1 2017-11-15 Martin Liska <mliska@suse.cz>
3 * tree-cfg.c (pass_warn_function_return::execute):
4 Compare warn_return_type for greater than zero.
6 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
10 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
11 to return true on Xeon and not on Xeon Phi.
12 (ix86_check_avx256_register): Changed to ...
13 (ix86_check_avx_upper_register): ... this. Add extra check for
14 VALID_AVX512F_REG_OR_XI_MODE.
15 (ix86_avx_u128_mode_needed): Changed
16 ix86_check_avx256_register to ix86_check_avx_upper_register.
17 (ix86_check_avx256_stores): Changed to ...
18 (ix86_check_avx_upper_stores): ... this. Changed
19 ix86_check_avx256_register to ix86_check_avx_upper_register.
20 (ix86_avx_u128_mode_after): Changed
21 avx_reg256_found to avx_upper_reg_found. Changed
22 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
23 (ix86_avx_u128_mode_entry): Changed
24 ix86_check_avx256_register to ix86_check_avx_upper_register.
25 (ix86_avx_u128_mode_exit): Ditto.
26 * config/i386/i386.h: (host_detect_local_cpu): New define.
28 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
30 * config/arm/xgene1.md (xgene1): Split into automatons
31 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
32 (xgene1_f_load): Adjust reservations and/or types.
33 (xgene1_f_store): Likewise.
34 (xgene1_load_pair): Likewise.
35 (xgene1_store_pair): Likewise.
36 (xgene1_fp_load1): Likewise.
37 (xgene1_load1): Likewise.
38 (xgene1_store1): Likewise.
39 (xgene1_move): Likewise.
40 (xgene1_alu): Likewise.
41 (xgene1_simd): Likewise.
42 (xgene1_bfm): Likewise.
43 (xgene1_neon_load1): Likewise.
44 (xgene1_neon_store1): Likewise.
45 (xgene1_neon_logic): Likewise.
46 (xgene1_neon_st1): Likewise.
47 (xgene1_neon_ld1r): Likewise.
48 (xgene1_alu_cond): Added.
49 (xgene1_shift_reg): Likwise.
50 (xgene1_bfx): Likewise.
51 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
53 2017-11-15 Jakub Jelinek <jakub@redhat.com>
56 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
58 (can_widen_mult_without_libcall): New function.
59 (expand_mul_overflow): If only checking unsigned mul overflow,
60 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
61 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
62 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
63 (expand_DIVMOD): Formatting fix.
64 * expmed.h (expand_mult): Add NO_LIBCALL argument.
65 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
66 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
68 PR tree-optimization/82977
69 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
70 constructed temporary to strlen_to_stridx.put.
72 2017-11-15 Martin Liska <mliska@suse.cz>
74 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
75 * configure: Regenerate.
77 2017-11-15 Martin Liska <mliska@suse.cz>
80 * config/sh/sh-mem.cc: Use proper probability for
83 2017-11-14 Jeff Law <law@redhat.com>
85 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
86 the red zone for stack_clash_protection_final_dynamic_probe targets
87 when the total dynamic stack size is zero bytes.
89 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
92 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
94 * dumpfile.h (TDF_COMPARE_DEBUG): New.
95 * final.c (rest_of_clean_state): Set it for the
97 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
98 class when TDF_COMPARE_DEBUG is set.
100 * dwarf2out.c (gen_producer_string): Discard
103 2017-11-15 Joseph Myers <joseph@codesourcery.com>
106 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
107 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
108 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
109 (__TGMATH_REAL_2_3): Remove macros.
110 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
111 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
113 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
115 (remquo): Define using __TGMATH_REAL_3.
117 2017-11-14 Jeff Law <law@redhat.com>
119 * vr-values.c: New file with contents extracted from tree-vrp.c.
120 * Makefile.in (OBJS): Add vr-values.o
121 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
122 (set_value_range, set_and_canonicalize_value_range): Likewise.
123 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
124 (value_range_constant_singleton, symbolic_range_p): Likewise.
125 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
126 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
127 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
128 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
129 (find_case_label_range, find_case_label_index): Likewise.
130 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
131 (range_int_cst_singleton_p, value_inside_range): Likewise.
132 (get_single_symbol): Likewise.
133 (switch_update): Move structure definition here.
