layout::count_stub_managed_regs): ...this.
22377 (xlogue_layout::is_stub_managed_reg): New function.
22378 (xlogue_layout::m_stub_names): Rename to...
22379 (xlogue_layout::s_stub_names): ...this, make static.
22380 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22381 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22382 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22383 xlogue_layout::s_stub_names): Instantiate statics.
22384 (stub_managed_regs): Remove.
22385 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22386 (disable_call_ms2sysv_xlogues): Rename to...
22387 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22388 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22390 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22391 change after reload_completed.
22392 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22394 (ix86_expand_prologue): Likewise.
22395 (ix86_expand_epilogue): Likewise.
22396 (ix86_expand_split_stack_prologue): Likewise.
22397 (ix86_compute_frame_layout): Remove frame parameter ...
22398 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22399 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22401 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22402 (ix86_frame): Move from here ...
22403 * config/i386/i386.h (ix86_frame): ... to here.
22404 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22405 complete ix86_frame data structure instead. Remove some_ld_name.
22407 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22409 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22410 symbols that hold a DECL_VALUE_EXPR.
22412 2017-06-01 Martin Jambor <mjambor@suse.cz>
22414 PR tree-optimization/80898
22415 * tree-sra.c (process_subtree_disqualification): Removed.
22416 (disqualify_candidate): Do not acll
22417 process_subtree_disqualification.
22418 (subtree_mark_written_and_enqueue): New function.
22419 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22420 RHS has been disqualified and re-queue LHS if necessary. Apart
22421 from that, ignore disqualified RHS.
22423 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22425 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22426 address fetch for z10 or later.
22428 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22430 * config/arc/arc.md (tst_movb): Add guard when splitting.
22432 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22434 * config/arc/arc.c (arc_can_eliminate): Test against
22435 arc_frame_pointer_needed.
22437 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22439 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22440 to prevent store reordering.
22441 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22442 (type): Add block type.
22443 (stack_tie): Define special instruction to be used in
22446 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22448 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22449 constraint. It is not valid for the pattern.
22450 (noncommutative_binary_comparison): Likewise.
22452 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22454 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22457 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22459 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22460 be used by the reg-alloc.
22462 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22464 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22465 reg-alloc when having mul64 or mul32x16 instructions.
22466 (mulsidi3): Likewise.
22467 (umulsidi3): Likewise.
22468 (mulsi32x16): New pattern.
22469 (mulsi64): Likewise.
22470 (mulsidi64): Likewise.
22471 (umulsidi64): Likewise.
22472 (MUL32x16_REG): Define.
22473 (mul64_600): Use MUL32x16_REG.
22474 (mac64_600): Likewise.
22475 (umul64_600): Likewise.
22476 (umac64_600): Likewise.
22478 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22480 * config/arc/arc.md (mulsi3_700): Make it commutative.
22482 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22484 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22486 (*sign_extendsidi2_insn): Likewise for movstosw.
22488 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22490 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22491 the type of the input discriminant value. Convert the
22492 discriminant value of signedness vary.
22494 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22496 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22497 Add to -Wall section.
22499 2017-06-01 Richard Biener <rguenther@suse.de>
22501 PR middle-end/66313
22502 * fold-const.c (fold_plusminus_mult_expr): If the factored
22503 factor may be zero use a wrapping type for the inner operation.
22504 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22505 and handle moved defs.
22506 (process_assignment): Properly guard the unary op case. Return a
22507 tri-state indicating that moving the stmt before the call may allow
22508 to continue. Pass through to_move.
22509 (find_tail_calls): Handle moving unrelated defs before
22512 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22515 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22516 splitter result in the canonical way.
22518 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22520 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22521 also for 32bit target. Update insn attributes.
22522 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22524 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22526 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22527 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22528 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22529 (_mm_maskz_min_ss): New intrinsics.
22531 2017-05-31 Martin Liska <mliska@suse.cz>
22533 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22534 Change comment style to one we normally use.
22535 (vectorizable_reduction): Likewise.
22536 (vectorizable_induction): Likewise.
22537 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22538 (vectorizable_call): Likewise.
22539 (vectorizable_simd_clone_call): Likewise.
22540 (vectorizable_conversion): Likewise.
22541 (vectorizable_assignment): Likewise.
22542 (vectorizable_shift): Likewise.
22543 (vectorizable_operation): Likewise.
22544 (vectorizable_store): Likewise.
22545 (vectorizable_load): Likewise.
22546 * tree-vectorizer.h: Likewise.
22548 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22550 * passes.c (emergency_dump_function): New.
22551 * tree-pass.h (emergency_dump_function): Declare.
22552 * plugin.c (plugins_internal_error_function): Remove.
22553 * plugin.h (plugins_internal_error_function): Remove declaration.
22554 * toplev.c (internal_error_function): New static function. Use it...
22555 (general_init): ...here.
22557 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22559 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22560 (arc_rtx_costs): Add costs for new patterns.
22561 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22562 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22564 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22566 * tree-ssa-strlen.c (get_next_strinfo): New function.
22567 (get_stridx_plus_constant): Use it.
22568 (zero_length_string): Likewise.
22569 (adjust_related_strinfos): Likewise.
22570 (adjust_last_stmt): Likewise.
22572 2017-05-31 Richard Biener <rguenther@suse.de>
22575 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22576 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22578 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22580 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22581 loop_vinfo argument and use of dependence distance vectors.
22582 Check instead whether the two references differ only in their
22583 initial value and assume that they have the same alignment if the
22584 difference is a multiple of the vector alignment.
22585 (vect_analyze_data_refs_alignment): Update call accordingly.
22587 2017-05-31 Martin Liska <mliska@suse.cz>
22590 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22592 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22594 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22595 (create_intersect_range_checks): Move from ...
22596 * tree-data-ref.c (create_intersect_range_checks_index)
22597 (create_intersect_range_checks): ... to here.
22598 (create_runtime_alias_checks): New function factored from ...
22599 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22600 here. Call above function.
22601 * tree-data-ref.h (create_runtime_alias_checks): New function.
22603 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22605 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22606 segment length for dr_b and compute it in wide_int.
22608 2017-05-31 Richard Biener <rguenther@suse.de>
22610 PR tree-optimization/80906
22611 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22612 and pass through iv_map.
22613 (copy_bb_and_scalar_dependences): Adjust.
22614 (translate_pending_phi_nodes): Likewise.
22615 (copy_loop_close_phi_args): Handle code-generating IVs instead
22618 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22620 * diagnostic-color.c (color_dict): Add "type-diff".
22621 (parse_gcc_colors): Update comment.
22622 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22623 -fdiagnostics-show-template-tree and -fno-elide-type.
22624 (GCC_COLORS): Add type-diff to example.
22626 (-fdiagnostics-show-template-tree): New.
22627 (-fno-elide-type): New.
22628 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22629 the pp_format_decoder callback. Call any m_format_postprocessor's
22631 (pretty_printer::pretty_printer): Initialize
22632 m_format_postprocessor.
22633 (pretty_printer::~pretty_printer): Delete any
22634 m_format_postprocessor.
22635 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22636 (class format_postprocessor): New class.
22637 (struct pretty_printer::format_decoder): Document the new parameters.
22638 (struct pretty_printer::m_format_postprocessor): New field.
22639 * tree-diagnostic.c (default_tree_printer): Update for new
22640 bool and const char ** params.
22641 * tree-diagnostic.h (default_tree_printer): Likewise.
22643 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22645 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22646 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22648 rs6000_gen_cell_microcode code.
22649 (rs6000_final_prescan_insn): Delete.
22650 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22651 "warn-cell-microcode" entries.
22652 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22653 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22654 throughout. Change cc_reg_not_micro_cr0_operand to
22655 cc_reg_not_cr0_operand throughout.
22656 (*extendhi<mode>2_noload): Delete.
22657 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22658 (mwarn-cell-microcode): Delete.
22659 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22660 -mgen-cell-microcode and -mwarn-cell-microcode.
22662 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22665 * config/i386/constraints.md (Yd): New constraint.
22667 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22668 and (?Yd, r) alternatives. Update insn attributes.
22669 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22670 and (?*Yd, r) alternatives. Update insn attributes.
22671 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22673 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22675 * gimplify.c (gimplify_modify_expr): Don't create a
22676 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22679 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22681 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22683 2017-05-30 Richard Biener <rguenther@suse.de>
22685 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22686 and reduc_def fields.
22687 (STMT_VINFO_REDUC_TYPE): New define.
22688 (STMT_VINFO_REDUC_DEF): Likewise.
22689 (vect_force_simple_reduction): Adjust prototype.
22690 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22691 (vect_is_simple_reduction): Remove check_reduction argument.
22692 (vect_force_simple_reduction): Adjust and set
22693 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22694 (vectorizable_reduction): Do not re-do reduction analysis
22695 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22696 * tree-parloops.c (gather_scalar_reductions): Adjust.
22698 2017-05-30 Richard Biener <rguenther@suse.de>
22700 PR middle-end/80901
22701 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22704 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22706 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22707 Introduce unknown_misalignment parameter and remove vf.
22708 (vect_peeling_hash_get_lowest_cost):
22709 Pass unknown_misalignment parameter.
22710 (vect_enhance_data_refs_alignment):
22711 Fix unsupportable data ref treatment.
22713 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22715 * tree-vect-data-refs.c (vect_get_data_access_cost):
22716 Workaround for SLP handling.
22717 (vect_enhance_data_refs_alignment):
22718 Compute costs for doing no peeling at all, compare to the best
22719 peeling costs so far and avoid peeling if cheaper.
22721 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22723 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22724 Return peeling info and set costs to zero for unlimited cost
22726 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22727 with unknown misalignment. Compute and costs for unknown
22728 misalignment, compare them to the costs for known misalignment
22729 and choose the cheapest for peeling.
22731 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22733 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22734 (vect_get_peeling_costs_all_drs): Create function.
22735 (vect_peeling_hash_get_lowest_cost):
22736 Use vect_get_peeling_costs_all_drs.
22737 (vect_peeling_supportable): Create function.
22738 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22740 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22742 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22743 DR_HAS_NEGATIVE_STEP.
22744 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22745 (vect_enhance_data_refs_alignment): Use.
22746 (vect_duplicate_ssa_name_ptr_info): Use.
22747 * tree-vectorizer.h (dr_misalignment): Use.
22748 (known_alignment_for_access_p): Use.
22750 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22753 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22755 (has_section_name): New function.
22757 2017-05-30 Martin Liska <mliska@suse.cz>
22760 * auto-profile.c (get_function_decl_from_block): Fix
22763 2017-05-30 Richard Biener <rguenther@suse.de>
22765 PR middle-end/80876
22766 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22768 2017-05-30 Martin Liska <mliska@suse.cz>
22770 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22771 * dumpfile.h (struct dump_file_info): Remove ctors.
22773 2017-05-30 Martin Liska <mliska@suse.cz>
22775 * predict.def: Fix GNU coding style.
22777 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22779 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22780 Mark 'to' argument with ATTRIBUTE_UNUSED.
22782 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22784 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22785 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22786 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22789 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22791 * doc/install.texi (Options specification): Restore entry of
22792 --enable-sjlj-exceptions.
22794 2017-05-27 Michael Eager <eager@eagercon.com>
22797 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22799 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22801 * config/microblaze/microblaze.h
22802 (FIXED_REGISTERS): Update in macro.
22803 (CALL_USED_REGISTERS): Update in macro.
22805 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22807 * doc/install.texi: Add links to macOS binary distributions.
22809 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22813 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22815 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22817 2017-05-26 Martin Liska <mliska@suse.cz>
22819 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22821 2017-05-26 Martin Liska <mliska@suse.cz>
22823 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22824 always leading ';; '.
22825 (dump_bb_info): Likewise.
22826 (brief_dump_cfg): Likewise.
22827 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22828 * dumpfile.c: Remove usage of TDF_VERBOSE.
22829 * dumpfile.h (enum dump_kind): Likewise.
22830 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22831 * print-tree.c (debug_verbose): Remove.
22832 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22833 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22834 * tree-diagnostic.c (default_tree_printer): Replace
22835 TDF_DIAGNOSTIC with TDF_SLIM.
22837 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22839 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22840 in parameter loop, rather than loop_vinfo.
22841 (create_intersect_range_checks): Ditto.
22842 (vect_create_cond_for_alias_checks): Update call to above functions.
22844 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22846 PR tree-optimization/80815
22847 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22848 for merging runtime alias checks. Handle negative DR_STEPs.
22850 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22852 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22854 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22855 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22856 out code pruning runtime alias checks.
22857 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22858 factored out from above.
22859 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22861 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22863 (prune_runtime_alias_test_list): New decalaration.
22865 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22867 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22868 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22869 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22870 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22871 (operator==, comp_dr_with_seg_len_pair): Ditto.
22872 (vect_prune_runtime_alias_test_list): Ditto.
22874 2017-05-26 Martin Liska <mliska@suse.cz>
22877 * params.def: Bound partial-inlining-entry-probability param.
22879 2017-05-26 Marek Polacek <polacek@redhat.com>
22882 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22885 2017-05-26 Richard Biener <rguenther@suse.de>
22887 PR tree-optimization/80842
22888 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22891 2017-05-26 Richard Biener <rguenther@suse.de>
22893 PR tree-optimization/80844
22894 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22896 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22898 * doc/md.texi (Machine Constraints): Update x86 family
22899 machine constraints section to match 'config/i386/constraints.md'.
22901 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22903 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22905 2017-05-25 Nathan Sidwell <nathan@acm.org>
22907 * doc/invoke.texi (--enable-languages): Update documentation.
22909 2017-05-25 Martin Liska <mliska@suse.cz>
22911 * dumpfile.c: Add TDF_FOLDING.
22912 * dumpfile.h (enum dump_kind): Likewise.
22913 * genmatch.c (dt_simplify::gen_1): Use it.
22915 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22917 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22919 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22921 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22922 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22924 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22926 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22927 * match.pd (X == C): Rewrite it here.
22928 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22929 with_certain_nonzero_bits2): New predicates.
22930 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22932 2017-05-24 Nathan Sidwell <nathan@acm.org>
22934 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22937 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22940 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22942 * config/powerpcspe: New port. Files are copied from the rs6000
22943 port, with "rs6000" in filenames replaced by "powerpcspe".
22945 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22947 PR rtl-optimization/80754
22948 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22950 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22952 * config/sparc/sparc.md (length): Return the correct value for -mflat
22953 sibcalls to match output_sibcall.
22955 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22959 * config/rs6000/rs6000.c (struct machine_function): Add new field
22961 (rs6000_get_separate_components): Init that field, use it.
22962 (rs6000_components_for_bb): Use the field.
22964 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22966 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22968 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22970 PR middle-end/80823
22971 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22973 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22976 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22977 address against address_operand predicate.
22978 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22980 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22982 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22983 parameters passed indirectly.
22985 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22987 * config/i386/i386.md (*movdi_internal): Remove SSE4
22988 alternative 18 (?r, *v). Update insn attributes.
22989 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22990 Update insn attributes.
22991 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22992 Update insn attributes.
22993 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22994 alternative 1 (r, v). Remove isa attribute.
22995 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22996 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22997 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22999 2017-05-23 Tom de Vries <tom@codesourcery.com>
23001 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23004 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23006 * cgraphunit.c (symbol_table::process_new_functions): Update.
23007 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23008 (inline_generate_summary): Rename to ...
23009 (ipa_fn_summary_generate): ... this one.
23010 (inline_read_summary): Rename to ...
23011 (ipa_fn_summary_read): ... this one.
23012 (inline_write_summary): Rename to ...
23013 (ipa_fn_summary_write): ... this one.
23014 (inline_free_summary): Rename to ...
23015 (ipa_free_fn_summary): ... this one.
23016 (pass_data_local_fn_summary, pass_local_fn_summary,
23017 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23018 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23019 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23020 make_pass_ipa_fn_summary): New.
23021 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23022 inline_write_summary, inline_free_summary): Remove.
23023 (ipa_free_fn_summary) : New.
23024 * ipa-inline.c (ipa_inline): Update.
23025 (pass_ipa_inline): Do not generate summaries.
23026 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23028 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23029 and add pass_ipa_fn_summary.
23030 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23032 (make_pass_inline_parameters): Remove.
23034 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23036 * omp-low.c (struct omp_context): Remove "default_kind" member.
23039 * omp-offload.c (execute_oacc_device_lower): Remove the
23040 parallelism dimensions function attributes for unparallelized
23041 OpenACC kernels constructs.
23043 2017-05-23 Martin Liska <mliska@suse.cz>
23045 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23047 (cgraph_edge::make_speculative): Likewise.
23048 (cgraph_edge::resolve_speculation): Likewise.
23049 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23050 (cgraph_node::dump): Likewise.
23051 * cgraph.h: Likewise.
23052 * cgraphunit.c (analyze_functions): Likewise.
23053 (symbol_table::compile): Likewise.
23054 * ipa-cp.c (print_all_lattices): Likewise.
23055 (determine_versionability): Likewise.
23056 (initialize_node_lattices): Likewise.
23057 (ipcp_verify_propagated_values): Likewise.
23058 (estimate_local_effects): Likewise.
23059 (update_profiling_info): Likewise.
23060 (create_specialized_node): Likewise.
23061 (perhaps_add_new_callers): Likewise.
23062 (decide_about_value): Likewise.
23063 (decide_whether_version_node): Likewise.
23064 (identify_dead_nodes): Likewise.
23065 (ipcp_store_bits_results): Likewise.
23066 * ipa-devirt.c (dump_targets): Likewise.
23067 (ipa_devirt): Likewise.
23068 * ipa-icf.c (sem_item::dump): Likewise.
23069 (sem_function::equals): Likewise.
23070 (sem_variable::equals): Likewise.
23071 (sem_item_optimizer::read_section): Likewise.
23072 (sem_item_optimizer::execute): Likewise.
23073 (congruence_class::dump): Likewise.
23074 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23075 (dump_inline_summary): Likewise.
23076 (estimate_node_size_and_time): Likewise.
23077 (inline_analyze_function): Likewise.
23078 * ipa-inline-transform.c (inline_call): Likewise.
23079 * ipa-inline.c (report_inline_failed_reason): Likewise.
23080 (want_early_inline_function_p): Likewise.
23081 (edge_badness): Likewise.
23082 (update_edge_key): Likewise.
23083 (inline_small_functions): Likewise.
23084 * ipa-profile.c (ipa_profile): Likewise.
23085 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23086 (ipa_make_edge_direct_to_target): Likewise.
23087 (remove_described_reference): Likewise.
23088 (ipa_impossible_devirt_target): Likewise.
23089 (propagate_controlled_uses): Likewise.
23090 (ipa_print_node_params): Likewise.
23091 (ipcp_transform_function): Likewise.
23092 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23093 (propagate_pure_const): Likewise.
23094 * ipa-reference.c (generate_summary): Likewise.
23095 (read_write_all_from_decl): Likewise.
23096 (propagate): Likewise.
23097 (ipa_reference_read_optimization_summary): Likewise.
23098 * ipa-utils.c (ipa_merge_profiles): Likewise.
23099 * ipa.c (walk_polymorphic_call_targets): Likewise.
23100 (symbol_table::remove_unreachable_nodes): Likewise.
23101 (ipa_single_use): Likewise.
23102 * passes.c (execute_todo): Likewise.
23103 * predict.c (drop_profile): Likewise.
23104 * symtab.c (symtab_node::get_dump_name): New function.
23105 (symtab_node::dump_name): Likewise.
23106 (symtab_node::dump_asm_name): Likewise.
23107 (symtab_node::dump_references): Likewise.
23108 (symtab_node::dump_referring): Likewise.
23109 (symtab_node::dump_base): Likewise.
23110 (symtab_node::debug_symtab): Likewise.
23111 * tree-sra.c (convert_callers_for_node): Likewise.
23112 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23113 * value-prof.c (init_node_map): Likewise.
23115 2017-05-23 Martin Liska <mliska@suse.cz>
23117 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23118 and symtab_node::debug_symtab to symbol_table::debug.
23119 * cgraphunit.c (analyze_functions): Use the renamed function.
23120 (symbol_table::compile): Likewise.
23121 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23122 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23123 * passes.c (execute_todo): Likewise.
23124 * symtab.c (symbol_table::dump): New function.
23125 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23127 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23129 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23130 that nonconst implies exec.
23132 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23134 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23135 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23136 (inline_edge_summary_vec): Turn into ...
23137 (ipa_call_summaries): ... this one.
23138 (redirect_to_unreachable, edge_set_predicate,
23139 evaluate_properties_for_edge, inline_summary_alloc,
23140 reset_ipa_call_summary, reset_inline_summary,
23141 inline_summary_t::duplicate): Update.
23142 (inline_edge_duplication_hook): Turn to ...
23143 (ipa_call_summary_t::duplicate): ... this one.
23144 (inline_edge_removal_hook): Turn to ...
23145 (ipa_call_summary_t::remove): ... this one.
23146 (dump_inline_edge_summary): Turn to ...
23147 (dump_ipa_call_summary): ... this one.
23148 (estimate_function_body_sizes): Update.
23149 (inline_update_callee_summaries): Update.
23150 (remap_edge_change_prob): Update.
23151 (remap_edge_summaries): Update.
23152 (inline_merge_summary): Update.
23153 (do_estimate_edge_time): Update.
23154 (inline_generate_summary): Update.
23155 (inline_read_section): Update.
23156 (inline_read_summary): Update.
23157 (inline_free_summary): Update.
23158 * ipa-inline.c (can_inline_edge_p): Update.
23159 (compute_inlined_call_time): Update.
23160 (want_inline_small_function_p): Update.
23161 (edge_badness): Update.
23162 (early_inliner): Update.
23163 * ipa-inline.h (inline_edge_summary): Turn to ...
23164 (ipa_call_summary): ... this one.
23165 (ipa_call_summary_t): New class.
23166 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23167 (ipa_call_summaries): New.
23168 (inline_edge_summary): Remove.
23169 (estimate_edge_growth): Update.
23170 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23171 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23172 * ipa-split.c (execute_split_functions): Update.
23173 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23175 2017-05-23 Tom de Vries <tom@codesourcery.com>
23177 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23178 attributes): Document rdrand effective target.
23180 2017-05-23 Tom de Vries <tom@codesourcery.com>
23182 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23183 attributes): Sort alphabetically.
23185 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23187 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23189 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23192 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23193 V2DF/V2DI splat into two separate patterns, one that handles
23194 registers, and the other that only handles memory. Drop support
23195 for splatting from a GPR on ISA 2.07 and then splitting the
23196 splat into direct move and splat.
23197 (vsx_splat_<mode>_reg): Likewise.
23198 (vsx_splat_<mode>_mem): Likewise.
23200 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23202 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23204 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23206 PR middle-end/80809
23207 * omp-low.c (finish_taskreg_remap): New function.
23208 (finish_taskreg_scan): If unit size of ctx->record_type
23209 is non-constant, unshare the size expression and replace
23210 decls in it with possible outer var refs.
23212 PR middle-end/80809
23213 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23214 GOVD_SHARED rather than GOVD_PRIVATE with it.
23215 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23216 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23218 PR middle-end/80853
23219 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23220 as last argument to build_outer_var_ref for pointer bases of array
23221 section reductions.
23223 2017-05-19 Martin Sebor <msebor@redhat.com>
23225 * print-tree.c (print_node): Print DECL_READ_P flag.
23227 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23229 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23230 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23231 * cgraph.c: Likewise.
23232 * cgraphunit.c: Likewise.
23233 * gengtype.c: Likewise.
23234 * ipa-cp.c: Likewise.
23235 * ipa-devirt.c: Likewise.
23236 * ipa-icf.c: Likewise.
23237 * ipa-predicate.c: Likewise.
23238 * ipa-profile.c: Likewise.
23239 * ipa-prop.c: Likewise.
23240 * ipa-split.c: Likewise.
23242 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23243 edge_predicate_pool, dump_inline_hints,
23244 inline_summary::account_size_time, redirect_to_unreachable,
23245 edge_set_predicate, set_hint_predicate,
23246 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23247 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23248 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23249 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23250 ipa_call_summary_t::remove, initialize_growth_caches,
23251 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23252 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23253 mark_modified, unmodified_parm_1, unmodified_parm,
23254 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23255 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23256 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23257 will_be_nonconstant_predicate, record_modified_bb_info,
23258 get_minimal_bb, record_modified, param_change_prob,
23259 phi_result_unknown_predicate, predicate_for_phi_result,
23260 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23261 estimate_function_body_sizes, compute_inline_parameters,
23262 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23263 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23264 inline_update_callee_summaries, remap_edge_change_prob,
23265 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23266 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23267 inline_analyze_function, inline_summary_t::insert,
23268 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23269 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23270 inline_free_summary): Move to ipa-fnsummary.h
23271 (predicate_t): Remove.
23272 * ipa-fnsummary.c: New file.
23273 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23274 (enum inline_hints_vals, inline_hints, agg_position_info,
23275 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23276 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23277 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23278 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23279 inline_read_summary, inline_write_summary, inline_free_summary,
23280 inline_analyze_function, initialize_inline_failed,
23281 inline_merge_summary, inline_update_overall_summary,
23282 compute_inline_parameters): Move to ipa-fnsummary.h
23283 * ipa-fnsummary.h: New file.
23284 * ipa-inline-transform.h: Include ipa-inline.h.
23285 * ipa-inline.c: LIkewise.
23287 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23289 * ipa-inline.c (edge_badness): Use inlined_time instead of
23290 inline_summaries->get.
23292 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23294 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23296 2017-05-22 Nathan Sidwell <nathan@acm.org>
23298 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23299 (fdump-lang): Document 'raw' option.
23300 * dumpfile.h (TDI_tu): Delete.
23301 * dumpfile.c (dump_files): Remove translation-unit.
23302 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23304 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23306 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23307 command option from $(AWK) call.
23308 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23309 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23310 [FORMAT]: Remove handling of variable.
23311 * config/avr/t-multilib: Regenerate.
23313 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23315 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23317 (dump_inline_summary): Do not print self_time.
23318 (estimate_function_body_sizes): Do not set self_time.
23319 (compute_inline_parameters): Likewise.
23320 (inline_read_section, inline_write_summary): Do not stream self_time.
23321 * ipa-inline.h (inline_summary): Drop self_time.
23323 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23325 * ipa-inline-analysis.c (account_size_time): Rename to ...
23326 (inline_summary::account_size_time): ... this one.
23327 (reset_ipa_call_summary): Turn to ...
23328 (ipa_call_summary::reset): ... this one.
23329 (reset_inline_summary): Turn to ...
23330 (inline_summary::reset): ... this one.
23331 (inline_summary_t::remove): Update.
23332 (inline_summary_t::duplicate): Update.
23333 (ipa_call_summary_t::remove): Update.
23334 (dump_inline_summary): Update.
23335 (estimate_function_body_sizes): Update.
23336 (compute_inline_parameters): Update.
23337 (estimate_node_size_and_time): Update.
23338 (inline_merge_summary): Update.
23339 (inline_update_overall_summary): Update.
23340 (inline_read_section): Update.
23341 (inline_write_summary): Update.
23342 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23343 add account_size_time and reset member functions.
23344 (ipa_call_summary): Add reset function.
23345 * ipa-predicate.h (predicate::operator &): Constify.
23347 2017-05-22 Richard Biener <rguenther@suse.de>
23349 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23351 2017-05-19 Jason Merrill <jason@redhat.com>
23353 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23355 2017-05-19 Marek Polacek <polacek@redhat.com>
23358 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23359 TYPE_OVERFLOW_WRAPS checks.
23361 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23363 * tree-core.h (enum omp_clause_default_kind): Add
23364 "OMP_CLAUSE_DEFAULT_PRESENT".
23365 * tree-pretty-print.c (dump_omp_clause): Handle it.
23366 * gimplify.c (enum gimplify_omp_var_data): Add
23367 "GOVD_MAP_FORCE_PRESENT".
23368 (gimplify_adjust_omp_clauses_1): Map it to
23369 "GOMP_MAP_FORCE_PRESENT".
23370 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23372 * gimplify.c (oacc_default_clause): Clarify.
23374 2017-05-19 Nathan Sidwell <nathan@acm.org>
23376 LANG_HOOK_REGISTER_DUMPS
23377 * toplev.c (general_init): Call register dump lang hook.
23378 * doc/invoke.texi: Document -fdump-lang option family.
23379 * dumpfile.c (dump_files): Remove class dump here.
23380 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23381 * dumpfile.h (tree_dump_index): Remove TDI_class.
23382 * langhooks-def.h (lhd_register_dumps): Declare.
23383 (LANG_HOOKS_REGISTER_DUMPS): Define.
23384 (LANG_HOOKS_INITIALIZER): Add it.
23385 * langhooks.c (lhd_register_dumps): Define.
23386 * langhooks.h (struct lang_hooks): Add register_dumps.
23388 2017-05-19 Nathan Sidwell <nathan@acm.org>
23390 * context.h (context::set_passes): New.
23391 * context.c (context::context): Do not create pass manager.
23392 * toplev.c (general_init): Create pass manager here.
23394 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23396 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23397 use this splitter if two add or or instructions would also work for
23398 the constant we want to generate.
23400 2017-05-19 Richard Biener <rguenther@suse.de>
23403 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23404 predicate evaluation.
23406 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23408 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23410 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23413 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23415 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23416 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23417 (inline_edge_summary_vec): Turn into ...
23418 (ipa_call_summaries): ... this one.
23419 (redirect_to_unreachable, edge_set_predicate,
23420 evaluate_properties_for_edge, inline_summary_alloc,
23421 reset_ipa_call_summary, reset_inline_summary,
23422 inline_summary_t::duplicate): Update.
23423 (inline_edge_duplication_hook): Turn to ...
23424 (ipa_call_summary_t::duplicate): ... this one.
23425 (inline_edge_removal_hook): Turn to ...
23426 (ipa_call_summary_t::remove): ... this one.
23427 (dump_inline_edge_summary): Turn to ...
23428 (dump_ipa_call_summary): ... this one.
23429 (estimate_function_body_sizes): Update.
23430 (inline_update_callee_summaries): Update.
23431 (remap_edge_change_prob): Update.
23432 (remap_edge_summaries): Update.
23433 (inline_merge_summary): Update.
23434 (do_estimate_edge_time): Update.
23435 (inline_generate_summary): Update.
23436 (inline_read_section): Update.
23437 (inline_read_summary): Update.
23438 (inline_free_summary): Update.
23439 * ipa-inline.c (can_inline_edge_p): Update.
23440 (compute_inlined_call_time): Update.
23441 (want_inline_small_function_p): Update.
23442 (edge_badness): Update.
23443 (early_inliner): Update.
23444 * ipa-inline.h (inline_edge_summary): Turn to ...
23445 (ipa_call_summary): ... this one.
23446 (ipa_call_summary_t): New class.
23447 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23448 (ipa_call_summaries): New.
23449 (inline_edge_summary): Remove.
23450 (estimate_edge_growth): Update.
23451 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23452 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23453 * ipa-split.c (execute_split_functions): Update.
23454 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23456 2017-05-19 Richard Biener <rguenther@suse.de>
23458 PR middle-end/80764
23459 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23461 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23463 * config/rs6000/rs6000.c (struct machine_function): Add field
23464 fpr_is_wrapped_separately.
23465 (rs6000_get_separate_components): Use 64 components. Handle the
23466 new FPR components.
23467 (rs6000_components_for_bb): Handle the FPR components.
23468 (rs6000_emit_prologue_components): Handle the FPR components.
23469 (rs6000_emit_epilogue_components): Handle the FPR components.
23470 (rs6000_set_handled_components): Handle the FPR components.
23471 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23472 that are already separately shrink-wrapped.
23473 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23474 that are already separately shrink-wrapped.
23476 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23479 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23482 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23483 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23484 cases where the register allocator uses a move and an offsettable
23485 memory operation to/from a FPR register on ISA 2.06/2.07.
23486 (define_peephole2 for Altivec d-form store): Likewise.
23488 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23491 * config/i386/mmx.md (*mov<mode>_internal): Enable
23492 alternatives 11, 12, 13 and 14 also for 32bit targets.
23493 Remove alternatives 15, 16, 17 and 18.
23494 * config/i386/sse.md (vec_concatv2di): Change
23495 alternative (!x, *y) to (x, ?!*Yn).
23497 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23499 * dumpfile.h (enum dump_kind): Remove stray comma.
23501 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23503 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23504 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23505 predicate::num_conditions
23506 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23507 (CHANGED): turn into predicate::changed.
23508 (agg_position_info): Move to ipa-predicate.h
23509 (add_condition, predicate::add_clause, predicate::operator &=,
23510 predicate::or_with, predicate::evaluate, predicate::probability,
23511 dump_condition, dump_clause, predicate::dump,
23512 predicate::remap_after_duplication, predicate::remap_after_inlining,
23513 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23514 (evaluate_conditions_for_known_args): Update.
23515 (set_cond_stmt_execution_predicate): Update.
23516 * ipa-inline.h: Include ipa-predicate.h
23517 (condition, inline_param_summary, conditions, agg_position_info,
23518 predicate): Move to ipa-predicate.h
23519 * ipa-predicate.c: New file.
23520 * ipa-predicate.h: New file.
23522 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23524 * final.c (leaf_function_p): Check we are not in a sequence.
23526 2017-05-18 Martin Liska <mliska@suse.cz>
23528 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23529 * dumpfile.c (dump_register): Use new enum dump_kind.
23530 (get_dump_file_name): Likewise.
23531 (dump_enable_all): Likewise.
23532 (dump_switch_p_1): Likewise.
23533 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23534 * dumpfile.h (enum dump_kind): New enum type.
23535 (struct dump_file_info): Create constructor and
23536 format fields and comments.
23537 * passes.c (pass_manager::register_one_dump_file):
23539 * statistics.c (statistics_early_init): Likewise.
23540 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23541 TDF_TREE with TDF_SLIM.
23542 (gather_memory_references_ref): Likewise.
23544 2017-05-18 Martin Liska <mliska@suse.cz>
23546 * vec.h (struct vnull): Use it.
23548 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23550 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23551 (true_predicate, false_predicate, true_predicate_p,
23552 false_predicate_p): Remove.
23553 (single_cond_predicate, not_inlined_predicate): Turn to member function
23555 (add_condition): Update.
23556 (add_clause): Turn to...
23557 (predicate::add_clause): ... this one; update; allow passing NULL
23559 (and_predicates): Turn to ...
23560 (predicate::operator &=): ... this one.
23561 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23562 (or_predicates): Turn to ...
23563 (predicate::or_with): ... this one.
23564 (evaluate_predicate): Turn to ...
23565 (predicate::evaluate): ... this one.
23566 (predicate_probability): Turn to ...
23567 (predicate::probability): ... this one.
23568 (dump_condition): Update.
23569 (dump_predicate): Turn to ...
23570 (predicate::dump): ... this one.
23571 (account_size_time): Update.
23572 (edge_set_predicate): Update.
23573 (set_hint_predicate): UPdate.
23574 (evaluate_conditions_for_known_args): Update.
23575 (evaluate_properties_for_edge): Update.
23576 (remap_predicate_after_duplication): Turn to...
23577 (predicate::remap_after_duplication): ... this one.
23578 (remap_hint_predicate_after_duplication): Update.
23579 (inline_summary_t::duplicate): UPdate.
23580 (dump_inline_edge_summary): Update.
23581 (dump_inline_summary): Update.
23582 (set_cond_stmt_execution_predicate): Update.
23583 (set_switch_stmt_execution_predicate): Update.
23584 (compute_bb_predicates): Update.
23585 (will_be_nonconstant_expr_predicate): Update.
23586 (will_be_nonconstant_predicate): Update.
23587 (phi_result_unknown_predicate): Update.
23588 (predicate_for_phi_result): Update.
23589 (array_index_predicate): Update.
23590 (estimate_function_body_sizes): Update.
23591 (estimate_node_size_and_time): Update.
23592 (estimate_ipcp_clone_size_and_time): Update.
23593 (remap_predicate): Rename to ...
23594 (predicate::remap_after_inlining): ... this one.
23595 (remap_hint_predicate): Update.
23596 (inline_merge_summary): Update.
23597 (inline_update_overall_summary): Update.
23598 (estimate_size_after_inlining): Update.
23599 (read_predicate): Rename to ...
23600 (predicate::stream_in): ... this one.
23601 (read_inline_edge_summary): Update.
23602 (write_predicate): Rename to ...
23603 (predicate::stream_out): ... this one.
23604 (write_inline_edge_summary): Update.
23605 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23606 (clause_t): Turn to uint32_t
23607 (predicate): Turn to class; implement constructor and operators
23609 (size_time_entry): Update.
23610 (inline_summary): Update.
23611 (inline_edge_summary): Update.
23613 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23615 * fold-const.c (fold_binary_loc): Move transformation...
23616 * match.pd (C - X CMP X): ... here.
23618 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23620 * config/sparc/sparc.c (sparc_option_override): Set function
23621 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23622 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23624 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23626 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23628 2017-05-18 Marek Polacek <polacek@redhat.com>
23631 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23632 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23633 gimple_assign_load_p.
23635 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23637 PR middle-end/80692
23638 * real.c (do_compare): Give decimal_do_compare preference over
23639 comparing just the signs.
23641 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23643 * doc/md.texi (Canonicalization of Instructions): Describe the
23644 canonical form of instructions that inherently set a condition
23647 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23649 PR middle-end/80775
23650 * tree-cfg.c: Move deletion of unreachable case statements to after
23651 the merging of consecutive case labels.
23653 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23655 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23656 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23657 restoring of callee-saved registers.
23659 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23661 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23662 * config/visium/visium.c (single_set_and_flags): Likewise.
23663 * config/visium/visium.md (Substitutions): Likewise.
23665 2017-05-17 Martin Liska <mliska@suse.cz>
23667 * cfg.c: Introduce dump_flags_t type and
23668 use it instead of int type.
23670 * cfghooks.c: Likewise.
23671 * cfghooks.h (struct cfg_hooks): Likewise.
23672 * cfgrtl.c: Likewise.
23673 * cfgrtl.h: Likewise.
23674 * cgraph.c (cgraph_node::get_body): Likewise.
23675 * coretypes.h: Likewise.
23676 * domwalk.c: Likewise.
23677 * domwalk.h: Likewise.
23678 * dumpfile.c (struct dump_option_value_info): Likewise.
23679 (dump_enable_all): Likewise.
23680 (dump_switch_p_1): Likewise.
23681 (opt_info_switch_p): Likewise.
23682 * dumpfile.h (enum tree_dump_index): Likewise.
23683 (struct dump_file_info): Likewise.
23684 * genemit.c: Likewise.
23685 * generic-match-head.c: Likewise.
23686 * gengtype.c (open_base_files): Likewise.
23687 * gimple-pretty-print.c: Likewise.
23688 * gimple-pretty-print.h: Likewise.
23689 * graph.c (print_graph_cfg): Likewise.
23690 * graphite-scop-detection.c (dot_all_sese): Likewise.
23691 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23692 * loop-unroll.c (report_unroll): Likewise.
23693 * passes.c (pass_manager::register_one_dump_file): Likewise.
23694 * print-tree.c: Likewise.
23695 * statistics.c: Likewise.
23696 * tree-cfg.c: Likewise.
23697 * tree-cfg.h: Likewise.
23698 * tree-dfa.c: Likewise.
23699 * tree-dfa.h: Likewise.
23700 * tree-dump.c (dump_function): Likewise.
23701 * tree-dump.h (struct dump_info): Likewise.
23702 * tree-pretty-print.c: Likewise.
23703 * tree-pretty-print.h: Likewise.
23704 * tree-ssa-live.c: Likewise.
23705 * tree-ssa-live.h: Likewise.
23706 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23707 * tree-vect-loop.c: Likewise.
23708 * tree-vect-slp.c: Likewise.
23710 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23711 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23713 PR tree-optimization/80457
23714 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23715 of all arguments to a statement as scalar_to_vec operations.
23716 (vectorizable_call): Adjust call to vect_model_simple_cost for
23718 (vectorizable_conversion): Likewise.
23719 (vectorizable_assignment): Likewise.
23720 (vectorizable_shift): Likewise.
23721 (vectorizable_operation): Likewise.
23722 (vectorizable_comparison): Likewise.
23723 (vect_is_simple_cond): Record the def types for operands.
23724 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23725 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23726 for statement argument count.
23728 2017-05-16 Carl Love <cel@us.ibm.com>
23730 * config/rs6000/rs6000-c: Add support for built-in functions
23731 vector unsigned long long vec_bperm (vector unsigned long long,
23732 vector unsigned char)
23733 vector signed long long vec_mule (vector signed int,
23735 vector unsigned long long vec_mule (vector unsigned int,
23736 vector unsigned int)
23737 vector signed long long vec_mulo (vector signed int,
23739 vector unsigned long long vec_mulo (vector unsigned int,
23740 vector unsigned int)
23741 vector signed char vec_sldw (vector signed char,
23742 vector signed char,
23744 vector unsigned char vec_sldw (vector unsigned char,
23745 vector unsigned char,
23747 vector signed short vec_sldw (vector signed short,
23748 vector signed short,
23750 vector unsigned short vec_sldw (vector unsigned short,
23751 vector unsigned short,
23753 vector signed int vec_sldw (vector signed int,
23756 vector unsigned int vec_sldw (vector unsigned int,
23757 vector unsigned int,
23759 vector signed long long vec_sldw (vector signed long long,
23760 vector signed long long,
23762 vector unsigned long long vec_sldw (vector unsigned long long,
23763 vector unsigned long long,
23765 * config/rs6000/rs6000-c: Add support for built-in functions
23766 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23767 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23768 * doc/extend.texi: Update the built-in documentation for the
23769 new built-in functions.