134 (to_remove_edges, to_update_switch_stmts): Provide externs.
135 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
136 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
137 (vrp_val_is_min, set_value_range): Likewise.
138 (set_and_canonicalize_value_range, copy_value_range): Likewise.
139 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
140 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
141 (range_is_nonnull, range_int_cst_p): Likewwise.
142 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
143 (get_single_symbol, operand_less_p): Likewise
144 (compare_values_warnv, compare_values): Likewise.
145 (value_inside_range, value_range_constant_singleton): Likewise.
146 (zero_nonzero_bitgs_from_vr): Likewise.
147 (extract_range_from_binary_expr_1): Likewise.
148 (overflow_comparison_p): Likewise.
149 (to_remove_edges, to_update_switch_stmts): Likewise.
150 (find_case_label-index, find_case_label_range): Likewise.
151 (switch_update, set_value_range_to_nonnegative): Remove.
152 (set_value_range_to_truthvalue): Likewise.
153 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
154 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
155 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
156 (find_case_label_ranges, test_for_singularity): Likewise.
157 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
158 (x_vr_values): Move to its remaining use site.
160 2017-11-10 Jeff Law <law@redhat.com>
162 * vr-values.h (VR_INITIALIZER): Move #define here.
163 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
164 * Makefile.in (OBJS): Add tree-evrp.o
165 * tree-vrp.h (assert_info): Move structure definition here.
166 (set_value_range_to_varying): Prototype.
167 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
168 (infer_value_range, register_edge_assert_for): Likewise.
169 (stmt_interesting_for_vrp): Likewise.
170 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
171 (set_value_range_to_varying): No longer static.
172 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
173 (infer_value_range, register_edge_assert_for): Likewise.
175 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
177 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
178 generate the XXBRD instruction.
180 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
181 (rs6000_builtin_type_compatible): Treat _Float128 and long double
182 as being compatible if -mabi=ieeelongdouble.
183 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
184 to setup float128 built-ins with hardware support.
185 (BU_FLOAT128_HW_2): Likewise.
186 (BU_FLOAT128_HW_3): Likewise.
187 (BU_FLOAT128_HW_VSX_1): Likewise.
188 (BU_FLOAT128_HW_VSX_2): Likewise.
189 (scalar_extract_expq): Change float128 built-in functions to
190 accommodate having both KFmode and TFmode functions. Use the
191 KFmode variant as the default.
192 (scalar_extract_sigq): Likewise.
193 (scalar_test_neg_qp): Likewise.
194 (scalar_insert_exp_q): Likewise.
195 (scalar_insert_exp_qp): Likewise.
196 (scalar_test_data_class_qp): Likewise.
197 (sqrtf128_round_to_odd): Delete processing the round to odd
198 built-in functions as special built-in functions, and define them
199 as float128 built-ins. Use the KFmode variant as the default.
200 (truncf128_round_to_odd): Likewise.
201 (addf128_round_to_odd): Likewise.
202 (subf128_round_to_odd): Likewise.
203 (mulf128_round_to_odd): Likewise.
204 (divf128_round_to_odd): Likewise.
205 (fmaf128_round_to_odd): Likewise.
206 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
207 support for KFmode and TFmode xststdcqp calls.
208 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
209 point, switch the built-in handlers for the get/set float128
210 exponent, get float128 mantissa, float128 test built-ins, and the
211 float128 round to odd built-in functions. Eliminate creating the
212 float128 round to odd built-in functions as special built-ins.
213 (rs6000_init_builtins): Eliminate special creation of the float128
214 round to odd built-in functions.
215 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
216 function insns to support both TFmode and KFmode variants.
217 (xsxsigqp_<mode>): Likewise.
218 (xsiexpqpf_<mode>): Likewise.
219 (xsiexpqp_<mode>): Likewise.
220 (xststdcqp_<mode>): Likewise.
221 (xststdcnegqp_<mode>): Likewise.
222 (xststdcqp_<mode>): Likewise.
224 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
226 * tree-ssa-threadupdate.c (compute_path_counts): Remove
227 unused path_in_freq_ptr parameter.
228 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
230 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
232 * ipa-inline.c (edge_badness): Dump sreal frequency.
233 (compute_inlined_call_time): Match natural implementaiton ...
234 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
235 forgotten division by CGRAPH_FREQ_BASE.
237 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
239 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
240 Solaris 11. Update comment.
241 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
243 * config/sol2.h (STARTFILE_SPEC): Likewise.
244 * configure: Regenerate.
246 2017-11-14 Carl Love <cel@us.ibm.com>
248 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
249 le_ and be_ prefixes to swap* variables. Remove
250 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
253 2017-11-14 Jason Merrill <jason@redhat.com>
255 Support GTY((cache)) on hash_map.
256 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
257 (ggc_cache_remove): Override it instead of ggc_mx.
258 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
259 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
260 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
261 (simple_cache_map_traits): Override maybe_mx.
262 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
263 (hash_map): Friend gt_cleare_cache.
264 (gt_cleare_cache): New.
265 * tree.h (tree_cache_traits): New hash_map traits class.
266 (tree_cache_map): New typedef.
268 2017-11-14 Richard Biener <rguenther@suse.de>
270 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
271 paramter and handling.
272 (cleanup_control_flow_bb): Likewise.
273 (cleanup_control_flow_pre): New helper performing a DFS walk
274 to call cleanup_control_flow_bb in PRE order.
275 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
276 via cleanup_control_flow_pre.
278 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
280 * config/aarch64/aarch64-simd.md
281 (aarch64_simd_bsl<mode>_internal): Remove DImode.
282 (*aarch64_simd_bsl<mode>_alt): Likewise.
283 (aarch64_simd_bsldi_internal): New.
284 (aarch64_simd_bsldi_alt): Likewise.
286 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
288 * tracer.c (better_p): Do not compare frequencies.
289 * reg-stack.c (better_edge): Likewise.
290 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
293 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
295 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
296 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
297 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
298 * lto-streamer-in.c (input_function): Use update_max_bb_count.
299 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
300 * predict.c (maybe_hot_frequency_p): Inline to ...
301 (maybe_hot_count_p): ... here; rewrite to counts.
302 (counts_to_freqs): Rename to ...
303 (update_max_bb_count): ... this one.
304 (expensive_function_p): Use counts.
305 (estimate_bb_frequencies): Update.
306 (rebuild_frequencies): Update.
307 * predict.h (counts_to_freqs): Rename to ...
308 (update_max_bb_count): ... this one.
309 * profile.c (compute_branch_probabilities): Add debug info
310 * tree-inline.c (expand_call_inline): Update debug info.
311 (optimize_inline_calls): Use update_max_bb_count..
312 (tree_function_versioning): Use update_max_bb_count..
313 * value-prof.c (gimple_value_profile_transformations):
314 Do not use update_max_bb_count.
316 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
318 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
319 always use frequencies.
321 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
323 * bb-reorder.c: Remove frequencies from comments.
324 (better_edge_p): Use profile counts.
325 (find_traces): Dump profile counts.
326 (rotate_loop): Use profile counts.
327 (find_traces_1_round): Likewise.
328 (connect_better_edge_p): Use counts instead of probabilities for
330 (copy_bb_p): Drop early check for non-0 frequency.
331 (sanitize_hot_paths): Update comments.
333 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
335 * ipa-split.c (struct split_point): Add count.
336 (consider_split): Do not compute incoming frequency; compute incoming
337 count and store it to split_point.
338 (split_function): Set count of the call to split part correctly.
340 2017-11-13 Carl Love <cel@us.ibm.com>
342 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
344 2017-11-13 Tom Tromey <tom@tromey.com>
346 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
348 2017-11-13 Carl Love <cel@us.ibm.com>
350 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
351 Add support for builtins:
352 unsigned int vec_first_{,miss}_match_{,or_eos}index,
353 vector {un,}signed {char,int,short},
354 vector {un,}signed {char,int,short}) arguments.
355 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
356 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
357 Add BU_P9V_AV_2 expansions for the builtins.
358 * config/rs6000/altivec.h (vec_first_match_index,
359 vec_first_mismatch_index, vec_first_match_or_eos_index,
360 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
361 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
362 new extern declaration.
363 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
364 * config/rs6000/vsx.md (first_match_index_<mode>,
365 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
366 first_mismatch_or_eos_index_<mode>): Add define expand.
367 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
368 * doc/extend.texi: Update the built-in documenation file for the new
371 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
373 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
374 into the min/max operations for _Float<N> and _Float<N>X types.