23771 2017-05-16 Marek Polacek <polacek@redhat.com>
23775 * tree.c (save_expr): Don't fold the expression.
23777 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23779 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23780 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23781 and (?*y,m). Update insn attributes.
23783 2017-05-16 Martin Liska <mliska@suse.cz>
23785 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23786 flags argument of print_gimple_stmt, print_gimple_expr,
23787 print_generic_stmt and print_generic_expr.
23788 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23789 * coretypes.h: Likewise.
23790 * except.c (dump_eh_tree): Likewise.
23791 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23792 * gimple-pretty-print.h: Likewise.
23793 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23794 (backprop::push_to_worklist): Likewise.
23795 (backprop::pop_from_worklist): Likewise.
23796 (backprop::process_use): Likewise.
23797 (backprop::intersect_uses): Likewise.
23798 (note_replacement): Likewise.
23799 * gimple-ssa-store-merging.c
23800 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23801 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23802 (pass_store_merging::execute): Likewise.
23803 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23804 (ssa_base_cand_dump_callback): Likewise.
23805 (dump_incr_vec): Likewise.
23806 (replace_refs): Likewise.
23807 (replace_mult_candidate): Likewise.
23808 (create_add_on_incoming_edge): Likewise.
23809 (create_phi_basis): Likewise.
23810 (insert_initializers): Likewise.
23811 (all_phi_incrs_profitable): Likewise.
23812 (introduce_cast_before_cand): Likewise.
23813 (replace_one_candidate): Likewise.
23814 * gimplify.c (gimplify_expr): Likewise.
23815 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23816 (set_rename): Likewise.
23817 (rename_uses): Likewise.
23818 (copy_loop_phi_nodes): Likewise.
23819 (add_close_phis_to_merge_points): Likewise.
23820 (copy_loop_close_phi_args): Likewise.
23821 (copy_cond_phi_args): Likewise.
23822 (graphite_copy_stmts_from_block): Likewise.
23823 (translate_pending_phi_nodes): Likewise.
23824 * graphite-poly.c (print_pdr): Likewise.
23825 (dump_gbb_cases): Likewise.
23826 (dump_gbb_conditions): Likewise.
23827 (print_scop_params): Likewise.
23828 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23829 (build_cross_bb_scalars_use): Likewise.
23830 (gather_bbs::before_dom_children): Likewise.
23831 * hsa-dump.c (dump_hsa_immed): Likewise.
23832 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23833 (get_replacement_map): Likewise.
23834 * ipa-inline-analysis.c (dump_condition): Likewise.
23835 (estimate_function_body_sizes): Likewise.
23836 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23837 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23838 * ipa-prop.c (ipa_dump_param): Likewise.
23839 (ipa_print_node_jump_functions_for_edge): Likewise.
23840 (ipa_modify_call_arguments): Likewise.
23841 (ipa_modify_expr): Likewise.
23842 (ipa_dump_param_adjustments): Likewise.
23843 (ipa_dump_agg_replacement_values): Likewise.
23844 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23845 * ipa-pure-const.c (check_stmt): Likewise.
23846 (pass_nothrow::execute): Likewise.
23847 * ipa-split.c (execute_split_functions): Likewise.
23848 * omp-offload.c (dump_oacc_loop_part): Likewise.
23849 (dump_oacc_loop): Likewise.
23850 * trans-mem.c (tm_log_emit): Likewise.
23851 (tm_memopt_accumulate_memops): Likewise.
23852 (dump_tm_memopt_set): Likewise.
23853 (dump_tm_memopt_transform): Likewise.
23854 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23855 (print_loop): Likewise.
23856 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23857 (chkp_gather_checks_info): Likewise.
23858 (chkp_get_check_result): Likewise.
23859 (chkp_remove_check_if_pass): Likewise.
23860 (chkp_use_outer_bounds_if_possible): Likewise.
23861 (chkp_reduce_bounds_lifetime): Likewise.
23862 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23863 (chkp_mark_completed_bounds): Likewise.
23864 (chkp_register_incomplete_bounds): Likewise.
23865 (chkp_mark_invalid_bounds): Likewise.
23866 (chkp_maybe_copy_and_register_bounds): Likewise.
23867 (chkp_build_returned_bound): Likewise.
23868 (chkp_get_bound_for_parm): Likewise.
23869 (chkp_build_bndldx): Likewise.
23870 (chkp_get_bounds_by_definition): Likewise.
23871 (chkp_generate_extern_var_bounds): Likewise.
23872 (chkp_get_bounds_for_decl_addr): Likewise.
23873 * tree-chrec.c (chrec_apply): Likewise.
23874 * tree-data-ref.c (dump_data_reference): Likewise.
23875 (dump_subscript): Likewise.
23876 (dump_data_dependence_relation): Likewise.
23877 (analyze_overlapping_iterations): Likewise.
23878 * tree-inline.c (expand_call_inline): Likewise.
23879 (tree_function_versioning): Likewise.
23880 * tree-into-ssa.c (dump_defs_stack): Likewise.
23881 (dump_currdefs): Likewise.
23882 (dump_names_replaced_by): Likewise.
23883 (dump_update_ssa): Likewise.
23884 (update_ssa): Likewise.
23885 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23886 * tree-parloops.c (build_new_reduction): Likewise.
23887 (try_create_reduction_list): Likewise.
23888 (ref_conflicts_with_region): Likewise.
23889 (oacc_entry_exit_ok_1): Likewise.
23890 (oacc_entry_exit_single_gang): Likewise.
23891 * tree-pretty-print.h: Likewise.
23892 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23893 (get_scalar_evolution): Likewise.
23894 (add_to_evolution): Likewise.
23895 (get_loop_exit_condition): Likewise.
23896 (analyze_evolution_in_loop): Likewise.
23897 (analyze_initial_condition): Likewise.
23898 (analyze_scalar_evolution): Likewise.
23899 (instantiate_scev): Likewise.
23900 (number_of_latch_executions): Likewise.
23901 (gather_chrec_stats): Likewise.
23902 (final_value_replacement_loop): Likewise.
23903 (scev_const_prop): Likewise.
23904 * tree-sra.c (dump_access): Likewise.
23905 (disqualify_candidate): Likewise.
23906 (create_access): Likewise.
23907 (reject): Likewise.
23908 (maybe_add_sra_candidate): Likewise.
23909 (create_access_replacement): Likewise.
23910 (analyze_access_subtree): Likewise.
23911 (analyze_all_variable_accesses): Likewise.
23912 (sra_modify_assign): Likewise.
23913 (initialize_constant_pool_replacements): Likewise.
23914 (find_param_candidates): Likewise.
23915 (decide_one_param_reduction): Likewise.
23916 (replace_removed_params_ssa_names): Likewise.
23917 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23918 * tree-ssa-copy.c (dump_copy_of): Likewise.
23919 (copy_prop_visit_cond_stmt): Likewise.
23920 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23921 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23922 (record_equivalences_from_stmt): Likewise.
23923 * tree-ssa-dse.c (compute_trims): Likewise.
23924 (delete_dead_call): Likewise.
23925 (delete_dead_assignment): Likewise.
23926 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23927 (forward_propagate_into_cond): Likewise.
23928 (pass_forwprop::execute): Likewise.
23929 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23930 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23932 (move_computations_worker): Likewise.
23933 (execute_sm): Likewise.
23934 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23935 (remove_exits_and_undefined_stmts): Likewise.
23936 (remove_redundant_iv_tests): Likewise.
23937 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23938 (adjust_iv_update_pos): Likewise.
23939 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23940 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23941 (value_replacement): Likewise.
23942 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23943 * tree-ssa-pre.c (print_pre_expr): Likewise.
23944 (get_representative_for): Likewise.
23945 (create_expression_by_pieces): Likewise.
23946 (insert_into_preds_of_block): Likewise.
23947 (eliminate_insert): Likewise.
23948 (eliminate_dom_walker::before_dom_children): Likewise.
23949 (eliminate): Likewise.
23950 (remove_dead_inserted_code): Likewise.
23951 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23952 * tree-ssa-reassoc.c (get_rank): Likewise.
23953 (eliminate_duplicate_pair): Likewise.
23954 (eliminate_plus_minus_pair): Likewise.
23955 (eliminate_not_pairs): Likewise.
23956 (undistribute_ops_list): Likewise.
23957 (eliminate_redundant_comparison): Likewise.
23958 (update_range_test): Likewise.
23959 (optimize_range_tests_var_bound): Likewise.
23960 (optimize_vec_cond_expr): Likewise.
23961 (rewrite_expr_tree): Likewise.
23962 (rewrite_expr_tree_parallel): Likewise.
23963 (linearize_expr): Likewise.
23964 (break_up_subtract): Likewise.
23965 (linearize_expr_tree): Likewise.
23966 (attempt_builtin_powi): Likewise.
23967 (attempt_builtin_copysign): Likewise.
23968 (transform_stmt_to_copy): Likewise.
23969 (transform_stmt_to_multiply): Likewise.
23970 (dump_ops_vector): Likewise.
23971 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23972 (print_scc): Likewise.
23973 (set_ssa_val_to): Likewise.
23974 (visit_reference_op_store): Likewise.
23975 (visit_use): Likewise.
23976 (sccvn_dom_walker::before_dom_children): Likewise.
23977 (run_scc_vn): Likewise.
23978 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23980 (expr_hash_elt::print): Likewise.
23981 (const_and_copies::pop_to_marker): Likewise.
23982 (const_and_copies::record_const_or_copy_raw): Likewise.
23983 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23984 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23985 (dump_predicates): Likewise.
23986 (find_uninit_use): Likewise.
23987 (warn_uninitialized_phi): Likewise.
23988 (pass_late_warn_uninitialized::execute): Likewise.
23989 * tree-ssa.c (verify_vssa): Likewise.
23990 (verify_ssa): Likewise.
23991 (maybe_optimize_var): Likewise.
23992 * tree-vrp.c (dump_value_range): Likewise.
23993 (dump_all_value_ranges): Likewise.
23994 (dump_asserts_for): Likewise.
23995 (register_edge_assert_for_2): Likewise.
23996 (vrp_visit_cond_stmt): Likewise.
23997 (vrp_visit_switch_stmt): Likewise.
23998 (vrp_visit_stmt): Likewise.
23999 (vrp_visit_phi_node): Likewise.
24000 (simplify_cond_using_ranges_1): Likewise.
24001 (fold_predicate_in): Likewise.
24002 (evrp_dom_walker::before_dom_children): Likewise.
24003 (evrp_dom_walker::push_value_range): Likewise.
24004 (evrp_dom_walker::pop_value_range): Likewise.
24005 (execute_early_vrp): Likewise.
24007 2017-05-16 Richard Biener <rguenther@suse.de>
24009 * dwarf2out.c (loc_list_from_tree_1): Do not create
24010 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24012 2017-05-16 Richard Biener <rguenther@suse.de>
24014 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24016 (note_variable_value_in_expr): If we resolved the decl ref
24017 do not push to the stack.
24019 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24021 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24022 operations in fast-math mode.
24023 (vaddq_f16): Likewise.
24024 (vmul_f16): Likewise.
24025 (vmulq_f16): Likewise.
24026 (vsub_f16): Likewise.
24027 (vsubq_f16): Likewise.
24028 * config/arm/neon.md (add<mode>3): New.
24030 (fma:<VH:mode>3): New. Also remove outdated comment.
24033 2017-05-16 Martin Liska <mliska@suse.cz>
24037 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24038 (odr_types_equivalent_p): Likewise.
24040 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24042 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24044 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24047 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24048 non-interunit SSE move alternatives with '?'.
24049 (zero-extendsidi peephole2): New peephole to skip intermediate
24050 general register in SSE zero-extend sequence.
24052 2017-05-15 Jeff Law <law@redhat.com>
24054 * reorg.c (relax_delay_slots): Create a new variable to hold
24055 the temporary target rather than clobbering TARGET_LABEL.
24057 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24058 missing argument to extract_bit_field call.
24059 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24061 2017-05-15 Martin Liska <mliska@suse.cz>
24064 * gcc.c (process_command): Do not allow empty argument of -o option.
24066 2017-05-15 Renlin Li <renlin.li@arm.com>
24068 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24069 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24070 * config/aarch64/constraints.md (Usf): Add long call check.
24071 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24072 (call_value): Likewise.
24073 (sibcall): Likewise.
24074 (sibcall_value): Likewise.
24076 (call_value_insn): New.
24077 (sibcall_insn): Update rtx pattern.
24078 (sibcall_value_insn): Likewise.
24079 (call_internal): Remove.
24080 (call_value_internal): Likewise.
24081 (sibcall_internal): Likewise.
24082 (sibcall_value_internal): Likewise.
24083 (call_reg): Likewise.
24084 (call_symbol): Likewise.
24085 (call_value_reg): Likewise.
24086 (call_value_symbol): Likewise.
24088 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24091 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24093 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24095 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24096 compatible with CCGOCmode and with CCZmode.
24098 2017-05-14 Martin Sebor <msebor@redhat.com>
24100 PR middle-end/77671
24101 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24102 (gimple_fold_builtin_snprintf): Same.
24103 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24104 (gimple_fold_builtin_snprintf): Same.
24105 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24106 of character types.
24107 (is_call_safe): New function.
24108 (try_substitute_return_value): Call it.
24109 (try_simplify_call): New function.
24110 (pass_sprintf_length::handle_gimple_call): Call it.
24112 2017-05-14 Martin Sebor <msebor@redhat.com>
24114 PR middle-end/80669
24115 * builtins.c (expand_builtin_stpncpy): Simplify.
24117 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24119 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24120 * config/i386/i386.h
24121 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24122 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24123 (struct machine_function): Add new members call_ms2sysv,
24124 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24125 (struct machine_frame_state): New fields sp_realigned and
24126 sp_realigned_offset.
24127 * config/i386/i386.c
24128 (enum xlogue_stub): New enum.
24129 (enum xlogue_stub_sets): New enum.
24130 (class xlogue_layout): New class.
24131 (struct ix86_frame): New fields stack_realign_allocate_offset,
24132 stack_realign_offset and outlined_save_offset. Modify comments to
24133 detail stack layout when using out-of-line stubs.
24134 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24135 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24136 -mcall-ms2sysv-xlogues.
24137 (stub_managed_regs): New static variable.
24138 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24139 registers managed by out-of-line stub.
24140 (disable_call_ms2sysv_xlogues): New function.
24141 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24142 m->call_ms2sysv when appropriate and compute frame layout for
24144 (sp_valid_at, fp_valid_at): New inline functions.
24145 (choose_basereg): New function.
24146 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24148 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24149 Use align parameter of choose_baseaddr to generated aligned SSE movs
24151 (pro_epilogue_adjust_stack): Modify to track
24152 machine_frame_state::sp_realigned.
24153 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24154 (ix86_nsaved_sseregs): Likewise.
24155 (ix86_emit_save_regs): Likewise.
24156 (ix86_emit_save_regs_using_mov): Likewise.
24157 (ix86_emit_save_sse_regs_using_mov): Likewise.
24158 (get_scratch_register_on_entry): Likewise.
24159 (gen_frame_set): New function.
24160 (gen_frame_load): Likewise.
24161 (gen_frame_store): Likewise.
24162 (emit_outlined_ms2sysv_save): Likewise.
24163 (emit_outlined_ms2sysv_restore): Likewise.
24164 (ix86_expand_prologue): Modify stack re-alignment code and call
24165 emit_outlined_ms2sysv_save when appropriate.
24166 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24167 parameter rtx_insn *insn, which allows the function to be used to only
24168 generate the notes.
24169 (ix86_expand_epilogue): Modify validity checks of frame and stack
24170 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24171 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24172 * config/i386/predicates.md
24173 (save_multiple): New predicate.
24174 (restore_multiple): Likewise.
24175 * config/i386/sse.md
24176 (save_multiple<mode>): New pattern.
24177 (save_multiple_realign<mode>): Likewise.
24178 (restore_multiple<mode>): Likewise.
24179 (restore_multiple_and_return<mode>): Likewise.
24180 (restore_multiple_leave_return<mode>): Likewise.
24181 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24183 2017-05-14 Julia Koval <julia.koval@intel.com>
24185 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24186 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24187 (__builtin_ia32_xsetbv): New builtins.
24188 * config/i386/i386.c (ix86_expand_special_args_builtin):
24190 (ix86_expand_builtin): Special expand for new intrinsics.
24191 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24192 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24193 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24195 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24197 * cfganal.c (inverted_post_order_compute): Change argument type
24199 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24200 * df-core.c (rest_of_handle_df_initialize): Adjust.
24201 (rest_of_handle_df_finish): Likewise.
24202 (df_analyze_1): Likewise.
24203 (df_analyze): Likewise.
24204 (loop_inverted_post_order_compute): Change argument to be a vec *.
24205 (df_analyze_loop): Adjust.
24206 (df_get_n_blocks): Likewise.
24207 (df_get_postorder): Likewise.
24208 * df.h (struct df_d): Change field to be a vec.
24209 * lcm.c (compute_laterin): Adjust.
24210 (compute_available): Likewise.
24211 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24212 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24213 * tree-ssa-pre.c (compute_antic): Likewise.
24215 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24217 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24218 (depth_first_search::depth_first_search): Change structure init
24219 function to this constructor.
24220 (depth_first_search::add_bb): Rename function to this member.
24221 (depth_first_search::execute): Likewise.
24222 (flow_dfs_compute_reverse_finish): Adjust.
24224 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24226 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24227 (longest_simple_path): Likewise.
24228 * shrink-wrap.c (spread_components): Likewise.
24229 (disqualify_problematic_components): Likewise.
24230 (emit_common_heads_for_components): Likewise.
24231 (emit_common_tails_for_components): Likewise.
24232 (insert_prologue_epilogue_for_components): Likewise.
24234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24236 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24239 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24241 * df-core.c (df_set_blocks): Start using auto_bitmap.
24242 (df_compact_blocks): Likewise.
24243 * df-problems.c (df_rd_confluence_n): Likewise.
24244 * df-scan.c (df_insn_rescan_all): Likewise.
24245 (df_process_deferred_rescans): Likewise.
24246 (df_update_entry_block_defs): Likewise.
24247 (df_update_exit_block_uses): Likewise.
24248 (df_entry_block_bitmap_verify): Likewise.
24249 (df_exit_block_bitmap_verify): Likewise.
24250 (df_scan_verify): Likewise.
24251 * lra-constraints.c (lra_constraints): Likewise.
24252 (undo_optional_reloads): Likewise.
24253 (lra_undo_inheritance): Likewise.
24254 * lra-remat.c (calculate_gen_cands): Likewise.
24255 (do_remat): Likewise.
24256 * lra-spills.c (assign_spill_hard_regs): Likewise.
24257 (spill_pseudos): Likewise.
24258 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24259 (bitmap_set_subtract_values): Likewise.
24261 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24263 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24264 management with auto_bitmap.
24265 (fix_inter_tick): Likewise.
24266 (fix_recovery_deps): Likewise.
24267 * ira.c (add_store_equivs): Likewise.
24268 (find_moveable_pseudos): Likewise.
24269 (split_live_ranges_for_shrink_wrap): Likewise.
24270 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24271 (rtx_reuse_manager::seen_def_p): Likewise.
24272 (rtx_reuse_manager::set_seen_def): Likewise.
24273 * print-rtl.h (class rtx_reuse_manager): Likewise.
24275 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24277 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24279 (migrate_btr_def): Likewise.
24280 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24281 * df-core.c (loop_post_order_compute): Likewise.
24282 (loop_inverted_post_order_compute): Likewise.
24283 * hsa-common.h: Likewise.
24284 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24285 * init-regs.c (initialize_uninitialized_regs): Likewise.
24286 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24287 (inline_small_functions): Likewise.
24288 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24289 * ira.c (combine_and_move_insns): Likewise.
24290 (build_insn_chain): Likewise.
24291 * loop-invariant.c (find_invariants): Likewise.
24292 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24293 * predict.c (tree_predict_by_opcode): Likewise.
24294 (predict_paths_leading_to): Likewise.
24295 (predict_paths_leading_to_edge): Likewise.
24296 (estimate_loops_at_level): Likewise.
24297 (estimate_loops): Likewise.
24298 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24299 (spread_components): Likewise.
24300 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24301 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24302 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24303 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24304 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24305 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24306 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24307 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24308 (mark_threaded_blocks): Likewise.
24309 (thread_through_all_blocks): Likewise.
24310 * tree-ssa.c (verify_ssa): Likewise.
24311 (execute_update_addresses_taken): Likewise.
24312 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24314 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24316 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24318 (post_order_compute): Likewise.
24319 (inverted_post_order_compute): Likewise.
24320 (pre_and_rev_post_order_compute_fn): Likewise.
24322 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24324 * genrecog.c (int_set::int_set): Explicitly construct our
24325 auto_vec base class.
24326 * vec.h (auto_vec::auto_vec): New constructor.
24328 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24330 * bitmap.h (class auto_bitmap): New constructor taking
24331 bitmap_obstack * argument.
24333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24335 * bitmap.h (class auto_bitmap): Change type of m_bits to
24336 bitmap_head, and adjust ctor / dtor and member operators.
24338 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24340 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24341 when returned register mode doesn't match original mode.
24343 2017-05-12 Jeff Law <law@redhat.com>
24344 Jakub Jelinek <jakub@redhat.com>
24346 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24347 we look for cc setter after the compare-elim changes.
24348 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24349 within the vector to match what compare-elim now expects.
24350 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24351 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24353 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24354 after the compare-elim changes.
24355 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24356 the vector to match what compare-elim now expects.
24357 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24358 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24359 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24360 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24361 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24363 * config/visium/visium.c (single_set_and_flags): Fix where
24364 we look for cc setter after the compare-elim changes.
24365 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24366 with the vector to match what compare-elim now expects.
24367 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24368 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24369 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24370 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24371 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24372 (neg<mode>2_insn_set_overflow): Likewise.
24374 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24376 PR middle-end/79794
24377 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24378 maybe_expand_insn call, set ops[0].target. If still set after call,
24379 set alt_rtl. Add extra arg to recursive calls.
24380 (extract_bit_field): Add alt_rtl argument. Pass to
24382 * expmed.h (extract_bit_field): Fix prototype.
24383 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24384 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24385 to extract_bit_field_calls.
24386 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24387 Pass alt_rtl to extract_bit_field calls.
24388 * calls.c (store_unaligned_arguments_into_psuedos)
24389 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24390 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24392 * optabs.h (struct expand_operand): Add target bitfield.
24394 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24396 * compare-elim.c (try_eliminate_compare): Canonicalize
24397 operation with embedded compare to
24398 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24399 (set (reg) (operation)].
24401 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24403 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24406 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24407 cost of adding a carry flag for ADC instruction.
24408 [case MINUS]: Ignore the cost of subtracting a carry flag
24409 for SBB instruction.
24411 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24413 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24415 * config/rs6000/bmiintrin.h: New file.
24416 * config/rs6000/bmi2intrin.h: New file.
24417 * config/rs6000/x86intrin.h: New file.
24419 2017-05-12 Jeff Law <law@redhat.com>
24421 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24424 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24426 PR middle-end/80707
24427 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24429 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24431 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24432 early expansion of vector divide builtins.
24433 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24434 builtins identified as having unsigned arguments.
24436 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24438 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24439 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24440 expansion of vector logical operations (and, andc, or, xor,
24443 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24445 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24446 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24448 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24450 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24451 early GIMPLE expansion of vector multiplies.
24453 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24455 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24456 TARGET_HAVE_MOVT conditional.
24457 (movt splitter): Likewise.
24459 2017-05-12 Richard Biener <rguenther@suse.de>
24461 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24462 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24463 Fold all stmts not inplace.
24465 2017-05-12 Richard Biener <rguenther@suse.de>
24467 PR tree-optimization/80713
24468 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24469 inserted_exprs bit for not removed stmts.
24471 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24473 PR middle-end/69921
24474 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24475 parallelized" attribute for parallelized OpenACC kernels.
24476 * omp-offload.c (execute_oacc_device_lower): Use it.
24478 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24479 Set "oacc kernels" attribute.
24480 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24481 parameter. Adjust all users.
24482 (oacc_fn_attrib_kernels_p): Remove function.
24483 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24484 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24485 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24486 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24487 assert "oacc kernels" attribute is set.
24489 2017-05-11 Carl Love <cel@us.ibm.com>
24491 * config/rs6000/rs6000-c: Add support for built-in functions
24492 vector unsigned char vec_popcnt (vector signed char)
24493 vector unsigned char vec_popcnt (vector unsigned char)
24494 vector unsigned short vec_popcnt (vector signed short)
24495 vector unsigned short vec_popcnt (vector unsigned short)
24496 vector unsigned int vec_popcnt (vector signed int)
24497 vector unsigned int vec_popcnt (vector unsigned int)
24498 vector unsigned long long vec_popcnt (vector signed long long)
24499 vector unsigned long long vec_popcnt (vector unsigned long long)
24500 vector signed long long vec_slo (vector signed long long,
24501 vector signed char)
24502 vector signed long long vec_slo (vector signed long long,
24503 vector unsigned char)
24504 vector unsigned long long vec_slo (vector unsigned long long,
24505 vector signed char)
24506 vector unsigned long long vec_slo (vector unsigned long long,
24507 vector unsigned char)
24508 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24509 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24510 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24511 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24512 * doc/extend.texi: Update the built-in documentation file for the
24513 new built-in functions.
24515 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24517 * attribs.h (sorted_attr_string): Move machine independent
24518 functions for target clone support from the i386 port to common
24519 code. Rename ix86_function_versions to common_function_versions.
24520 Rename make_name to make_unique_name.
24521 (common_function_versions): Likewise.
24522 (make_unique_name): Likewise.
24523 (make_dispatcher_decl): Likewise.
24524 (is_function_default_version): Likewise.
24525 * attribs.c (attr_strcmp): Likewise.
24526 (sorted_attr_string): Likewise.
24527 (common_function_versions): Likewise.
24528 (make_unique_name): Likewise.
24529 (make_dispatcher_decl): Likewise.
24530 (is_function_default_version): Likewise.
24531 * config/i386/i386.c (attr_strcmp): Likewise.
24532 (sorted_attr_string): Likewise.
24533 (ix86_function_versions): Likewise.
24534 (make_name): Likewise.
24535 (make_dispatcher_decl): Likewise.
24536 (is_function_default_version): Likewise.
24537 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24539 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24542 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24543 Account for direct move costs for vec_construct of integer
24546 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24549 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24550 (UNSPEC_STX_ATOMIC): Ditto.
24551 (loaddi_via_sse): New insn.
24552 (storedi_via_sse): Ditto.
24553 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24554 Update corresponding peephole2 patterns.
24555 (atomic_storedi_fpu): Ditto.
24557 2017-05-11 Julia Koval <julia.koval@intel.com>
24559 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24560 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24562 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24563 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24564 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24566 2017-05-11 Nathan Sidwell <nathan@acm.org>
24568 * graphite-poly.c: Include dumpfile.h.
24570 * dumpfle.h (dump_function): Declare here ...
24571 * tree-dump.h (dump_function): ... not here.
24572 * dumpfile.c: #include tree-cfg.h.
24573 (dump_function): Move here from ...
24574 * tree-dump.c (dump_function): ... here.
24575 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24576 * graphite-poly.c: Don't include tree-dump.h.
24577 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24578 * print-tree.c: Likewise.
24579 * stor-layout.c: Likewise.
24580 * tree-nested.c: Likewise.
24582 * dumpfile.c (dump_start): Use TDF_FLAGS.
24583 (dump_enable_all): Fix TDF_KIND check thinko.
24585 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24587 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24588 array entries to represent two legal parameterizations of the
24589 overloaded __builtin_cmpb function, as represented by the
24590 P6_OV_BUILTIN_CMPB constant.
24591 (altivec_resolve_overloaded_builtin): Add special case handling
24592 for the __builtin_cmpb function, as represented by the
24593 P6_OV_BUILTIN_CMPB constant.
24594 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24595 (BU_P6_64BIT_2): New macro.
24596 (BU_P6_OVERLOAD_2): New macro
24597 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24598 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24599 (CMPB): Add overload support to represent both 32-bit and 64-bit
24600 compare-bytes function.
24601 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24602 support for TARGET_CMPB.
24603 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24604 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24605 documentation of the __builtin_cmpb overloaded built-in function.
24607 2017-05-11 Richard Biener <rguenther@suse.de>
24609 PR tree-optimization/80705
24610 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24611 bases are not vectorizable.
24613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24615 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24616 when counting register pressure.
24618 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24620 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24621 (iv_ca_more_deps): Renamed to ...
24622 (iv_ca_compare_deps): ... this.
24623 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24625 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24627 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24629 (determine_group_iv_costs): ... here.
24630 (find_inv_vars_cb): Record inv var if it's not recorded before.
24632 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24634 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24635 (get_shiftadd_cost): Ditto.
24637 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24639 * tree-ssa-address.c: Include header file.
24640 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24642 (add_to_parts): Refactor.
24643 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24644 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24647 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24649 PR tree-optimization/53090
24650 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24652 (extract_cond_operands): Detect condition with IV on both sides
24653 and return COMP_IV_EXPR_2.
24654 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24655 (rewrite_use_compare): Simplify by removing call to function
24656 extract_cond_operands.
24658 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24660 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24661 (extract_cond_operands): Detect condition comparing against non-
24662 invariant bound and return appropriate enum value.
24663 (find_interesting_uses_cond): Update use of extract_cond_operands.
24664 Handle its return value accordingly.
24665 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24669 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24670 nonlinear iv_use computation in loop invariant sensitive way.
24672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24674 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24675 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24677 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24679 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24680 (dump_cand): Support iv_cand.inv_exprs.
24681 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24683 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24688 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24690 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24691 as local function. Include necessary header files.
24692 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24694 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24696 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24698 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24700 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24701 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24702 RSHIFT_EXPR and BIT_NOT_EXPR.
24704 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24706 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24707 (adjust_setup_cost): New parameter supporting round up adjustment.
24708 (struct address_cost_data): Delete.
24709 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24710 (split_address_cost, ptr_difference_cost): Delete.
24711 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24712 (struct ainc_cost_data): New struct.
24713 (get_address_cost_ainc): New function.
24714 (get_address_cost, get_computation_cost): Reimplement.
24715 (determine_group_iv_cost_address): Record inv_expr for all uses of
24717 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24718 (iv_ca_has_deps): Reimplemented to ...
24719 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24721 (iv_ca_extend): Call iv_ca_more_deps.
24723 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24725 * tree-ssa-address.c (struct mem_address): Move to header file.
24726 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24727 * tree-ssa-address.h (struct mem_address): Move from C file.
24728 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24732 * tree-affine.h (aff_combination_type): New interface.
24733 (aff_combination_zero_p): Remove static.
24734 (aff_combination_const_p): New interface.
24735 (aff_combination_singleton_var_p): New interfaces.
24737 2017-05-11 Richard Biener <rguenther@suse.de>
24739 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24740 Skip unreachable blocks and destinations.
24741 (eliminate): Move stmt removal and fixup ...
24742 (fini_eliminate): ... here. Skip inserted exprs.
24743 (pass_pre::execute): Move fini_pre after fini_eliminate.
24744 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24745 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24746 PRE to get rid of dead code that has invalid SSA form and
24747 split critical edges again.
24749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24751 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24753 2017-05-11 Richard Biener <rguenther@suse.de>
24755 * passes.c (execute_function_todo): Verify loops if they are
24756 said to be up-to-date.
24757 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24758 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24760 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24763 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24764 handle calling assemble_external ourself.
24767 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24768 modes with zero size. Enhance comment.
24770 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24772 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24773 built-ins for vec_xl and vec_xst with short and char pointer
24776 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24778 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24779 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24780 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24781 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24782 (_mm_maskz_min_round_ss): New intrinsics.
24783 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24784 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24785 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24786 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24787 (__builtin_ia32_minss_mask_round): New builtins.
24788 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24789 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24790 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24792 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24793 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24795 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24798 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24800 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24801 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24802 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24803 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24804 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24805 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24806 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24807 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24808 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24809 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24810 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24811 (__builtin_ia32_mulss_mask_round): New builtins.
24812 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24813 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24814 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24816 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24817 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24819 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24822 2017-05-10 Julia Koval <julia.koval@intel.com>
24824 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24825 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24826 (_mm256_setr_m128i): New intrinsics.
24828 2017-05-10 Julia Koval <julia.koval@intel.com>
24830 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24831 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24832 (_mm_maskz_rcp14_ss): New intrinsics.
24833 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24834 (__builtin_ia32_rcp14ss_mask): New builtins.
24835 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24837 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24839 PR tree-optimization/51513
24840 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24841 (assert_unreachable_fallthru_edge_p): Use it.
24842 (group_case_labels_stmt): Likewise.
24843 * tree-cfg.h: Prototype it.
24844 * stmt.c: Include cfghooks.h and tree-cfg.h.
24845 (emit_case_dispatch_table) <gap_label>: New local variable.
24846 Use it to fill dispatch table gaps.
24847 Test for default_label before updating probabilities.
24848 (expand_case) <default_label>: Remove unneeded initialization.
24849 Test for unreachable default case statement and remove its edge.
24850 Set default_label accordingly.
24851 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24853 2017-05-10 Carl Love <cel@us.ibm.com>
24855 * config/rs6000/rs6000-c: Add support for built-in functions
24856 vector signed char vec_neg (vector signed char)
24857 vector signed short int vec_neg (vector short int)
24858 vector signed int vec_neg (vector signed int)
24859 vector signed long long vec_neg (vector signed long long)
24860 vector float vec_neg (vector float)
24861 vector double vec_neg (vector double)
24862 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24864 * config/rs6000/altivec.h: Add define for vec_neg
24865 * doc/extend.texi: Update the built-in documentation for the
24866 new built-in functions.
24868 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24870 PR tree-optimization/77644
24871 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24873 2017-05-10 Nathan Sidwell <nathan@acm.org>
24875 * dumpfile.h (TDI_lang_all): New.
24876 (TDF_KIND): New. Renumber others
24877 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24879 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24881 (get_dump_file_name): Adjust suffix generation.
24882 (dump_enable_all): Use TDF_KIND.
24883 * doc/invoke.texi (-fdump-lang-all): Document.
24885 * dumpfile.h: Tabify.
24887 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24890 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24891 Move member access before delete.
24893 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24895 * tree-inline.c (expand_call_inline): Split block at stmt
24898 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24901 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24902 are now unused after splitting mov{sf,sd}_hardfloat.
24903 (f32_lr2): Likewise.
24904 (f32_lm): Likewise.
24905 (f32_lm2): Likewise.
24906 (f32_li): Likewise.
24907 (f32_li2): Likewise.
24908 (f32_lv): Likewise.
24909 (f32_sr): Likewise.
24910 (f32_sr2): Likewise.
24911 (f32_sm): Likewise.
24912 (f32_sm2): Likewise.
24913 (f32_si): Likewise.
24914 (f32_si2): Likewise.
24915 (f32_sv): Likewise.
24916 (f32_dm): Likewise.
24917 (f32_vsx): Likewise.
24918 (f32_av): Likewise.
24919 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24920 For movsf, order stores so the VSX stores occur before the GPR
24921 store which encourages the register allocator to use a traditional
24922 FPR instead of a GPR. For movsd, order the stores so that the GPR
24923 store comes before the VSX stores to allow the power6 to work.
24924 This is due to the power6 not having a 32-bit integer store
24925 instruction from a FPR.
24926 (movsf_hardfloat): Likewise.
24927 (movsd_hardfloat): Likewise.
24929 2017-05-09 Martin Sebor <msebor@redhat.com>
24931 PR translation/80280
24932 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24935 PR translation/80280
24936 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24937 data member added in r247778.
24938 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24940 2017-05-09 Nathan Sidwell <nathan@acm.org>
24942 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24944 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24947 2017-05-09 Marek Polacek <polacek@redhat.com>
24949 * doc/invoke.texi: Fix typo.
24951 2017-05-09 Richard Biener <rguenther@suse.de>
24953 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24954 (vrp_val_is_min): Likewise.
24955 (set_value_range_to_value): Likewise.
24956 (set_value_range_to_nonnegative): Likewise.
24957 (gimple_assign_nonzero_p): Likewise.
24958 (gimple_stmt_nonzero_p): Likewise.
24959 (vrp_int_const_binop): Likewise. Remove unreachable case.
24960 (adjust_range_with_scev): Adjust comments.
24961 (compare_range_with_value): Likewise.
24962 (extract_range_from_phi_node): Likewise.
24963 (test_for_singularity): Likewise.
24965 2017-05-09 Richard Biener <rguenther@suse.de>
24967 * tree-vrp.c (get_single_symbol): Add assert that we don't
24968 get overflowed constants as invariant part.
24969 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24970 checks. Use wi::cmp instead of recursing for integer constants.
24971 (compare_values): Just ignore whether we assumed undefined
24972 overflow instead of failing the compare.
24973 (extract_range_for_var_from_comparison_expr): Add comment before the
24974 TREE_NO_WARNING sets.
24975 (test_for_singularity): Likewise.
24976 (extract_range_from_comparison): Do not disable optimization
24977 when we assumed undefined overflow.
24978 (extract_range_basic): Remove init of unused var.
24980 2017-05-09 Richard Biener <rguenther@suse.de>
24982 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24983 (extract_range_from_multiplicative_op_1): Adjust.
24984 (extract_range_from_binary_expr_1): Use int_const_binop.
24986 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24989 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24990 rs6000_store_data_bypass_p in seven define_bypass directives and
24991 in several comments.
24992 * config/rs6000/rs6000-protos.h: Add prototype for
24993 rs6000_store_data_bypass_p function.
24994 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24995 function implements slightly different (rs6000-specific) semantics
24996 than store_data_bypass_p, returning false rather than aborting
24997 with assertion error when arguments do not satisfy the
24998 requirements of store data bypass.
24999 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25000 rs6000_store_data_bypass_p.
25002 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25004 * config/xtensa/xtensa-protos.h
25005 (xtensa_initial_elimination_offset): New declaration.
25006 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25007 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25008 macro definition, add case for FRAME_POINTER_REGNUM when
25009 FRAME_GROWS_DOWNWARD.
25010 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25011 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25012 xtensa_initial_elimination_offset.
25014 2017-05-08 Nathan Sidwell <nathan@acm.org>
25016 * doc/invoke.texi: Alphabetize -fdump options.
25018 2017-05-08 Martin Sebor <msebor@redhat.com>
25020 PR translation/80280
25021 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25023 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25025 * target.def (compute_frame_layout): New optional target hook.
25026 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25027 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25028 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25030 * reload1.c (verify_initial_elim_offsets): Likewise.
25031 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25032 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25033 (arm_get_frame_offsets): Split up into this ...
25034 (arm_compute_frame_layout): ... and this function.
25036 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25038 * config/aarch64/constraints.md (Usa): New constraint.
25039 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25041 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25043 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25044 with_multilib_list after it has been checked.
25046 2017-05-08 Richard Biener <rguenther@suse.de>
25048 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25049 (bitmap_set_subtract_values): Likewise.
25051 2017-05-08 Richard Biener <rguenther@suse.de>
25053 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25054 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25056 (gimple_stmt_nonzero_warnv_p): Rename to ...
25057 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25059 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25060 (extract_range_basic): Adjust, do not disable propagation on
25061 strict overflow sensitive simplification.
25062 (vrp_visit_cond_stmt): Likewise.
25064 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25066 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25067 body size unconditionally.
25069 2017-05-07 Jeff Law <law@redhat.com>
25072 2017-05-06 Jeff Law <law@redhat.com>
25073 PR tree-optimization/78496
25074 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25077 PR tree-optimization/78496
25078 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25079 (simplify_stmt_using_ranges): Call it.
25080 (vrp_dom_walker::before_dom_children): Extract equivalences
25081 from an ASSERT_EXPR with an equality comparison against a
25084 2017-05-06 Jeff Law <law@redhat.com>
25086 PR tree-optimization/78496
25087 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25090 PR tree-optimization/78496
25091 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25092 (simplify_stmt_using_ranges): Call it.
25093 (vrp_dom_walker::before_dom_children): Extract equivalences
25094 from an ASSERT_EXPR with an equality comparison against a
25097 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25099 * lra-constraints.c (lra_copy_reg_equiv): New function.
25100 (split_reg): Use it to copy equivalence information from the
25101 original register to the spill register.
25103 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25105 PR rtl-optimization/75964
25106 * simplify-rtx.c (simplify_const_relational_operation): Remove
25107 invalid handling of comparisons of integer ABS.
25109 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25111 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25112 initialize to zero.
25113 (init_regs): Remove declaration.
25114 (function_arg_advance_32): Initialize error_p as boolean variable.
25116 2017-05-05 Nathan Sidwell <nathan@acm.org>
25118 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25119 lines. Use for (;;).
25121 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25123 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25124 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25125 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25126 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25127 VF=2 that require versioning.
25129 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25131 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25134 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25136 * diagnostic.h (diagnostic_override_option_index): Convert from
25137 macro to inline function.
25139 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25141 * diagnostic.c (last_module_changed_p): New function.
25142 (set_last_module): New function.
25143 (diagnostic_report_current_module): Convert macro usage to
25144 the above functions.
25145 * diagnostic.h (diagnostic_context::last_module): Strengthen
25146 from const line_map * to const line_map_ordinary *.
25147 (diagnostic_last_module_changed): Delete macro.
25148 (diagnostic_set_last_module): Delete macro.
25150 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25152 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25153 with diagnostic_report_diagnostic.