376 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
379 * dwarf2out.c (do_eh_frame): New static variable.
380 (dwarf2out_begin_prologue): Set it.
381 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
383 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
385 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
387 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
389 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
392 * debug.h (dwarf2out_do_eh_frame): Declare.
393 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
394 (dwarf2out_do_frame): Use it.
395 (dwarf2out_do_cfi_asm): Likewise.
396 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
397 (dwarf2out_assembly_start): Likewise.
398 (dwarf2out_begin_prologue): Fix comment.
399 * toplev.c (compile_file): Always call dwarf2out_frame_finish
400 if the target needs either debug or unwind DWARF2 info.
401 * lto-opts.c (lto_write_options): Do not save -fexceptions,
402 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
403 -ftrapping-math, -ftrapv and -fwrapv.
405 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
407 * cgraph.c (cgraph_edge::sreal_frequency): New function.
408 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
409 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
410 (estimate_edge_size_and_time): Likewise.
411 (ipa_merge_fn_summary_after_inlining): Likewise.
412 * ipa-inline.c (cgraph_freq_base_rec): Remove.
413 (compute_uninlined_call_time): Use sreal_frequency.
414 (compute_inlined_call_time): Likewise.
415 (ipa_inline): Do not initialize cgraph_freq_base_rec.
416 * profile-count.c: Include sreal.h.
417 (profile_count::to_sreal_scale): New.
418 * profile-count.h: Forward declare sreal.
419 (profile_count::to_sreal_scale): Declare.
421 2017-11-13 Nathan Sidwell <nathan@acm.org>
423 * diagnostic.c (maybe_line_and_column): New.
424 (diagnostic_get_location_text): Use it.
425 (diagnostic_report_current_module): Likewise.
426 (test_diagnostic_get_location_text): Add tests.
428 2017-11-13 Luis Machado <luis.machado@linaro.org>
430 * doc/md.texi (Specifying processor pipeline description): Fix
431 incorrect latency for the div instruction example.
433 2017-11-13 Jakub Jelinek <jakub@redhat.com>
435 PR tree-optimization/78821
436 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
437 that bit_not_p is the same.
438 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
439 (split_group): Count precisely bit_not_p bits in each statement.
440 (invert_op): New function.
441 (imm_store_chain_info::output_merged_store): Use invert_op to
442 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
443 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
445 2017-11-13 Martin Liska <mliska@suse.cz>
447 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
448 (struct source_info): Likewise.
449 (add_branch_counts): Likewise.
450 (add_line_counts): Likewise.
451 (function_summary): Likewise.
452 (output_intermediate_line): Likewise.
453 (generate_results): Likewise.
455 2017-11-13 Martin Liska <mliska@suse.cz>
457 * gcov.c (struct block_info): Remove typedef for block_t.
458 (struct line_info): Likewise.
459 (line_info::has_block): Likewise.
460 (EXIT_BLOCK): Likewise.
463 (get_cycles_count): Likewise.
464 (process_file): Likewise.
465 (read_graph_file): Likewise.
466 (solve_flow_graph): Likewise.
467 (find_exception_blocks): Likewise.
468 (add_line_counts): Likewise.
469 (accumulate_line_info): Likewise.
470 (output_line_details): Likewise.
472 2017-11-13 Martin Liska <mliska@suse.cz>
474 * gcov.c (struct arc_info): Remove typedef for arc_t.
475 (struct line_info): Likewise.
476 (add_branch_counts): Likewise.
477 (output_branch_count): Likewise.
478 (function_info::~function_info): Likewise.
480 (output_intermediate_line): Likewise.
481 (read_graph_file): Likewise.
482 (solve_flow_graph): Likewise.
483 (find_exception_blocks): Likewise.
484 (add_line_counts): Likewise.
485 (accumulate_line_info): Likewise.
486 (output_line_details): Likewise.
487 (output_function_details): Likewise.
489 2017-11-13 Martin Liska <mliska@suse.cz>
491 * gcov.c (struct function_info): Remove typedef for function_t.
492 (struct source_info): Likewise.
493 (source_info::get_functions_at_location): Likewise.
494 (solve_flow_graph): Likewise.
495 (find_exception_blocks): Likewise.
496 (add_line_counts): Likewise.