25154 (diagnostic_n_impl_richloc): Likewise.
25155 * diagnostic.h (report_diagnostic): Delete macro.
25156 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25157 with diagnostic_report_diagnostic.
25158 * substring-locations.c (format_warning_va): Likewise.
25160 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25162 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25163 save/restor of format_spec. Move option-printing code to...
25164 (print_option_information): ...this new function, and
25165 reimplement by simply printing to the pretty_printer,
25166 rather than appending to the format string.
25168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25170 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25171 handling logic into...
25172 (update_effective_level_from_pragmas): ...this new function.
25174 2017-05-04 Andrew Waterman <andrew@sifive.com>
25176 * config/riscv/riscv.opt (mstrict-align): New option.
25177 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25178 (SLOW_UNALIGNED_ACCESS): Define.
25179 (riscv_slow_unaligned_access): Declare.
25180 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25182 (riscv_slow_unaligned_access): New variable.
25183 (rocket_tune_info): Set slow_unaligned_access to true.
25184 (optimize_size_tune_info): Set slow_unaligned_access to false.
25185 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25186 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25187 (riscv_option_override): Set riscv_slow_unaligned_access.
25188 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25190 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25192 * config/riscv/riscv.md: Unify indentation.
25194 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25199 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25201 (extendsi<mode>2): Add support for doing sign extension via
25202 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25203 don't have ISA 3.0 instructions.
25204 (extendsi<mode>2 splitter): Likewise.
25205 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25206 generate the normal insns since SImode can now go in vector
25207 registers. Disallow the special UNSPECs needed for previous
25208 machines to hide SImode being used. Add new insns
25209 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25210 (fix_trunc<mode>si2_stfiwx): Likewise.
25211 (fix_trunc<mode>si2_internal): Likewise.
25212 (fixuns_trunc<mode>si2): Likewise.
25213 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25214 (fctiw<u>z_<mode>_smallint): Likewise.
25215 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25216 of floating point to 32-bit integer from doing a direct move to
25217 the GPR registers to do a store.
25218 (fctiwz_<mode>): Break long line.
25220 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25222 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25223 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25224 (addr_list, addr_offset_valid_p): New.
25225 (split_address_groups): Check offset validity with above function.
25226 (gt-tree-ssa-loop-ivopts.h): Include header file.
25228 2017-05-05 Nathan Sidwell <nathan@acm.org>
25230 * config.gcc (arm*-*-*): Add missing 'fi'.
25232 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25234 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25235 included in -fopt-info does not matter.
25236 * doc/optinfo.texi (-fopt-info): Fix description of default
25237 behavour. Explicitly say order of options included in -fopt-info
25240 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25242 * config.gcc: Allow combinations of aprofile and rmprofile values for
25243 --with-multilib-list.
25244 * config/arm/t-multilib: New file.
25245 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25246 variables. Remove setting of ISA and floating-point ABI in
25247 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25248 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25249 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25250 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25252 * config/arm/t-rmprofile: Likewise except for the matches changes.
25253 * doc/install.texi (--with-multilib-list): Document the combination of
25254 aprofile and rmprofile values and warn about pitfalls in doing that.
25256 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25258 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25259 (movdi_aarch64): Likewise.
25261 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25263 PR tree-optimization/80632
25264 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25266 (build_arrays): Initialize it for virtual phis.
25267 (fix_phi_nodes): Use it for virtual phis.
25269 PR tree-optimization/80558
25270 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25271 [x, y] op z into [x op, y op z] for op & or | if conditions
25274 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25275 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25278 * config/arm/arm.md (use_literal_pool): Remove.
25279 (64-bit immediate split): No longer takes cost into consideration
25280 if arm_disable_literal_pool is enabled.
25281 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25282 used when arm_disable_literal_pool is enabled.
25283 (arm_max_const_double_inline_cost): Remove use of
25284 arm_disable_literal_pool.
25285 (push_minipool_fix): Add assert.
25286 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25287 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25288 (no_literal_pool_sf_immediate): New.
25290 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25292 PR tree-optimization/80613
25293 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25294 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25296 2017-05-05 Richard Biener <rguenther@suse.de>
25298 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25300 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25302 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25303 of this flag from insn conditions due to removal from r247495.
25305 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25307 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25309 (arm_early_store_addr_dep_ptr): Likewise.
25310 * config/arm/aarch-common-protos.h
25311 (arm_early_load_addr_dep_ptr): Add prototype.
25312 (arm_early_store_addr_dep_ptr): Likewise.
25313 * config/arm/cortex-a53.md: Add new bypasses.
25315 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25317 * tree.c (next_type_uid): Change type to unsigned.
25318 (type_hash_canon): Decrement back next_type_uid if
25319 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25320 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25323 2017-05-04 Martin Sebor <msebor@redhat.com>
25325 * builtins.c: Fix a trivial typo in a comment.
25327 PR middle-end/79234
25328 * builtins.c (check_sizes): Adjust to handle reading past the end.
25329 Avoid printing excessive upper bound of ranges. Use %E to print
25330 tree nodes instead of converting them to %wu.
25331 (expand_builtin_memchr): New function.
25332 (compute_dest_size): Rename...
25333 (compute_objsize): ...to this.
25334 (expand_builtin_memcpy): Adjust.
25335 (expand_builtin_mempcpy): Adjust.
25336 (expand_builtin_strcat): Adjust.
25337 (expand_builtin_strcpy): Adjust.
25338 (check_strncat_sizes): Adjust.
25339 (expand_builtin_strncat): Adjust.
25340 (expand_builtin_strncpy): Adjust and simplify.
25341 (expand_builtin_memset): Adjust.
25342 (expand_builtin_bzero): Adjust.
25343 (expand_builtin_memcmp): Adjust.
25344 (expand_builtin): Handle memcmp.
25345 (maybe_emit_chk_warning): Check strncat just once.
25347 2017-05-04 Martin Sebor <msebor@redhat.com>
25349 PR preprocessor/79214
25350 PR middle-end/79222
25351 PR middle-end/79223
25352 * builtins.c (check_sizes): Add inlining context and issue
25353 warnings even when -Wno-system-headers is set.
25354 (check_strncat_sizes): Same.
25355 (expand_builtin_strncat): Same.
25356 (expand_builtin_memmove): New function.
25357 (expand_builtin_stpncpy): Same.
25358 (expand_builtin): Handle memmove and stpncpy.
25360 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25362 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25363 which is not used any more.
25365 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25367 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25369 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25371 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25372 (cortexa53_tunings): Likewise.
25373 (cortexa57_tunings): Likewise.
25374 (cortexa72_tunings): Likewise.
25375 (cortexa73_tunings): Likewise.
25377 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25379 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25380 Set loop alignment to 8.
25382 2017-05-04 Martin Sebor <msebor@redhat.com>
25384 PR translation/80280
25385 * builtins.c (expand_builtin_object_size): Add missing quoting to
25386 %D and like directives.
25387 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25388 (hsa_type_for_tree_type): Same.
25389 (verify_function_arguments): Same.
25390 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25391 * varasm.c (get_section): Same.
25394 2017-05-04 Martin Sebor <msebor@redhat.com>
25396 PR translation/80280
25397 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25399 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25401 * config/aarch64/aarch64.c (generic_addrcost_table):
25402 Change HI/TI mode setting.
25404 2017-05-04 Martin Jambor <mjambor@suse.cz>
25406 PR tree-optimization/80622
25407 * tree-sra.c (comes_initialized_p): New function.
25408 (build_accesses_from_assign): Only set write lazily when
25409 comes_initialized_p is false.
25410 (analyze_access_subtree): Use comes_initialized_p.
25411 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25412 instead of testing for PARM_DECL.
25414 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25416 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25417 constraint on operand 0 to allow more general addressing modes.
25418 Adjust output template.
25419 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25421 * config/aarch64/aarch64-protos.h
25422 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25423 * config/aarch64/constraints.md (Dp): New address constraint.
25424 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25427 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25429 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25430 update use of estimate_ipcp_clone_size_and_time.
25431 (estimate_local_effects): Update use of
25432 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25433 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25434 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25435 Return nonspecialized time.
25437 2017-05-04 Richard Biener <rguenther@suse.de>
25439 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25440 for the last VUSE which def dominates the PHI. Directly call
25442 (get_continuation_for_phi_1): Remove.
25444 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25446 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25447 to explain the use of truncating division. Cap the number of
25448 iterations to the maximum given by nb_iterations_upper_bound,
25451 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25453 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25454 * configure: Regenerate.
25455 * config.in: Regenerate.
25456 * config/i386/driver-mingw32.c: new file.
25457 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25458 * config.host: Link driver-mingw32.o on MinGW host.
25459 * doc/install.texi: Document new --enable-mingw-wildcard configure
25462 2017-05-04 Marek Polacek <polacek@redhat.com>
25464 PR tree-optimization/80612
25465 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25467 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25468 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25470 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25471 (movt splitter): Likewise.
25472 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25473 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25474 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25475 block for Thumb-1 with MOVT.
25476 (thumb2_legitimate_address_p): Move code block ...
25477 (can_avoid_literal_pool_for_label_p): ... into this new function.
25478 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25480 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25481 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25482 "M-profile targets with the MOVT instruction".
25484 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25486 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25487 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25488 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25490 2017-05-04 Martin Liska <mliska@suse.cz>
25492 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25493 variable cond_code.
25495 2017-05-04 Richard Biener <rguenther@suse.de>
25497 * tree.c (array_at_struct_end_p): Handle arrays at struct
25498 end with flexarrays more conservatively. Refactor and treat
25499 arrays of arrays or aggregates more strict. Fix
25500 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25501 * tree.c (array_at_struct_end_p): Adjust prototype.
25502 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25503 * gimple-fold.c (get_range_strlen): Likewise.
25504 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25506 2017-05-04 Richard Biener <rguenther@suse.de>
25508 PR tree-optimization/31130
25509 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25511 (supports_overflow_infinity): Likewise.
25512 (is_negative_overflow_infinity): Likewise.
25513 (is_positive_overflow_infinity): Likewise.
25514 (is_overflow_infinity): Likewise.
25515 (stmt_overflow_infinity): Likewise.
25516 (overflow_infinity_range_p): Likewise.
25517 (usable_range_p): Remove as always returning true.
25518 (make_overflow_infinity): Remove.
25519 (negative_overflow_infinity): Likewise.
25520 (positive_overflow_infinity): Likewise.
25521 (avoid_overflow_infinity): Likewise.
25522 (set_value_range): Adjust accordingly.
25523 (set_value_range_to_nonnegative): Likewise, remove now unused
25524 overflow_infinity arg.
25525 (vrp_operand_equal_p): Adjust.
25526 (update_value_range): Likewise.
25527 (range_int_cst_singleton_p): Likewise.
25528 (operand_less_p): Likewise.
25529 (compare_values_warnv): Likewise.
25530 (extract_range_for_var_from_comparison_expr): Likewise.
25531 (vrp_int_const_binop): Likewise.
25532 (zero_nonzero_bits_from_vr): Likewise.
25533 (extract_range_from_multiplicative_op_1): Likewise.
25534 (extract_range_from_binary_expr_1): Likewise.
25535 (extract_range_from_unary_expr): Likewise.
25536 (extract_range_from_comparison): Likewise.
25537 (extract_range_basic): Likewise.
25538 (adjust_range_with_scev): Likewise.
25539 (compare_ranges): Likewise.
25540 (compare_range_with_value): Likewise.
25541 (dump_value_range): Likewise.
25542 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25544 (simplify_cond_using_ranges): Adjust.
25546 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25548 * brig-builtins.def: Added a builtin for class_f64.
25549 * builtin-types.def: Added a builtin type needed by class_f64.
25551 2017-05-03 Jason Merrill <jason@redhat.com>
25553 * timevar.def: Add TV_CONSTEXPR.
25555 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25557 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25559 2017-05-03 Martin Jambor <mjambor@suse.cz>
25561 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25562 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25563 * ipa-cp.c (ipcp_propagate_stage): Do not call
25564 ipa_update_after_lto_read.
25565 * ipa-inline.c (ipa_inline): Likewise.
25567 2017-05-03 Martin Jambor <mjambor@suse.cz>
25569 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25570 tag. Added a default constructor and a destructor.
25571 (ipa_edge_args_sum_t): New class;
25572 (ipa_edge_args_sum): Declare.
25573 (ipa_edge_args_vector): Remove declaration.
25574 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25575 (ipa_free_edge_args_substructures): Remove declaration.
25576 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25577 (ipa_edge_args_info_available_for_edge_p): Likewise.
25578 * ipa-prop.c (ipa_edge_args_vector): Removed.
25579 (edge_removal_hook_holder): Likewise.
25580 (edge_duplication_hook_holder): Likewise.
25581 (ipa_edge_args_sum): New variable.
25582 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25583 ipa_edge_args_vector.
25584 (ipa_free_edge_args_substructures): Likewise.
25585 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25586 ipa_edge_args_vector.
25587 (ipa_edge_removal_hook): Turned into method
25588 ipa_edge_args_sum_t::remove.
25589 (ipa_edge_duplication_hook): Turned into method
25590 ipa_edge_args_sum_t::duplicate.
25591 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25592 registering edge hooks.
25593 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25594 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25595 ipa_edge_args_sum instead of ipa_edge_args_vector.
25596 * ipa-profile.c (ipa_profile): Likewise.
25598 2017-05-03 Martin Jambor <mjambor@suse.cz>
25600 * symbol-summary.h (function_summary): New method exists.
25601 (function_summary::symtab_removal): Deallocate through release.
25602 (call_summary): New class.
25603 (gt_ggc_mx): New overload.
25604 (gt_pch_nx): Likewise.
25605 (gt_pch_nx): Likewise.
25607 2017-05-03 Jeff Law <law@redhat.com>
25609 PR tree-optimization/78496
25610 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25611 from simplify_cond_using_ranges. Split off code to walk
25612 backwards through casts into ...
25613 (simplify_cond_using_ranges_2): New function.
25614 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25615 (execute_vrp): After identifying jump threads, call
25616 simplify_cond_using_ranges_2.
25618 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25621 * ipa-inline.h (inline_summary): Add ctor.
25622 (create_ggc): Do not use ggc_cleared_alloc.
25624 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25625 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25627 * gcc.c (handle_braces): Support escaping in switch matching
25629 * doc/invoke.texi (Spec Files): Document it.
25630 Remove superfluous @code markup in items.
25632 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25634 * diagnostic-show-locus.c (struct column_range): New struct.
25635 (get_affected_columns): New function.
25636 (get_printed_columns): New function.
25637 (struct correction): New struct.
25638 (correction::ensure_capacity): New function.
25639 (correction::ensure_terminated): New function.
25640 (struct line_corrections): New struct.
25641 (line_corrections::~line_corrections): New dtor.
25642 (line_corrections::add_hint): New function.
25643 (layout::print_trailing_fixits): Reimplement in terms of the new
25645 (selftest::test_overlapped_fixit_printing): New function.
25646 (selftest::diagnostic_show_locus_c_tests): Call it.
25648 2017-05-03 Nathan Sidwell <nathan@acm.org>
25650 Canonicalize canonical type hashing
25651 * tree.h (type_hash_canon_hash): Declare.
25652 * tree.c (type_hash_list, attribute_hash_list): Move into
25653 type_hash_canon_hash.
25654 (build_type_attribute_qual_variant): Break out hash code calc into
25655 type_hash_canon_hash.
25656 (type_hash_canon_hash): New. Generic type hash computation.
25657 (build_range_type_1, build_array_type_1, build_function_type,
25658 build_method_type_directly, build_offset_type, build_complex_type,
25659 make_vector_type): Call it.
25661 2017-05-03 Richard Biener <rguenther@suse.de>
25663 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25664 When all DRs have unknown misaligned do not always peel
25665 when there is a store but apply the same costing model as if
25666 there were only loads.
25668 2017-05-03 Richard Biener <rguenther@suse.de>
25671 PR tree-optimization/80492
25672 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25673 compare_base_decls returning dont-know properly.
25675 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25677 * config/arm/iterators.md (CCSI): New mode iterator.
25678 (arch): New mode attribute.
25679 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25680 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25681 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25682 code iterator for success result mode.
25683 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25684 the corresponding new insn generators.
25686 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25689 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25690 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25692 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25694 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25695 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25696 (DDR_A): Wrap DDR argument in brackets.
25697 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25698 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25699 (DDR_REVERSED_P): Likewise.
25701 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25703 PR tree-optimization/79472
25704 * tree-switch-conversion.c (struct switch_conv_info): Add
25705 contiguous_range and default_case_nonstandard fields.
25706 (collect_switch_conv_info): Compute contiguous_range and
25707 default_case_nonstandard fields, don't clear final_bb if
25708 contiguous_range and only the default case doesn't have the required
25710 (check_all_empty_except_final): Set default_case_nonstandard instead
25711 of failing if contiguous_range and the default case doesn't have empty
25713 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25714 and only the default case doesn't have the required constants. Skip
25716 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25717 if default_case_nonstandard.
25718 (build_constructors): Build constant 1 just once. Assert that default
25719 values aren't inserted in between cases if contiguous_range. Skip
25721 (build_arrays): Skip virtual phis.
25722 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25723 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25724 Handle virtual phis.
25725 (gen_inbound_check): Handle default_case_nonstandard case.
25726 (process_switch): Adjust check_final_bb caller. Call
25727 gather_default_values with the first non-default case instead of
25728 default case if default_case_nonstandard.
25730 2017-05-02 Nathan Sidwell <nathan@acm.org>
25732 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25733 check. Fix formatting.
25735 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25737 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25738 errors when comparing specialized and unspecialized times.
25740 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25742 * diagnostic-show-locus.c
25743 (layout::should_print_annotation_line_p): Make private.
25744 (layout::print_annotation_line): Make private.
25745 (layout::annotation_line_showed_range_p): Make private.
25746 (layout::show_ruler): Make private.
25747 (layout::print_source_line): Make private. Pass in line and
25748 line_width, rather than calling location_get_source_line. Drop
25750 (layout::print_leading_fixits): New method.
25751 (layout::print_any_fixits): Rename to...
25752 (layout::print_trailing_fixits): ...this, and make private.
25753 Don't print newline fixits.
25754 (diagnostic_show_locus): Move logic for printing one row into...
25755 (layout::print_line): ...this new function. Move the
25756 location_get_source_line call and error-handling from
25757 print_source_line to here. Call print_leading_fixits, and rename
25758 print_any_fixits to print_trailing_fixits.
25759 (selftest::test_fixit_insert_containing_newline): Update now that
25760 newlines are partially supported.
25761 (selftest::test_fixit_insert_containing_newline_2): New test.
25762 (selftest::test_fixit_replace_containing_newline): Update comments.
25763 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25764 * edit-context.c (class added_line): New class.
25765 (class edited_line): Describe newline handling in comment.
25766 (edited_line::actually_edited_p): New method.
25767 (edited_line::print_content): Delete redundant decl.
25768 (edited_line::m_predecessors): New field.
25769 (edited_file::print_content): Call edited_line::print_content.
25770 (edited_file::print_diff): Update to support newlines.
25771 (edited_file::print_diff_hunk): Likewise.
25772 (edited_file::print_run_of_changed_lines): New function.
25773 (edited_file::print_diff_line): Convert to...
25774 (print_diff_line): ...this.
25775 (edited_file::get_effective_line_count): New function.
25776 (edited_line::edited_line): Initialize new field m_predecessors.
25777 (edited_line::~edited_line): Clean up m_predecessors.
25778 (edited_line::apply_fixit): Handle newlines.
25779 (edited_line::get_effective_line_count): New function.
25780 (edited_line::print_content): New function.
25781 (edited_line::print_diff_lines): New function.
25782 (selftest::test_applying_fixits_insert_containing_newline): New
25784 (selftest::test_applying_fixits_replace_containing_newline): New
25786 (selftest::insert_line): New function.
25787 (selftest::test_applying_fixits_multiple_lines): Add example of
25789 (selftest::edit_context_c_tests): Call the new tests.
25791 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25793 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25794 parameter cand. Update dump information.
25795 (get_computation_cost): Update uses.
25797 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25799 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25800 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25801 (get_computation_at, rewrite_use_address): Update use of
25802 get_computation_aff.
25804 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25806 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25807 (get_computation): Delete.
25808 (get_computation_cost): Implement like get_computation_cost_at.
25809 Use get_computation_at.
25810 (get_computation_cost_at): Delete.
25811 (rewrite_use_nonlinear_expr): Use get_computation_at.
25812 (rewrite_use_compare, remove_unused_ivs): Ditto.
25814 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25816 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25818 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25820 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25821 (ivopts_global_cost_for_size): Rename parameter and update uses.
25822 (iv_ca_recount_cost): Update uses.
25823 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25824 candidates seperately in n_invs and n_cands.
25825 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25827 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25829 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25830 (find_inv_vars_cb): New.
25831 (find_depends): Renamed to ...
25832 (find_inv_vars): ... this.
25833 (add_candidate_1, force_var_cost): Call find_inv_vars.
25834 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25838 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25839 inv_vars. Add inv_exprs.
25840 (struct iv_cand): Rename depends_on to inv_vars.
25841 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25842 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25843 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25844 (dump_cand): Dump inv_vars.
25845 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25846 (record_invariant, find_depends, add_candidate_1): Ditto.
25847 (set_group_iv_cost, force_var_cost): Ditto.
25848 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25849 (get_computation_cost_at, get_computation_cost): Ditto.
25850 (determine_group_iv_cost_generic): Ditto.
25851 (determine_group_iv_cost_address): Ditto.
25852 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25853 (determine_group_iv_costs): Ditto.
25854 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25855 (iv_ca_set_remove_invariants): Renamed to ...
25856 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25857 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25858 (iv_ca_set_add_invariants): Renamed to ...
25859 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25860 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25861 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25862 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25863 (create_new_ivs): Remove useless dump.
25865 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25867 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25869 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25870 (iv_ca_set_no_cp, create_new_iv): Ditto.
25872 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25874 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25876 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25878 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25879 function tree_check2.
25881 2017-05-02 Martin Liska <mliska@suse.cz>
25883 * doc/gcov.texi: Add missing preposition.
25884 * gcov.c (function_info::function_info): Properly fill up
25885 all member variables.
25887 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25889 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25891 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25893 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25895 2017-05-02 Martin Liska <mliska@suse.cz>
25898 * lto-streamer-in.c (lto_read_tree_1): Remove
25899 LTO_STREAMER_DEBUG.
25900 * lto-streamer.c (struct tree_hash_entry): Likewise.
25901 (struct tree_entry_hasher): Likewise.
25902 (tree_entry_hasher::hash): Likewise.
25903 (tree_entry_hasher::equal): Likewise.
25904 (lto_streamer_init): Likewise.
25905 (lto_orig_address_map): Likewise.
25906 (lto_orig_address_get): Likewise.
25907 (lto_orig_address_remove): Likewise.
25908 * lto-streamer.h: Likewise.
25909 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25910 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25912 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25914 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25915 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25916 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25917 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25918 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25919 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25920 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25921 (mm_maskz_sub_ss): New intrinsics.
25922 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25923 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25924 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25925 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25926 (__builtin_ia32_subss_mask_round): New builtins.
25927 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25928 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25929 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25931 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25932 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25934 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25937 2017-05-02 Martin Jambor <mjambor@suse.cz>
25939 PR tree-optimization/78687
25940 * tree-sra.c (access): New field parent.
25941 (process_subtree_disqualification): New function.
25942 (disqualify_candidate): Call it.
25943 (build_accesses_from_assign): Reset write flag if creating an
25945 (build_access_subtree): Fill in parent field and also prpagate
25946 down grp_write flag.
25947 (create_artificial_child_access): New parameter set_grp_write, set
25948 grp_write to its value.
25949 (propagate_subaccesses_across_link): Also propagate grp_write flag
25951 (propagate_all_subaccesses): Push the closest parent back to work
25952 queue if add_access_to_work_queue returned true.
25954 2017-05-02 Richard Biener <rguenther@suse.de>
25956 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25957 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25958 -fstrict-overflow documentation.
25959 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25960 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25961 flag_strict_overflow.
25962 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25963 * lto-opts.c (lto_write_options): Do not stream it.
25964 * lto-wrapper.c (merge_and_complain): Do not handle it.
25965 * opts.c (default_options_table): Do not set -fstrict-overflow.
25966 (finish_options): Likewise do not clear it when sanitizing.
25967 * simplify-rtx.c (simplify_const_relational_operation): Do not
25968 test flag_strict_overflow.
25970 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25972 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25973 using enabled attribute.
25974 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25975 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25976 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25977 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25978 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25979 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25980 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25981 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25982 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25983 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25985 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25987 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25989 2017-05-02 Richard Biener <rguenther@suse.de>
25991 PR tree-optimization/80591
25993 2017-04-10 Richard Biener <rguenther@suse.de>
25995 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25998 2017-05-02 Richard Biener <rguenther@suse.de>
26000 PR tree-optimization/80549
26001 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26002 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26003 headers if they do not have a preheader.
26005 2017-05-02 Martin Liska <mliska@suse.cz>
26008 * common.opt: Fix typo.
26009 * doc/invoke.texi: Likewise.
26011 2017-05-01 Jan Beulich <jbeulich@suse.com>
26013 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26014 swapping, add (x,x,m,x,n) alternative.
26016 2017-05-01 Nathan Sidwell <nathan@acm.org>
26018 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26019 unnecessary unadjusted_alignment check.
26021 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26024 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26026 * gimplify.c (gimplify_cilk_detach): New function.
26027 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26028 * tree-core.h: Document EXPR_CILK_SPAWN.
26029 * tree.h (EXPR_CILK_SPAWN): Define.
26031 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26033 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26034 to use new fixit_hint representation, using the "replace" logic.
26035 (get_line_span_for_fixit_hint): Likewise.
26036 (layout::print_any_fixits): Likewise.
26037 (selftest::test_one_liner_many_fixits): Rename to...
26038 (selftest::test_one_liner_many_fixits_1): ...this, and update
26039 comment and expected output to reflect that the multiple fix-it
26040 hints are now consolidated into one insertion.
26041 (selftest::test_one_liner_many_fixits_2): New test.
26042 (selftest::test_diagnostic_show_locus_one_liner): Update for
26044 (selftest::test_fixit_consolidation): Update for fix-it API
26046 * diagnostic.c (print_parseable_fixits): Likewise.
26047 * edit-context.c (edited_line::m_line_events): Convert from
26048 auto_vec <line_event *> to auto_vec <line_event>.
26049 (class line_event): Convert from abstract base class to a concrete
26050 class, taking over the role of replace_event.
26051 (class insert_event): Delete.
26052 (class replace_event): Rename to class line_event. Convert to
26054 (edit_context::add_fixits): Reimplement.
26055 (edit_context::apply_insert): Delete.
26056 (edit_context::apply_replace): Rename to...
26057 (edit_context::apply_fixit): ...this. Convert to half-open range.
26058 (edited_file::apply_insert): Delete.
26059 (edited_file::apply_replace): Rename to...
26060 (edited_file::apply_fixit): ...this.
26061 (edited_line::~edited_line): Drop deletion of events.
26062 (edited_line::apply_insert): Delete.
26063 (edited_line::apply_replace): Rename to...
26064 (edited_line::apply_fixit): ...this. Convert to half-open range.
26065 Update for change to type of m_line_events.
26066 * edit-context.h (edit_context::apply_insert): Delete.
26067 (edit_context::apply_replace): Rename to...
26068 (edit_context::apply_fixit): ...this.
26070 2017-05-01 Martin Sebor <msebor@redhat.com>
26072 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26075 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26078 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26079 __get_cpuid_max returns 0.
26080 (__get_cpuid_count): Ditto.
26082 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26084 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26085 replacement expression is another instance of one of its arguments.
26087 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26090 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26091 check for stack push/pop autoinc.
26092 * config/i386/i386.c (ix86_agi_dependent): Return false
26093 if the only reason why modified_in_p returned true is that
26094 addr is SP based and set_insn is a push or pop.
26096 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26098 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26101 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26104 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26105 (account_size_time): Use two predicates - exec_pred and
26107 (evaluate_conditions_for_known_args): Compute both clause and
26109 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26110 (inline_summary_t::duplicate): Update.
26111 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26113 (compute_inline_parameters): Likewise.
26114 (estimate_edge_size_and_time): Update caluclation of time.
26115 (estimate_node_size_and_time): Compute both time and nonspecialized
26117 (estimate_ipcp_clone_size_and_time): Update.
26118 (inline_merge_summary): Update.
26119 (do_estimate_edge_time): Update.
26120 (do_estimate_edge_size): Update.
26121 (do_estimate_edge_hints): Update.
26122 (inline_read_section, inline_write_summary): Stream both new predicates.
26123 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26125 (compute_inlined_call_time): Cleanup.
26126 (big_speedup_p): Update.
26127 (edge_badness): Update.
26128 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26129 (size_time_entry): Replace predicate by exec_predicate and
26130 nonconst_predicate.
26131 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26132 (estimate_edge_time): Return also nonspec_time.
26133 (reset_edge_growth_cache): Update.
26135 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26137 PR rtl-optimization/80491
26138 * ifcvt.c (noce_process_if_block): When looking for x setter
26139 with missing else_bb, don't check only the insn right before
26140 cond_earliest, but look for the last insn that x is modified in
26141 within the same bb.
26143 PR rtl-optimization/80491
26144 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26146 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26148 PR tree-optimization/80487
26149 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26151 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26153 PR tree-optimization/79697
26154 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26155 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26156 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26158 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26159 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26161 2017-04-28 Martin Sebor <msebor@redhat.com>
26163 PR tree-optimization/80523
26164 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26165 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26167 (maybe_warn, format_directive, parse_directive): Use new functions.
26168 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26170 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26172 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26174 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26176 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26177 target_header_dir): Set correctly.
26178 * configure: Regenerated.
26179 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26180 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26181 instead of SYSTEM_HEADER_DIR.
26183 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26185 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26186 (estimate_local_effects): Likewise.
26187 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26188 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26189 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26190 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26191 do_estimate_edge_time, estimate_edge_time): Likewise.
26192 * ipa-inline-analysis.c (estimate_node_size_and_time,
26193 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26194 (estimate_time_after_inlining): Remove.
26196 2017-04-28 Martin Liska <mliska@suse.cz>
26198 * doc/gcov.texi: Enhance documentation of gcov.
26200 2017-04-28 Martin Liska <mliska@suse.cz>
26202 * doc/gcov.texi: Sort options in alphabetic order.
26203 * doc/gcov-dump.texi: Likewise.
26204 * doc/gcov-tool.texi: Likewise.
26205 * gcov.c (print_usage): Likewise.
26206 * gcov-dump.c (print_usage): Likewise.
26207 * gcov-tool.c (print_merge_usage_message): Likewise.
26208 (print_rewrite_usage_message): Likewise.
26209 (print_overlap_usage_message): Likewise.
26211 2017-04-28 Martin Liska <mliska@suse.cz>
26213 PR gcov-profile/53915
26214 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26216 2017-04-28 Martin Liska <mliska@suse.cz>
26218 PR gcov-profile/79891
26219 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26220 is marked by compiler as living on a line.
26221 (get_cycles_count): Remove usage of the union.
26222 (output_intermediate_file): Likewise.
26223 (find_source): Fix GNU coding style.
26224 (accumulate_line_counts): Remove old non-all block mode.
26225 (output_lines): Remove usage of the union.
26226 * profile.c (output_location): Include all BBs, even if
26227 belonging to a same line (and file) as a previous BB.
26229 2017-04-28 Martin Liska <mliska@suse.cz>
26231 * gcov.c (process_args): Handle new argument 'w'.
26232 (read_graph_file): Assign ID to BBs.
26233 (output_branch_count): Display BB # if verbose flag is set.
26234 (output_lines): Likewise for arcs.
26235 (print_usage): Add '--verbose' option help.
26236 * doc/gcov.texi: Document --verbose (-w) option.
26238 2017-04-28 Martin Liska <mliska@suse.cz>
26240 * gcov.c (struct block_location_info): New struct.
26241 (process_file): Fill up the new structure.
26242 (read_graph_file): Replace usage of encoding by the newly added
26244 (add_line_counts): Likewise.
26245 (accumulate_line_counts): Remove usage of the union.
26246 (function_info::function_info): New function.
26247 (function_info::~function_info): Likewise.
26248 (process_file): Call delete instead of release_function.
26249 (release_function): Release the function.
26250 (release_structures): Call delete instead of release_function.
26251 (solve_flow_graph): Replace usage of num_blocks.
26252 (find_exception_blocks): Likewise.
26253 (output_lines): Fix GNU coding style.
26255 2017-04-28 Martin Liska <mliska@suse.cz>
26258 * coverage.c (coverage_remove_note_file): New function.
26259 * coverage.h: Declare the function.
26260 * toplev.c (finalize): Clean if an error has been seen.
26262 2017-04-28 Martin Liska <mliska@suse.cz>
26264 PR gcov-profile/80031
26265 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26266 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26267 * gcov.c (read_graph_file): Read just number of blocks.
26268 * profile.c (branch_prob): Do not stream 0 flags per a basic
26271 2017-04-28 Martin Liska <mliska@suse.cz>
26273 * gcov-dump.c (tag_*): Add new argument to declarations.
26274 (dump_gcov_file): Likewise.
26275 (tag_blocks): Add and use new argument depth.
26276 (tag_arcs): Likewise.
26277 (tag_lines): Likewise.
26278 (tag_counters): Likewise.
26279 (tag_summary): Likewise.
26280 (dump_working_sets): Use depth to do a proper indentation.
26282 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26285 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26286 * symtab.c (symtab_node::debug_symtab): Move definition here.
26288 2017-04-28 Richard Biener <rguenther@suse.de>
26290 * lto-streamer.h (LTO_major_version): Bump to 7.
26292 2017-04-28 Richard Biener <rguenther@suse.de>
26294 * tree-vrp.c (assert_info): New struct.
26295 (add_assert_info): New helper.
26296 (register_edge_assert_for_2): Refactor to add asserts to a vector
26298 (register_edge_assert_for_1): Likewise.
26299 (register_edge_assert_for): Likewise.
26300 (finish_register_edge_assert_for): New helper actually registering
26301 asserts where live on edge.
26302 (find_conditional_asserts): Adjust.
26303 (find_switch_asserts): Likewise.
26304 (evrp_dom_walker::try_find_new_range): Generalize.
26305 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26307 2017-04-27 Marek Polacek <polacek@redhat.com>
26310 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26311 arg10 and arg11 to itype.
26313 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26315 * doc/extend.texi (Object Size Checking): Improve grammar.
26317 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26320 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26321 that the logic for permitting reciprocal estimates matches that
26324 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26327 * tree.c (type_cache_hasher::equal): Only compare
26328 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26329 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26330 non-aggregate element types.
26331 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26332 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26334 2017-04-27 Richard Biener <rguenther@suse.de>
26336 PR middle-end/80533
26337 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26338 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26339 keeping a reference to a trailing array.
26341 2017-04-27 Richard Biener <rguenther@suse.de>
26343 PR middle-end/80539
26344 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26345 being in loop-closed SSA form conservatively.
26346 (chrec_fold_multiply_poly_poly): Likewise.
26348 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26350 PR middle-end/79665
26351 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26352 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26354 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26357 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26358 (aarch64_function_arg_alignment): Return unsigned int again, but still
26359 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26360 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26361 Don't emit -Wpsabi note.
26362 (aarch64_function_arg_boundary): Likewise.
26363 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26366 2017-04-26 Nathan Sidwell <nathan@acm.org>
26368 * tree.h (crc32_unsigned_n): Declare.
26369 (crc32_unsigned, crc32_unsigned): Make inline.
26370 * tree.c (crc32_unsigned_bits): Replace with ...
26371 (crc32_unsigned_n): ... this.
26372 (crc32_unsigned, crc32_byte): Remove.
26373 (crc32_string): Remove unnecessary braces.
26375 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26377 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26378 * ipa-inline-analysis.c (MAX_TIME): Remove.
26379 (account_size_time): Use sreal for time.
26380 (dump_inline_summary): Update.
26381 (estimate_function_body_sizes): Update.
26382 (estimate_edge_size_and_time): Update.
26383 (estimate_calls_size_and_time): Update.
26384 (estimate_node_size_and_time): Update.
26385 (inline_merge_summary): Update.
26386 (inline_update_overall_summary): Update.
26387 (estimate_time_after_inlining): Update.
26388 (inline_read_section): Update.
26389 (inline_write_summary): Update.
26390 * ipa-inline.c (compute_uninlined_call_time): Update.
26391 (compute_inlined_call_time): Update.
26392 (recursive_inlining): Update.
26393 (inline_small_functions): Update.
26394 (dump_overall_stats): Update.
26395 * ipa-inline.h: Include sreal.h.
26396 (size_time_entry): Turn time to sreal.
26397 (inline_summary): Turn self_time nad time to sreal.
26399 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26401 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26403 (sreal::stream_out, sreal::stream_in): New.
26404 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26406 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26408 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26411 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26414 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26416 (dimode_scalar_chain::compute_convert_gain): Ditto.
26417 (dimode_scalar_chain::make_vector_copies): Ditto.
26418 (dimode_scalar_chain::convert_reg): Ditto.
26419 (dimode_scalar_chain::convert_insn): Ditto.
26420 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26421 (VI248_AVX512BW_1): New mode iterator.
26422 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26423 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26426 2017-04-25 Martin Sebor <msebor@redhat.com>
26428 PR tree-optimization/80497
26429 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26430 constants are representable in HOST_WIDE_INT.
26431 (parse_directive): Ditto.
26433 2017-04-25 Martin Sebor <msebor@redhat.com>
26436 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26437 (new_zero_array): Adjust signature.
26438 (dom_info::dom_init): Used unsigned rather that size_t.
26439 (dom_info::dom_info): Same.
26441 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26442 Jakub Jelinek <jakub@redhat.com>
26445 * config/arm/arm.c: Include gimple.h.
26446 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26447 returns negative, increment ncrn only if it returned positive.
26448 (arm_needs_doubleword_align): Return int instead of bool,
26449 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26450 members, but if there is any such non-FIELD_DECL
26451 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26452 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26453 returns negative, increment nregs only if it returned positive.
26454 (arm_setup_incoming_varargs): Likewise.
26455 (arm_function_arg_boundary): Emit -Wpsabi note if
26456 arm_needs_doubleword_align returns negative, return
26457 DOUBLEWORD_ALIGNMENT only if it returned positive.
26459 2017-04-25 Marek Polacek <polacek@redhat.com>
26462 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26463 first argument to type.
26465 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26468 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26469 type checks to test for compatibility instead of equality.
26471 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26472 Jakub Jelinek <jakub@redhat.com>
26475 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26477 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26478 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26479 the alignment computation, but return their maximum in warn_alignment.
26480 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26481 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26483 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26484 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26487 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26489 * config/arc/simdext.md (dmpyh): Fix typo.
26491 2017-04-25 Richard Biener <rguenther@suse.de>
26493 PR tree-optimization/80492
26494 * alias.c (compare_base_decls): Handle registers with asm
26495 specification conservatively.
26496 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26497 compare_base_decls returning dont-know properly.
26499 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26501 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26502 (legitimate_offset_address_p): New function.
26503 (arc_legitimate_address_p): Use above function.
26505 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26507 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26509 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26511 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26512 ACCH registers whenever they are available.
26514 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26516 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26517 double regs fix when not used.
26519 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26521 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26523 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26524 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26526 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26528 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26529 when emitting short ADD instructions.
26531 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26533 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26535 (cmpsi_cc_c_insn): Likewise.
26536 (cbranchsi4_scratch): Compute proper instruction length using
26537 compact_hreg_operand.
26538 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26540 2017-04-25 Richard Biener <rguenther@suse.de>
26542 PR middle-end/80509
26543 * passes.c (pass_manager::pass_manager): Initialize
26544 m_name_to_pass_map.
26546 2017-04-25 Richard Biener <rguenther@suse.de>
26548 PR tree-optimization/79201
26549 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26551 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26554 * config/s390/vector.md: Split MEM->GPR vector moves for
26555 non-s_operand addresses.
26557 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26560 * config/s390/predicates.md (reload_const_wide_int_operand): New
26562 * config/s390/s390.md ("movti"): Remove d/P alternative.
26563 ("movti_bigconst"): New pattern definition.
26565 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26568 * s390-protos.h (s390_expand_cs_hqi): Removed.
26569 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26570 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26571 modes as well as CCZ1mode and CCZmode.
26572 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26573 signature of s390_emit_compare_and_swap.
26574 (s390_expand_cs_hqi): Likewise, make static.
26575 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26576 compare-and-swap, in some cases.
26577 (s390_expand_cs): Wrapper function.
26578 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26580 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26581 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26582 patterns for small and large integers. Forbid symref memory operands.
26583 Move expander to s390.c. Require cc register.
26584 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26585 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26586 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26587 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26588 symref memory operands. Remove CC mode and call s390_match_ccmode
26590 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26592 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26594 * config/s390/s390.md (define_peephole2): New peephole to help
26595 combining the load-and-test pattern with volatile memory.
26597 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26599 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26600 with CCZmode for TARGET_Z196.
26602 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26604 PR rtl-optimization/80501
26605 * combine.c (make_compound_operation_int): Set subreg_code to SET
26606 even for AND with mask of the sign bit of mode.
26608 PR rtl-optimization/80500
26609 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26610 sum's initial value.
26612 2017-04-25 Julian Brown <julian@codesourcery.com>
26613 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26615 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26617 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26619 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26621 2017-04-25 Julian Brown <julian@codesourcery.com>
26622 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26624 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26625 (thunderx2t99_sha): New Reservation.
26627 2017-04-25 Julian Brown <julian@codesourcery.com>
26628 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26630 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26631 type for 1-element load.
26633 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26635 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26637 2017-04-24 Martin Jambor <mjambor@suse.cz>
26639 PR tree-optimization/80293
26640 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26641 char arrays not totally scalarizable if it is false.
26642 (analyze_all_variable_accesses): Pass correct value in the new
26643 parameter. Add a statistics counter.
26645 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26647 PR middle-end/79931
26648 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26650 2017-04-24 Richard Biener <rguenther@suse.de>
26652 PR tree-optimization/80494
26653 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26654 out for complex types.
26656 2017-04-24 Richard Biener <rguenther@suse.de>
26658 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26659 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26660 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26661 (DFS): Adjust and never fail.
26662 (sccvn_dom_walker::fail): Remove.
26663 (sccvn_dom_walker::before_dom_children): Adjust.
26664 (run_scc_vn): Likewise and never fail.
26665 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26666 (pass_fre::execute): Likewise.
26668 2017-04-24 Richard Biener <rguenther@suse.de>
26670 PR tree-optimization/79725
26671 * tree-ssa-sink.c (statement_sink_location): Return whether
26672 failure reason was zero uses. Move that check later.
26673 (sink_code_in_bb): Deal with zero uses by removing the stmt
26676 2017-04-24 Richard Biener <rguenther@suse.de>
26679 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26680 pointer-based references.
26682 2017-04-24 Richard Biener <rguenther@suse.de>
26685 * pass_manager.h (pass_manager::operator new): Remove.
26686 (pass_manager::operator delete): Likewise.
26687 * passes.c (pass_manager::operator new): Remove.
26688 (pass_manager::operator delete): Likewise.
26689 (pass_manager::pass_manager): Zero individual pass members.
26691 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26694 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26695 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26696 Check "XEXP (src, 1)" operand here.
26697 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26698 Check "XEXP (src, 1)" operand here.
26699 (dimode_scalar_chain::make_vector_copies): Detect count register
26700 of a shift instruction. Zero extend count register from QImode
26701 to DImode to satisfy vector shift pattern count operand predicate.
26702 Substitute vector shift count operand with a DImode copy.
26703 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26706 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26708 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26709 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26710 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26711 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26712 (UNSPEC_NOREX_MEM): Remove definition.
26714 2017-04-21 Richard Biener <rguenther@suse.de>
26716 PR tree-optimization/79547
26717 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26718 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26719 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26720 without any constraints.
26722 2017-04-21 Richard Biener <rguenther@suse.de>
26724 PR tree-optimization/78847
26725 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26727 2017-04-21 Richard Biener <rguenther@suse.de>
26729 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26730 (build_distinct_type_copy): Likewise.
26731 (build_variant_type_copy): Likewise.
26732 * tree.c (build_qualified_type): Pass down mem-stat info.
26733 (build_distinct_type_copy): Likewise.
26734 (build_variant_type_copy): Likewise.
26736 2017-04-21 Richard Biener <rguenther@suse.de>
26738 PR tree-optimization/80237
26739 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26741 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26742 for a simplified result.
26744 2016-04-21 Richard Biener <rguenther@suse.de>
26746 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26747 sth as strict as a simple_iv but a chrec without symbols and an
26748 operand defined in the loop we are peeling (and not some subloop).
26749 (propagate_constants_for_unrolling): Propagate all constants.
26751 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26754 * config/i386/i386.c (print_reg): Remove assert for disalowed
26755 regno values, call output_operand_lossage instead.
26757 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26760 * config/i386/constraints.md (Yc): New register constraint.
26761 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26762 Use Yc constraint for alternative 2 of operand 0. Remove
26763 preferred_for_speed attribute.
26765 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26767 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26768 lastprivate clauses in SIMT case.
26770 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26772 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26774 2017-04-20 Richard Biener <rguenther@suse.de>
26776 PR tree-optimization/57796
26777 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26778 as N scalar stores.
26779 (vect_model_load_cost): Cost gathers as N scalar loads.
26781 2017-04-20 Richard Biener <rguenther@suse.de>
26783 * ggc-page.c (ggc_allocated_p): Rename to ...
26784 (safe_lookup_page_table_entry): ... this and return the lookup
26786 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26788 2017-04-20 Richard Biener <rguenther@suse.de>
26790 PR tree-optimization/80453
26791 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26792 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26793 from the conditions.
26794 (vn_phi_eq): Pass them down.
26795 (vn_phi_lookup): Record them.
26796 (vn_phi_insert): Likewise.
26798 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26800 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26801 uninitialized variable warning to avoid buffer overrun.
26803 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26806 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26807 is suppressed for '{ 0 }' in C.
26809 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26811 * BASE-VER: Set to 8.0.0.
26813 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26815 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26816 priority .init_array and .fini_array section with SECTION_NOTYPE
26819 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26821 PR middle-end/80423
26822 * tree.h (build_array_type): Add typeless_storage default argument.
26823 * tree.c (type_cache_hasher::equal): Also compare
26824 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26825 (build_array_type): Add typeless_storage argument, set
26826 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26828 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26829 (build_array_type): Likewise. Add typeless_storage argument.
26831 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26832 Jakub Jelinek <jakub@redhat.com>
26834 PR tree-optimization/80426
26835 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26836 operation on symbolic operands, also compute the overflow for the
26837 invariant part when the operation degenerates into a negation.
26839 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26842 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26843 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26846 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26848 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26851 * config/avr/avr.c (tree.h): Include it.
26852 (cgraph.h): Include it.
26853 (avr_encode_section_info): Don't warn for uninitialized progmem
26854 variable if it's just an alias.
26856 2017-04-19 Richard Biener <rguenther@suse.de>
26859 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26860 when needed by AutoPGO.
26862 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26865 * doc/lto.texi: Remove an extra 'that'.
26867 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26869 PR rtl-optimization/80429
26870 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26871 are only used in debug insns.
26873 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26874 Vladimir Makarov <vmakarov@redhat.com>
26876 * config/sparc/predicates.md (input_operand): Add comment. Return
26877 true for any memory operand when LRA is in progress.
26878 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26880 2017-04-18 Jeff Law <law@redhat.com>
26883 * mips.md ({return,simple_return}_internal): Do not overwrite
26886 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26888 PR tree-optimization/80443
26889 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26890 instead of adding 1, subtract -1 and similarly instead of subtracting
26893 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26895 PR rtl-optimization/80357
26896 * haifa-sched.c (tmp_bitmap): New variable.
26897 (model_recompute): Handle duplicate use records.
26898 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26899 (free_global_sched_pressure_data): Free it.
26901 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26904 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26905 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26906 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26907 instead of SYSTEM_HEADER_DIR.
26909 2017-04-18 Jeff Law <law@redhat.com>
26911 PR middle-end/80422
26912 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26913 predecessors after walking up the insn chain.
26915 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26918 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26919 sizetype type into debug info.
26921 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26924 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26925 unneeded test for TARGET_UPPER_REGS_SF.
26926 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26928 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26931 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26932 instead of gsi_after_labels.
26934 2017-04-18 Jeff Law <law@redhat.com>
26936 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26940 2017-04-13 Jeff Law <law@redhat.com>
26941 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26942 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26944 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26947 * config/avr/avr.c (intl.h): Include it.
26948 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26950 2017-04-18 Martin Liska <mliska@suse.cz>
26952 PR gcov-profile/78783
26953 * gcov-tool.c (gcov_output_files): Validate that destination
26954 file is either removed by the tool or by a user.
26956 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26957 Guy Benyei <guybe@mellanox.com>
26959 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26960 block, and do not negate it, the stored id is already negative.
26962 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26964 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26966 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26969 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26970 masks of options that should be turned off if the VSX vector
26971 options are turned off.
26972 (OTHER_P8_VECTOR_MASKS): Likewise.
26973 (OTHER_VSX_VECTOR_MASKS): Likewise.
26974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26975 rs6000_disable_incompatible_switches to validate no type switches
26977 (rs6000_incompatible_switch): New function to disallow turning on
26978 other vector options if -mno-vsx, -mno-power8-vector, or
26979 -mno-power9-vector are specified.
26981 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26983 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26985 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26987 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26988 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26989 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26990 (ARG_POINTER_CFA_OFFSET): Likewise.
26992 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26994 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26995 conditions to take advantage of various optimizations.
26997 2017-04-13 Jeff Law <law@redhat.com>
26999 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27000 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27001 (zero_extendsidi2_dext): Likewise.
27003 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27006 * fold-const.c (fold_ternary_loc): Revert
27007 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27010 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27012 PR rtl-optimization/80343
27013 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27014 new scratch pseudo.
27016 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27019 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27020 to ubsan_encode_value.
27022 2017-04-13 Jeff Law <law@redhat.com>
27024 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27025 appearing in DEBUG_INSNs.
27027 2017-04-13 Martin Liska <mliska@suse.cz>
27029 PR gcov-profile/80413
27030 * gcov-io.c (gcov_write_string): Copy to buffer just when
27031 allocated size is greater than zero.
27033 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27036 * dwarf2out.c (decls_for_scope): Ignore declarations of
27037 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27039 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27042 * ipa-visibility.c (non_local_p): Fix typos.
27043 (localize_node): When localizing symbol in same comdat group,
27044 dissolve the group only when we know external symbols are going
27046 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27048 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27050 PR tree-optimization/79390
27051 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27052 order does not result in usable sequence, retry with reversed operand
27058 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27059 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27060 op0 instead of fold_convert_loc (loc, type, arg0).
27062 2017-04-12 Jeff Law <law@redhat.com>
27064 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27065 has a delay slot in the generated code.
27067 * config/cris/cris.md (cris_preferred_reload_class): Return
27068 GENNONACR_REGS rather than GENERAL_REGS.
27070 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27073 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27074 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27075 signedness of the result type.
27077 2017-04-12 Richard Biener <rguenther@suse.de>
27078 Jeff Law <law@redhat.com>
27080 PR tree-optimization/80359
27081 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27082 trim stores to TARGET_MEM_REFs.
27084 2017-04-12 Richard Biener <rguenther@suse.de>
27086 PR tree-optimization/79390
27087 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27088 threading case even more.
27090 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27093 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27094 for quad_address_p for TImode, instead of just not indexed_address.
27096 2017-04-12 Richard Biener <rguenther@suse.de>
27097 Bernd Edlinger <bernd.edlinger@hotmail.de>
27099 PR middle-end/79671
27100 * alias.c (component_uses_parent_alias_set_from): Handle
27101 TYPE_TYPELESS_STORAGE.
27102 (get_alias_set): Likewise.
27103 * tree-core.h (tree_type_common): Add typeless_storage flag.
27104 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27105 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27106 for types containing members with TYPE_TYPELESS_STORAGE.
27107 (place_field): Likewise.
27108 (layout_type): Likewise for ARRAY_TYPE.
27109 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27110 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27111 TYPE_TYPELESS_STORAGE.
27112 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27114 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27117 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27118 first argument to type.
27120 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27124 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27125 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27126 (rs6000_expand_binop_builtin): Likewise.
27127 (rs6000_expand_ternop_builtin): Likewise; also add missing
27128 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27129 vshasigma built-ins.
27130 * doc/extend.texi: Document that vec_xxpermdi's third argument
27131 must be a constant.
27133 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27135 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27136 Use shift_const cost parameter when calculating gain of STV shifts.
27138 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27140 PR rtl-optimization/70478
27141 * lra-constraints.c (process_alt_operands): Check memory for
27142 disfavoring memory insn operand.
27144 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27146 PR middle-end/80100
27147 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27148 left shift in unsigned HOST_WIDE_INT type.
27150 PR rtl-optimization/80385
27151 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27152 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27155 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27156 if they have any depend clauses.
27158 2017-04-11 Martin Liska <mliska@suse.cz>
27161 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27162 * ipa-split.c (split_function): Create a local comdat symbol
27163 if caller is in a comdat group.
27165 2017-04-11 Martin Liska <mliska@suse.cz>
27168 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27171 2017-04-11 Martin Sebor <msebor@redhat.com>
27173 PR middle-end/80364
27174 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27175 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27177 (directive::set_width, directive::set_precision, format_character):
27179 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27182 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27185 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27186 conflict, set target->arch_name instead of target->cpu_name.
27188 2017-04-11 Richard Biener <rguenther@suse.de>
27190 PR tree-optimization/80374
27191 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27192 build_zero_cst, remove fold_convertible_p check again.
27194 2017-04-11 Martin Liska <mliska@suse.cz>
27197 * ubsan.c (instrument_object_size): Do not instrument register
27200 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27203 * config/i386/i386-builtin-types.def
27204 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27205 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27206 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27207 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27208 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27209 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27210 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27211 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27212 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27213 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27214 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27215 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27216 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27217 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27218 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27219 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27220 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27221 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27222 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27223 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27224 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27225 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27226 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27227 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27228 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27229 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27230 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27231 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27232 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27233 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27234 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27235 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27236 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27237 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27238 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27239 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27240 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27241 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27242 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27243 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27244 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27245 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27246 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27247 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27248 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27249 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27251 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27252 flag to second_arg_count, handle 4 argument function type _COUNT
27253 aliases, handle second_arg_count on second argument rather than last.
27255 2017-04-10 Jeff Law <law@redhat.com>
27257 PR tree-optimization/80374
27258 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27259 record anything if we can not convert integer_zero_node to the
27262 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27265 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27266 Enhance special handling given to the TARGET_P9_MINMAX option in
27267 relation to certain other options.
27269 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27271 PR tree-optimization/80153
27272 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27273 remove POINTER_PLUS_EXPR's base part directly, rather than through
27276 2017-04-10 Richard Biener <rguenther@suse.de>
27277 Bin Cheng <bin.cheng@arm.com>
27279 PR tree-optimization/80153
27280 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27281 the first element of pointer type aff_tree. Build result expr in
27283 (add_elt_to_tree): Convert to type unconditionally. Remove other
27284 fold_convert calls.
27285 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27286 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27288 2017-04-10 Richard Biener <rguenther@suse.de>
27290 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27293 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27295 PR rtl-optimization/70478
27296 * lra-constraints.c (curr_small_class_check): New.
27297 (update_and_check_small_class_inputs): New.
27298 (process_alt_operands): Update curr_small_class_check. Disfavor
27299 alternative insn memory operands. Check available regs for small
27302 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27305 * config/mips/mips.opt (-mvirt): Update description.
27306 * doc/invoke.texi (-mvirt): Likewise.
27308 2017-04-10 Richard Biener <rguenther@suse.de>
27310 PR middle-end/80362
27311 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27312 looking for NEGATE_EXPR in -A / -B to A / B folding.
27314 2017-04-10 Martin Liska <mliska@suse.cz>
27316 PR gcov-profile/80224
27317 * gcov.c (print_usage): Fix usage string.
27318 (get_gcov_intermediate_filename): Remove.
27319 (output_gcov_file): Use both for normal and intermediate format.
27320 (generate_results): Do not initialize special file for
27321 intermediate format.
27323 2017-04-10 Richard Biener <rguenther@suse.de>
27325 PR tree-optimization/80304
27326 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27329 2017-04-10 Nathan Sidwell <nathan@acm.org>
27332 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27333 (rs6000_init_builtins): Use it.
27335 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27337 * config/arm/arm.md (<mrc>): Add mode to SET source.
27338 (<mrrc>): Likewise.
27340 2017-04-10 Richard Biener <rguenther@suse.de>
27342 PR middle-end/80344
27343 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27345 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27348 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27349 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27350 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27351 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27352 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27353 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27354 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27355 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27356 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27357 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27358 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27359 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27360 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27361 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27362 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27363 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27364 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27365 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27366 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27367 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27368 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27369 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27370 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27372 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27374 PR rtl-optimization/70478
27375 * lra-constraints.c: Reverse the last patch.
27377 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27379 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27380 Add comment for WCHAR_T.
27382 2017-04-08 Martin Liska <mliska@suse.cz>
27385 2017-04-07 Martin Liska <mliska@suse.cz>
27388 * ipa-split.c (split_function): Add function part to a same comdat
27391 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27394 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27396 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27398 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27399 (vec_pairing): ...to this.
27400 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27401 (rs6000_sched_init): Adjust for name change.
27402 (struct rs6000_sched_context): Likewise.
27403 (rs6000_init_sched_context): Likewise.
27404 (rs6000_set_sched_context): Likewise.
27406 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27412 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27414 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27415 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27416 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27418 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27420 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27422 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27424 PR rtl-optimization/70703
27425 * ira-color.c (update_conflict_hard_regno_costs): Use
27426 int64_t instead of HOST_WIDE_INT.
27428 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27430 PR rtl-optimization/70478
27431 * lra-constraints.c (process_alt_operands): Disfavor alternative
27432 insn memory operands.
27434 2017-04-07 Jeff Law <law@redhat.com>
27436 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27437 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27439 2017-04-07 Martin Liska <mliska@suse.cz>
27442 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27443 Show error message instead of an ICE.
27445 2017-04-07 Martin Liska <mliska@suse.cz>
27448 * ipa-split.c (split_function): Add function part to a same comdat
27451 2017-04-07 Richard Biener <rguenther@suse.de>
27453 PR middle-end/80341
27454 * tree.c (get_unwidened): Also handle ! for_type case for
27456 * convert.c (do_narrow): Split out from ...
27457 (convert_to_integer_1): ... here. Do not pass final truncation
27458 type to get_unwidened for TRUNC_DIV_EXPR.
27460 2017-04-07 Richard Biener <rguenther@suse.de>
27462 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27464 (aff_combination_const): Adjust.
27465 (aff_combination_scale): Likewise.
27466 (aff_combination_add_elt): Likewise.
27467 (aff_combination_add_cst): Likewise.
27468 (aff_combination_convert): Likewise.
27469 (add_elt_to_tree): Likewise. Remove unused argument.
27470 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27472 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27474 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27476 * config/arm/arm.c (arm_default_short_enums): Use
27477 ARM_DEFAULT_SHORT_ENUMS.
27478 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27480 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27483 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27484 members with redundant out-of-class redeclaration.
27486 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27489 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27490 * config/i386/i386.md (*zero_extendsidi2):
27491 Add (?*x,*x) and (?*v,*v) alternatives.
27493 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27496 * config/i386/i386.c (ix86_expand_builtin)
27497 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27498 mode from insn data. Convert operands to insn operand mode.
27499 Copy operands that don't satisfy insn predicate to a register.
27501 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27503 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27506 2017-04-06 Richard Biener <rguenther@suse.de>
27508 PR tree-optimization/80334
27509 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27510 preserve alignment of accesses.
27512 2017-04-06 Richard Biener <rguenther@suse.de>
27514 PR tree-optimization/80262
27515 * tree-sra.c (build_ref_for_offset): Preserve address-space
27517 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27518 Drop useless address-space information on MEM_REF offsets.
27520 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27522 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27524 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27526 PR rtl-optimization/70703
27527 * ira-color.c (update_conflict_hard_regno_costs): Use
27528 HOST_WIDE_INT instead of long.
27530 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27533 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27534 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27536 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27537 for x86_64 target. Handle -m3dnowa option.
27539 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27541 PR rtl-optimization/70703
27542 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27543 (update_conflict_hard_regno_costs): Use long instead of unsigned
27544 arithmetic for cost calculation.
27546 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27547 Bernd Edlinger <bernd.edlinger@hotmail.de>
27550 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27553 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27556 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27557 ptr_mode with Pmode throughout.
27558 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27559 into probe_stack_range and use DImode.
27561 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27564 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27565 call_eh_return is true.
27567 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27569 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27570 Initialize last_match_fntype_index.
27572 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27575 * tree-nvr.c: Include internal-fn.h.
27576 (pass_return_slot::execute): Ignore internal calls without
27579 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27580 Richard Biener <rguenther@suse.de>
27583 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27584 captures used multiple times, except for the last use.
27585 * generic-match-head.c: Include gimplify.h.
27587 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27589 PR tree-optimization/79390
27590 * target.h (struct noce_if_info): Declare.
27591 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27592 * target.def (noce_conversion_profitable_p): New target hook.
27593 * ifcvt.h (struct noce_if_info): New type, moved from ...
27594 * ifcvt.c (struct noce_if_info): ... here.
27595 (noce_conversion_profitable_p): Renamed to ...
27596 (default_noce_conversion_profitable_p): ... this. No longer
27598 (noce_try_store_flag_constants, noce_try_addcc,
27599 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27600 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27601 instead of noce_conversion_profitable_p.
27602 * config/i386/i386.c: Include ifcvt.h.
27603 (ix86_option_override_internal): Don't override
27604 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27605 (ix86_noce_conversion_profitable_p): New function.
27606 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27607 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27608 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27609 * doc/tm.texi: Regenerated.
27611 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27616 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27619 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27620 instructions for small multiply cores.
27622 2017-04-04 Jeff Law <law@redhat.com>
27624 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27626 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27627 that are not set by the loop over the elements.
27629 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27632 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27633 int mode, convert_modes it to mode as unsigned, otherwise use
27634 lowpart_subreg to mode rather than SImode.
27635 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27636 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27637 Use DImode instead of SImode for the shift count operand.
27638 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27641 2017-04-04 Richard Biener <rguenther@suse.de>
27643 PR middle-end/80281
27644 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27645 arithmetic done for the negate or the plus. Simplify.
27646 (A - (-B) -> A + B): Likewise.
27647 * fold-const.c (split_tree): Make sure to not negate pointers.
27649 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27651 PR rtl-optimization/60818
27652 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27653 a compare of comparisons with the thing compared if this results
27654 in a different machine mode.
27656 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27658 * alias.c (base_alias_check): Fix typo in comment.
27659 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27660 * cgraphunit.c (symbol_table::compile): Likewise.
27661 * collect2.c (maybe_run_lto_and_relink): Likewise.
27662 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27663 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27664 * config/avr/avr.c (avr_map_op_t): Likewise.
27665 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27666 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27667 * config/epiphany/epiphany.md (movcc): Likewise.
27668 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27669 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27671 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27672 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27673 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27674 * config/sh/sh.c (sh_rtx_costs): Likewise.
27675 * fold-const.c (fold_truth_andor): Likewise.
27676 * genautomata.c (collapse_flag): Likewise.
27677 * gengtype.h (struct type::u::s): Likewise.
27678 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27679 * input.c (FORMAT_AMOUNT): Likewise.
27680 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27681 (known_aggs_to_agg_replacement_list): Likewise.
27682 * ipa-inline-analysis.c: Likewise.
27683 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27684 * ipa-polymorphic-call.c
27685 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27686 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27687 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27689 * modulo-sched.c (apply_reg_moves): Likewise.
27690 * omp-expand.c (build_omp_regions_1): Likewise.
27691 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27692 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27693 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27694 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27695 * value-prof.c: Likewise.
27696 * var-tracking.c (val_reset): Likewise.
27698 2017-04-03 Richard Biener <rguenther@suse.de>
27700 PR tree-optimization/80275
27701 * fold-const.c (split_address_to_core_and_offset): Handle
27704 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27706 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27707 descriptors is at least equal to that of functions.
27709 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27711 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27713 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27716 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27717 (mov<IMOD4:mode>): New expander.
27718 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27720 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27722 PR rtl-optimization/79405
27723 * fwprop.c (propagations_left): New variable.
27724 (forward_propagate_into): Decrement it.
27725 (fwprop_init): Initialize it.
27726 (fw_prop): If the variable has reached zero, stop propagating.
27727 (fwprop_addr): Ditto.
27729 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27732 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27733 a FUNCTION_DECL, pass it as decl instead of origin to
27736 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27738 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27741 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27744 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27745 TARGET_VSX_SMALL_INTEGER.
27747 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27749 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27750 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27752 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27754 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27755 extraction from odd-numbered MSA register.
27757 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27759 PR middle-end/80173
27760 * expmed.c (store_bit_field_1): Don't attempt to create
27761 a word subreg out of hard registers wider than word if they
27762 have HARD_REGNO_NREGS of 1 for their mode.
27764 PR middle-end/80163
27765 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27766 conversions to integer types wider than word and pointer.
27769 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27770 (rtx_equal_for_cselib_p): Pass 0 to it.
27771 * cselib.c (cselib_hasher::equal): Likewise.
27772 (rtx_equal_for_cselib_1): Add depth argument. If depth
27773 is 128, don't look up VALUE locs and punt. Increment
27774 depth in recursive calls when walking VALUE locs.
27776 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27778 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27779 (make_gcov_file_name): Use the canonical path name for generating
27781 (read_line): Fix handling of files with ascii null bytes.
27783 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27785 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27786 to initialise a vector register instead
27787 of using a const_int.
27789 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27791 PR translation/80189
27792 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27793 diagnostic messages.
27795 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27798 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27799 (dfp_diex_<mode>): Update mode of operand 1.
27800 * doc/extend.texi (dxex, dxexq): Document change to return type.
27801 (diex, diexq): Document change to argument type.
27803 2017-03-30 Martin Jambor <mjambor@suse.cz>
27806 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27807 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27808 it reflects the signature changes performed at the callee side.
27809 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27810 to cgraph_build_function_type_skip_args.
27811 (build_function_decl_skip_args): Adjust call to the above function.
27813 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27816 * config/i386/sse.md
27817 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27818 register as dest whenever it is a MEM not rtx_equal_p to the
27819 corresponding dup operand, and when forcing into reg move the
27820 reg into the memory afterwards.
27821 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27822 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27823 for the force_reg mode.
27824 (avx512vl_vextractf128<mode>): Use register as dest either
27825 always when a MEM, or when it is a MEM not rtx_equal_p to the
27826 corresponding dup operand, or even not when it is a CONST_VECTOR
27827 depending on the mode and lo vs. hi.
27828 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27830 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27831 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27832 Likewise. Require that operands[2] is even.
27833 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27834 Remove extraneous parens. Require that operands[2] is a multiple
27836 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27837 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27838 disallow memory then.
27840 2017-03-30 Richard Biener <rguenther@suse.de>
27842 PR tree-optimization/77498
27843 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27844 to non-constants over backedges.
27846 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27848 PR rtl-optimization/80233
27849 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27850 as last_combined_insn. Do not test for BARRIER_P separately.
27852 2017-03-29 Andreas Schwab <schwab@suse.de>
27855 * calls.c (prepare_call_address): Convert funexp to Pmode before
27856 copying to temp reg.
27858 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27860 PR tree-optimization/80158
27861 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27862 Handle possible future case of more than one alternate
27864 (replace_rhs_if_not_dup): Likewise.
27865 (replace_one_candidate): Likewise.
27867 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27869 PR rtl-optimization/80193
27870 * ira.c (ira): Do not check allocation for LRA.
27872 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27874 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27875 (nvptx_output_simt_exit): Declare.
27876 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27877 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27878 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27879 (nvptx_declare_function_name): ...here. Emit declaration of local
27880 memory space buffer for omp_simt_enter insn.
27881 (nvptx_output_unisimt_switch): New.
27882 (nvptx_output_softstack_switch): New.
27883 (nvptx_output_simt_enter): New.
27884 (nvptx_output_simt_exit): New.
27885 * config/nvptx/nvptx.h (struct machine_function): New fields
27886 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27887 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27888 (UNSPECV_SIMT_EXIT): Ditto.
27889 (omp_simt_enter_insn): New insn.
27890 (omp_simt_enter): New expansion.
27891 (omp_simt_exit): New insn.
27892 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27894 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27895 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27896 (expand_GOMP_SIMT_EXIT): New.
27897 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27898 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27899 (GOMP_SIMT_EXIT): Ditto.
27900 * target-insns.def (omp_simt_enter): New insn.
27901 (omp_simt_exit): Ditto.
27902 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27904 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27905 (lower_rec_input_clauses): Likewise.
27906 (lower_lastprivate_clauses): Handle SIMT privatization.
27908 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27909 (ompdevlow_adjust_simt_enter): New.
27910 (find_simtpriv_var_op): New.
27911 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27912 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27914 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27915 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27916 (copy_decl_for_dup_finish): Ditto.
27918 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27920 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27923 * config/i386/i386.c (ix86_option_override_internal): Always
27924 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27926 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27928 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27930 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27932 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27933 mark new edge's irreducible flag accordign to it.
27934 (vect_do_peeling): Check loop preheader edge's irreducible flag
27935 and pass it to function slpeel_add_loop_guard.
27937 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27939 PR tree-optimization/80218
27940 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27941 Update block frequencies and counts.
27943 2017-03-28 Richard Biener <rguenther@suse.de>
27945 PR tree-optimization/78644
27946 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27947 of a simplification result we may not use it at all.
27949 2017-03-28 Richard Biener <rguenther@suse.de>
27952 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27953 without arguments, generate default definition of a SSA name.
27955 2017-03-28 Richard Biener <rguenther@suse.de>
27957 PR middle-end/80222
27958 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27959 TYPE_REF_CAN_ALIAS_ALL references.
27960 * fold-const.c (fold_indirect_ref_1): Likewise.
27962 2017-03-28 Martin Liska <mliska@suse.cz>
27965 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27966 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27968 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27969 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27971 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27972 (EXTRA_SPECS): Define.
27973 (SUBTARGET_EXTRA_SPECS): Likewise.
27974 (SUBTARGET_CPP_SPEC): Likewise.
27975 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27976 SUBTARGET_EXTRA_SPECS.
27977 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27979 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27981 * config/arc/simdext.md (vst64_insn): Update pattern.
27982 (vld32wh_insn): Likewise.
27983 (vld32wl_insn): Likewise.
27984 (vld64_insn): Likewise.
27985 (vld32_insn): Likewise.
27987 2017-03-28 Marek Polacek <polacek@redhat.com>
27990 * fold-const.c (fold_comparison): Use protected_set_expr_location
27991 instead of SET_EXPR_LOCATION.
27993 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27995 * tree.c (add_expr): Avoid name lookup warning.
27997 2017-03-27 Jeff Law <law@redhat.com>
27999 PR tree-optimization/80216
28000 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28001 function name. Limit recursion depth.
28002 (record_temporary_equivalences): Corresponding changes.
28004 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28006 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28009 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28012 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28014 * cfgcleanup.c (reg_note_cfa_p): New array.
28015 (insns_have_identical_cfa_notes): New function.
28016 (old_insns_match_p): Don't cross-jump in between /f
28017 and non-/f instructions. If both i1 and i2 are frame related,
28018 verify all CFA notes, their order and content.
28020 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28023 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28024 HImode and SImode with zero extend to DImode to one insn.
28025 (bswap<mode>2_extenddi): Likewise.
28026 (bswapsi2_extenddi): Likewise.
28027 (bswaphi2_extendsi): Likewise.
28028 (bswaphi2): Combine bswap HImode and SImode into one insn.
28029 Separate memory insns from swapping register.
28030 (bswapsi2): Likewise.
28031 (bswap<mode>2): Likewise.
28032 (bswaphi2_internal): Delete, no longer used.
28033 (bswapsi2_internal): Likewise.
28034 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28035 store, and gpr<-gpr swap insns.
28036 (bswap<mode>2_store): Likewise.
28037 (bswaphi2_reg): Register only splitter, combine with the splitter.
28038 (bswaphi2 splitter): Likewise.
28039 (bswapsi2_reg): Likewise.
28040 (bswapsi2 splitter): Likewise.
28041 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28042 the insns into load, store, and register/register insns.
28043 (bswapdi2_ldbrx): Likewise.
28044 (bswapdi2_load): Likewise.
28045 (bswapdi2_store): Likewise.
28046 (bswapdi2_reg): Likewise.
28048 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28050 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28051 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28053 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28056 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28058 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28059 special handling for target option conflicts between dform
28060 options (-mpower9-dform, -mpower9-dform-vector,
28061 -mpower9-dform-scalar) and -mno-direct-move.
28063 2017-03-27 Richard Biener <rguenther@suse.de>
28065 PR tree-optimization/80181
28066 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28068 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28070 * config/arc/predicates.md (move_double_src_operand): Replace the
28071 call to move_double_src_operand with a call to address_operand.
28073 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28075 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28076 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28077 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28079 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28081 * config/arc/predicates.md (long_immediate_loadstore_operand):
28082 Consider scaled addresses cases.
28084 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28086 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28087 restored when in interrupt.
28088 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28089 doesn't have delay slot.
28091 2017-03-27 Richard Biener <rguenther@suse.de>
28094 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28095 inlined thunk clones.
28097 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28100 * asan.c (instrument_derefs): Copy over last operand from
28101 original COMPONENT_REF to the new COMPONENT_REF with
28102 DECL_BIT_FIELD_REPRESENTATIVE.
28103 * ubsan.c (instrument_object_size): Likewise.
28105 2017-03-27 Richard Biener <rguenther@suse.de>
28107 PR tree-optimization/80170
28108 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28109 sure DR/SCEV didnt fold in constants we do not see when looking
28110 at the reference base alignment.
28112 2017-03-27 Richard Biener <rguenther@suse.de>
28114 PR middle-end/80171
28115 * gimple-fold.c (fold_ctor_reference): Properly guard against
28116 NULL return value from canonicalize_constructor_val.
28118 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28121 * config/i386/i386.c (ix86_expand_builtin)
28122 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28123 flags reg setting and flags reg using instructions.
28124 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28125 clobbering instructions to zero extend op2.
28127 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28129 * doc/install.texi (Configuration) <--with-aix-soname>:
28130 Update link to AIX ld.
28132 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28134 PR rtl-optimization/80160
28135 PR rtl-optimization/80159
28136 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28137 reg_alternate_class into account.
28139 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28142 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28143 to consider in curr_insn_transform.
28145 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28147 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28148 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28149 and emit_mode_inner.
28151 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28153 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28154 argument to the overloaded builtin variants. Use the new flag to
28155 deprecate certain builtin variants.
28156 * config/s390/s390-builtin-types.def: Add new builtin types.
28157 * config/s390/s390-builtins.h: Support new flags field for
28158 overloaded builtins.
28159 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28160 (s390_macro_to_expand): Enable vector float data type.
28161 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28162 builtins by incrementing the __VEC__ version number.
28163 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28165 (s390_resolve_overloaded_builtin): Emit error messages depending
28166 on the builtin flags.
28167 * config/s390/s390.c (s390_expand_builtin): Support additional
28168 flags argument. Change error message to match the messages
28169 emitted in s390-c.c.
28170 * config/s390/s390.md: New UNSPEC_* constants.
28171 (op_type): Add new instruction types.
28172 * config/s390/vecintrin.h: Add new builtins and test data class
28174 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28175 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28176 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28177 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28178 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28179 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28181 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28182 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28183 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28184 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28186 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28187 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28188 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28189 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28191 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28192 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28193 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28194 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28195 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28196 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28197 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28199 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28200 ("vec_scatter_element<V_HW_4:mode>_DI")
28201 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28202 ("vec_fpint<mode>", "vflls")
28203 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28204 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28205 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28206 ("*vec_cmphe<mode>_cc"): ... these.
28208 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28209 mode constant instead of magic value.
28211 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28213 * config/s390/s390.c (s390_expand_vec_compare): Support other
28214 vector floating point modes than just V2DF.
28215 (s390_expand_vcond): Likewise.
28216 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28217 (s390_cannot_change_mode_class): Prevent mode changes between TF
28218 and V1TF in vector registers.
28219 * config/s390/s390.md (DF, SF): New mode attributes.
28220 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28221 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28222 SFmode support for VRs.
28223 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28225 (VFT, VF_HW): New mode iterators.
28226 (vw, sdx): New mode attributes.
28227 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28228 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28229 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28230 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28231 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28232 also the new vector floating point modes. Renaming to ...
28234 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28235 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28236 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28237 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28238 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28239 ("vec_unordered<mode>"): ... these.
28241 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28242 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28243 ("*vec_extendv2df"): New insn definitions.
28245 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28247 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28248 ("mulditi3_2", "*muldi3_sign"): New patterns.
28249 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28250 rename the pattern definition.
28252 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28254 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28256 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28258 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28260 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28261 instruction if possible.
28262 * config/s390/vector.md (vec_halfnumelts): New mode
28264 ("*vec_vllezlf<mode>"): New pattern.
28266 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28268 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28269 ("popcountv4si2", "popcountv2di2"): Rename to ...
28270 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28271 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28273 ("popcount<mode>2_vxe"): New pattern.
28275 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28277 * common/config/s390/s390-common.c (processor_flags_table): Add
28279 * config.gcc: Add arch12.
28280 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28281 Default to arch12 for unknown CPU model numbers.
28282 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28283 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28284 PROCESSOR_max sanity check.
28285 * config/s390/s390-opts.h (enum processor_type): Add
28287 * config/s390/s390.c (processor_table): Add arch12.
28288 (s390_expand_builtin): Add check for B_VXE flag.
28289 (s390_issue_rate): Add PROCESSOR_ARCH12.
28290 (s390_get_sched_attrmask): Likewise.
28291 (s390_get_unit_mask): Likewise.
28292 (s390_sched_score): Enable z13 scheduling for arch12.
28293 (s390_sched_reorder): Likewise.
28294 (s390_sched_variable_issue): Likewise.
28295 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28297 (s390_tune_attr): Use z13 scheduling also for arch12.
28298 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28299 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28300 (TARGET_VXE_P): New macros.
28301 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28302 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28303 * config/s390/s390.opt: Add arch12 as processor_type.
28305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28307 * config/s390/s390.md
28308 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28309 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28310 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28312 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28313 Rename expanders to ...
28315 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28316 ("fixuns_truncdddi2_emu"): ... these.
28318 ("fixuns_trunc<mode>si2_emu"): New expander.
28320 ("*fixuns_truncdfdi2_z13"): Rename to ...
28321 ("*fixuns_truncdfdi2_vx"): ... this.
28323 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28325 * config/s390/2964.md: Remove the single element vector compare
28326 instructions which are no longer used.
28327 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28329 (s390_canonicalize_comparison): Remove handling of DFmode
28331 (s390_expand_vec_compare_scalar): Remove function.
28332 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28333 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28335 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28337 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28339 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28341 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28342 will anyway by matched by mov<mode>_64dfp.
28344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28346 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28347 vlef/vstef. Add missing operand to vleif.
28349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28351 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28352 pair for all vector types with 64 bit elements.
28353 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28354 * config/s390/vector.md (V_HW_64): ... here.
28355 (V_128_NOSINGLE): New mode iterator.
28356 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28357 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28358 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28359 ("*vec_load_pairv2di"): Change to ...
28360 ("*vec_load_pair<mode>"): ... this one.
28362 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28364 * config/s390/constraints.md: Add comments.
28365 (jKK): Reject element sizes > 8 bytes.
28366 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28368 * config/s390/s390.md: Add the s_operand checks formerly in
28369 s390_split_ok_p to various splitters where they are still
28371 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28372 for 128 bit vectors. Plus two splitters.
28374 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28376 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28379 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28382 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28383 error if the boundary argument is not constant.
28385 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28387 PR rtl-optimization/80112
28388 * loop-doloop.c (doloop_condition_get): Don't check condition
28389 if cmp isn't SET with IF_THEN_ELSE src.
28391 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28393 PR tree-optimization/80158
28394 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28395 replacing a candidate statement, also replace it for the
28396 candidate's alternate interpretation.
28397 (replace_rhs_if_not_dup): Likewise.
28398 (replace_one_candidate): Likewise.
28400 2017-03-24 Richard Biener <rguenther@suse.de>
28402 PR tree-optimization/80167
28403 * graphite-isl-ast-to-gimple.c
28404 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28406 (translate_isl_ast_to_gimple::get_rename): Likewise.
28408 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28410 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28411 handling of certain combinations of target options, including the
28412 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28413 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28415 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28418 * config/arm/arm.md (*load_multiple): Add reload_completed to
28419 matching condition.
28421 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28422 Richard Biener <rguenth@suse.de>
28424 PR tree-optimization/79908
28425 PR tree-optimization/80136
28426 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28427 been cast away, gimplify_and_add suffices.
28429 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28431 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28433 2017-03-23 Richard Biener <rguenther@suse.de>
28435 PR tree-optimization/80032
28436 * gimplify.c (gimple_push_cleanup): Forced unconditional
28437 cleanups still have to go to the conditional_cleanups
28440 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28442 PR tree-optimization/80072
28443 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28445 (next_operand_entry_id): Change type to unsigned int.
28446 (sort_by_operand_rank): Make sure to return the right return value
28447 even if unsigned fields are bigger than INT_MAX.
28448 (struct oecount): Change cnt and id type to unsigned int.
28449 (oecount_hasher::equal): Formatting fix.
28450 (oecount_cmp): Make sure to return the right return value
28451 even if unsigned fields are bigger than INT_MAX.
28452 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28455 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28456 TREE_READONLY on result if writing it more than once.
28459 * doc/invoke.texi (-fsanitize=thread): Document that with
28460 -fnon-call-exceptions atomics are not able to throw
28464 * tsan.c: Include tree-eh.h.
28465 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28466 maybe_clean_or_replace_eh_stmt where needed.
28467 (instrument_memory_accesses): Add cfg_changed argument.
28468 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28469 if it returned true.
28470 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28472 PR rtl-optimization/63191
28473 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28474 wrapper function, moved the whole old content into ...
28475 (ix86_delegitimize_address_1): ... this. New inline function.
28476 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28477 true as last argument instead of ix86_delegitimize_address.
28479 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28481 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28482 cortexa57_branch_cost.
28484 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28486 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28488 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28491 * doc/md.texi (Constraints): Document wA constraint.
28492 * config/rs6000/constraints.md (wA): New.
28493 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28494 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28495 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28496 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28498 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28501 * gimplify.c (is_oacc_declared): New function.
28502 (oacc_default_clause): Use it to set default flags for acc declared
28503 variables inside parallel regions.