497 (output_intermediate_file): Likewise.
498 (process_file): Likewise.
499 (generate_results): Likewise.
500 (release_structures): Likewise.
501 (read_graph_file): Likewise.
502 (read_count_file): Likewise.
503 (accumulate_line_counts): Likewise.
504 (output_lines): Likewise.
506 2017-11-13 Martin Liska <mliska@suse.cz>
508 * gcov.c (function_info::function_info): Remove num_counts
509 and add vector<gcov_type>.
510 (function_info::~function_info): Use the vector.
511 (process_file): Likewise.
512 (read_graph_file): Likewise.
513 (read_count_file): Likewise.
514 (solve_flow_graph): Likewise.
516 2017-11-13 Martin Liska <mliska@suse.cz>
518 * gcov.c (function_info::is_artificial): New function.
519 (process_file): Erase all artificial early.
520 (generate_results): Skip as all artificial are already
523 2017-11-13 Martin Liska <mliska@suse.cz>
525 * gcov.c (read_graph_file): Store to global vector of functions.
526 (read_count_file): Iterate the vector.
527 (process_file): Likewise.
528 (generate_results): Likewise.
529 (release_structures): Likewise.
531 2017-11-13 Jakub Jelinek <jakub@redhat.com>
533 PR tree-optimization/82954
534 * gimple-ssa-store-merging.c
535 (imm_store_chain_info::coalesce_immediate_stores): If
536 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
538 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
540 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
541 Upddate call to ENDIAN_LANE_N.
542 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
543 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
544 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
545 and use aarch64_endian_lane_rtx.
546 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
548 2017-11-12 Tom de Vries <tom@codesourcery.com>
550 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
552 2017-11-12 Tom de Vries <tom@codesourcery.com>
554 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
556 2017-11-12 Tom de Vries <tom@codesourcery.com>
558 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
560 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
561 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
562 * defaults.h (ASM_OUTPUT_LABELREF): Same.
564 2017-11-11 Martin Sebor <msebor@redhat.com>
567 * doc/extend.texi (attribute nonstring): Remove spurious argument.
570 * prefic.c (translate_name): Replace strncpy with memcpy to
571 avoid -Wstringop-truncation.
573 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
575 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
577 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
579 * predict.c (maybe_hot_frequency_p): Do not use cfun.
581 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
583 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
586 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
588 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
589 enable generating XXBRH if the value is in a vector register.
590 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
591 value is in a vector register.
592 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
593 register to register bswap64's instead of doing the GPR sequence
594 used on previous machines.
595 (bswapdi2_xxbrd): New insn.
596 (bswapdi2_reg): Disallow on ISA 3.0.
597 (register to register bswap64 splitter): Do not split the insn on
598 ISA 3.0 systems that use XXBRD.
600 2017-11-10 Martin Sebor <msebor@redhat.com>
603 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
605 (find_subframework_file): Same.
607 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
609 * auto-profile.c (afdo_indirect_call): Drop frequency.
610 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
611 (cgraph_node::create_edge): Drop frequency argument.
612 (cgraph_node::create_indirect_edge): Drop frequency argument.
613 (cgraph_edge::make_speculative): Drop frequency arguments.
614 (cgraph_edge::resolve_speculation): Do not update frequencies
615 (cgraph_edge::dump_edge_flags): Do not dump frequency.
616 (cgraph_node::dump): Check consistency in IPA mode.
617 (cgraph_edge::maybe_hot_p): Use IPA counter.
618 (cgraph_edge::verify_count_and_frequency): Rename to ...
619 (cgraph_edge::verify_count): ... this one; drop frequency checking.
620 (cgraph_node::verify_node): Update.
621 * cgraph.h (struct cgraph_edge): Drop frequency.
622 (cgraph_edge::frequency): New function.
623 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
625 (cgraph_edge::rebuild_edges): Likewise.
626 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
627 (duplicate_thunk_for_node): Do not pass frequency.
628 (cgraph_node::create_clone): Scale only counts.
629 (cgraph_node::create_virtual_clone): Do not pass frequency.
630 (cgraph_node::create_edge_including_clones): Do not pass frequency.
631 (cgraph_node::create_version_clone): Do not pass frequency.
632 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
633 (cgraph_node::expand_thunk): Do not pass frequency.
634 (cgraph_node::create_wrapper): Do not pass frequency.