28504 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28505 declared variables.
28506 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28507 declare attribute to any decl as necessary.
28509 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28512 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28513 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28514 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28515 (arm_arch_lpae): This.
28516 * config/arm/arm.c (arm_arch7ve): Rename into ...
28517 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28518 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28521 2017-03-22 Martin Liska <mliska@suse.cz>
28524 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28525 error message instead of an ICE.
28527 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28529 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28531 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28533 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28538 2017-03-21 Martin Sebor <msebor@redhat.com>
28540 * doc/extend.texi: Use "cannot" instead of "can't."
28541 * doc/hostconfig.texi: Same.
28542 * doc/install.texi: Same.
28543 * doc/invoke.texi: Same.
28544 * doc/loop.texi: Same.
28545 * doc/md.texi: Same.
28546 * doc/objc.texi: Same.
28547 * doc/rtl.texi: Same.
28548 * doc/tm.texi: Same.
28549 * doc/tm.texi.in: Same.
28550 * doc/trouble.texi: Same.
28552 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28555 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28556 (collect_checksum_attributes): Set it.
28557 (die_checksum_ordered): Use it.
28559 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28561 PR tree-optimization/79908
28562 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28563 change: For a VA_ARG whose LHS has been cast away, use
28564 force_gimple_operand to construct the side effects.
28566 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28568 PR translation/80001
28569 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28570 more amenable to translation.
28571 (oacc_loop_auto_partitions): Likewise.
28573 2017-03-21 Marek Polacek <polacek@redhat.com>
28574 Martin Sebor <msebor@redhat.com>
28576 PR tree-optimization/80109
28577 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28578 on INTEGRAL_TYPE_P.
28580 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28581 Segher Boessenkool <segher@kernel.crashing.org>
28584 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28585 check reg_used_between_p between insn and one of succ or succ2
28586 depending on if succ is artificial insn not inserted into insn
28589 2017-03-21 Martin Liska <mliska@suse.cz>
28591 PR gcov-profile/80081
28592 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28593 * doc/gcc.texi: Include gcov-dump stuff.
28594 * doc/gcov-dump.texi: New file.
28596 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28598 PR rtl-optimization/79150
28599 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28600 conditional jump, if the jump is the last insn of the loop.
28602 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28603 Richard Biener <rguenth@suse.de>
28605 PR tree-optimization/79908
28606 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28607 been cast away, use force_gimple_operand to construct the side
28610 2017-03-21 Martin Liska <mliska@suse.cz>
28612 PR libfortran/79956
28613 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28616 2017-03-21 Brad Spengler <spender@grsecurity.net>
28619 * plugin.c (htab_hash_plugin): New function.
28620 (add_new_plugin): Use it and adjust.
28621 (parse_plugin_arg_opt): Adjust.
28622 (init_one_plugin): Likewise.
28624 2017-03-21 Richard Biener <rguenther@suse.de>
28626 PR tree-optimization/80032
28627 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28628 if set force the cleanup to happen unconditionally.
28629 (gimplify_target_expr): Push inserted clobbers with force_uncond
28630 to avoid them being removed by control-dependent DCE.
28632 2017-03-21 Richard Biener <rguenther@suse.de>
28634 PR tree-optimization/80122
28635 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28636 va_arg_pack_len when the inlined call stmt requires pack
28638 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28640 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28643 * tsan.c (instrument_builtin_call): If the memory model argument
28644 is not a constant, assume it is valid.
28647 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28650 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28652 PR rtl-optimization/79910
28653 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28654 if its dest is used by an insn before I2 (other than the combined
28655 insns themselves, which are properly handled already).
28657 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28660 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28662 * combine.c (record_used_regs): New static function.
28663 (try_combine): Handle situations where there is an additional
28664 instruction between I2 and I3 which needs to have a LOG_LINK
28668 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28670 * combine.c (try_combine): Delete redundant i1 test. Call
28671 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28673 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28676 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28677 alternatives 13/14.
28679 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28681 PR tree-optimization/80054
28682 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28683 the optimization if a PHI or any of its arguments is not dominated
28684 by the candidate's basis. Use gphi* rather than gimple* as
28686 (replace_profitable_candidates): Clean up a gimple* variable that
28687 should be a gphi* variable.
28689 2017-03-20 Martin Sebor <msebor@redhat.com>
28692 * doc/extend.texi (attribute constructor): Document present limitation.
28694 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28697 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28698 __POWER9_VECTOR__ #ifdef control, change template definition to
28699 use Power9-specific built-in function.
28700 (vec_any_eq): Likewise.
28701 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28702 to control outcomes from this test.
28703 (vector_ae_<mode>p): For VEC_F modes, likewise.
28705 2017-03-20 Ian Lance Taylor <iant@google.com>
28707 * config/i386/i386.c (ix86_function_regparm): Save an extra
28708 register for -fsplit-stack with DECL_STATIC_CHAIN.
28710 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28713 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28714 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28716 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28718 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28720 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28723 2017-03-20 Marek Polacek <polacek@redhat.com>
28726 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28728 2017-03-20 Richard Biener <rguenther@suse.de>
28730 PR tree-optimization/80113
28731 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28732 allocate extra SSA name for PHI def.
28733 (add_close_phis_to_outer_loops): Likewise.
28734 (add_close_phis_to_merge_points): Likewise.
28735 (copy_loop_close_phi_args): Likewise.
28736 (copy_cond_phi_nodes): Likewise.
28738 2017-03-20 Martin Liska <mliska@suse.cz>
28740 PR middle-end/79753
28741 * tree-chkp.c (chkp_build_returned_bound): Do not build
28742 returned bounds for a LHS that's not a BOUNDED_P type.
28744 2017-03-20 Martin Liska <mliska@suse.cz>
28748 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28749 COMPLEX_CST and VECTOR_CST.
28751 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28754 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28755 target operand. A new splitter adds the clobber statement in case
28756 the target operand is dead anyway.
28758 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28760 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28761 to age-old versions of binutils and glibc.
28763 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28765 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28767 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28769 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28771 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28773 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28774 requirement for binutils 2.13.
28776 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28778 * combine.c (try_combine): Delete redundant i1 test. Call
28779 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28781 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28783 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28784 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28786 <riscv64-*-elf>: Re-arrange section
28787 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28788 <riscv32-*-linux>: Likewise.
28789 <riscv64-*-elf>: Likewise
28790 <riscv64-*-linux>: Likewise.
28792 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28795 * aarch64.opt(verbose-cost-dump): Fix typo.
28797 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28800 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28801 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28803 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28805 * reload.c (find_reloads): When reloading a nonoffsettable address,
28806 use RELOAD_OTHER for it and its address reloads.
28808 PR rtl-optimization/79910
28809 * combine.c (record_used_regs): New static function.
28810 (try_combine): Handle situations where there is an additional
28811 instruction between I2 and I3 which needs to have a LOG_LINK
28814 2017-03-17 Jeff Law <law@redhat.com>
28816 PR tree-optimization/71437
28817 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28818 conditional in the hash table first.
28819 (vrp_dom_walker::before_dom_children): Extract condition from
28820 ASSERT_EXPR. Record condition, its inverion and any implied
28821 conditions as well.
28823 2017-03-17 Marek Polacek <polacek@redhat.com>
28824 Markus Trippelsdorf <markus@trippelsdorf.de>
28826 PR tree-optimization/80079
28827 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28830 2017-03-17 Richard Biener <rguenther@suse.de>
28832 PR middle-end/80075
28833 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28834 Properly verify the LHS before the RHS possibly claims to be
28836 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28839 2017-03-17 Martin Jambor <mjambor@suse.cz>
28841 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28842 (List of -O2 options): Likewise.
28843 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28846 2017-03-17 Tom de Vries <tom@codesourcery.com>
28848 * gcov-dump.c (print_usage): Print bug_report_url.
28850 2017-03-17 Richard Biener <rguenther@suse.de>
28852 PR middle-end/80050
28853 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28854 (parser::peek): Likewise.
28856 2017-03-17 Richard Biener <rguenther@suse.de>
28858 PR tree-optimization/80048
28859 * sese.c (free_sese_info): Properly release rename_map and
28860 copied_bb_map elements.
28862 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28864 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28865 Add linked-list forward and backlinks. Insert on
28866 construction, remove on destruction.
28867 (class pass_store_merging): Add m_stores_head field.
28868 (pass_store_merging::terminate_and_process_all_chains):
28869 Iterate over m_stores_head list.
28870 (pass_store_merging::terminate_all_aliasing_chains):
28872 (pass_store_merging::execute): Check for debug stmts first.
28873 Push new chains onto the m_stores_head stack.
28875 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28878 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28879 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28880 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28882 2017-03-16 Jeff Law <law@redhat.com>
28884 PR tree-optimization/71437
28885 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28886 member function. Implementation moved into after_dom_children
28887 member function and into the threader's thread_outgoing_edges
28889 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28890 some code into new thread_outgoing_edges.
28891 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28892 definition. Simplify marker handling (do it here). Assume we always
28893 have the available expression and the const/copies tables.
28894 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28896 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28897 * tree-vrp.c (equiv_stack): No longer file scoped.
28898 (vrp_dom_walker): New class.
28899 (vrp_dom_walker::before_dom_children): New member function.
28900 (vrp_dom_walker::after_dom_children): Likewise.
28901 (identify_jump_threads): Setup domwalker. Use it rather than
28902 walking edges in a random order by hand. Simplify setup/finalization.
28903 (finalize_jump_threads): Remove.
28904 (vrp_finalize): Do not call identify_jump_threads here.
28905 (execute_vrp): Do it here instead and call thread_through_all_blocks
28908 PR tree-optimization/71437
28909 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28911 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28913 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28914 (dom_opt_dom_walker::thread_across_edge): Remove
28915 handle_dominating_asserts argument. All callers changed.
28916 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28917 changes. Remove calls to lhs_of_dominating_assert. Other
28918 uses of handle_dominating_asserts turn into unconditional code
28919 (simplify_control_stmt_condition_1): Likewise.
28920 (simplify_control_stmt_condition): Likewise.
28921 (thread_through_normal_block, thread_across_edge): Likewise.
28922 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28923 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28924 object if it is not an SSA_NAME.
28925 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28926 before calling into the VRP specific simplifiers.
28927 (identify_jump_threads): Remove handle_dominating_asserts
28930 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28933 * tree-diagnostic.c (default_tree_printer): No longer static.
28934 * tree-diagnostic.h (default_tree_printer): New prototype.
28936 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28938 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28939 Change ins into fmov.
28941 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28943 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28944 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28945 Use h_con constraint for operand 1.
28946 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28947 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28949 2017-03-15 Jeff Law <law@redhat.com>
28951 PR tree-optimization/71437
28952 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28953 (record_temporary_equivalences): Use it.
28955 PR tree-optimization/71437
28956 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28957 tree-ssa-scopedtables.
28958 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28959 (record_conditions, record_cond, vuse_eq): Likewise.
28960 (record_edge_info): Adjust to API tweak of record_conditions.
28961 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28962 (record_temporary_equivalences, optimize_stmt): Likewise.
28963 (eliminate_redundant_computations): Likewise.
28964 (record_equivalences_from_stmt): Likewise.
28965 * tree-ssa-scopedtables.c: Include options.h and params.h.
28966 (vuse_eq): New function, moved from tree-ssa-dom.c
28967 (build_and_record_new_cond): Likewise.
28968 (record_conditions): Likewise. Accept vector of conditions rather
28969 than edge_equivalence structure for first argument.
28970 for the first argument.
28971 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28972 from tree-ssa-dom.c.
28973 (avail_exprs_stack::record_cond): Likewise.
28974 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28975 from tree-ssa-dom.c.
28976 (avail_exprs_stack): Add new member functions lookup_avail_expr
28978 (record_conditions): Declare.
28980 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28983 * lra-constraints.c (process_alt_operands): Increase reject for
28984 reloading an input/output operand.
28986 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28989 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28990 insns to convert from signed/unsigned char/short to IEEE 128-bit
28992 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28994 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28997 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28998 subreg of inner mode for values already in registers.
29000 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29002 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29003 iteration reg is used after the loop.
29005 2017-03-14 Martin Sebor <msebor@redhat.com>
29007 PR tree-optimization/79800
29008 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29009 precision in negative-positive range.
29010 (format_floating): Call non-const overload with adjusted precision.
29012 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29015 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29018 2017-03-14 Martin Sebor <msebor@redhat.com>
29020 PR middle-end/80020
29021 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29022 * builtins.def (aligned_alloc): Use it.
29025 * Makefile.in (GTFILES): Add calls.c.
29026 * calls.c: Include "gt-calls.h".
29028 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29030 PR rtl-optimization/79728
29031 * regs.h (struct target_regs): New field
29032 x_contains_allocatable_regs_of_mode.
29033 (contains_allocatable_regs_of_mode): New macro.
29034 * reginfo.c (init_reg_sets_1): Initialize it, and change
29035 contains_reg_of_mode so it includes global regs as well.
29036 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29037 rather than contains_regs_of_mode.
29039 2017-03-14 Martin Liska <mliska@suse.cz>
29041 * doc/invoke.texi: Document options that can't be combined with
29042 -fcheck-pointer-bounds.
29044 2017-03-14 Martin Liska <mliska@suse.cz>
29046 PR middle-end/79831
29047 * doc/invoke.texi (-Wchkp): Document the option.
29049 2017-03-14 Martin Liska <mliska@suse.cz>
29051 * Makefile.in: Install gcov-dump.
29053 2017-03-14 Martin Liska <mliska@suse.cz>
29055 * multiple_target.c (expand_target_clones): Bail out for
29056 an invalid attribute.
29058 2017-03-14 Richard Biener <rguenther@suse.de>
29060 * alias.c (struct alias_set_entry): Pack properly.
29061 * cfgloop.h (struct loop): Likewise.
29062 * cse.c (struct set): Likewise.
29063 * ipa-utils.c (struct searchc_env): Likewise.
29064 * loop-invariant.c (struct invariant): Likewise.
29065 * lra-remat.c (struct cand): Likewise.
29066 * recog.c (struct change_t): Likewise.
29067 * rtl.h (struct address_info): Likewise.
29068 * symbol-summary.h (function_summary): Likewise.
29069 * tree-loop-distribution.c (struct partition): Likewise.
29070 * tree-object-size.c (struct object_size_info): Likewise.
29071 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29072 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29073 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29074 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29075 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29076 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29077 (struct _stmt_vec_info): Likewise.
29079 2017-03-14 Martin Liska <mliska@suse.cz>
29082 * multiple_target.c (create_dispatcher_calls): Check that
29083 a target can create a function dispatcher.
29085 2017-03-14 Martin Liska <mliska@suse.cz>
29088 * multiple_target.c (expand_target_clones): Drop local.local
29089 flag for default implementation.
29091 2017-03-14 Richard Biener <rguenther@suse.de>
29093 PR tree-optimization/80030
29094 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29096 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29098 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29099 gcc_fallthrough() instead of __attribute__((fallthrough));
29101 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29103 * doc/gcc.texi: Remove "up" link to (DIR).
29104 * doc/gccint.texi: Ditto.
29106 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29108 * doc/install.texi (Specific) <avr>: Remove reference to
29111 2017-03-13 Jeff Law <law@redhat.com>
29113 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29114 attribute rather than comments.
29116 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29117 match_scratch operand is highest.
29119 2017-03-13 Martin Liska <mliska@suse.cz>
29121 PR middle-end/78339
29122 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29123 is a CHKP clone, use original declaration.
29125 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29127 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29128 (arc_conditional_register_usage): Use a different allocation order
29129 when optimizing for size.
29130 * common/config/arc/arc-common.c (arc_option_optimization_table):
29131 Section anchors default on when optimizing for size.
29133 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29135 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29137 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29139 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29140 * config/arc/arc.md (cpu_facility): Add cd variant.
29141 (*movqi_insn): Add code density variant.
29142 (*movhi_insn): Likewise.
29143 (*movqi_insn): Likewise.
29144 (*addsi3_mixed): Likewise.
29145 (subsi3_insn): Likewise.
29147 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29149 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29151 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29153 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29154 expressions with MINUS and UNARY ops.
29156 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29159 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29161 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29162 between vec_select and vector argument.
29163 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29164 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29165 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29166 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29167 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29168 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29170 2017-03-13 Richard Biener <rguenther@suse.de>
29173 * params.def (vect-max-peeling-for-alignment): Fix typo.
29175 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29177 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29178 issue that only occurred with binutils below 2.18.
29180 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29182 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29183 refer to binutils 2.11/2.12 minimum.
29185 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29187 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29188 ftp.kernel.org and simplify binutils requirement.
29190 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29192 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29194 (Optimize Options): Ditto. Also remove redundancy.
29196 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29198 PR translation/79848
29199 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29201 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29202 to G_ to avoid double translation.
29204 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29206 PR translation/79923
29207 * auto-profile.c (get_combined_location): Convert leading
29208 character of diagnostics to lower case and remove trailing period.
29209 (read_profile): Likewise for various diagnostics.
29210 * config/arm/arm.c (arm_option_override): Remove trailing period
29211 from various diagnostics.
29212 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29213 (msp430_expand_delay_cycles): Likewise.
29215 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29218 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29219 full command-line argument, rather than just "str".
29220 (aarch64_validate_march): Likewise.
29221 (aarch64_validate_mtune): Likewise.
29223 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29225 PR rtl-optimization/78911
29226 * lra-assigns.c (must_not_spill_p): New function.
29227 (spill_for): Use it.
29229 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29231 PR tree-optimization/79981
29232 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29233 ATOMIC_COMPARE_EXCHANGE ifn result.
29234 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29235 IFN_ATOMIC_COMPARE_EXCHANGE.
29237 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29240 * opts.c (parse_sanitizer_options): Add missing question mark to
29241 "did you mean" message.
29243 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29245 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29247 (VMULEUH_UNS): Likewise.
29248 (VMULOUB_UNS): Likewise.
29249 (VMULOUH_UNS): Likewise.
29250 * config/rs6000/rs6000.c (builtin_function_type): Remove
29251 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29253 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29256 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29257 x with result of extra_parsing_for_operand_code_0.
29258 (function_reader::extra_parsing_for_operand_code_0): Convert
29259 return type from void to rtx, returning x. When reading
29260 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29261 larger size containing struct block_symbol.
29263 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29266 -mfloat128-hardware without -m64.
29268 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29271 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29272 entries to the case statement that marks unsigned arguments to
29273 overloaded functions.
29275 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29277 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29278 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29280 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29283 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29284 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29286 2017-03-10 Martin Liska <mliska@suse.cz>
29290 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29291 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29294 2017-03-10 Olivier Hainque <hainque@adacore.com>
29296 * tree-switch-conversion (array_value_type): Start by resetting
29297 candidate type to it's main variant.
29299 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29301 PR rtl-optimization/79909
29302 * combine.c (try_combine): Use simplify_replace_rtx on individual
29303 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29304 of the whole CALL_INSN_FUNCTION_USAGE.
29306 PR tree-optimization/79972
29307 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29308 get_range_info on SSA_NAMEs. Formatting fixes.
29310 2017-03-10 Richard Biener <rguenther@suse.de>
29311 Jakub Jelinek <jakub@redhat.com>
29313 PR tree-optimization/77975
29314 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29315 edge to be constant.
29316 (get_val_for): For constant x return it. Formatting fix.
29317 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29318 would use the same bases as the current one.
29320 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29322 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29323 instead of vec_select for V1TImode.
29324 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29326 (VSX_LE_128): Add V1TI to this mode iterator.
29327 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29328 (*vsx_le_perm_store_<mode>): Likewise.
29329 (pre-reload splitter for VSX stores): Likewise.
29330 (post-reload splitter for VSX stores): Likewise.
29331 (*vsx_xxpermdi2_le_<mode>): Likewise.
29332 (*vsx_lxvd2x2_le_<mode>): Likewise.
29333 (*vsx_stxvd2x2_le_<mode>): Likewise.
29335 2017-03-09 Michael Eager <eager@eagercon.com>
29337 Correct failures with --enable-checking=yes,rtl.
29339 * config/microblaze/microblaze.c (microblaze_expand_shift):
29340 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29341 test for const0_rtx.
29342 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29343 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29345 2017-03-09 Richard Biener <rguenther@suse.de>
29347 PR tree-optimization/79977
29348 * graphite-scop-detection.c (scop_detection::merge_sese):
29349 Handle the case of extra exits to blocks dominating the entry.
29351 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29353 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29356 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29358 PR rtl-optimization/79949
29359 * lra-constraints.c (process_alt_operands): Check memory when
29360 trying to predict a cycle. Print about the overall increase.
29362 2017-03-09 Richard Biener <rguenther@suse.de>
29364 PR middle-end/79971
29365 * gimple-expr.c (useless_type_conversion_p): Preserve
29366 TYPE_SATURATING for fixed-point types.
29368 2017-03-09 Richard Biener <rguenther@suse.de>
29371 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29372 alignment of BLKmode params.
29374 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29377 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29378 (VALL_NO_V2Q): Likewise.
29380 * config/aarch64/aarch64-simd.md
29381 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29383 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29384 VALL_NO_V2Q mode iterator.
29385 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29387 2017-03-09 Martin Liska <mliska@suse.cz>
29389 PR tree-optimization/79631
29390 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29391 tree_int_cst_sign_bit just for INTEGER constants.
29393 2017-03-09 Martin Liska <mliska@suse.cz>
29397 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29400 2017-03-09 Marek Polacek <polacek@redhat.com>
29403 * tree.c (inchash::add_expr): Handle TREE_VEC.
29405 2017-03-09 Martin Liska <mliska@suse.cz>
29408 (chkp_narrow_size_and_offset): New function.
29409 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29410 (void chkp_parse_bit_field_ref): New function.
29411 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29412 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29414 2017-03-09 Martin Liska <mliska@suse.cz>
29417 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29418 (chkp_find_bounds_1): Remove gcc_unreachable.
29420 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29423 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29424 BUILT_IN_SYNC*, determine the access type from the size suffix and
29425 always build a MEM_REF with that type. Handle forgotten
29426 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29429 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29430 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29431 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29432 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29433 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29434 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29435 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29436 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29437 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29438 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29439 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29440 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29441 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29442 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29443 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29444 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29445 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29446 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29447 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29448 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29449 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29450 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29451 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29452 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29453 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29454 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29455 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29456 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29457 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29458 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29459 definitions outside of __OPTIMIZE__ guarded section.
29462 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29463 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29464 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29465 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29468 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29470 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29471 ("vfenez<mode>"): Add missing constraints.
29473 2017-03-08 Martin Sebor <msebor@redhat.com>
29476 * config/nds32/nds32.c (nds32_option_override):
29477 Fix misspelled diagnostic.
29479 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29482 * gimplify.c (gimplify_omp_for): Replace index var in outer
29483 taskloop statement with an artificial variable and add
29484 OMP_CLAUSE_PRIVATE clause for it.
29486 2017-03-08 Richard Biener <rguenther@suse.de>
29488 PR tree-optimization/79955
29489 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29490 for accesses that are completely outside of the variable.
29492 2017-03-08 Andrew Haley <aph@redhat.com>
29494 PR tree-optimization/79943
29495 * tree-ssa-loop-split.c (compute_new_first_bound): When
29496 calculating the new upper bound, (END-BEG) should be added, not
29499 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29501 * config/avr/avr.md (setmemhi): Make sure match_dup
29502 operand number comes before match_scratch.
29504 2017-03-08 Richard Biener <rguenther@suse.de>
29506 PR tree-optimization/79920
29507 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29508 with ncopies == 1 to ...
29509 (vect_transform_slp_perm_load): ... here. Properly compute
29510 all element loads by iterating VF times over the group. Do
29511 not handle ncopies (computed in a broken way) in
29512 vect_create_mask_and_perm.
29514 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29517 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29518 is a uniform vector, use uniform_vector_p return value instead of
29519 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29521 2017-03-07 Marek Polacek <polacek@redhat.com>
29523 PR middle-end/79809
29524 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29525 (alloca_call_type): Likewise.
29527 2017-03-07 Martin Liska <mliska@suse.cz>
29529 * gcov.c (process_args): Put comment to correct location.
29531 2017-03-07 Martin Liska <mliska@suse.cz>
29533 PR middle-end/68270
29534 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29535 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29536 (chkp_narrow_bounds_for_field): Likewise.
29537 (chkp_parse_array_and_component_ref): Pass one more argument to
29540 2017-03-07 Richard Biener <rguenther@suse.de>
29542 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29545 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29547 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29548 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29550 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29553 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29554 to end of description.
29555 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29557 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29559 PR rtl-optimization/79901
29560 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29562 (*avx512f_<code><mode>3<mask_name>): ... this.
29563 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29564 iterator instead of VI8_AVX2_AVX512BW.
29566 PR rtl-optimization/79901
29567 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29568 min/max expander, expand it using expand_vec_cond_expr.
29571 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29574 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29577 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29578 to void * for PCH reasons.
29579 * dwarf2out.c (output_loc_operands, output_die): Cast
29580 v.val_vec.array to unsigned char *.
29582 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29585 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29588 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29590 PR rtl-optimization/79571
29591 * lra-constraints.c (process_alt_operands): Calculate static
29592 reject and subtract it from overall when only addresses will be
29595 2017-03-06 Julia Koval <julia.koval@intel.com>
29598 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29599 incoming stack boundary to 128 for 64-bit targets.
29601 2017-03-06 Richard Biener <rguenther@suse.de>
29603 PR tree-optimization/79894
29604 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29605 to NULL after folding it.
29607 2017-03-06 Richard Biener <rguenther@suse.de>
29609 PR tree-optimization/79824
29610 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29611 check disabling peeling for gaps.
29613 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29615 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29616 attributes): Document gettimeofday.
29618 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29620 * config/s390/s390.c (s390_option_override_internal): Set
29621 PARAM_MIN_VECT_LOOP_BOUND
29623 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29625 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29626 * config/s390/s390.md: Likewise.
29628 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29631 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29632 (<avx2_avx512>_perm<mode>): Rename to ...
29633 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29635 (<avx512>_perm<mode>_mask): Rename to ...
29636 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29638 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29639 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29640 instead of VI8F_256_512.
29641 (avx512f_perm<mode>): New define_expand.
29642 (avx512f_perm<mode>_mask): Likewise.
29643 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29644 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29646 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29648 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29649 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29651 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29653 2017-03-06 Martin Liska <mliska@suse.cz>
29656 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29657 when having a SSA NAME w/o VAR_DECL assigned to it.
29659 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29661 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29662 msa_dpsub_<su>_d): Fix MODE for vec_select.
29664 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29666 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29668 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29670 2017-03-06 Richard Biener <rguenther@suse.de>
29672 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29673 * plugin.c (register_plugin_info): Likewise.
29674 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29676 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29678 * config/i386/sse.md (sse_storehps, sse_storelps,
29679 avx_<castmode><avxsizesuffix>_<castmode>,
29680 avx512f_<castmode><avxsizesuffix>_<castmode>,
29681 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29682 in condition that at least one operand is not a MEM.
29684 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29686 PR middle-end/79805
29687 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29688 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29690 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29691 gimple_call_nothrow_p flag based on whether original builtin can throw.
29692 If it can, emit following stmts on the fallthrough edge.
29693 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29694 don't create new bb if inserting just debug stmts on the edge, try to
29695 insert them on the fallthru bb or just reset debug stmts.
29697 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29700 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29701 restore recog_data (including the operand rtxes inside it) around
29702 the call to get_insn_template.
29704 2017-03-03 Martin Sebor <msebor@redhat.com>
29706 PR tree-optimization/79699
29707 * context.c (context::~context): Free MPFR caches to avoid
29708 a memory leak on program exit.
29710 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29712 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29713 Use wide_int::ulow () instead of .elt (0).
29715 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29717 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29718 (*pushxf): Limit oF constraint to 32bit targets and add oC
29719 constraint for 64bit targets.
29720 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29721 (*pushdf): Change rmF constraint to rmC.
29723 2017-03-03 Martin Liska <mliska@suse.cz>
29725 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29726 Remove unused variable.
29728 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29731 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29732 is a memory operand, increase num_memory.
29733 (ix86_expand_args_builtin): Likewise.
29735 2017-03-03 Jan Hubicka <jh@suse.cz>
29738 * ipa-devirt.c (maybe_record_node): Properly handle
29739 __cxa_pure_virtual visibility.
29741 2017-03-03 Martin Liska <mliska@suse.cz>
29743 PR tree-optimization/79803
29744 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29746 (pass_loop_prefetch::execute): Disabled optimization if an
29747 assumption about L1 cache size is not met.
29749 2017-03-03 Martin Liska <mliska@suse.cz>
29751 PR rtl-optimization/79574
29752 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29753 (hash_scan_set): Likewise.
29754 (dump_hash_table): Likewise.
29755 (hoist_code): Likewise.
29757 2017-03-03 Richard Biener <rguenther@suse.de>
29759 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29760 in place of uhigh ().
29761 (fixed_convert_from_real): Likewise.
29763 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29766 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29768 2017-03-03 Richard Biener <rguenther@suse.de>
29770 PR middle-end/79818
29771 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29772 TYPE_OVERFLOW_UNDEFINED check.
29774 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29776 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29778 (vector_ae_<mode>_p): Likewise.
29779 (vector_nez_<mode>_p): Likewise.
29780 (vector_ne_v2di_p): Likewise.
29781 (vector_ae_v2di_p): Likewise.
29782 (vector_ne_<mode>_p): Likewise.
29783 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29785 (vsx_tsqrt<mode>2_fe): Likewise.
29787 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29790 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29792 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29794 PR rtl-optimization/79780
29795 * cprop.c (one_cprop_pass): When second and further conditional trap
29796 in a single basic block is turned into an unconditional trap, turn it
29797 into a deleted note to avoid RTL verification failures.
29799 2017-03-02 Richard Biener <rguenther@suse.de>
29801 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29803 2017-03-02 Richard Biener <rguenther@suse.de>
29805 PR tree-optimization/79345
29807 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29808 param and abort the walk, returning -1 if it is hit.
29809 (walk_aliased_vdefs): Take a limit param and pass it on.
29810 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29811 defaulting to 0 and return a signed int.
29812 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29813 (check_defs): New helper.
29814 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29815 about uninitialized memory.
29816 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29817 bogus uninitialized warning.
29818 (fixed_convert_from_real): Likewise.
29820 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29822 PR tree-optimization/66768
29823 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29824 iv_use if base object can't be determined.
29826 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29828 PR tree-optimization/79345
29829 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29830 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29831 (get_pattern_stats): Initialize it.
29832 * genemit.c (gen_expand): Verify match_scratch numbers come after
29833 match_operand/match_dup numbers.
29834 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29835 match_scratch numbers.
29836 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29838 * config/s390/s390.md (trunctdsd2): Likewise.
29840 2017-03-02 Richard Biener <rguenther@suse.de>
29842 * wide-int.h (wide_int_storage::operator=): Implement in terms
29845 2017-03-02 Richard Biener <rguenther@suse.de>
29847 PR tree-optimization/79777
29848 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29849 the to insert expression to sth existing.
29851 2017-03-01 Martin Sebor <msebor@redhat.com>
29853 PR middle-end/79692
29854 * gimple-ssa-sprintf.c
29855 (directive::known_width_and_precision): New function.
29856 (format_integer): Use it.
29857 (get_mpfr_format_length): Consider the full range of precision
29858 when computing %g output with the # flag. Set the likely byte
29859 count to 3 rather than 1 when precision is indeterminate.
29860 (format_floating): Correct the lower bound of precision.
29862 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29864 * doc/invoke.texi: Document default code model for 64-bit Linux.
29866 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29869 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29870 udiv rather than div since input pattern is unsigned.
29872 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29874 * config/i386/i386.c (print_reg): Warn for values of
29875 unsupported size in integer register.
29877 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29880 * config/rs6000/predicates.md (current_file_function_operand): Do
29881 not allow self calls to be local if the function is replaceable.
29883 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29886 * config/rs6000/altivec.h (vec_ctz and others): Change the
29887 preprocessor macro that controls conditional compilation from
29888 _ARCH_PWR9 to __POWER9_VECTOR__.
29889 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29890 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29891 control (instead of _ARCH_PWR9 control) so that template
29892 definition uses power9-specific function.
29893 (vec_any_eq): Likewise.
29894 (vec_all_ne): Change macro definition to use a power9-specific
29895 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29896 _ARCH_PWR9 control).
29897 (vec_any_eq) Likewise.
29898 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29899 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29900 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29901 support for xvcmpnedp instruction.
29902 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29903 macro expansion so that Power9 implementation of vec_all_ne does
29904 not use the AltiVec predicate framework.
29905 (VCMPNEH_P): Likewise.
29906 (VCMPNEW_P): Likewise.
29907 (VCMPNED_P): Likewise.
29908 (VCMPNEFP_P): Likewise.
29909 (VCMPNEDP_P): Likewise.
29910 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29911 implementation of vec_any_eq to not use AltiVec predicate
29913 (VCMPAEH_P): Likewise.
29914 (VCMPAEW_P): Likewise.
29915 (VCMPAED_P): Likewise.
29916 (VCMPAEFP_P): Likewise.
29917 (VCMPAEDP_P): Likewise.
29918 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29919 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29920 not use the AltiVec predicate framework.
29921 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29922 of vec_any_eq to not use AltiVec predicate framework.
29923 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29924 support for predefined __POWER9_VECTOR__ macro to indicate that
29925 Power9 instruction selection is enabled.
29926 (altivec_overloaded_builtins): Remove extraneous
29927 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29928 function argument types RS6000_BTI_bool_V16QI and
29929 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29930 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29931 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29932 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29933 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29934 Power9 for implementations of vec_cmpne. Change the signature for
29935 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29936 (representing vec_all_ne) to remove the previously described first
29937 argument of type RS6000_BTI_INTSI, as this was an artifact of
29938 reliance on the AltiVec predicate framework, which is no longer
29939 used in the implementation of these functions. Add
29940 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29941 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29942 since, unlike the AltiVec predicate framework implementation, we
29943 do not share function descriptors between vec_alle and vec_anyeq.
29944 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29945 set of modes that receive special treatment even when
29946 TARGET_P9_VECTOR is true. The special treatment emits code that
29947 does not depend on Power9 instructions.
29948 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29949 define_expand to not rely on AltiVec predicate framework.
29950 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29952 (vector_ne_v2di_p): Change this define_expand to not rely on
29953 AltiVec predicate framework.
29954 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29956 (vector_ne_<mode>_p): Change this define_expand to not rely on
29957 AltiVec predicate framework.
29958 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29960 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29961 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29962 define_insn pattern.
29963 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29964 define_insn pattern because the xvcmpne<VSs>. instruction is not
29966 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29967 instruction is not supported.
29969 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29971 * config/nvptx/nvptx.c: Include intl.h.
29973 2017-03-01 Martin Jambor <mjambor@suse.cz>
29976 * ipa-prop.h (ipa_bits): Removed field known.
29977 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29978 to pointers. Adjusted their comments to warn about their sharing.
29979 (ipcp_transformation_summary): Change bits to a vector of pointers.
29980 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29981 (ipa_get_ipa_bits_for_value): Declare.
29982 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29983 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29984 (ipa_bits_hash_table): Likewise.
29985 (ipa_vr_ggc_hash_traits): Likewise.
29986 (ipa_vr_hash_table): Likewise.
29987 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29988 being pointers and vr_known being removed.
29989 (ipa_set_jf_unknown): Likewise.
29990 (ipa_get_ipa_bits_for_value): New function.
29991 (ipa_set_jfunc_bits): Likewise.
29992 (ipa_get_value_range): New overloaded functions.
29993 (ipa_set_jfunc_vr): Likewise.
29994 (ipa_compute_jump_functions_for_edge): Use the above functions to
29995 construct bits and vr parts of jump functions.
29996 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29997 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29999 (ipcp_grow_transformations_if_necessary): Also allocate
30000 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30002 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30003 them. Fix too long lines.
30004 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30005 vr_known being removed.
30006 (ipa_read_jump_function): Use new setter functions to construct bits
30007 and vr parts of jump functions or set them to NULL.
30008 (write_ipcp_transformation_info): Adjust for bits being pointers.
30009 (read_ipcp_transformation_info): Likewise.
30010 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30012 Include gt-ipa-prop.h.
30013 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30015 (ipcp_store_bits_results): Likewise.
30016 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30017 Do not write to existing jump functions but use a temporary instead.
30019 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30022 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30023 attempt to use its first operand as BIT_FIELD_REF base.
30025 2017-03-01 Richard Biener <rguenther@suse.de>
30027 PR middle-end/79721
30028 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30029 interpolating formula in wrapping arithmetic.
30030 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30032 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30034 PR tree-optimization/79734
30035 * tree-vect-generic.c (expand_vector_condition): Optimize
30036 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30037 Handle VEC_COND_EXPR where comparison has different inner width from
30038 type's inner width.
30040 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30042 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30043 markup, and similar issues. Remove @opindex entries for things
30044 that aren't options. Add missing -mmpy-option entries.
30046 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30048 PR tree-optimization/79737
30049 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30050 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30051 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30052 instead of byte_size. Formatting fix.
30053 (shift_bytes_in_array_right): Formatting fix.
30055 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30058 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30059 condition on optimize for the leaf function test.
30061 2017-02-28 Martin Liska <mliska@suse.cz>
30064 * read-rtl-function.c (function_reader::handle_unknown_directive):
30065 Bail out when one uses -flto.
30067 2017-02-28 Martin Liska <mliska@suse.cz>
30069 * common.opt: Replace space with tabular for options of <number>
30071 * config/i386/i386.opt: Show <number> value for
30072 -mlarge-data-threshold.
30073 * opts.c (print_filtered_help): Do not display number in hexadecimal
30076 2017-02-28 Martin Liska <mliska@suse.cz>
30078 * common.opt: Fix --help=option -Q for options which are of
30081 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30083 * config/i386/i386.c (print_reg): Error out for values
30084 of 8-bit size in invalid integer register.
30086 2017-02-28 Martin Sebor <msebor@redhat.com>
30088 PR tree-optimization/79691
30089 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30091 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30094 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30095 gcc_unreachable with output_operand_lossage.
30097 2017-02-28 Richard Biener <rguenther@suse.de>
30099 PR tree-optimization/79740
30100 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30102 (visit_nary_op): Insert the nary into the hashtable if we
30103 pattern-matched sth.
30104 * tree-ssa-pre.c (eliminate_insert): Robustify.
30106 2017-02-28 Richard Biener <rguenther@suse.de>
30108 PR middle-end/79731
30109 * fold-const.c (decode_field_reference): Reject out-of-bound
30112 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30114 * config/i386/i386.c: Include intl.h.
30115 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30116 instead of just cond ? "..." : "...".
30117 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30118 * coverage.c (read_counts_file): Likewise.
30119 * omp-offload.c: Include intl.h.
30120 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30121 of just cond ? "..." : "...".
30122 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30123 of just cond ? "..." : "...".
30125 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30128 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30130 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30131 'tune for' CPU name.
30132 * config/arm/arm-cpu-data.h: Regenerated.
30134 2017-02-28 Richard Biener <rguenther@suse.de>
30136 PR tree-optimization/79732
30137 * tree-inline.c (expand_call_inline): Do not shadow var.
30139 2017-02-28 Richard Biener <rguenther@suse.de>
30141 PR tree-optimization/79723
30142 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30143 address-space properly.
30145 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30147 * doc/optinfo.texi (Optimization groups): Fix option used for
30149 * doc/invoke.texi (-fopt-info): Document "omp".
30150 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30151 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30152 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30153 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30154 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30156 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30158 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30159 OPTGROUP_OMP with "omp".
30161 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30164 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30165 for arithmetic shift of unsigned V2DI.
30167 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30169 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30170 arc/linux.h headers.
30171 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30172 (LINK_SPEC): Likewise.
30173 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30174 (EXTRA_SPECS): Likewise.
30175 (STARTFILE_SPEC): Likewise.
30176 (ENDFILE_SPEC): Likewise.
30177 (LIB_SPEC): Likewise.
30178 (TARGET_SDATA_DEFAULT): Likewise.
30179 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30180 (MULTILIB_DEFAULTS): Likewise.
30181 (DWARF2_UNWIND_INFO): Likewise.
30182 * config/arc/big.h: New file.
30183 * config/arc/elf.h: Likewise.
30184 * config/arc/linux.h: Likewise.
30185 * config/arc/t-uClibc: Remove.
30187 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30189 PR tree-optimization/77536
30190 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30191 (tree_transform_and_unroll_loop): Use above function to compute the
30192 estimated niter of unrolled loop and use it when scaling profile.
30193 Also use count info rather than frequency if it's non-zero.
30194 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30195 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30196 (vect_transform_loop): Call above function.
30198 2017-02-27 Richard Biener <rguenther@suse.de>
30200 PR tree-optimization/45397
30201 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30202 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30203 (visit_nary_op): Add pattern matching for CSEing sign-changed
30204 or truncated operations with wider ones.
30206 2017-02-27 Richard Biener <rguenther@suse.de>
30208 PR tree-optimization/79690
30209 * tree-vect-stmts.c (vectorizable_store): Use vector type
30210 built from the DR with address-space.
30212 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30214 * doc/invoke.texi (Optimize Options): Refine the description
30215 of asan-use-after-return.
30217 2017-02-25 Alan Modra <amodra@gmail.com>
30219 PR rtl-optimization/79584
30220 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30221 base, not ad->base_term, the reg within base. Remove assertion
30222 that ad->base == ad->base_term. Replace gen_int_mode using
30223 bogus mode with const0_rtx.
30225 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30227 PR middle-end/79396
30228 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30229 FMA_EXPR like tcc_binary or tcc_unary.
30231 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30234 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30236 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30237 (output_loc_operands): Handle DW_OP_call_ref and
30238 DW_OP_GNU_variable_value.
30239 (struct variable_value_struct): New type.
30240 (struct variable_value_hasher): Likewise.
30241 (variable_value_hash): New variable.
30242 (string_types): Remove.
30243 (copy_loc_descr): New function.
30244 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30245 (prepend_loc_descr_to_each): New function.
30246 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30247 instead of add_loc_descr_to_each if the first argument is single
30248 location list and the second has multiple.
30249 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30250 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30251 when looking for variable value which doesn't have other location info.
30252 (loc_list_from_tree): Formatting fix.
30253 (gen_array_type_die): Simplify DW_AT_string_length handling.
30254 (adjust_string_types): Remove.
30255 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30256 string_types. Call resolve_variable_values.
30257 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30258 (resolve_addr_in_expr): Likewise. Add A argument.
30259 (copy_deref_exprloc): Remove deref argument. Adjust for the
30260 original expression being DW_OP_GNU_variable_value with optionally
30261 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30262 optionally after it.
30263 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30264 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30265 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30266 (variable_value_hasher::hash, variable_value_hasher::equal): New
30268 (resolve_variable_value_in_expr, resolve_variable_value,
30269 resolve_variable_values, note_variable_value_in_expr,
30270 note_variable_value): New functions.
30271 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30274 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30277 * opts.h (handle_generated_option): Add GENERATED_P argument.
30278 * opts-common.c (handle_option): Adjust function comment.
30279 (handle_generated_option): Add GENERATED_P argument, pass it to
30281 (control_warning_option): Pass false to handle_generated_option
30283 * opts.c (maybe_default_option): Pass true to handle_generated_option
30285 * optc-gen.awk: Likewise.
30287 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30289 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30290 a REG, look at the REG it is a SUBREG of.
30291 (splitter for cmpeqsi_t): Ditto.
30293 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30295 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30296 the special USEs with the pattern of the insn, not the insn itself.
30298 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30301 * doc/invoke.texi: Document -mload-store-pairs.
30303 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30304 Sandra Loosemore <sandra@codesourcery.com>
30306 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30307 argument isn't a CONST_INT.
30308 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30309 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30310 (nios2_validate_compare): Bypass alternate compare logic if *op2
30311 is not a CONST_INT.
30312 (ldstwm_operation_p): Return false if first_base is not a REG or
30313 if first_offset is not a CONST_INT.
30315 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30317 * config/cris/cris.md: Use correct operand in a define_peephole2.
30319 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30321 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30323 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30325 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30326 this_insn if it is an INSN or JUMP_INSN.
30327 (force_offsettable): Look at base, not at addr.
30328 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30329 on things that aren't necessarily CONST_INTs.
30331 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30333 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30334 -mfpmath=sse is the default also for x86-32 targets with SSE2
30335 instruction set when @option{-ffast-math} is enabled
30337 2017-02-24 Jeff Law <law@redhat.com>
30339 PR rtl-optimizatoin/79286
30340 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30343 2017-02-24 Richard Biener <rguenther@suse.de>
30345 PR tree-optimization/79389
30346 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30349 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30351 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30352 function comment to reflect reality.
30353 (loop_exits_before_overflow): Fix typo in function description.
30355 2017-02-24 Richard Biener <rguenther@suse.de>
30357 PR tree-optimization/79389
30358 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30359 properly that a threading opportunity exists. Detect conditional
30360 copy/constant propagation opportunities.
30362 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30364 * config/visium/visium.md (type): Add trap.
30365 (b): New mode attribute.
30366 (*btst): Rename into...
30367 (*btst<mode>): ...this and adjust.
30368 (*cbranchsi4_btst_insn): Rename into...
30369 (*cbranch<mode>4_btst_insn): ...this and adjust.
30370 (trap): New define_insn.
30372 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30374 PR tree-optimization/79389
30375 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30376 (noce_reversed_cond_code): New function.
30377 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30378 reversed_comparison_code. Formatting fix.
30379 (noce_try_store_flag): Test rev_cond != NULL in addition to
30380 reversed_comparison_code.
30381 (noce_try_store_flag_constants): Likewise.
30382 (noce_try_store_flag_mask): Likewise.
30383 (noce_try_addcc): Use rev_cond if non-NULL instead of
30384 reversed_comparison_code.
30385 (noce_try_cmove_arith): Likewise. Formatting fixes.
30386 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30387 (noce_find_if_block): Initialize rev_cond.
30388 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30389 instead of false as last argument never attempt to reverse it
30392 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30394 PR tree-optimization/79663
30395 * tree-predcom.c (combine_chains): Process refs in reverse order
30396 only for ZERO length chains, and add explaining comment.
30398 2017-02-23 Jeff Law <law@redhat.com>
30400 PR tree-optimization/79578
30401 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30402 in call to operand_equal_p.
30404 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30407 * config/i386/cpuid.h: Fix another undefined behavior.
30409 2017-02-23 Richard Biener <rguenther@suse.de>
30411 PR tree-optimization/79683
30412 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30413 vector types for data-refs.
30415 2017-02-23 Martin Liska <mliska@suse.cz>
30417 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30419 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30421 PR middle-end/79665
30422 * internal-fn.c (get_range_pos_neg): Moved to ...
30423 * tree.c (get_range_pos_neg): ... here. No longer static.
30424 * tree.h (get_range_pos_neg): New prototype.
30425 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30426 are known to be in between 0 and signed maximum inclusive, try to
30427 expand both unsigned and signed divmod and use the cheaper one from
30430 2017-02-22 Jeff Law <law@redhat.com>
30432 PR tree-optimization/79578
30433 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30434 to compare base operands.
30436 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30439 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30440 gpc_reg_operand instead of fpr_reg_operand.
30442 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30444 * config/mips/mips.c (mips_return_in_memory): Force FP
30445 vector types to be returned in memory for o32 ABI.
30447 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30449 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30450 instead of DW_TAG_member for static data member declarations and don't
30451 set no_linkage_name for static inline data members.
30452 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30455 2017-02-22 Martin Liska <mliska@suse.cz>
30457 * doc/invoke.texi: Replace inequality signs with square brackets
30460 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30463 * lra-constraints.c (simplify_operand_subreg): Handle
30464 WORD_REGISTER_OPERATIONS targets.
30466 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30469 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30470 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30471 elimination by swapping fld*.
30473 2017-02-22 Richard Biener <rguenther@suse.de>
30475 PR tree-optimization/79673
30476 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30477 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30478 irrelevant address-space qualifiers and avoiding a
30479 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30481 2017-02-22 Richard Biener <rguenther@suse.de>
30483 PR tree-optimization/79666
30484 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30485 to not symbolically negate if that may introduce undefined
30488 2017-02-22 Martin Liska <mliska@suse.cz>
30491 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30492 * data-streamer-out.c (streamer_write_gcov_count_stream):
30494 * value-prof.c (stream_out_histogram_value): Make assert more
30495 precise based on type of counter.
30497 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30500 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30501 Use nonimmediate_operand instead of memory_operand for operand 1.
30502 (float-extend standard_x87sse_constant_load splitter): Ditto.
30504 2017-02-21 Jeff Law <law@redhat.com>
30506 PR tree-optimization/79621
30507 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30508 blocks with edges to themselves.
30510 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30513 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30514 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30515 Use gimple_call_builtin_p.
30518 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30519 on temporarily removed DEBUG_INSNs.
30521 PR tree-optimization/79649
30522 * tree-loop-distribution.c (classify_partition): Give up on
30523 non-generic address space loads/stores.
30525 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30527 * doc/loop.texi (Loop manipulation): Remove nonexistent
30528 tree_ssa_loop_version from the documentation.
30529 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30531 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30534 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30535 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30536 * config/rs6000/rs6000.c: Include except.h.
30537 (rs6000_expand_split_stack_prologue): Call
30538 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30540 2017-02-21 Martin Jambor <mjambor@suse.cz>
30543 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30544 have been analyzed.
30546 2017-02-21 Martin Jambor <mjambor@suse.cz>
30548 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30549 for backward compatibility only.
30550 * doc/invoke.texi (Option Summary): Remove all references to
30551 -fipa-cp-alignment.
30553 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30557 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30559 * lra-constraints.c (curr_insn_transform): Handle
30560 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30562 2017-02-21 Martin Liska <mliska@suse.cz>
30564 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30566 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30568 PR translation/79638
30569 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30571 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30574 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30575 (arm_function_ok_for_sibcall): Return false for an indirect call by
30576 descriptor if all the argument registers are used.
30577 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30578 alignment of the function.
30580 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30582 PR tree-optimization/61441
30583 * simplify-rtx.c (simplify_const_unary_operation): For
30584 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30585 the sNaN unmodified.
30587 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30589 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30590 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30591 instead of SYSTEM_HEADER_DIR.
30593 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30594 Martin Liška <mliska@suse.cz>
30596 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30597 Fix typos and grammar, use active voice, and clarify.
30599 2017-02-20 Marek Polacek <polacek@redhat.com>
30601 PR middle-end/79537
30602 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30605 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30607 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30610 * config/i386/i386.c (ix86_expand_builtin): Handle
30611 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30612 ix86_builtins_isa[fcode].isa as a requirement of those
30613 flags and any other flag in the bitmask.
30614 (ix86_init_mmx_sse_builtins): Use 0 instead of
30615 ~OPTION_MASK_ISA_64BIT as mask.
30616 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30617 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30618 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30619 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30621 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30624 * lra-constraints.c (split_reg): Check requested split mode
30625 is supported by the register.
30627 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30629 * lra-constraints.c (simplify_operand_subreg): Remove early
30632 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30635 * lra-constraints.c (curr_insn_transform): Tighten condition
30636 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30638 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30641 * lra-constraints.c (curr_insn_transform): Handle
30642 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30644 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30647 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30649 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30651 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30654 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30657 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30659 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30662 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30664 * final.c (last_columnnum, override_columnnum): New variables.
30665 (final_start_function): Set last_columnnum, pass it to begin_prologue
30666 hook and pass 0 to dwarf2out_begin_prologue.
30667 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30668 to source_line debug hook.
30669 (notice_source_line): Compute last_columnnum and for debug_column_info
30670 return true on column changes.
30671 * debug.h (struct gcc_debug_hooks): Add column argument to
30672 source_line and begin_prologue hooks.
30673 (debug_nothing_int_charstar_int_bool): Remove prototype.
30674 (debug_nothing_int_int_charstar,
30675 debug_nothing_int_int_charstar_int_bool): New prototypes.
30676 (dwarf2out_begin_prologue): Add column argument.
30677 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30678 begin_prologue hooks.
30679 (debug_nothing_int_charstar_int_bool): Remove.
30680 (debug_nothing_int_int_charstar,
30681 debug_nothing_int_int_charstar_int_bool): New functions.
30682 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30683 through to dwarf2out_source_line.
30684 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30685 (dwarf2out_source_line): Add column argument, emit it if requested.
30686 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30688 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30689 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30690 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30691 through to dwarf2out_begin_prologue.
30692 (vmsdbgout_source_line): Add column argument, pass it through to
30693 dwarf2out_source_line.
30694 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30695 dbxout_source_line caller.
30696 (dbxout_source_line): Add column argument.
30698 * common.opt (gno-column-info, gcolumn-info): New options.
30699 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30700 (check_die): Also test for multiple DW_AT_decl_column attributes.
30701 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30702 DW_AT_decl_column if requested.
30703 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30705 (gen_variable_die): Likewise.
30706 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30707 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30710 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30711 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30712 (ix86_handle_option): Handle OPT_m3dnowa.
30713 * doc/invoke.texi (-m3dnowa): Document.
30714 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30715 -m3dnowa instead of -m3dnow -march=athlon.
30718 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30719 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30721 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30724 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30725 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30726 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30727 generator for vsx_xxpermdi_<mode>_be.
30728 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30729 force big-endian semantics.
30730 (vsx_xxpermdi_<mode>_be): New define_expand with same
30731 implementation as previous version of vsx_xxpermdi_<mode>.
30733 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30735 PR tree-optimization/79327
30736 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30737 variable, its initialization and use.
30739 2017-02-17 Julia Koval <julia.koval@intel.com>
30741 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30742 (OPTION_MASK_ISA_PKU_UNSET): New.
30743 (ix86_handle_option): Handle -mrdpid.
30744 * config/i386/cpuid.h (bit_RDPID): New.
30745 * config/i386/driver-i386.c (host_detect_local_cpu):
30746 Detect RDPID feature.
30747 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30748 * config/i386/i386-c.c (ix86_target_macros_internal):
30750 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30751 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30752 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30753 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30754 * config/i386/i386.md (define_insn "rdpid"): New.
30755 * config/i386/i386.opt Add -mrdpid.
30756 * config/i386/immintrin.h (_rdpid_u32): New.
30758 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30760 PR rtl-optimization/79541
30761 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30762 instead of transforming it into USE.
30764 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30766 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30767 If HONOR_SNANS (SFmode) force the input to a register.
30768 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30769 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30770 an frsp or similar insn.
30772 2017-02-17 Martin Liska <mliska@suse.cz>
30774 PR rtl-optimization/79577
30775 * params.def (selsched-max-sched-times): Increase minimum to 1.
30777 2017-02-17 Martin Liska <mliska@suse.cz>
30779 PR rtl-optimization/79574
30780 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30782 2017-02-17 Martin Liska <mliska@suse.cz>
30784 PR tree-optimization/79529
30785 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30786 ssa_defined_default_def_p to handle cases which are implicitly
30788 * tree-ssa.c (ssa_defined_default_def_p): New function.
30789 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30790 which are implicitly defined.
30791 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30793 2017-02-17 Richard Biener <rguenther@suse.de>
30795 PR middle-end/79576
30796 * params.def (max-ssa-name-query-depth): Limit to 10.
30798 2017-02-17 Richard Biener <rguenther@suse.de>
30800 PR tree-optimization/79552
30801 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30804 2017-02-17 Richard Biener <rguenther@suse.de>
30807 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30809 2017-02-17 Marek Polacek <polacek@redhat.com>
30811 PR middle-end/79536
30812 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30813 (fold_negate_expr): New wrapper.
30815 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30817 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30818 Correct terminology and de-emphasize pre-standard behavior.
30820 2017-02-16 Alan Modra <amodra@gmail.com>
30822 PR rtl-optimization/79286
30823 * ira.c (def_dominates_uses): New function.
30824 (update_equiv_regs): Don't create an equivalence for insns that
30825 may trap where the register def does not dominate the use.
30827 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30829 PR rtl-optimization/78127
30830 * lra.c (lra): Call lra_eliminate before finish the loop after
30833 2017-02-16 Richard Biener <rguenther@suse.de>
30835 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30837 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30838 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30839 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30840 (isl_val_int_from_wi): New function.
30841 (extract_affine_gmp): Rename to ...
30842 (extract_affine_wi): ... this, take a widest_int.
30843 (extract_affine_int): Just wrap extract_affine_wi.
30844 (add_param_constraints): Use isl_val_int_from_wi.
30845 (add_loop_constraints): Likewise, and extract_affine_wi.
30847 2017-02-15 Jeff Law <law@redhat.com>
30849 PR middle-end/79521
30850 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30851 ira_init_register_move_cost_if_necessary.
30853 2017-02-15 Martin Sebor <msebor@redhat.com>
30855 PR middle-end/32003
30856 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30857 removed in a prior commit.
30859 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30861 PR tree-optimization/79347
30862 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30863 counters during peeling.
30865 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30867 * Makefile.in (site.exp): Remove "set ISLVER".
30869 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30872 * real.c (real_from_integer): Call real_convert even for decimal.
30874 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30877 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30879 2017-02-14 Andrew Pinski <apinski@cavium.com>
30881 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30882 cores and change the partno/implementer to be correct.
30883 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30884 the 'B" as the implementer.
30885 * config/aarch64/aarch64-tune.md: Regenerate.
30887 2017-02-14 Carl Love <cel@us.ibm.com>
30889 * config/rs6000/rs6000.c: Add case statement entry to make the
30890 xvcvuxdsp built-in argument unsigned.
30891 * config/rs6000/vsx.md: Fix the source and return operand types so they
30892 match the instruction definitions from the ISA document. Fix typo
30893 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30896 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30899 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30900 member early_clobber_alts.
30901 * lra-lives.c (reg_early_clobber_p): New.
30902 (process_bb_lives): Use it.
30903 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30904 (debug_operand_data): Initialize early_clobber_alts.
30905 (setup_operand_alternative): Set up early_clobber_alts.
30906 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30907 alternatives to new_insn_reg.
30908 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30910 (lra_update_insn_regno_info): Pass the new arg.
30912 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30914 PR middle-end/79505
30915 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30916 (new_oacc_loop_raw): Don't clear already cleared fields.
30919 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30920 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30921 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30923 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30926 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30928 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30931 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30932 the extra instruction to the right place to store 128-bit constant
30935 2017-02-14 Martin Sebor <msebor@redhat.com>
30937 PR middle-end/79448
30938 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30939 warning for strings of unknown length.
30941 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30943 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30945 2017-02-14 Jeff Law <law@redhat.com>
30948 * ira-costs.c (scan_one_insn): Initialize register move costs
30949 for pseudos seen in USE/CLOBBER insns.
30951 PR tree-optimization/79095
30952 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30953 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30954 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30955 if the operands are known to be not equal, then the resulting range
30957 (intersect_ranges): If the new range is ~[0,0] and the old range is
30958 wide, then prefer ~[0,0].
30959 * tree-vrp.c (overflow_comparison_p_1): New function.
30960 (overflow_comparison_p): New function.
30961 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30962 if NAME is used in an overflow test.
30963 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30964 overflow check that can be expressed as an equality test, then adjust
30965 ops to be that equality test.
30967 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30969 * config/s390/s390-builtin-types.def: Remove flags argument.
30970 * config/s390/s390.c (s390_init_builtins): Likewise.
30972 2017-02-14 Martin Liska <mliska@suse.cz>
30974 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30975 vector. Fix trailing white spaces.
30977 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30979 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30982 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30984 PR rtl-optimization/68664
30985 * config/arm/arm.c (arm_sched_can_speculate_insn):
30986 New function. Declare prototype.
30987 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30989 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30991 PR rtl-optimization/68664
30992 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30994 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30996 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30998 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30999 max skip bytes for function, loop and jump.
31001 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31003 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31004 ABS_EXPR for gimple dump.
31006 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31009 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31011 PR tree-optimization/79408
31012 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31013 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31014 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31015 also if rhs1 is INTEGER_CST.
31017 2017-02-14 Richard Biener <rguenther@suse.de>
31019 PR middle-end/79432
31020 * tree-into-ssa.c (insert_phi_nodes): When the function can
31021 have abnormal edges rewrite SSA names with broken use-def
31022 dominance out of SSA and register them for PHI insertion.
31024 2017-02-13 Martin Sebor <msebor@redhat.com>
31026 PR middle-end/79496
31027 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31028 clearing info.nowrite flag when snprintf size argument is a range.
31030 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31032 * cprop.c (cprop_jump): Add missing space in string literal.
31033 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31034 (get_constraint_for_component_ref): Likewise.
31035 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31036 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31037 * lra-constraints.c (process_alt_operands): Likewise.
31038 * ipa-inline.c (inline_small_functions): Likewise.
31039 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31040 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31041 * trans-mem.c (diagnose_tm_1_op): Likewise.
31042 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31043 (grid_parallel_clauses_gridifiable): Likewise.
31045 * config/nvptx/mkoffload.c (process): Add space in between
31048 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31049 "MOD4_SSE_REGS" and "ALL_REGS".
31051 * spellcheck.c (test_data): Add , in between "foo" and "food".
31053 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31056 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31057 boundary crossing check and subsequent code generation agree.
31059 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31061 * config/aarch64/aarch64.c (has_memory_op): Delete.
31062 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31065 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31067 PR rtl-optimization/79388
31068 PR rtl-optimization/79450
31069 * combine.c (distribute_notes): When removing TEM_INSN for which
31070 corresponding dest has last value recorded, invalidate that last
31073 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31075 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31076 of explicit '@'. Add missing assembly comment marker on branch costs
31079 2017-02-13 Nathan Sidwell <nathan@acm.org>
31081 * gengtype-lex.l (<in_struct>): Add '/'.
31083 2017-02-13 Martin Liska <mliska@suse.cz>
31086 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31088 2017-02-13 Richard Biener <rguenther@suse.de>
31090 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31092 * configure: Re-generate.
31093 * config.in: Likewise.
31094 * graphite-dependences.c: Simplify as if
31095 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31096 * graphite-isl-ast-to-gimple.c: Likewise.
31097 * graphite-optimize-isl.c: Likewise.
31098 * graphite-poly.c: Likewise.
31099 * graphite-sese-to-poly.c: Likewise.
31100 * graphite.h: Likewise.
31101 * toplev.c: Include isl/version.h and use isl_version () for
31102 printing the ISL version.
31103 * doc/install.texi: Update ISL requirement.
31105 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31107 * doc/standards.texi (Standards): Update reference to
31110 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31112 * doc/extend.texi (Named Address Spaces): sourceware.org now
31114 * doc/install.texi (Binaries): Ditto.
31117 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31119 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31120 standard terminology "stringize"/"stringizing" throughout.
31121 * doc/cppinternals.texi: Likewise.
31123 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31125 * doc/extend.texi: Fix some spelling mistakes and typos.
31126 * doc/invoke.texi: Likewise.
31128 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31131 * params.def (inline-min-speedup) Change from 10 to 8.
31133 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31135 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31138 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31141 * ipa-inline-analysis.c (get_minimal_bb): New function.
31142 (record_modified): Use it.
31143 (remap_edge_change_prob): Handle also ancestor functions.
31145 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31147 * doc/contrib.texi (Contributors): Remove broken link into
31148 the Mauve CVS repository.
31150 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31152 PR middle-end/79454
31153 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31154 result computation whenever lhs doesn't have vector mode, not
31155 just when it has BLKmode.
31157 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31159 * doc/makefile.texi (profiledbootstrap): Refer to the
31160 installation instructions only in textual form.
31162 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31165 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31167 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31169 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31170 (Specific): Update mingw-w64 reference.
31172 (Specific): Remove broken link to Renesas RX processor.
31174 2017-02-10 Richard Biener <rguenther@suse.de>
31176 * toplev.c (process_options): Do not mention obsolete graphite
31177 options when printing sorry message about missing graphite support.
31178 Mention -floop-nest-optimize.
31180 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31182 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31183 (vtst_p16): Likewise.
31184 (vtstq_p8): Likewise.
31185 (vtstq_p16): Likewise.
31187 (vtstq_p64): Likewise.
31188 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31189 (vset_lane_p64): New.
31190 (vsetq_lane_p64): New.
31192 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31194 PR tree-optimization/79411
31195 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31196 stmt operands are SSA_NAMEs used in abnormal phis.
31197 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31200 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31203 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31206 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31208 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31210 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31212 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31216 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31217 not arbitrary TREE_CONSTANT.
31220 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31221 "omp declare target link" attribute unless is_global_var.
31222 * omp-offload.c (find_link_var_op): Likewise.
31224 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31225 Chung-Lin Tang <cltang@codesourcery.com>
31227 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31229 (gimplify_adjust_omp_clauses): Don't delete TILE.
31230 (gimplify_omp_for): Deal with TILE.
31231 * internal-fn.c (expand_GOACC_TILE): New function.
31232 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31234 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31235 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31237 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31238 avoid DIV for outermost collapse var.
31239 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31240 Remove out of date comments, fix whitespace.
31241 * omp-general.c (omp_extract_for_data): Deal with tiling.
31242 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31243 adjust OLF_DIM_BASE value.
31244 (struct omp_for_data): Add tiling field.
31245 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31246 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31247 for auto loops. Remove default auto determining, moved to
31248 oacc_loop_fixed_partitions.
31249 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31250 stmts, add e_mask field.
31251 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31252 (oacc_thread_numbers): Use oacc_dim_call.
31253 (oacc_xform_tile): New.
31254 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31255 (finish_oacc_loop): Adjust for ifns vector.
31256 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31257 add case for GOACC_TILE fns.
31258 (oacc_loop_xform_loop): Delete.
31259 (oacc_loop_process): Iterate over call list directly, and add
31260 handling for GOACC_TILE fns.
31261 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31263 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31264 vector partitioning to outer loops. Assign 2 partitions to loops
31265 when available. Add TILE handling.
31266 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31267 (execite_oacc_device_lower): Process GOACC_TILE fns,
31268 ignore unknown specs.
31269 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31270 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31271 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31273 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31275 * configure.ac (ACX_BUGURL): Update.
31276 * configure: Regenerate.
31278 2017-02-09 Richard Biener <rguenther@suse.de>
31280 PR tree-optimization/69823
31281 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31282 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31284 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31286 * config/arc/arc-c.def: Add __NPS400__ definition.
31287 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31288 (TARGET_NPS400): Define.
31290 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31292 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31294 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31295 pointer, arch_info.
31296 (arc_cpu_types): Fill the arch_info field with a pointer into the
31297 arc_arch_types table.
31298 (arc_selected_cpu): Declare.
31299 * config/arc/arc.c (arc_selected_cpu): Make global.
31300 (arc_selected_arch): Delete.
31301 (arc_base_cpu): Delete.
31302 (arc_override_options): Remove references to deleted variables,
31303 update access to arch information.
31304 (ARC_OPT): Update access to arch information.
31305 (ARC_OPTX): Likewise.
31306 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31307 (TARGET_ARC600): Update access to arch information.
31308 (TARGET_ARC601): Likewise.
31309 (TARGET_ARC700): Likewise.
31310 (TARGET_EM): Likewise.
31311 (TARGET_HS): Likewise.
31312 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31315 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31318 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31319 condition/operands for integer GE/LE/GEU/LEU operations.
31321 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31323 PR translation/79397
31324 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31327 2017-02-08 Martin Jambor <mjambor@suse.cz>
31330 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31331 whether allocation happened.
31332 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31333 nothing was allocated.
31335 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31337 PR tree-optimization/79408
31338 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31339 constant, but SSA_NAME with a known integer range, use the minimum
31340 of that range instead of op1 to determine if modulo can be replaced
31341 with its first operand.
31343 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31345 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31347 2017-02-08 Richard Biener <rguenther@suse.de>
31349 PR tree-optimization/71824
31350 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31351 Check all loops contained in the merged region.
31353 2017-02-07 Andrew Pinski <apinski@cavium.com>
31355 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31357 2017-02-07 Andrew Pinski <apinski@cavium.com>
31359 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31360 (thunderxt88): Likewise.
31361 (thunderxt81): Disable LSE and change v8.1 to v8.
31362 (thunderxt83): Likewise.
31364 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31365 Richard Biener <rguenther@suse.de>
31367 PR middle-end/79399
31368 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31369 type from int to size_t.
31370 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31372 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31374 PR rtl-optimization/79386
31375 * cprop.c (bypass_conditional_jumps): Initialize
31376 bypass_last_basic_block already before splitting bbs after
31377 unconditional traps...
31378 (bypass_conditional_jumps): ... rather than here.
31381 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31382 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31383 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31384 fix -masm=intel patterns.
31386 2017-02-07 Richard Biener <rguenther@suse.de>
31388 PR tree-optimization/79256
31389 PR middle-end/79278
31390 * builtins.c (get_object_alignment_2): Use min_align_of_type
31391 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31392 and ADJUST_FIELD_ALIGN.
31394 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31396 * doc/tm.texi: Regenerate.
31397 * stor-layout.c (layout_decl): Adjust.
31398 (update_alignment_for_field): Likewise.
31399 (place_field): Likewise.
31400 (min_align_of_type): Likewise.
31401 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31402 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31403 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31404 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31405 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31406 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31407 * config/i386/i386.c (x86_field_alignment): Likewise.
31408 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31409 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31410 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31411 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31412 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31413 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31417 2017-01-30 Richard Biener <rguenther@suse.de>
31419 PR tree-optimization/79256
31420 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31421 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31424 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31426 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31427 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31428 builtins to SImode and emit a zero-extend, if necessary.
31430 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31432 * docs/invoke.texi (RISC-V Options): Alphabetize.
31434 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31436 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31439 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31441 * config/riscv/riscv.c: New file.
31442 * common/config/riscv/riscv-common.c: Likewise.
31443 * config.gcc: Likewise.
31444 * config/riscv/constraints.md: Likewise.
31445 * config/riscv/elf.h: Likewise.
31446 * config/riscv/generic.md: Likewise.
31447 * config/riscv/linux.h: Likewise.
31448 * config/riscv/multilib-generator: Likewise.
31449 * config/riscv/peephole.md: Likewise.
31450 * config/riscv/pic.md: Likewise.
31451 * config/riscv/predicates.md: Likewise.
31452 * config/riscv/riscv-builtins.c: Likewise.
31453 * config/riscv/riscv-c.c: Likewise.
31454 * config/riscv/riscv-ftypes.def: Likewise.
31455 * config/riscv/riscv-modes.def: Likewise.
31456 * config/riscv/riscv-opts.h: Likewise.
31457 * config/riscv/riscv-protos.h: Likewise.
31458 * config/riscv/riscv.h: Likewise.
31459 * config/riscv/riscv.md: Likewise.
31460 * config/riscv/riscv.opt: Likewise.
31461 * config/riscv/sync.md: Likewise.
31462 * config/riscv/t-elf-multilib: Likewise.
31463 * config/riscv/t-linux: Likewise.
31464 * config/riscv/t-linux-multilib: Likewise.
31465 * config/riscv/t-riscv: Likewise.
31466 * configure.ac: Likewise.
31467 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31468 Waterman as RISC-V maintainers.
31469 * doc/install.texi: Add RISC-V entries.
31470 * doc/invoke.texi: Add RISC-V options section.
31471 * doc/md.texi: Add RISC-V constraints section.
31472 * configure: Regenerated.
31474 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31477 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31478 false values to be constant vectors with all 0 or all 1 bits set.
31479 (vcondu<mode><mode>): Likewise.
31480 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31482 (fpmask_comparison_operator): Update comment.
31483 (vecint_comparison_operator): New predicate.
31484 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31485 vector conditionals when the true and false values are constant
31486 vectors with all 0 bits or all 1 bits set.
31488 2017-02-06 Martin Sebor <msebor@redhat.com>
31490 PR tree-optimization/79376
31491 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31493 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31495 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31496 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31497 to simplify split condition.
31499 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31501 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31502 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31505 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31507 PR rtl-optimization/68664
31508 * target.def (can_speculate_insn): New hook.
31509 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31510 * doc/tm.texi: Regenerate.
31511 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31512 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31513 (rs6000_sched_can_speculate_insn): New function.
31515 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31517 PR tree-optimization/79284
31518 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31519 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31520 vectorizable_mask_load_store, vectorizable_operation,
31521 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31522 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31523 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31524 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31525 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31526 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31527 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31528 is_gimple_assign (stmt). Replace another such test with
31529 is_gimple_assign (stmt).
31531 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31534 * config/avr/avr.c (rtl-iter.h): Include it.
31535 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31536 (avr_legitimate_combined_insn): ...and implementation.
31538 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31540 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31541 * config/s390/s390.c (s390_const_operand_ok)
31542 (s390_canonicalize_comparison, s390_extract_part)
31543 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31544 (s390_contiguous_bitmask_p, s390_rtx_costs)
31545 (legitimize_pic_address): Likewise.
31546 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31547 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31548 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31549 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31550 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31552 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31554 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31555 REGNO($0) == REGNO($1).
31557 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31559 * config/s390/linux.h(SIZE_TYPE): Add comment.
31561 2017-02-06 Julian Brown <julian@codesourcery.com>
31562 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31563 Virendra Pathak <virendra.pathak@broadcom.com>
31565 * config/aarch64/aarch64-cores.def: Change the scheduler
31567 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31568 * config/aarch64/thunderx2t99.md: New file.
31570 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31572 * doc/standards.texi (Go Language): Update link to language
31575 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31577 * tree-eh.c (lower_resx): Sanitize profile.
31578 (cleanup_empty_eh_move_lp): Likewise.
31580 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31583 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31585 * cfgloopmanip.h (loop_version): Update prototype.
31586 * modulo-sched.c (sms_schedule): Update call of loop_version.
31587 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31588 * tree-parloops.c (gen_parallel_loop): Likewise.
31589 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31590 * tree-ssa-loop-split.c (split_loop): Likewise.
31591 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31592 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31594 2017-02-05 Martin Liska <mliska@suse.cz>
31597 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31599 (print_operand_address): Initialize a struct to zero.
31601 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31603 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31604 garbage collector only in textual form.
31606 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31608 * doc/extend.texi (x86 specific memory model extensions for
31609 transactional memory): Simplify a phrase.
31611 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31614 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31615 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31616 (atomic_storedi_1): Likewise.
31618 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31620 PR tree-optimization/79338
31621 * tree-parloops.c (gather_scalar_reductions): Don't call
31622 vect_analyze_loop_form for loop->inner before destroying loop's
31625 2017-02-03 Martin Sebor <msebor@redhat.com>
31627 PR tree-optimization/79327
31628 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31629 when precision has resulted in leading zeros.
31630 (format_integer): Adjust the likely counter to assume an unknown
31631 argument that may be zero is non-zero.
31633 2017-02-03 Jason Merrill <jason@redhat.com>
31636 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31637 avoid copying non-taken branch.
31639 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31641 PR tree-optimization/79340
31642 * tree-vect-loop.c (vectorizable_reduction): Release
31643 vec_defs elements after safe_splicing them into other vectors.
31646 PR tree-optimization/79327
31647 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31648 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31650 (format_integer): Use wide_int_to_tree instead of build_int_cst
31651 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31652 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31653 of shortest and longest sequence.
31655 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31657 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31658 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31660 2017-02-03 Walter Lee <walt@tilera.com>
31663 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31664 after initial stackframe link reg save.
31665 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31667 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31670 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31671 wu for stxssp alternative.
31673 2017-02-03 Martin Sebor <msebor@redhat.com>
31675 PR tree-optimization/79352
31676 * gimple-fold.c (get_range_strlen): Add argument.
31677 (get_range_strlen): Change return type to bool.
31678 (get_maxval_strlen): Pass in a dummy argument.
31679 * gimple-fold.h (get_range_strlen): Change return type to bool.
31680 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31681 * tree.h (array_at_struct_end_p): Add argument.
31682 * tree.c (array_at_struct_end_p): Handle it.
31684 2017-02-03 Martin Liska <mliska@suse.cz>
31687 * multiple_target.c (create_dispatcher_calls): Redirect edge
31688 from a caller of a dispatcher.
31689 (expand_target_clones): Make the clones local.
31690 (ipa_target_clone): Do both target clones and resolvers.
31691 (ipa_dispatcher_calls): Remove the pass.
31692 (pass_dispatcher_calls::gate): Likewise.
31693 (make_pass_dispatcher_calls): Likewise.
31694 * passes.def (pass_target_clone): Put as very first IPA early
31697 2017-02-03 Martin Liska <mliska@suse.cz>
31699 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31700 in case of a function with ifunc attribute.
31702 2017-02-03 Martin Liska <mliska@suse.cz>
31704 * cgraph.c (cgraph_node::dump): Dump function version info.
31705 * symtab.c (symtab_node::dump_base): Add missing new line.
31707 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31709 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31710 (ifcombine_ifandif): Use it.
31712 2017-02-03 Martin Liska <mliska@suse.cz>
31714 * doc/invoke.texi: Document default value for
31715 use-after-scope-direct-emission-threshold.
31717 2017-02-03 Martin Liska <mliska@suse.cz>
31719 PR tree-optimization/79339
31720 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31721 (format_floating): Likewise.
31723 2017-02-03 Martin Liska <mliska@suse.cz>
31726 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31728 (ipa_node_params_t::remove): Likewise.
31729 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31730 initialization from removed ipa_node_params_t::insert.
31731 (ipa_node_params::~ipa_node_params): Move from removed
31732 ipa_node_params_t::release.
31733 * symbol-summary.h (symbol_summary::m_released): New member.
31734 Do not release a summary twice. Do not allow to call finalizer
31735 for types of a summary that live in GGC memory.
31737 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31739 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31742 2017-02-02 Martin Sebor <msebor@redhat.com>
31744 PR middle-end/79275
31745 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31746 (format_string): Tighten up the range of output for non-constant
31747 strings and correct the expected range for wide non-constant strings.
31749 2017-02-02 Martin Sebor <msebor@redhat.com>
31751 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31753 PR middle-end/32003
31754 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31756 (-fdump-tree-@var): Add to index and document how to come up
31757 with pass-specific option and dump file names.
31758 (-fdump-passes): Clarify where to look for output.
31760 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31762 PR middle-end/77445
31763 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31764 statistics of the analyzed path; allow threading for speed when
31765 any of BBs along the path are optimized for speed.
31767 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31769 PR middle-end/78468
31770 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31771 settings of the virtual registers.
31774 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31776 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31777 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31780 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31782 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31783 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31785 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31787 * config/s390/s390.md: Add missing comments with the expanded
31789 * config/s390/vector.md: Likewise.
31790 * config/s390/vx-builtins.md: Likewise.
31792 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31795 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31796 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31797 conditions on a single line.
31799 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31801 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31802 __S390_VX__ to __VX__.
31804 2017-02-01 Andrew Pinski <apinski@cavium.com>
31806 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31807 stmt_info to record_stmt_cost.
31808 (vect_get_known_peeling_cost): Pass stmt_info if known to
31810 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31811 cpu_vector_cost field into
31812 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31813 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31814 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31815 splitting of scalar_stmt_cost and vec_stmt_cost.
31816 (thunderx_vector_cost): Likewise.
31817 (cortexa57_vector_cost): LIkewise.
31818 (exynosm1_vector_cost): Likewise.
31819 (xgene1_vector_cost): Likewise.
31820 (thunderx2t99_vector_cost): Improve after the splitting of the two
31822 (aarch64_builtin_vectorization_cost): Update for the splitting of
31823 scalar_stmt_cost and vec_stmt_cost.
31825 2017-02-01 Torvald Riegel <triegel@redhat.com>
31826 Richard Henderson <rth@redhat.com>
31828 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31829 conditional on existance of a fast atomic load.
31830 * optabs-query.c (can_atomic_load_p): New function.
31831 * optabs-query.h (can_atomic_load_p): Declare it.
31832 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31833 no fast atomic load is available for the particular size of access.
31834 (expand_atomic_compare_and_swap): Likewise.
31835 (expand_atomic_load): Likewise.
31836 (expand_atomic_store): Likewise.
31837 (expand_atomic_fetch_op): Likewise.
31838 * testsuite/lib/target-supports.exp
31839 (check_effective_target_sync_int_128): Remove x86 because it provides
31840 no fast atomic load.
31841 (check_effective_target_sync_int_128_runtime): Likewise.
31843 2017-02-01 Richard Biener <rguenther@suse.de>
31845 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31846 (graphite_transform_loops): Provide opt-info for optimized nests.
31847 * tree-parloop.c (parallelize_loops): Provide opt-info for
31848 parallelized loops.
31850 2017-02-01 Richard Biener <rguenther@suse.de>
31852 PR middle-end/79315
31853 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31854 was not set before.
31856 2017-02-01 Richard Biener <rguenther@suse.de>
31858 PR tree-optimization/71824
31859 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31860 Verify the loops are valid in the merged SESE region.
31861 (scop_detection::can_represent_loop_1): Check analyzing the
31862 evolution of the number of iterations in the region succeeds.
31864 2017-01-31 Ian Lance Taylor <iant@golang.org>
31866 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31867 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31869 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31871 PR preprocessor/79210
31872 * input.c (get_substring_ranges_for_loc): Replace line_width
31873 assertion with error-handling.
31875 2017-01-31 Richard Biener <rguenther@suse.de>
31877 PR tree-optimization/77318
31878 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31879 (create_pw_aff_from_tree): Take loop parameter.
31880 (add_condition_to_pbb): Pass loop of the condition to
31881 create_pw_aff_from_tree.
31883 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31885 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31886 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31888 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31892 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31894 (convert_int_to_float128): Likewise.
31895 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31896 (convert_int_to_float128): Likewise.
31897 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31898 (UNSPEC_IEEE128_CONVERT): Likewise.
31899 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31900 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31901 Use local variables for IBM extended format.
31902 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31903 (fix_trunc<mode>si2_fprs): Likewise.
31904 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31905 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31906 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31907 to know that we can now have integers of all sizes in vector
31909 (fix<uns>_<mode>di2_hw): Likewise.
31910 (float<uns>_<mode>si2_hw): Likewise.
31911 (fix_<mode>si2_hw): Likewise.
31912 (fixuns_<mode>si2_hw): Likewise.
31913 (float<uns>_<mode>di2_hw): Likewise.
31914 (float_<mode>di2_hw): Likewise.
31915 (float_<mode>si2_hw): Likewise.
31916 (floatuns_<mode>di2_hw): Likewise.
31917 (floatuns_<mode>si2_hw): Likewise.
31918 (xscvqp<su>wz_<mode>): Delete, no longer used.
31919 (xscvqp<su>dz_<mode>): Likewise.
31920 (xscv<su>dqp_<mode>): Likewise.
31921 (ieee128_mfvsrd_64bit): Likewise.
31922 (ieee128_mfvsrd_32bit): Likewise.
31923 (ieee128_mfvsrwz): Likewise.
31924 (ieee128_mtvsrw): Likewise.
31925 (ieee128_mtvsrd_64bit): Likewise.
31926 (ieee128_mtvsrd_32bit): Likewise.
31928 2017-01-31 Martin Liska <mliska@suse.cz>
31931 * ipa-prop.c (ipa_free_all_node_params): Call release method
31932 instead of ~sumbol_summary to not to trigger double times
31935 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31937 PR tree-optimization/71691
31938 * bitmap.h (class auto_bitmap): New.
31939 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31940 is_maybe_undefined instead of ssa_undefined_value_p.
31942 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31944 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31945 __S390_ARCH_LEVEL__ to __ARCH__.
31947 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31949 PR tree-optimization/79267
31950 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31951 if should_remove_lhs_p is true.
31953 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31956 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31957 (add_alignment_attribute): New.
31958 (base_type_die): Add alignment attribute.
31959 (subrange_type_die): Likewise.
31960 (modified_type_die): Likewise.
31961 (gen_array_type_die): Likewise.
31962 (gen_descr_array_type_die: Likewise.
31963 (gen_enumeration_type_die): Likewise.
31964 (gen_subprogram_die): Likewise.
31965 (gen_variable_die): Likewise.
31966 (gen_field_die): Likewise.
31967 (gen_ptr_to_mbr_type_die): Likewise.
31968 (gen_struct_or_union_type_die): Likewise.
31969 (gen_subroutine_type_die): Likewise.
31970 (gen_typedef_die): Likewise.
31971 (base_type_cmp): Compare alignment attribute.
31973 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31976 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31977 (setb_unsigned) New pattern for setb with CCUNS.
31978 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31979 subfc./subfe sequence to avoid overflow problems. Generate a
31980 shorter sequence with cmpld/setb for power9.
31981 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31982 for generating subfc. instruction.
31983 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31984 now uses this instruction.
31986 2017-01-30 Ian Lance Taylor <iant@google.com>
31989 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31990 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31992 2017-01-30 Martin Sebor <msebor@redhat.com>
31994 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31995 Move constant to the right of a relational operator.
31996 (get_mpfr_format_length, format_character, format_string): Ditto.
31997 (should_warn_p, maybe_warn): Same.
31999 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32001 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32004 * asan.c (get_translation_unit_decl): Remove function.
32005 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32007 2017-01-30 Martin Liska <mliska@suse.cz>
32009 PR gcov-profile/79259
32010 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32011 -fprofile-generate.
32013 2017-01-30 Martin Liska <mliska@suse.cz>
32016 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32017 Initialize variables with NULL value.
32019 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32022 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32024 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32026 2017-01-30 Richard Biener <rguenther@suse.de>
32028 PR tree-optimization/79276
32029 * tree-vrp.c (process_assert_insertions): Properly adjust common
32030 when removing a duplicate.
32032 2017-01-30 Richard Biener <rguenther@suse.de>
32034 PR tree-optimization/79256
32035 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32036 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32038 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32040 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32043 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32044 ("*r<noxa>sbg_<mode>_sll_bitmask")
32045 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32046 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32047 Use contiguous_bitmask_nowrap_operand.
32049 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32052 * config/rs6000/altivec.h (vec_xl): Revise #define.
32053 (vec_xst): Likewise.
32055 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32057 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32059 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32061 PR rtl-optimization/79194
32062 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32063 traps before call to bypass_conditional_jumps.
32065 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32067 PR tree-optimization/71374
32068 * lra-constraints.c (check_conflict_input_operands): New.
32069 (match_reload): Use it.
32071 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32074 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32075 account to calculate conflict_set.
32077 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32079 PR rtl-optimization/78559
32080 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32081 other_insn in combine.
32083 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32085 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32086 uint16_type_node for BT_UINT16.
32088 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32090 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32091 "RTL Tests" to menu.
32092 (GIMPLE Tests): New node.
32093 (RTL Tests): New node.
32095 2017-01-27 Richard Biener <rguenther@suse.de>
32097 PR tree-optimization/79245
32098 * tree-loop-distribution.c (distribute_loop): Apply cost
32099 modeling also to detected patterns.
32101 2017-01-27 Richard Biener <rguenther@suse.de>
32103 PR tree-optimization/71433
32104 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32105 (compare_assert_loc): New function.
32106 (process_assert_insertions): Sort and optimize assert locations
32107 to remove duplicates and push down identical assertions on
32108 edges to their destination block.
32110 2017-01-27 Richard Biener <rguenther@suse.de>
32112 PR tree-optimization/79244
32113 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32114 out SSA names even if abnormal.
32116 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32118 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32119 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32120 instead of MPFR_RNDN.
32122 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32125 * arm.c (arm_option_override): Don't call build_target_option_node
32126 until after doing all option overrides.
32127 (arm_valid_target_attribute_tree): Likewise.
32129 2017-01-27 Martin Liska <mliska@suse.cz>
32131 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32132 for {cd}tors and C++ {cd}tors.
32134 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32136 * config/s390/s390.md ("*setmem_long_and")
32137 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32139 2017-01-26 Martin Sebor <msebor@redhat.com>
32141 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32144 2017-01-26 Martin Sebor <msebor@redhat.com>
32146 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32147 HAVE_DFmode before using XFmode or DFmode.
32148 (parse_directive): Avoid using the z length modifier to avoid
32149 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32151 PR middle-end/78703
32152 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32153 to accept adjustment as an array.
32154 (get_int_range): New function.
32155 (struct directive): Make width and prec arrays.
32156 (directive::set_width, directive::set_precision): Call get_int_range.
32157 (format_integer, format_floating): Handle width and precision ranges.
32158 (format_string, parse_directive): Same.
32160 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32163 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32164 comdat_type_p set, just clone them, but keep the children in the
32168 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32169 which have direct callers with -fvar-tracking-assignments enabled
32171 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32172 inside of type units.
32174 2017-01-26 Martin Sebor <msebor@redhat.com>
32176 PR middle-end/78703
32177 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32179 (struct format_result): Replace number_chars, number_chars_min,
32180 and number_chars_max with a single member of struct result_range.
32182 (format_result::operator+=): Adjust.
32183 (struct fmtresult): Remove bounded. Handle likely and unlikely
32185 (fmtresult::adjust_for_width_or_precision): New function.
32186 (fmtresult:type_max_digits): New function.
32187 (bytes_remaining): Handle likely and unlikely counters.
32188 (min_bytes_remaining): Remove.
32189 (format_percent): Simplify.
32190 (format_integer, format_floating): Set likely and unlikely counters.
32191 (get_string_length, format_character, format_string): Same.
32192 (format_plain, should_warn_p): New function.
32193 (maybe_warn): Call should_warn_p. Update diagnostic messages
32194 and handle those for all directives, including plain strings.
32195 (format_directive): Handle likely and unlikely counters.
32196 Remove unnecessary quoting from diagnostics. Add an informational
32198 (add_bytes): Remove.
32199 (pass_sprintf_length::compute_format_length): Simplify.
32200 (try_substitute_return_value): Handle likely and unlikely counters.
32202 2017-01-26 Carl Love <cel@us.ibm.com>
32204 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32205 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32207 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32210 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32211 endianess for subregs into account.
32212 * lra-constraints.c (lra_constraints): Do risky transformations
32213 always on the first iteration.
32214 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32215 last_call_used_reg_set.
32216 (process_bb_lives): Define and use last_call_used_reg_set.
32217 * lra.c (lra): Always continue after lra_constraints on the first
32220 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32222 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32224 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32226 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32228 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32229 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32230 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32231 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32232 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32233 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32234 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32235 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32236 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32238 2017-01-26 Marek Polacek <polacek@redhat.com>
32241 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32242 for the third operand.
32244 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32246 PR middle-end/79236
32247 * omp-low.c (struct omp_context): Add simt_stmt field.
32248 (scan_omp_for): Return omp_context *.
32249 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32250 context to the _simt_ SIMD stmt.
32251 (lower_omp_for): For combined SIMD with sibling _simt_
32252 SIMD, make sure to use the same decls in _looptemp_
32253 clauses as in the sibling.
32255 2017-01-26 David Sherwood <david.sherwood@arm.com>
32257 PR middle-end/79212
32258 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32261 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32264 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32265 emit fld b; fld a; if possible.
32267 * brig-builtins.def: Update copyright years.
32268 * config/arm/arm_acle_builtins.def: Update copyright years.
32270 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32273 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32274 constraint instead of o for the stxsd instruction.
32276 2017-01-25 Carl Love <cel@us.ibm.com>
32278 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32279 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32281 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32283 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32285 2017-01-25 Richard Biener <rguenther@suse.de>
32287 PR tree-optimization/69264
32288 * target.def (vector_alignment_reachable): Improve documentation.
32289 * doc/tm.texi: Regenerate.
32290 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32292 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32293 earlier changes with respect to TYPE_USER_ALIGN.
32294 (vector_alignment_reachable_p): Likewise. Improve dumping.
32296 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32299 * config/arm/arm.md (xordi3): Force constant operand into a register
32302 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32304 * doc/invoke.texi (-fstore-merging): Correct default optimization
32305 levels at which it is enabled.
32306 (-O): Move -fstore-merging from list to...
32309 2017-01-25 Richard Biener <rguenther@suse.de>
32312 * omp-expand.c: Include debug.h.
32313 (expand_omp_taskreg): Make sure to generate early debug before
32314 outlining anything from a function.
32315 (expand_omp_target): Likewise.
32316 (grid_expand_target_grid_body): Likewise.
32318 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32321 * asan.c (get_translation_unit_decl): New function.
32322 (asan_add_global): Extract modules file name from globals
32323 TRANSLATION_UNIT_DECL name.
32325 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32328 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32329 for long calls with APCS frame and VFP.
32331 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32333 * cfg.c (original_copy_tables_initialized_p): New function.
32334 * cfg.h (original_copy_tables_initialized_p): New decl.
32335 * cfgrtl.c (relink_block_chain): Guard the call to
32336 free_original_copy_tables with a call to
32337 original_copy_tables_initialized_p.
32338 * cgraph.h (symtab_node::native_rtl_p): New decl.
32339 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32340 (symtab_node::needed_p): Don't assert for early assembly output
32341 for __RTL functions.
32342 (cgraph_node::finalize_function): Set "force_output" for __RTL
32344 (cgraph_node::analyze): Bail out early for __RTL functions.
32345 (analyze_functions): Update assertion to support __RTL functions.
32346 (cgraph_node::expand): Bail out early for __RTL functions.
32347 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32349 * function.h (struct function): Update comment for field
32351 * gimple-expr.c: Include "tree-pass.h".
32352 (gimple_has_body_p): Return false for __RTL functions.
32353 * Makefile.in (OBJS): Add run-rtl-passes.o.
32354 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32356 (gcc::pass_manager::get_clean_slate): New accessor.
32357 * passes.c: Include "insn-addr.h".
32358 (should_skip_pass_p): Add logging. Update logic for running
32359 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32360 property-provider override so it is only done for gimple passes.
32362 (skip_pass): New function.
32363 (execute_one_pass): Call skip_pass when skipping passes.
32364 * read-md.c (md_reader::read_char): Support filtering
32365 the input to a subset of line numbers.
32366 (md_reader::md_reader): Initialize fields
32367 m_first_line and m_last_line.
32368 (md_reader::read_file_fragment): New function.
32369 * read-md.h (md_reader::read_file_fragment): New decl.
32370 (md_reader::m_first_line): New field.
32371 (md_reader::m_last_line): New field.
32372 * read-rtl-function.c (function_reader::create_function): Only
32373 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32374 curr_properties. Set DECL_INITIAL to a dummy block.
32375 (read_rtl_function_body_from_file_range): New function.
32376 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32378 * run-rtl-passes.c: New file.
32379 * run-rtl-passes.h: New file.
32381 2017-01-24 Jeff Law <law@redhat.com>
32383 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32386 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32388 PR tree-optimization/79159
32389 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32390 (record_nonwrapping_iv): Improve boundary using above function if no
32391 value range information.
32393 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32394 Martin Jambor <mjambor@suse.cz>
32396 * brig-builtins.def: New file.
32397 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32398 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32399 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32400 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32401 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32402 * builtin-types.def (BT_INT8): New.
32403 (BT_INT16): Likewise.
32404 (BT_UINT8): Likewise.
32405 (BT_UINT16): Likewise.
32406 (BT_FN_ULONG): Likewise.
32407 (BT_FN_UINT_INT): Likewise.
32408 (BT_FN_UINT_ULONG): Likewise.
32409 (BT_FN_UINT_LONG): Likewise.
32410 (BT_FN_UINT_PTR): Likewise.
32411 (BT_FN_ULONG_PTR): Likewise.
32412 (BT_FN_INT8_FLOAT): Likewise.
32413 (BT_FN_INT16_FLOAT): Likewise.
32414 (BT_FN_UINT32_FLOAT): Likewise.
32415 (BT_FN_UINT16_FLOAT): Likewise.
32416 (BT_FN_UINT8_FLOAT): Likewise.
32417 (BT_FN_UINT64_FLOAT): Likewise.
32418 (BT_FN_UINT16_UINT32): Likewise.
32419 (BT_FN_UINT32_UINT16): Likewise.
32420 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32421 (BT_FN_INT_PTR_INT): Likewise.
32422 (BT_FN_UINT_PTR_UINT): Likewise.
32423 (BT_FN_LONG_PTR_LONG): Likewise.
32424 (BT_FN_ULONG_PTR_ULONG): Likewise.
32425 (BT_FN_VOID_UINT64_UINT64): Likewise.
32426 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32427 (BT_FN_INT8_INT8_INT8): Likewise.
32428 (BT_FN_INT16_INT16_INT16): Likewise.
32429 (BT_FN_INT_INT_INT): Likewise.
32430 (BT_FN_UINT_FLOAT_UINT): Likewise.
32431 (BT_FN_FLOAT_UINT_UINT): Likewise.
32432 (BT_FN_ULONG_UINT_UINT): Likewise.
32433 (BT_FN_ULONG_UINT_PTR): Likewise.
32434 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32435 (BT_FN_UINT_UINT_UINT): Likewise.
32436 (BT_FN_VOID_UINT_PTR): Likewise.
32437 (BT_FN_UINT_UINT_PTR: Likewise.
32438 (BT_FN_UINT32_UINT64_PTR): Likewise.
32439 (BT_FN_INT_INT_UINT_UINT): Likewise.
32440 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32441 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32442 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32443 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32444 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32445 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32446 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32447 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32448 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32449 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32450 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32451 * doc/frontends.texi: List BRIG FE.
32452 * doc/install.texi (Testing): Add BRIG tesring requirements.
32453 * doc/invoke.texi (Overall Options): Mention BRIG.
32454 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32456 2017-01-24 Richard Biener <rguenther@suse.de>
32458 PR translation/79208
32459 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32461 2017-01-24 Martin Jambor <mjambor@suse.cz>
32464 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32465 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32466 and known_contexts.
32468 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32470 PR middle-end/79123
32471 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32472 casts from signed to unsigned really don't have a range.
32474 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32476 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32477 GMP_RNDx for compatiblity.
32479 2017-01-24 Martin Liska <mliska@suse.cz>
32482 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32483 that would prevent us to call alloca with -1 as argument.
32485 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32487 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32488 Avoid -Wformat-security warning.
32490 2017-01-23 Andrew Pinski <apinski@cavium.com>
32492 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32495 2017-01-23 Martin Sebor <msebor@redhat.com>
32497 PR middle-end/78703
32498 * gimple-ssa-sprintf.c (warn_level): New global.
32499 (format_integer): Use it here and throughout the rest of the file.
32500 Use the same switch to compute sign as base.
32501 (maybe_warn): New function.
32502 (format_directive): Factor out warnings into maybe_warn.
32503 Add debugging output. Use warn_level.
32504 (add_bytes): Use warn_level.
32505 (pass_sprintf_length::compute_format_length): Add debugging output.
32506 (try_substitute_return_value): Same.
32507 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32509 PR middle-end/78703
32510 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32511 (struct fmtresult, format_integer, format_floating): Adjust.
32512 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32513 (get_string_length, format_string,format_directive): Same.
32514 (pass_sprintf_length::compute_format_length): Same.
32515 (try_substitute_return_value): Simplify slightly.
32517 PR middle-end/78703
32518 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32519 (fmtresult::operator+=): Outlined.
32520 (struct fmtresult): Add ctors.
32521 (struct conversion_spec): Rename...
32522 (struct directive): ...to this. Add and remove data members.
32523 (directive::set_width, directive::set_precision): New functions.
32524 (format_percent): Use fmtresult ctor.
32525 (get_width_and_precision): Remove.
32526 (format_integer): Make naming changes. Avoid computing width and
32528 (format_floating): Same. Adjust indentation.
32529 (format_character, format_none): New functions.
32530 (format_string): Moved character handling to format_character.
32531 (format_directive): Remove arguments, change return type.
32532 (parse_directive): New function.
32533 (pass_sprintf_length::compute_format_length): Move directive
32534 parsing to parse_directive.
32536 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32538 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32539 (assign_assembler_name_if_needed): ... this.
32540 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32541 (assign_assembler_name_if_needed): ... this.
32542 (free_lang_data_in_cgraph): Adjust callers.
32543 * cgraphunit.c (cgraph_node::analyze): Likewise.
32544 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32547 2017-01-23 Richard Biener <rguenther@suse.de>
32549 PR tree-optimization/79088
32550 PR tree-optimization/79188
32551 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32552 resetting loop bounds after last path deletion. Reset loop
32553 bounds of the target loop, make code match the comments.
32554 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32555 Make sure loops need no fixups.
32557 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32559 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32560 exponent support with double type for first argument.
32561 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32562 type returned by __builtin_vec_extract_sig,
32563 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32564 functions from "vector int" to "vector unsigned int" or from
32565 "vector long long int" to "vector unsigned long long int".
32566 Changed type returned by __builtin_vec_extract_exp,
32567 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32568 functions from "vector int" to "vector unsigned int" or from
32569 "vector long long int" to "vector unsigned long long int".
32570 Changed return type of __builtin_vec_test_data_class,
32571 __builtin_vec_test_data_class_sp, and
32572 __builtin_vec_test_data_class_dp from "vector int" to
32573 "vector bool int" or from "vector long long int" to "vector bool
32574 long long int" and changed second argument type from "unsigned
32575 int" to "int". Added new overloaded function forms "vector float
32576 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32577 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32578 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32579 double, vector unsigned long long int)" and "vector double
32580 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32581 long int)". Changed return type of
32582 __builtin_scalar_test_data_class and
32583 __builtin_scalar_test_data_class_sp and
32584 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32585 int" and changed second argument from "unsigned int" to "int".
32586 Changed type returned by __builtin_scalar_test_neg,
32587 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32588 from "int" to "bool int". Added new overloaded function form
32589 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32590 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32591 exponent double-precision with floating point first argument.
32592 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32593 documentation of scalar_test_data_class, scalar_test_neg,
32594 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32595 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32596 vec_test_data_class built-in functions to reflect refinements in
32597 their type signatures.
32599 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32601 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32603 (aarch64_elf_asm_destructor): Likewise.
32605 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32607 PR rtl-optimization/78634
32608 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32609 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32610 * ifcvt.c (noce_try_cmove): Add missing cost check.
32612 PR rtl-optimization/71724
32613 * combine.c (if_then_else_cond): Look for situations where it is
32614 beneficial to undo the work of one of the recursive calls.
32616 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32618 PR tree-optimization/70754
32619 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32620 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32621 combined stmt before it if not NULL.
32622 (combine_chains): Process refs reversely and compute dominance point
32625 2017-01-23 Martin Liska <mliska@suse.cz>
32627 PR tree-optimization/79196
32628 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32629 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32631 (strlen_optimize_stmt): Call the renamed function.
32633 2017-01-23 Michael Matz <matz@suse.de>
32635 PR tree-optimization/78384
32636 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32638 2017-01-23 Richard Biener <rguenther@suse.de>
32640 PR tree-optimization/79186
32641 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32642 both incoming edges before moving an assert.
32644 2017-01-23 Martin Jambor <mjambor@suse.cz>
32646 * ipa-prop.c (load_from_param_1): Removed.
32647 (load_from_unmodified_param): Bits from load_from_param_1 put back
32649 (load_from_param): Removed.
32650 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32651 with stmt. Reverted back to use of load_from_unmodified_param.
32653 2017-01-23 Martin Jambor <mjambor@suse.cz>
32656 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32657 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32658 field a pointer to garbage collected vector, mark lattices and
32659 ipcp_orig_node with GTY((skip)).
32660 (ipa_get_param_count): Adjust to descriptors being a pointer.
32661 (ipa_get_param): Likewise.
32662 (ipa_get_type): Likewise.
32663 (ipa_get_param_move_cost): Likewise.
32664 (ipa_set_param_used): Likewise.
32665 (ipa_get_controlled_uses): Likewise.
32666 (ipa_set_controlled_uses): Likewise.
32667 (ipa_is_param_used): Likewise.
32668 (ipa_node_params_t): Move into garbage collector. New methods insert
32670 (ipa_node_params_sum): Annotate wth GTY(()).
32671 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32673 (ipa_load_from_parm_agg): Adjust declaration.
32674 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32675 * ipa-profile.c (ipa_profile): Likewise.
32676 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32677 (ipa_populate_param_decls): Make descriptors parameter garbage
32679 (ipa_dump_param): Adjust to descriptors being a pointer.
32680 (ipa_alloc_node_params): Likewise.
32681 (ipa_initialize_node_params): Likewise.
32682 (load_from_param_1): Make descriptors parameter garbage collected.
32683 (load_from_unmodified_param): Likewise.
32684 (load_from_param): Likewise.
32685 (ipa_load_from_parm_agg): Likewise.
32686 (ipa_node_params::~ipa_node_params): Removed.
32687 (ipa_free_all_node_params): Remove call to delete operator.
32688 (ipa_node_params_t::insert): New.
32689 (ipa_node_params_t::remove): Likewise.
32690 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32691 copy known_csts and known_contexts vectors.
32692 (ipa_read_node_info): Adjust to descriptors being a pointer.
32693 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32695 (ipcp_transform_function): Make descriptors variable garbage
32698 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32700 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32701 * config/i386/avx512dqintrin.h: Ditto.
32702 * config/i386/avx512fintrin.h: Ditto.
32703 * config/i386/i386.c: Handle new builtins.
32704 * config/i386/i386-builtin.def: Add new builtins.
32705 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32706 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32708 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32709 Martin Liska <mliska@suse.cz>
32711 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32712 * asan.c (asan_expand_poison_ifn): Support stores and use
32713 appropriate ASAN report function.
32714 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32715 * internal-fn.def (ASAN_POISON_USE): Declare.
32716 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32717 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32718 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32719 ASAN_POISON calls w/o LHS.
32720 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32721 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32722 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32723 * gimplify.c (asan_poison_variables): Add attribute
32724 use_after_scope_memory to variables that really needs to live
32726 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32727 having the attribute.
32729 2017-01-23 Martin Liska <mliska@suse.cz>
32731 * asan.c (create_asan_shadow_var): New function.
32732 (asan_expand_poison_ifn): Likewise.
32733 * asan.h (asan_expand_poison_ifn): New declaration.
32734 * internal-fn.c (expand_ASAN_POISON): Likewise.
32735 * internal-fn.def (ASAN_POISON): New builtin.
32736 * sanopt.c (pass_sanopt::execute): Expand
32737 asan_expand_poison_ifn.
32738 * tree-inline.c (copy_decl_for_dup_finish): Make function
32740 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32741 * tree-ssa.c (is_asan_mark_p): New function.
32742 (execute_update_addresses_taken): Rewrite local variables
32743 (identified just by use-after-scope as addressable) into SSA.
32745 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32747 * doc/install.texi (Specific): opensource.apple.com uses https
32748 now. Remove trailing slash.
32750 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32752 * README.Portability: Remove note on an Irix compatibility issue.
32754 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32756 * gcov.c (INCLUDE_ALGORITHM): Define.
32757 (INCLUDE_VECTOR): Define.
32758 No longer include <vector> and <algorithm> directly.
32760 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32762 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32764 * doc/invoke.texi (Code Gen Options): Ditto.
32766 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32769 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32771 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32773 rtl-optimization/79125
32774 * cprop.c (local_cprop_pass): Handle cases where we make an
32775 unconditional trap.
32777 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32781 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32782 read from, for big endian.
32784 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32786 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32787 register pauth builtins for LP64 only.
32789 2017-01-20 Marek Polacek <polacek@redhat.com>
32792 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32795 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32797 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32799 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32800 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32801 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32803 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32806 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32807 in big-endian mode when they are not a single duplicated value.
32809 2017-01-20 Richard Biener <rguenther@suse.de>
32811 * BASE-VER: Bump to 7.0.1.
32813 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32815 * omp-low.c (omplow_simd_context): New struct. Use it...
32816 (lower_rec_simd_input_clauses): ...here and...
32817 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32818 references to idx, lane, max_vf, is_simt.
32820 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32822 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32825 2017-01-20 Martin Jambor <mjambor@suse.cz>
32827 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32828 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32830 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32831 (GTFILES): Rename hsa.c to hsa-common.c.
32832 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32833 * hsa-dump.c: Likewise.
32834 * hsa-gen.c: Likewise.
32835 * hsa-regalloc.c: Likewise.
32836 * ipa-hsa.c: Likewise.
32837 * omp-expand.c: Likewise.
32838 * omp-low.c: Likewise.
32839 * toplev.c: Likewise.
32841 2017-01-20 Marek Polacek <polacek@redhat.com>
32844 * doc/invoke.texi: Document -Wduplicated-branches.
32845 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32846 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32847 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32848 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32849 return 0 only when not OEP_LEXICOGRAPHIC.
32850 (fold_build_cleanup_point_expr): Use the expression
32851 location when building CLEANUP_POINT_EXPR.
32852 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32853 * tree.c (add_expr): Handle error_mark_node.
32855 2017-01-20 Martin Liska <mliska@suse.cz>
32858 * tree-profile.c (init_ic_make_global_vars): Do not call
32860 (gimple_init_gcov_profiler): Likewise.
32862 2017-01-20 Martin Liska <mliska@suse.cz>
32865 * cgraph.h (maybe_create_reference): Remove argument and
32867 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32869 * ipa-cp.c (create_specialized_node): Likewise.
32870 * symtab.c (symtab_node::maybe_create_reference): Handle
32871 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32873 2017-01-20 Martin Liska <mliska@suse.cz>
32875 * read-rtl-function.c (function_reader::create_function): Use
32876 build_decl instread of build_decl_stat.
32878 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32880 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32881 * config/i386/avx512dqintrin.h: Ditto.
32882 * config/i386/avx512fintrin.h: Ditto.
32883 * config/i386/i386-builtin-types.def: Add new types.
32884 * config/i386/i386.c: Handle new types.
32885 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32886 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32887 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32888 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32889 (__builtin_ia32_kshiftridi): New.
32890 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32892 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32896 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32897 define to rs6000_init_stack_protect_guard.
32898 (rs6000_init_stack_protect_guard): New function.
32900 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32901 Yunqiang Su <yunqiang.su@imgtec.com>
32903 * config.gcc (supported_defaults): Add madd4.
32904 (with_madd4): Add validation.
32905 (all_defaults): Add madd4.
32906 * config/mips/mips.opt (mmadd4): New option.
32907 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32909 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32911 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32912 (ISA_HAS_FUSED_MADD4): Likewise.
32913 * doc/invoke.texi (-mmadd4): Document the new option.
32914 * doc/install.texi (--with-madd4): Document the new option.
32916 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32918 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32919 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32920 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32921 (aarch64_init_pauth_hint_builtins): New.
32922 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32923 (aarch64_expand_builtin): Expand new builtins.
32925 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32927 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32928 * combine-stack-adj.c (no_unhandled_cfa): Handle
32929 REG_CFA_TOGGLE_RA_MANGLE.
32930 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32931 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32932 info for return address signing.
32933 (aarch64_expand_epilogue): Likewise.
32935 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32937 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32938 * config/aarch64/aarch64-protos.h
32939 (aarch64_return_address_signing_enabled): New declaration.
32940 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32942 (aarch64_expand_prologue): Sign return address before it's pushed onto
32944 (aarch64_expand_epilogue): Authenticate return address fetched from
32946 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32947 (aarch64_attributes): New function attributes for "sign-return-address".
32948 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32949 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32950 ("*do_return"): Generate combined instructions according to key index.
32951 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32952 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32954 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32955 * config/aarch64/aarch64.opt (msign-return-address=): New.
32956 * doc/extend.texi (AArch64 Function Attributes): Documents
32957 "sign-return-address=".
32958 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32960 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32962 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32963 overall option summary.
32965 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32967 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32968 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32969 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32970 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32972 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32974 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32975 -mpower9-minmax by default for -mcpu=power9.
32976 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32977 128-bit floating point.
32979 2017-01-20 Alan Modra <amodra@gmail.com>
32981 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32982 optimizing for size.
32984 2017-01-20 Alan Modra <amodra@gmail.com>
32987 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32988 for strcmp and strncmp from corresponding builtin decl.
32990 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32992 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32993 instead of i386/rtems-64.h.
32994 * config/i386/rtems-64.h: Remove.
32996 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33000 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33002 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33004 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33006 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33007 Change int to HOST_WIDE_INT.
33008 * config/aarch64/aarch64-protos.h
33009 (aarch64_simd_gen_const_vector_dup): Likewise.
33010 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33012 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33014 * langhooks-def.h (lhd_type_for_size): New decl.
33015 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33016 * langhooks.c (lhd_type_for_size): New function, taken from
33019 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33021 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33022 define_bypass for CR latency.
33023 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33024 (power9-alu2): Add define_bypass for CR latency.
33026 (power9-mul): Update insn latency.
33027 (power9-mul-compare): Update insn latency, bypass latency and remove
33030 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33032 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33034 * config/aarch64/aarch64.md
33035 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33036 aarch64_nopcrelative_literal_loads.
33037 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33039 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33041 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33042 TARGET_LOONGSON_3A.
33043 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33045 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33048 * config.gcc (supported_defaults): Add lxc1-sxc1.
33049 (with_lxc1_sxc1): Add validation.
33050 (all_defaults): Add lxc1-sxc1.
33051 * config/mips/mips.opt (mlxc1-sxc1): New option.
33052 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33054 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33055 __mips_no_lxc1_sxc1.
33056 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33057 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33058 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33060 2017-01-19 Richard Biener <rguenther@suse.de>
33062 PR tree-optimization/72488
33063 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33064 sure to restore SSA info.
33065 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33067 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33069 PR rtl-optimization/79121
33070 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33071 of the inner type when shifting an extended value.
33073 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33076 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33077 interposable aliases.
33079 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33082 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33083 Use the evmergelohi instruction.
33084 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33085 (mov_sitf_e500_subreg8_2_be): Likewise.
33086 (mov_sitf_e500_subreg12_2_le): Likewise.
33087 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33088 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33089 (mov_sitf_e500_subreg8_2_le): Likewise.
33090 (mov_sitf_e500_subreg12_2_be): Likewise.
33092 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33094 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33095 attribute from vecsimple to vecperm.
33096 (altivec_vbpermq2): Likewise.
33098 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33101 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33103 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33104 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33105 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33106 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33107 case where N arg is SIZE_MAX.
33108 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33109 (cmpstrsi): Add pattern.
33111 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33114 __builtin_vec_revb builtins.
33115 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33116 built-in functions to support generation of the ISA 3.0 XXBR<x>
33117 vector byte reverse instructions.
33118 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33119 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33120 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33121 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33122 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33123 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33124 (P9V_BUILTIN_VEC_REVB): Likewise.
33125 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33126 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33127 (p9_xxbrq_v16qi): Likewise.
33128 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33129 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33130 (p9_xxbrh_v8hi): Likewise.
33131 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33132 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33133 vec_revb built-in functions.
33135 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33137 PR rtl-optimization/78952
33138 * config/i386/i386.md (any_extract): New code iterator.
33139 (*insvqi_2): Use any_extract for source operand.
33140 (*insvqi_3): Use any_shiftrt for source operand.
33142 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33144 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33146 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33148 2017-01-18 Matthias Klose <doko@ubuntu.com>
33150 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33152 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33154 * config/rs6000/altivec.h (vec_bperm): Change #define.
33155 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33156 (altivec_vbpermq2): New define_insn.
33157 (altivec_vbpermd): Likewise.
33158 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33159 function interface.
33160 (VBPERMD): Likewise.
33161 (VBPERM): New polymorphic function interface.
33162 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33163 Add entries for P9V_BUILTIN_VEC_VBPERM.
33164 * doc/extend.texi: Add interfaces for vec_bperm.
33166 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33168 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33169 first letter of error messages.
33170 (s390_resolve_overloaded_builtin): Likewise.
33171 * config/s390/s390.c (s390_expand_builtin): Likewise.
33172 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33173 (s390_valid_target_attribute_inner_p): Likewise.
33174 * config/s390/s390.md ("tabort"): Likewise.
33176 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33178 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33179 (ISA_AVOID_DIV_HILO): New macro.
33180 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33181 (ISA_HAS_DDIV): Likewise.
33183 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33185 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33187 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33189 * doc/invoke.texi (fabi-version): Spelling fix.
33191 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33194 * doc/invoke.texi (fabi-version): Mention mangling fix for
33197 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33200 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33202 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33205 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33206 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33208 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33210 (stack_protect_test): Ditto.
33211 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33212 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33214 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33215 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33216 -mstack-protector-guard-offset=.
33217 (RS/6000 and PowerPC Options): Ditto.
33219 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33221 * config/i386/i386.h (MASK_CLASS_P): New define.
33222 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33223 there are no registers from different register sets also when
33224 mask registers are used. Update function comment.
33225 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33226 to (*k/*r) and (*k/*km) alternatives.
33228 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33230 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33231 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33232 (EH_RETURN_HANDLER_RTX): New define.
33233 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33234 Force frame pointer in EH return functions.
33235 (aarch64_expand_epilogue): Add barrier for eh_return.
33236 (aarch64_final_eh_return_addr): Remove.
33237 (aarch64_eh_return_handler_rtx): New function.
33238 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33240 (aarch64_eh_return_handler_rtx): New prototype.
33242 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33244 * config/rs6000/altivec.h (vec_rlmi): New #define.
33245 (vec_vrlnm): Likewise.
33246 (vec_rlnm): Likewise.
33247 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33248 (UNSPEC_VRLNM): Likewise.
33249 (VIlong): New mode iterator.
33250 (altivec_vrl<VI_char>mi): New define_insn.
33251 (altivec_vrl<VI_char>nm): Likewise.
33252 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33254 (VRLDNM): Likewise.
33255 (RLNM): New polymorphic function entry.
33256 (VRLWMI): New monomorphic function entry.
33257 (VRLDMI): Likewise.
33258 (RLMI): New polymorphic function entry.
33259 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33260 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33261 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33264 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33267 * dwarf2out.c (field_byte_offset): Restore the
33268 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33269 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33270 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33273 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33276 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33278 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33281 * dwarf2out.c (add_data_member_location_attribute): For constant
33282 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33283 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33284 DW_AT_byte_size attributes.
33286 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33288 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33289 after forcing to constant memory when the code model is medium.
33291 2017-01-17 Julia Koval <julia.koval@intel.com>
33294 * config/i386/avx512fintrin.h
33295 (_mm512_i32gather_ps): Change __addr type to void const*.
33296 (_mm512_mask_i32gather_ps): Ditto.
33297 (_mm512_i32gather_pd): Ditto.
33298 (_mm512_mask_i32gather_pd): Ditto.
33299 (_mm512_i64gather_ps): Ditto.
33300 (_mm512_mask_i64gather_ps): Ditto.
33301 (_mm512_i64gather_pd): Ditto.
33302 (_mm512_mask_i64gather_pd): Ditto.
33303 (_mm512_i32gather_epi32): Ditto.
33304 (_mm512_mask_i32gather_epi32): Ditto.
33305 (_mm512_i32gather_epi64): Ditto.
33306 (_mm512_mask_i32gather_epi64): Ditto.
33307 (_mm512_i64gather_epi32): Ditto.
33308 (_mm512_mask_i64gather_epi32): Ditto.
33309 (_mm512_i64gather_epi64): Ditto.
33310 (_mm512_mask_i64gather_epi64): Ditto.
33311 (_mm512_i32scatter_ps): Change __addr type to void*.
33312 (_mm512_mask_i32scatter_ps): Ditto.
33313 (_mm512_i32scatter_pd): Ditto.
33314 (_mm512_mask_i32scatter_pd): Ditto.
33315 (_mm512_i64scatter_ps): Ditto.
33316 (_mm512_mask_i64scatter_ps): Ditto.
33317 (_mm512_i64scatter_pd): Ditto.
33318 (_mm512_mask_i64scatter_pd): Ditto.
33319 (_mm512_i32scatter_epi32): Ditto.
33320 (_mm512_mask_i32scatter_epi32): Ditto.
33321 (_mm512_i32scatter_epi64): Ditto.
33322 (_mm512_mask_i32scatter_epi64): Ditto.
33323 (_mm512_i64scatter_epi32): Ditto.
33324 (_mm512_mask_i64scatter_epi32): Ditto.
33325 (_mm512_i64scatter_epi64): Ditto.
33326 (_mm512_mask_i64scatter_epi64): Ditto.
33327 * config/i386/avx512pfintrin.h
33328 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33329 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33330 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33331 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33332 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33333 (_mm512_prefetch_i32scatter_ps): Ditto.
33334 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33335 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33336 (_mm512_prefetch_i64scatter_pd): Ditto.
33337 (_mm512_prefetch_i64scatter_ps): Ditto.
33338 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33339 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33340 * config/i386/avx512vlintrin.h
33341 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33342 (_mm_mmask_i32gather_ps): Ditto.
33343 (_mm256_mmask_i32gather_pd): Ditto.
33344 (_mm_mmask_i32gather_pd): Ditto.
33345 (_mm256_mmask_i64gather_ps): Ditto.
33346 (_mm_mmask_i64gather_ps): Ditto.
33347 (_mm256_mmask_i64gather_pd): Ditto.
33348 (_mm_mmask_i64gather_pd): Ditto.
33349 (_mm256_mmask_i32gather_epi32): Ditto.
33350 (_mm_mmask_i32gather_epi32): Ditto.
33351 (_mm256_mmask_i32gather_epi64): Ditto.
33352 (_mm_mmask_i32gather_epi64): Ditto.
33353 (_mm256_mmask_i64gather_epi32): Ditto.
33354 (_mm_mmask_i64gather_epi32): Ditto.
33355 (_mm256_mmask_i64gather_epi64): Ditto.
33356 (_mm_mmask_i64gather_epi64): Ditto.
33357 (_mm256_i32scatter_ps): Change __addr type to void*.
33358 (_mm256_mask_i32scatter_ps): Ditto.
33359 (_mm_i32scatter_ps): Ditto.
33360 (_mm_mask_i32scatter_ps): Ditto.
33361 (_mm256_i32scatter_pd): Ditto.
33362 (_mm256_mask_i32scatter_pd): Ditto.
33363 (_mm_i32scatter_pd): Ditto.
33364 (_mm_mask_i32scatter_pd): Ditto.
33365 (_mm256_i64scatter_ps): Ditto.
33366 (_mm256_mask_i64scatter_ps): Ditto.
33367 (_mm_i64scatter_ps): Ditto.
33368 (_mm_mask_i64scatter_ps): Ditto.
33369 (_mm256_i64scatter_pd): Ditto.
33370 (_mm256_mask_i64scatter_pd): Ditto.
33371 (_mm_i64scatter_pd): Ditto.
33372 (_mm_mask_i64scatter_pd): Ditto.
33373 (_mm256_i32scatter_epi32): Ditto.
33374 (_mm256_mask_i32scatter_epi32): Ditto.
33375 (_mm_i32scatter_epi32): Ditto.
33376 (_mm_mask_i32scatter_epi32): Ditto.
33377 (_mm256_i32scatter_epi64): Ditto.
33378 (_mm256_mask_i32scatter_epi64): Ditto.
33379 (_mm_i32scatter_epi64): Ditto.
33380 (_mm_mask_i32scatter_epi64): Ditto.
33381 (_mm256_i64scatter_epi32): Ditto.
33382 (_mm256_mask_i64scatter_epi32): Ditto.
33383 (_mm_i64scatter_epi32): Ditto.
33384 (_mm_mask_i64scatter_epi32): Ditto.
33385 (_mm256_i64scatter_epi64): Ditto.
33386 (_mm256_mask_i64scatter_epi64): Ditto.
33387 (_mm_i64scatter_epi64): Ditto.
33388 (_mm_mask_i64scatter_epi64): Ditto.
33389 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33390 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33391 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33392 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33393 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33394 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33395 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33396 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33397 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33398 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33399 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33400 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33401 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33402 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33403 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33404 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33405 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33406 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33407 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33408 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33409 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33410 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33411 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33412 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33413 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33414 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33415 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33416 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33417 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33418 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33419 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33420 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33421 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33422 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33423 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33424 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33425 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33426 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33427 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33428 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33429 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33430 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33431 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33432 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33433 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33434 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33435 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33436 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33437 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33438 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33439 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33440 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33441 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33442 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33443 definitions accordingly.
33445 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33446 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33449 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33452 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33455 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33456 hard regs for inner regno.
33458 2017-01-17 Martin Liska <mliska@suse.cz>
33461 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33462 assumption and add comment.
33464 2017-01-17 Nathan Sidwell <nathan@acm.org>
33466 * ipa-visibility.c (localize_node): New function, broken out of ...
33467 (function_and_variable_visibility): ... here. Call it.
33469 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33471 PR middle-end/77445
33472 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33473 correctly set frequency of oudgoing edge.
33474 (duplicate_thread_path): Fix profile updating.
33476 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33479 * configure.ac: Add GCC_BASE_VER.
33480 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33481 version from BASE-VER file.
33482 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33483 (gcc.o): Depend on $(BASEVER).
33484 * common.opt (dumpfullversion): New option.
33485 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33486 * doc/invoke.texi: Document -dumpfullversion.
33487 * doc/install.texi: Document --with-gcc-major-version-only.
33488 * configure: Regenerated.
33490 2017-01-17 Richard Biener <rguenther@suse.de>
33492 PR tree-optimization/71433
33493 * tree-vrp.c (register_new_assert_for): Merge same asserts
33494 on all incoming edges.
33495 (process_assert_insertions_for): Handle insertions at the
33498 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33500 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33501 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33503 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33506 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33509 2017-01-17 Alan Modra <amodra@gmail.com>
33512 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33513 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33514 symbolic stack limit when pic.
33516 2017-01-16 Martin Sebor <msebor@redhat.com>
33518 PR tree-optimization/78608
33519 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33521 2017-01-16 Jeff Law <law@redhat.com>
33524 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33525 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33526 for several include directories that may be relative to sysroot.
33527 * config/i386/x-mingw32 (gplus_includedir): Define.
33528 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33529 (native_system_includedir): Likewise.
33530 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33531 override if TARGET_SYSTEM_ROOT is defined.
33532 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33534 PR tree-optimization/79090
33535 PR tree-optimization/33562
33536 PR tree-optimization/61912
33537 PR tree-optimization/77485
33538 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33539 and computed trims into the dump file.
33541 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33543 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33545 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33548 * gimplify.c (gimplify_init_constructor): If want_value and
33549 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33553 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33554 sequence. Formatting fixes.
33555 (doloop_optimize): Formatting fixes.
33558 * gcc.c (debug_level_greater_than_spec_func): New function.
33559 (static_spec_functions): Add debug-level-gt spec function.
33560 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33562 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33563 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33564 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33565 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33566 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33567 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33569 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33571 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33572 QImode fixups to general and mask registers only.
33574 2017-01-16 Carl Love <cel@us.ibm.com>
33576 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33577 for built-in functions
33578 vector signed char vec_nabs (vector signed char)
33579 vector signed short vec_nabs (vector signed short)
33580 vector signed int vec_nabs (vector signed int)
33581 vector signed long long vec_nabs (vector signed long long)
33582 vector float vec_nabs (vector float)
33583 vector double vec_nabs (vector double)
33584 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33586 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33587 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33588 * doc/extend.texi: Update the documentation file for the new built-in
33591 2017-01-16 Martin Sebor <msebor@redhat.com>
33593 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33596 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33598 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33599 UNSPEC_VSX__XXSPLTD to require special splat handling.
33601 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33604 * system.h: Poison strndup.
33606 2017-01-16 Alan Modra <amodra@gmail.com>
33609 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33612 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33614 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33616 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33618 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33619 call recog here. Assert that INSN_CODE (insn) is non-negative.
33621 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33624 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33626 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33627 in the currently scheduled RTL fragment.
33629 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33631 PR rtl-optimization/78751
33632 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33635 2017-01-14 Jeff Law <law@redhat.com>
33637 PR tree-optimization/79090
33638 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33639 variable length stores.
33640 (compute_trims): Delete dead assignment to *trim_tail.
33641 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33644 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33646 PR rtl-optimization/78626
33647 PR rtl-optimization/78727
33648 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33649 of a block, and split such blocks after everything else is finished.
33651 2017-01-14 Alan Modra <amodra@gmail.com>
33654 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33655 target legitimate_combined_insn.
33656 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33657 (rs6000_legitimate_combined_insn): New function.
33658 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33660 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33661 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33662 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33664 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33666 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33668 2017-01-13 Jeff Law <law@redhat.com>
33670 PR tree-optimization/33562
33671 PR tree-optimization/61912
33672 PR tree-optimization/77485
33673 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33675 (delete_dead_assignment): Likewise.
33676 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33677 statement to delete_dead_call and delete_dead_assignment.
33679 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33682 * substring-locations.c (format_warning_va): Strengthen case 1 so
33683 that both endpoints of the substring must be within the format
33684 range for just the substring to be printed.
33686 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33688 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33689 * config/i386/i386.c (ix86_target_string): Add missing options
33690 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33691 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33692 flags_other and ix86_target_other to flags2_other. Display unknown
33694 (ix86_valid_target_attribute_inner_p): Add missing options and
33695 reorder options by implied ISAs, as in ix86_target_string.
33697 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33699 * hash-table.h (hash_table::too_empty_p): New function.
33700 (hash_table::expand): Use it.
33701 (hash_table::traverse): Likewise.
33702 (hash_table::empty_slot): Use sizeof (value_type) instead of
33703 sizeof (PTR) to convert bytes to elements. Shrink the table
33704 if the current size is excessive for the current number of
33707 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33709 * ira-costs.c (record_reg_classes): Break from the inner loop
33710 early once alt_fail is known to be true. Update outer loop
33711 handling accordingly.
33713 2017-01-13 Jeff Law <law@redhat.com>
33715 * tree-ssa-dse.c (decrement_count): New function.
33716 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33717 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33718 when we know the partially dead statement is a mem* function.
33720 PR tree-optimization/61912
33721 PR tree-optimization/77485
33722 * tree-ssa-dse.c: Include expr.h.
33723 (maybe_trim_constructor_store): New function.
33724 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33726 PR tree-optimization/33562
33727 PR tree-optimization/61912
33728 PR tree-optimization/77485
33729 * doc/invoke.texi: Document new dse-max-object-size param.
33730 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33731 * tree-ssa-dse.c: Include params.h.
33732 (dse_store_status): New enum.
33733 (initialize_ao_ref_for_dse): New, partially extracted from
33735 (valid_ao_ref_for_dse, normalize_ref): New.
33736 (setup_live_bytes_from_ref, compute_trims): Likewise.
33737 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33738 (maybe_trim_partially_dead_store): Likewise.
33739 (maybe_trim_complex_store): Likewise.
33740 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33741 Track what bytes live from the original store. Return tri-state
33742 for dead, partially dead or live.
33743 (dse_dom_walker): Add constructor, destructor and new private members.
33744 (delete_dead_call, delete_dead_assignment): New extracted from
33746 (dse_optimize_stmt): Make a member of dse_dom_walker.
33747 Use initialize_ao_ref_for_dse.
33749 PR tree-optimization/33562
33750 PR tree-optimization/61912
33751 PR tree-optimization/77485
33752 * sbitmap.h (bitmap_count_bits): Prototype.
33753 (bitmap_clear_range, bitmap_set_range): Likewise.
33754 * sbitmap.c (bitmap_clear_range): New function.
33755 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33757 2017-01-13 Martin Liska <mliska@suse.cz>
33760 * function.c (set_cfun): Add new argument force.
33761 * function.h (set_cfun): Likewise.
33762 * ipa-inline-transform.c (inline_call): Use the function when
33763 strict alising from is dropped for function we inline to.
33765 2017-01-13 Richard Biener <rguenther@suse.de>
33767 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33768 for dumping GIMPLE INTEGER_CSTs.
33770 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33772 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33773 to 201112L since C++17.
33775 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33778 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33779 if -fsanitize=kernel-address is present.
33781 2017-01-13 Richard Biener <rguenther@suse.de>
33783 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33784 as _Literal ( type ) number in case usual suffixes do not
33785 preserve all information.
33787 2017-01-13 Richard Biener <rguenther@suse.de>
33789 PR tree-optimization/77283
33790 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33791 and ssa-iterators.h.
33792 (is_feasible_trace): Implement a cost model based on joiner
33795 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33798 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33799 char or short to __float128/_Float128 directly.
33801 2017-01-12 Martin Sebor <msebor@redhat.com>
33803 to -Wformat-overflow.
33804 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33805 (min_bytes_remaining): Same.
33806 (get_string_length): Same.
33807 (format_string): Same.
33808 (format_directive): Same.
33810 (pass_sprintf_length::handle_gimple_call): Same.
33812 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33814 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33815 info.nowrite calls with no lhs that can't throw. Return bool
33816 whether gsi_remove has been called or not.
33817 (pass_sprintf_length::handle_gimple_call): Return bool whether
33818 try_substitute_return_value called gsi_remove. Formatting fix.
33819 (pass_sprintf_length::execute): Don't use gsi_remove if
33820 handle_gimple_call returned true.
33823 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33824 be removed due to side-effects, don't remove following barrier nor
33825 turn the successor edge into fallthru edge.
33827 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33830 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33831 element-reversing loads and stores as not swappable.
33833 2017-01-12 Nathan Sidwell <nathan@acm.org>
33834 Nicolai Stange <nicstange@gmail.com>
33836 * combine.c (try_combine): Don't ignore result of overlap checking
33837 loop. Combine overlap & asm check into single loop.
33839 2017-01-12 Richard Biener <rguenther@suse.de>
33841 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33842 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33844 2017-01-12 Richard Biener <rguenther@suse.de>
33846 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33847 and TS_TARGET_OPTION directly derive from TS_BASE.
33848 * tree-core.h (tree_optimization_option): Derive from tree_base.
33849 (tree_target_option): Likewise.
33851 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33853 * config/i386/i386.c (memory_address_length): Increase len
33854 only when rip_relative_addr_p returns false.
33856 2017-01-11 Julia Koval <julia.koval@intel.com>
33858 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33859 (OPTION_MASK_ISA_SGX_SET): New.
33860 (ix86_handle_option): Handle OPT_msgx.
33861 * config.gcc: Added sgxintrin.h.
33862 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33863 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33864 * config/i386/i386.c (ix86_target_string): Add -msgx.
33866 (ix86_option_override_internal): Handle new options.
33867 (ix86_valid_target_attribute_inner_p): Add sgx.
33868 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33869 * config/i386/i386.opt: Add msgx.
33870 * config/i386/sgxintrin.h: New file.
33871 * config/i386/x86intrin.h: Add sgxintrin.h.
33873 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33876 * fold-const.c (maybe_nonzero_address): Return 1 for function
33878 (tree_single_nonzero_warnv_p): Don't handle function local objects
33882 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33885 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33888 * opts.c: Include "spellcheck.h"
33889 (struct string_fragment): New struct.
33890 (struct edit_distance_traits<const string_fragment &>): New
33892 (get_closest_sanitizer_option): New function.
33893 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33895 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33897 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33899 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33900 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33901 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33902 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33903 for initial die_offset if dwarf_split_debug_info.
33904 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33905 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33907 (output_skeleton_debug_sections): Formatting fix. Use
33908 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33909 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33911 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33913 * config/arm/cortex-a53.md: Add bypasses for
33914 cortex_a53_r2f_cvt.
33915 (cortex_a53_r2f): Only use for transfers.
33916 (cortex_a53_f2r): Likewise.
33917 (cortex_a53_r2f_cvt): Add reservation for conversions.
33918 (cortex_a53_f2r_cvt): Likewise.
33920 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33922 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33923 to all inlined functions, change static to extern.
33925 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33928 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33930 (arm_assemble_integer): Likewise.
33932 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33934 * config.gcc: Use new awk script to check CPU, FPU and architecture
33935 parameters for --with-... options.
33936 * config/arm/parsecpu.awk: New file
33937 * config/arm/arm-cpus.in: New file.
33938 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33940 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33942 * config/arm/t-arm: Update dependency rules.
33943 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33944 of processing .def files.
33945 * config/arm/genopt.sh: Deleted.
33946 * config/arm/gentune.sh: Deleted.
33947 * config/arm/arm-cores.def: Deleted.
33948 * config/arm/arm-arches.def: Deleted.
33949 * config/arm/arm-fpus.def: Deleted.
33950 * config/arm/arm-tune.md: Regenerated.
33951 * config/arm/arm-tables.opt: Regenerated.
33952 * config/arm/arm-cpu.h: New generated file.
33953 * config/arm/arm-cpu-data.h: New generated file.
33954 * config/arm/arm-cpu-cdata.h: New generated file.
33956 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33959 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33961 (input_varpool_node): Unpack dynamically_initialized bit.
33963 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33965 PR rtl-optimization/79032
33966 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33967 the alignment of the adjusted memory reference against that of MODE,
33968 instead of the alignment of the original memory reference.
33970 2017-01-11 Martin Jambor <mjambor@suse.cz>
33972 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33974 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33975 decorated functions.
33977 2017-01-11 Richard Biener <rguenther@suse.de>
33979 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33980 set range/nonnull info for PHI results. Do not set it on
33981 stmts marked for removal.
33983 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33985 * expr.c (store_field): In the bitfield case, fetch the return value
33986 from the registers before applying a single big-endian adjustment.
33987 Always do a final load for a BLKmode value not larger than a word.
33989 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33992 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33993 that we correctly handle column numbers greater than
33994 LINE_MAP_MAX_COLUMN_NUMBER.
33996 2017-01-10 Martin Sebor <msebor@redhat.com>
33998 PR middle-end/78245
33999 * gimple-ssa-sprintf.c (get_destination_size): Call
34000 {init,fini}object_sizes.
34001 * tree-object-size.c (addr_object_size): Adjust.
34002 (pass_through_call): Adjust.
34003 (pass_object_sizes::execute): Adjust.
34004 * tree-object-size.h (fini_object_sizes): Declare.
34006 2017-01-10 Martin Sebor <msebor@redhat.com>
34008 PR tree-optimization/78775
34009 * builtins.c (get_size_range): Move...
34010 * calls.c: ...to here.
34011 (alloc_max_size): Accept zero argument.
34012 (operand_signed_p): Remove.
34013 (maybe_warn_alloc_args_overflow): Call get_size_range.
34014 * calls.h (get_size_range): Declare.
34016 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34018 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34019 from TI's devices.csv file as of September 2016.
34020 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34022 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34024 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34025 * doc/invoke.texi: Likewise.
34026 * doc/md.texi: Likewise.
34027 * doc/objc.texi: Likewise.
34029 2017-01-10 Joshua Conner <joshconner@google.com>
34031 * config/arm/fuchsia-elf.h: New file.
34032 * config/fuchsia.h: New file.
34033 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34034 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34036 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34038 2016-01-10 Richard Biener <rguenther@suse.de>
34040 PR tree-optimization/79034
34041 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34042 Propagate out degenerate PHIs in the joiner.
34044 2017-01-10 Martin Liska <mliska@suse.cz>
34046 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34047 (sort_congruence_classes_by_decl_uid): Likewise.
34048 (sort_congruence_class_groups_by_decl_uid): Likewise.
34049 (sem_item_optimizer::merge_classes): Sort class, groups in these
34050 classes and members in the groups by DECL_UID of declarations.
34051 This would make merge operations stable.
34053 2017-01-10 Martin Liska <mliska@suse.cz>
34055 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34056 usage of m_classes_vec.
34057 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34058 (sem_item_optimizer::get_group_by_hash): Likewise.
34059 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34060 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34061 (sem_item_optimizer::verify_classes): Likewise.
34062 (sem_item_optimizer::process_cong_reduction): Likewise.
34063 (sem_item_optimizer::dump_cong_classes): Likewise.
34064 (sem_item_optimizer::merge_classes): Likewise.
34065 * ipa-icf.h (congruence_class_hash): Rename from
34066 congruence_class_group_hash. Remove declaration of m_classes_vec.
34068 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34070 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34071 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34072 * config.gcc: Add avx512vpopcntdqintrin.h.
34073 * config/i386/avx512vpopcntdqintrin.h: New.
34074 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34075 * config/i386/i386-builtin-types.def: Add new types.
34076 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34077 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34078 __builtin_ia32_vpopcountq_v8di_mask): New.
34079 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34080 __AVX512VPOPCNTDQ__.
34081 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34082 (PTA_AVX512VPOPCNTDQ): Define.
34083 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34084 TARGET_AVX512VPOPCNTDQ_P): Define.
34085 * config/i386/i386.opt: Add mavx512vpopcntdq.
34086 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34087 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34089 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34091 PR middle-end/77484
34092 * predict.def (PRED_CALL): Set to 67.
34094 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34096 * expr.c (store_field): In the bitfield case, if the value comes from
34097 a function call and is of an aggregate type returned in registers, do
34098 not modify the field mode; extract the value in all cases if the mode
34099 is BLKmode and the size is not larger than a word.
34101 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34104 * config/i386/cpuid.h: Fix undefined behavior.
34106 2017-01-04 Jeff Law <law@redhat.com>
34108 PR tree-optimization/79007
34109 PR tree-optimization/67955
34110 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34111 conservative for pt.null when flag_non_call_exceptions is on.
34113 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34115 PR translation/79019
34116 PR translation/79020
34117 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34118 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34119 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34121 * config/avr/avr.opt (maccumulate-args): Likewise.
34122 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34123 * common.opt (freport-bug): Likewise.
34124 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34125 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34126 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34127 translatable string.
34128 * config/i386/i386.c (function_value_32): Likewise.
34129 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34130 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34132 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34133 * common/config/msp430/msp430-common.c (msp430_handle_option):
34135 * symtab.c (symtab_node::verify_base): Likewise.
34136 * opts.c (set_debug_level): Likewise.
34137 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34139 missing whitespace to translatable strings.
34140 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34141 * config/sh/superh.h: Likewise.
34142 * config/i386/xopintrin.h: Likewise.
34143 * config/i386/znver1.md: Likewise.
34144 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34145 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34146 * double-int.h (struct double_int): Likewise.
34147 * double-int.c (div_and_round_double): Likewise.
34148 * wide-int.cc: Likewise.
34149 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34150 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34151 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34152 (crossjumps_occurred): ... this.
34153 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34156 PR tree-optimization/78899
34157 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34158 returning bool return struct loop *, NULL for failure and the new
34160 (versionable_outer_loop_p): Don't version outer loop if it has
34161 dont_vectorized bit set.
34162 (tree_if_conversion): When versioning outer loop, ensure
34163 tree_if_conversion is performed also on the inner loop of the
34164 non-vectorizable outer loop copy.
34165 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34166 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34167 prevent vectorization of it.
34168 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34169 the outer loop vectorization of the non-scalar version is attempted
34170 before vectorization of the inner loop in scalar version. If
34171 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34172 vectorization of its inner loop.
34173 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34174 has 2 inner loops, rename also on edges from bb whose single pred
34175 is outer_loop->header. Fix typo in function comment.
34177 2017-01-09 Martin Sebor <msebor@redhat.com>
34180 * asan.c (asan_emit_stack_protection): Increase local buffer size
34181 to avoid snprintf truncation warning.
34183 2017-01-09 Andrew Pinski <apinski@cavium.com>
34185 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34186 to reference thunderx2t99 for the tuning structure
34187 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34189 (thunderx2t99_extra_costs): This.
34190 * config/aarch64/aarch64-tune.md: Regenerate.
34191 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34192 (vulcan_addrcost_table): This.
34193 (vulcan_regmove_cost): Rename to ...
34194 (thunderx2t99_regmove_cost): This.
34195 (vulcan_vector_cost): Rename to ...
34196 (thunderx2t99_vector_cost): this.
34197 (vulcan_branch_cost): Rename to ...
34198 (thunderx2t99_branch_cost): This.
34199 (vulcan_tunings): Rename to ...
34200 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34201 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34203 2017-01-09 Martin Jambor <mjambor@suse.cz>
34207 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34208 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34209 (propagate_vr_accross_jump_function): Use the above function for all
34210 value range computations for pass-through jump functions and type
34211 converasion from explicit value range values.
34212 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34213 parameters from TYPE_ARG_TYPES.
34214 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34215 (ipa_write_node_info): Stream type of the actual argument.
34216 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34218 2017-01-09 Martin Liska <mliska@suse.cz>
34221 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34222 (lookup_compiler): Do not show error message with have_E.
34224 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34226 PR tree-optimization/78938
34227 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34228 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34229 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34230 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34233 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34235 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34238 2017-01-09 Richard Biener <rguenther@suse.de>
34240 PR tree-optimization/78997
34241 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34242 name condition properly.
34244 2017-01-09 Richard Biener <rguenther@suse.de>
34247 * dwarf2out.c (is_cxx): New overload with context.
34248 (is_naming_typedef_decl): Use it.
34250 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34252 * invoke.texi (Option Summary): Correct spacing in option lists
34253 and add line breaks to fix over-long lines.
34255 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34257 PR middle-end/17660
34259 * extend.texi (Common Variable Attributes): Add xref to GCC
34260 Internals manual to explain mode attribute keywords.
34262 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34265 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34266 and Preprocessor Options.
34267 (Options for Linking): Document -pthread here....
34268 (RS/6000 and PowerPC Options): ...not here.
34269 (Solaris 2 Options): ...or here.
34270 * doc/cppopts.texi: Document -pthread.
34272 2017-01-08 Martin Sebor <msebor@redhat.com>
34274 PR middle-end/77708
34275 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34276 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34277 New member functions.
34278 (format_directive): Used them.
34280 (pass_sprintf_length::handle_gimple_call): Same.
34281 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34282 to avoid truncation for any argument.
34283 (extract_affine_mul): Same.
34284 * tree.c (get_file_function_name): Same.
34286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34288 PR middle-end/77484
34289 * predict.def (PRED_INDIR_CALL): Set to 86.
34291 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34293 PR preprocessor/54124
34294 * doc/cppopts.texi: Reformat -d subtable to list the full name
34295 of the options. Add cross-reference to the docs for the general
34296 compiler -d options.
34297 * doc/invoke.texi (Developer Options): Add cross-reference to the
34298 preprocessor-specific -d option documentation.
34300 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34302 PR preprocessor/13498
34303 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34304 redudant material, and reflect new command-line options.
34305 (System Headers): Likewise.
34307 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34309 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34310 -isystem, and -idirafter. Copy-edit.
34311 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34312 default for -ftrack-macro-expansion. Delete obsolete and
34313 badly-formatted implementation details about -fdebug-cpp output.
34314 * doc/cppwarnopts.texi: Copy-edit.
34316 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34319 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34320 that the transition from a max line width >= 1<<10 to narrower
34321 lines works correctly.
34323 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34325 * doc/options.texi (PerFunction): New.
34326 * opt-functions.awk (switch_flags): Map both Optimization and
34327 PerFunction to CL_OPTIMIZATION.
34328 * opth-gen.awk: Test for PerFunction flag along with
34330 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34331 it only when the latter is present. Skip those that don't in
34332 the hash function generator.
34333 * common.opt (fvar-tracking): Mark as PerFunction instead of
34335 (fvar-tracking-assignments): Likewise.
34336 (fvar-tracking-assignments-toggle): Likewise.
34337 (fvar-tracking-uninit): Likewise.
34339 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34341 PR translation/79018
34342 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34345 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34348 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34349 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34350 TARGET_LONG_JUMP_TABLE_OFFSETS.
34351 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34352 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34353 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34354 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34355 * config/m68k/m68k.md (tablejump expander): Likewise.
34356 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34357 TARGET_LONG_JUMP_TABLE_OFFSETS.
34358 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34359 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34361 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34362 David Holsgrove <david.holsgrove@xilinx.com>
34364 * common/config/microblaze/microblaze-common.c
34365 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34366 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34368 * config/microblaze/microblaze.c (microblaze_must_save_register)
34369 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34371 (microblaze_eh_return): New function.
34372 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34373 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34374 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34375 * config/microblaze/microblaze.md (eh_return): New pattern.
34377 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34379 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34380 GCC_DIAGNOSTIC_STRINGIFY): Define.
34382 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34384 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34386 * config/arm/arm.md (<mcrr>): New.
34388 * config/arm/arm.c (arm_arch5te): New.
34389 (arm_option_override): Set arm_arch5te.
34390 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34392 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34393 (arm_mcrr_qualifiers): ... this. New.
34394 (MRRC_QUALIFIERS): Define to...
34395 (arm_mrrc_qualifiers): ... this. New.
34396 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34398 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34399 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34400 (MRRCI, mrrc, MRRC): New.
34401 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34402 VUNSPEC_MRRC2): New.
34404 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34406 * config/arm/arm.md (<mcr>): New.
34408 * config/arm/arm.c (arm_coproc_builtin_available): Add
34409 support for mcr, mrc, mcr2 and mrc2.
34410 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34411 (arm_mcr_qualifiers): ... this. New.
34412 (MRC_QUALIFIERS): Define to ...
34413 (arm_mrc_qualifiers): ... this. New.
34414 (MCR_QUALIFIERS): Define to ...
34415 (arm_mcr_qualifiers): ... this. New.
34416 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34418 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34419 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34420 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34421 VUNSPEC_MRC2): New.
34423 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34425 * config/arm/arm.md (*ldc): New.
34429 * config/arm/arm.c (arm_coproc_builtin_available): Add
34430 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34431 (arm_coproc_ldc_stc_legitimate_address): New.
34432 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34433 'qualifier_const_pointer'.
34434 (LDC_QUALIFIERS): Define to...
34435 (arm_ldc_qualifiers): ... this. New.
34436 (STC_QUALIFIERS): Define to...
34437 (arm_stc_qualifiers): ... this. New.
34438 * config/arm/arm-protos.h
34439 (arm_coproc_ldc_stc_legitimate_address): New.
34440 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34441 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34442 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34443 stc2, stcl, stc2l): New.
34444 * config/arm/constraints.md (Uz): New.
34445 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34446 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34447 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34448 VUNSPEC_STC2L): New.
34450 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34452 * config/arm/arm.md (<cdp>): New.
34453 * config/arm/arm.c (neon_const_bounds): Rename this ...
34454 (arm_const_bounds): ... this.
34455 (arm_coproc_builtin_available): New.
34456 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34457 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34458 (CDP_QUALIFIERS): Define to...
34459 (arm_cdp_qualifiers): ... this. New.
34461 (arm_expand_builtin_args): Add case for 6 arguments.
34462 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34463 (arm_const_bounds): ... this.
34464 (arm_coproc_builtin_available): New.
34465 * config/arm/arm_acle.h (__arm_cdp): New.
34467 * config/arm/arm_acle_builtins.def (cdp): New.
34469 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34470 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34471 'arm_const_bounds'.
34472 * config/arm/types.md (coproc): New.
34473 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34474 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34475 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34476 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34478 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34480 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34481 (UBINOP_QUALIFIERS): New.
34483 (acle_builtin_data): New. Change comment.
34484 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34485 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34486 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34487 arm_acle_builtins.def.
34488 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34489 (arm_init_acle_builtins): New.
34490 (CRC32_BUILTIN): Remove.
34491 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34492 crc32cb, crc32ch and crc32cw.
34493 (arm_init_crc32_builtins): Remove.
34494 (arm_init_builtins): Use arm_init_acle_builtins rather
34495 than arm_init_crc32_builtins.
34496 (arm_expand_acle_builtin): New.
34497 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34498 * config/arm/arm_acle_builtins.def: New.
34500 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34502 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34503 (arm_builtin_datum): ... this.
34504 (arm_init_neon_builtin): Rename to ...
34505 (arm_init_builtin): ... this. Add a new parameters PREFIX
34506 and USE_SIG_IN_NAME.
34507 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34508 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34509 'arm_builtin_datum'.
34510 (arm_init_vfp_builtins): Likewise.
34511 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34512 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34513 (arm_expand_neon_args): Rename to ...
34514 (arm_expand_builtin_args): ... this. Rename builtin_arg
34515 enum values and differentiate between ARG_BUILTIN_MEMORY
34516 and ARG_BUILTIN_NEON_MEMORY.
34517 (arm_expand_neon_builtin_1): Rename to ...
34518 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34519 values, arm_expand_builtin_args and add bool parameter NEON.
34520 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34521 (arm_expand_vfp_builtin): Likewise.
34522 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34524 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34526 PR middle-end/77484
34527 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34528 * predict.c (tree_estimate_probability_bb): Reverse direction of
34529 polymorphic call predictor.
34531 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34533 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34534 (determine_pass_name_match): ...this new function and...
34535 (should_skip_pass_p): ...this new function.
34537 2017-01-06 Nathan Sidwell <nathan@acm.org>
34539 * ipa-visibility.c (function_and_variable_visibility): Reformat
34540 comments and long lines. Remove extrneous if.
34541 * symtab.c (symtab_node::make_decl_local): Fix code format.
34542 (symtab_node::set_section_for_node): Fix comment typo.
34544 2017-01-06 Martin Liska <mliska@suse.cz>
34547 * lra-constraints.c: Rename invariant to lra_invariant.
34548 * predict.c (set_even_probabilities): Initialize e to NULL.
34550 2017-01-05 Martin Sebor <msebor@redhat.com>
34552 PR tree-optimization/78910
34553 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34554 (format_integer): Correct off-by-one error in the handling
34555 of precision with negative numbers in signed conversions..
34557 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34559 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34561 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34563 PR tree-optimization/71016
34564 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34565 factor_out_conditional_conversion. Formatting fix.
34566 (factor_out_conditional_conversion): Add cond_stmt argument.
34567 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34568 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34571 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34573 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34574 read-rtl-function.o, and selftest-rtl.o.
34575 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34576 (selftest::aarch64_test_loading_full_dump): New function.
34577 (selftest::aarch64_run_selftests): New function.
34578 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34579 selftest::aarch64_run_selftests.
34580 * config/i386/i386.c
34581 (selftest::ix86_test_loading_dump_fragment_1): New function.
34582 (selftest::ix86_test_loading_call_insn): New function.
34583 (selftest::ix86_test_loading_full_dump): New function.
34584 (selftest::ix86_test_loading_unspec): New function.
34585 (selftest::ix86_run_selftests): Call the new functions.
34586 * emit-rtl.c (maybe_set_max_label_num): New function.
34587 * emit-rtl.h (maybe_set_max_label_num): New decl.
34588 * function.c (instantiate_decls): Guard call to
34589 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34590 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34592 * gensupport.c (gen_reader::gen_reader): Pass "false"
34593 for new "compact" param of rtx_reader.
34594 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34595 rather than an empty string for NULL strings.
34596 * read-md.c: Potentially include config.h rather than bconfig.h.
34597 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34598 (have_error): New global, copied from errors.c.
34599 (md_reader::read_name): Rename to...
34600 (md_reader::read_name_1): ...this, adding "out_loc" param,
34601 and converting "missing name or number" to returning false, rather
34603 (md_reader::read_name): Reimplement in terms of read_name_1.
34604 (md_reader::read_name_or_nil): New function.
34605 (md_reader::read_string): Handle "(nil)" by returning NULL.
34606 (md_reader::md_reader): Add new param "compact".
34607 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34608 (md_reader::read_file): New method.
34609 * read-md.h (md_reader::md_reader): Add new param "compact".
34610 (md_reader::read_file): New method.
34611 (md_reader::is_compact): New accessor.
34612 (md_reader::read_name): Convert return type from void to file_location.
34613 (md_reader::read_name_or_nil): New decl.
34614 (md_reader::read_name_1): New decl.
34615 (md_reader::m_compact): New field.
34616 (noop_reader::noop_reader): Pass "false" for new "compact" param
34618 (rtx_reader::rtx_reader): Add new "compact" param.
34619 (rtx_reader::read_rtx_operand): Make virtual and convert return
34620 type from void to rtx.
34621 (rtx_reader::read_until): New decl.
34622 (rtx_reader::handle_any_trailing_information): New virtual function.
34623 (rtx_reader::postprocess): New virtual function.
34624 (rtx_reader::finalize_string): New virtual function.
34625 (rtx_reader::m_in_call_function_usage): New field.
34626 (rtx_reader::m_reuse_rtx_by_id): New field.
34627 * read-rtl-function.c: New file.
34628 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34629 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34630 (selftest::verify_three_block_rtl_cfg): New decl.
34631 * read-rtl-function.h: New file.
34632 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34633 For host, include function.h, memmodel.h, and emit-rtl.h.
34634 (one_time_initialization): New function.
34635 (struct compact_insn_name): New struct.
34636 (compact_insn_names): New array.
34637 (find_code): Handle insn codes in compact dumps.
34638 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34639 (bind_subst_iter_and_attr): Likewise.
34640 (add_condition_to_string): Likewise.
34641 (add_condition_to_rtx): Likewise.
34642 (apply_attribute_uses): Likewise.
34643 (add_current_iterators): Likewise.
34644 (apply_iterators): Likewise.
34645 (initialize_iterators): Guard usage of apply_subst_iterator with
34646 #ifdef GENERATOR_FILE.
34647 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34648 (md_reader::read_mapping): Likewise.
34649 (add_define_attr_for_define_subst): Likewise.
34650 (add_define_subst_attr): Likewise.
34651 (read_subst_mapping): Likewise.
34652 (check_code_iterator): Likewise.
34653 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34655 (one_time_initialization): New function.
34656 (rtx_reader::read_until): New method.
34657 (read_flags): New function.
34658 (parse_reg_note_name): New function.
34659 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34660 Handle reuse_rtx ids.
34661 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34662 Add parsing support for RTL dumps, mirroring the special-cases in
34663 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34664 values, and calling handle_any_trailing_information.
34665 (rtx_reader::read_rtx_operand): Convert return type from void
34666 to rtx, returning return_rtx. Handle case 'e'. Call
34667 finalize_string on XSTR and XTMPL fields.
34668 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34669 "(nil)" values were omitted. Call the postprocess vfunc on the
34671 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34672 class ctor. Initialize m_in_call_function_usage. Call
34673 one_time_initialization.
34674 * rtl-tests.c (selftest::test_uncond_jump): Call
34675 set_new_first_and_last_insn.
34676 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34677 * selftest-rtl.c: New file.
34678 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34679 (selftest::get_insn_by_uid): New decl.
34680 * selftest-run-tests.c (selftest::run_tests): Call
34681 read_rtl_function_c_tests.
34682 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34683 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34686 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34688 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34689 operands in a special way. Assert that pos+len <= mode precision.
34691 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34693 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34694 3 argument Alias with unlimited for the negative form.
34695 (fno-vect-cost-model): Removed.
34697 2017-01-05 Martin Liska <mliska@suse.cz>
34699 * hsa-gen.c (gen_hsa_divmod): New function.
34700 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34702 2017-01-05 Martin Liska <mliska@suse.cz>
34705 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34708 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34710 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34711 small constant length operands.
34713 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34715 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34716 between loop iterations.
34718 2017-01-05 Martin Liska <mliska@suse.cz>
34721 * gimplify.c (gimplify_decl_expr): Compare to
34722 asan_poisoned_variables instread of checking flags.
34723 (gimplify_target_expr): Likewise.
34724 (gimplify_expr): Likewise.
34725 (gimplify_function_tree): Conditionally initialize
34726 asan_poisoned_variables.
34728 2017-01-04 Jeff Law <law@redhat.com>
34730 PR tree-optimizatin/78812
34731 * rtl.h (contains_mem_rtx_p): Prototype.
34732 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34733 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34734 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34735 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34736 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34738 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34740 * input.c (assert_char_at_range): Default-initialize actual_range.
34742 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34744 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34745 to match the caller.
34747 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34749 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34750 insns after final jump in test to emit dummy move.
34752 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34754 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34755 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34757 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34759 * multiple_target.c (create_dispatcher_calls): Init e_next.
34760 * tree-ssa-loop-split.c (split_loop): Init border.
34761 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34764 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34769 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34770 (altivec_register_operand): Do not return true if the operand
34771 contains a SUBREG mixing SImode and SFmode.
34772 (vsx_register_operand): Likewise.
34773 (vsx_reg_sfsubreg_ok): New predicate.
34774 (vfloat_operand): Do not return true if the operand contains a
34775 SUBREG mixing SImode and SFmode.
34776 (vint_operand): Likewise.
34777 (vlogical_operand): Likewise.
34778 (gpc_reg_operand): Likewise.
34779 (int_reg_operand): Likewise.
34780 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34781 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34782 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34784 (rs6000_emit_move_si_sf_subreg): New helper function.
34785 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34786 fixup SUBREGs involving SImode and SFmode.
34787 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34788 numbers for the new peephole2 optimization.
34789 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34790 the GLIBC math library that do AND/IOR/XOR operations on single
34791 precision floating point.
34792 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34793 target macros to say whether we need to avoid SUBREGs mixing
34795 (TARGET_ALLOW_SF_SUBREG): Likewise.
34796 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34797 (UNSPEC_SI_FROM_SF): Likewise.
34798 (iorxor): Change spacing.
34799 (and_ior_xor): New iterator for AND, IOR, and XOR.
34800 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34801 (movdi_from_sf_zero_ext): Likewise.
34802 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34803 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34804 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34805 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34806 (fms<mode>4): Likewise.
34807 (fnma<mode>4): Likewise.
34808 (fnms<mode>4): Likewise.
34809 (nfma<mode>4): Likewise.
34810 (nfms<mode>4): Likewise.
34812 2017-01-04 Marek Polacek <polacek@redhat.com>
34815 * doc/invoke.texi: Document -Wpointer-compare.
34817 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34819 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34822 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34823 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34826 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34829 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34830 documentation of the powerpc_popcntb_ok attribute.
34831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34832 code to issue warning messages if a requested CPU configuration is
34833 not supported by the binary (assembler and loader) toolchain.
34834 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34835 made to define a built-in function that has been disabled.
34836 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34837 made to define a built-in function that has been disabled.
34838 (altivec_init_builtins): Add comment explaining why definition
34839 of the DST built-in functions is not preceded by an assertion
34840 check. Add assertions to prevent ICE if attempts are made to
34841 define an altivec predicate or an abs* built-in function that has
34843 (htm_init_builtins): Add comment explaining why definition of the
34844 htm built-in functions is not preceded by an assertion check.
34846 2017-01-04 Jeff Law <law@redhat.com>
34848 PR tree-optimizatin/67955
34849 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34850 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34851 the points-to solution does not include pt_null. Use DECL_PT_UID
34854 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34856 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34857 Use gen_int_mode instead of gen_lopwart for const_int operands.
34859 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34861 PR tree-optimization/71563
34862 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34863 out of range value - has low bits known to be zero.
34865 2017-01-04 Alan Modra <amodra@gmail.com>
34867 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34868 * configure: Regenerate.
34869 * config.in: Regenerate.
34871 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34874 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34875 a substring of the message, but strcmp with the whole message. Ifdef
34876 ENABLE_NLS, translate the message first using dgettext.
34878 2017-01-03 Jeff Law <law@redhat.com>
34880 PR tree-optimizatin/78856
34881 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34882 (mark_threaded_blocks): Remove code to truncate thread paths that
34883 cross multiple loop headers. Instead invalidate the cached loop
34884 iteration information and handle case of a thread path walking
34885 into an irreducible region.
34887 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34890 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34891 assertions. Add support for doing the signbit if the IEEE 128-bit
34892 floating point value is in a GPR.
34893 * config/rs6000/rs6000.md (Fsignbit): Delete.
34894 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34895 Update the length attribute if the value is in a GPR.
34896 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34897 the sign or zero extension instruction, since the value is always 0/1.
34898 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34901 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34902 extracting SImode to a GPR register so that we can generate a
34903 store, limit the vector to be in a traditional Altivec register
34904 for the vextuwrx instruction.
34906 2017-01-03 Ian Lance Taylor <iant@google.com>
34908 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34910 2017-01-03 Martin Sebor <msebor@redhat.com>
34912 PR tree-optimization/78696
34913 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34914 precision. Use MPFR for %f for greater fidelity. Correct handling
34916 (pass_sprintf_length::compute_format_length): Set width and precision
34917 specified by asrerisk to void_node for vararg functions.
34918 (try_substitute_return_value): Adjust dump output.
34920 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34922 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34924 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34926 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34927 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34928 -mlra/-mno-lra was passed to the compiler.
34930 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34932 PR rtl-optimization/65618
34933 * emit-rtl.c (try_split): Move initialization of "before" and
34934 "after" to just before the call to emit_insn_after_setloc.
34936 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34938 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34940 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34942 * dwarf2out.c (gen_enumeration_type_die): When
34943 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34945 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34947 PR tree-optimization/78965
34948 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34949 Change first argument from const call_info & to call_info &. For %n
34950 set info.nowrite to false.
34952 PR middle-end/78901
34953 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34954 possibly throwing calls.
34956 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34957 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34958 and fns handling, rather than in a separate case SSA_NAME.
34960 2017-01-02 Jeff Law <law@redhat.com>
34962 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34963 fixes for first_period and second_period variables.
34965 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34968 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34969 (*insvqi_1): New insn pattern.
34970 (*insvqi_1_mem_rex64): Ditto.
34971 (*insvqi_2): Ditto.
34972 (*insvqi_3): Rename from *insvqi.
34974 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34976 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34978 * doc/cfg.texi (Edges): Remove reference to Java.
34979 (Maintaining the CFG): Ditto.
34981 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34983 PR middle-end/77674
34984 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34985 transparent aliases.
34987 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34989 PR middle-end/77484
34990 * predict.def (PRED_CALL): Update hitrate.
34991 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34992 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34993 into direct/indirect/polymorphic variants.
34995 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34997 Update copyright years.
34999 * gcc.c (process_command): Update copyright notice dates.
35000 * gcov-dump.c (print_version): Ditto.
35001 * gcov.c (print_version): Ditto.
35002 * gcov-tool.c (print_version): Ditto.
35003 * gengtype.c (create_file): Ditto.
35004 * doc/cpp.texi: Bump @copying's copyright year.
35005 * doc/cppinternals.texi: Ditto.
35006 * doc/gcc.texi: Ditto.
35007 * doc/gccint.texi: Ditto.
35008 * doc/gcov.texi: Ditto.
35009 * doc/install.texi: Ditto.
35010 * doc/invoke.texi: Ditto.
35012 Copyright (C) 2017 Free Software Foundation, Inc.
35014 Copying and distribution of this file, with or without modification,
35015 are permitted in any medium without royalty provided the copyright
35016 notice and this notice are preserved.