635 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
637 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
638 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
639 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
640 (gather_caller_stats): Use frequency function.
641 (ipcp_cloning_candidate_p): Use frequency function.
642 (ipcp_propagate_stage): Use frequency function.
643 (get_info_about_necessary_edges): Use frequency function.
644 (update_profiling_info): Update only IPA profile.
645 (update_specialized_profile): Use frequency functoin.
646 (perhaps_add_new_callers): Update only IPA profile.
647 * ipa-devirt.c (ipa_devirt): Use IPA profile.
648 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
649 (dump_ipa_call_summary): Use frequency function.
650 (estimate_edge_size_and_time): Use frequency function.
651 (ipa_merge_fn_summary_after_inlining): Use frequency function.
652 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
653 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
654 (update_noncloned_counts): ... ths one; scale counts only.
655 (clone_inlined_nodes): Do not scale frequency.
656 (inline_call): Do not pass frequency.
657 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
658 (compute_inlined_call_time): Use IPA profile.
659 (want_inline_small_function_p): Use IPA profile.
660 (want_inline_self_recursive_call_p): Use IPA profile.
661 (edge_badness): Use IPA profile.
662 (lookup_recursive_calls): Use IPA profile.
663 (recursive_inlining): Do not pass frequency.
664 (resolve_noninline_speculation): Do not update frequency.
665 (inline_small_functions): Collect max of IPA profile.
666 (dump_overall_stats): Dump IPA porfile.
667 (dump_inline_stats): Dump IPA porfile.
668 (ipa_inline): Collect IPA stats.
669 * ipa-inline.h (clone_inlined_nodes): Update prototype.
670 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
671 (ipa_propagate_frequency): Use frequency function.
672 (ipa_profile): Cleanup.
673 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
674 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
675 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
676 (input_node): Do not stream frequency.
677 (input_edge): Do not stream frequency.
678 (merge_profile_summaries): Scale only IPA profiles.
679 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
680 * predict.c (drop_profile): Do not recompute frequency.
681 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
682 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
683 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
684 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
685 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
686 * tree-inline.c (copy_bb): Do not scale frequency.
687 (expand_call_inline): Do not scale frequency.
688 (tree_function_versioning): Do not scale frequency.
689 * ubsan.c (ubsan_create_edge): Do not pass frequency.
691 2017-11-10 Julia Koval <julia.koval@intel.com>
693 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
694 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
695 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
696 _mm256_maskz_gf2p8affine_epi64_epi8,
697 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
698 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
699 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
700 __builtin_ia32_vgf2p8affineqb_v32qi,
701 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
702 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
704 2017-11-10 Tamar Christina <tamar.christina@arm.com>
708 (arm_option_override): Refactor.
709 (arm_option_reconfigure_globals): New.
710 (arm_options_perform_arch_sanity_checks): New.
711 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
712 (arm_options_perform_arch_sanity_checks): Likewise
714 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
716 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
718 * rs6000/rs6000.md ("type" attr): Add qmul.
719 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
720 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
721 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
723 2017-11-10 Martin Sebor <msebor@redhat.com>
726 * builtins.c (compute_objsize): Handle arrays that
727 compute_builtin_object_size likes to fail for. Make extern.
728 * builtins.h (compute_objsize): Declare.
729 (check_strncpy_sizes): New function.
730 (expand_builtin_strncpy): Call check_strncpy_sizes.
731 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
732 -Wstringop-truncation.
733 (gimple_fold_builtin_strncat): Same.
734 * gimple.c (gimple_build_call_from_tree): Set call location.
735 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
736 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
737 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
738 (handle_builtin_strlen): Use strlen_to_stridx.
739 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
741 Use strlen_to_stridx.
742 (pass_strlen::execute): Release strlen_to_stridx.
743 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
744 (-Wstringop-truncation): Document new option.
746 2017-11-10 Martin Liska <mliska@suse.cz>
748 PR gcov-profile/82702
749 * gcov.c (main): Handle intermediate files in a different
751 (get_gcov_intermediate_filename): New function.
752 (output_gcov_file): Remove support of intermediate files.
753 (generate_results): Allocate intermediate file.
754 (release_structures): Clean-up properly fn_end.
755 (output_intermediate_file): Start iterating with line 1.
757 2017-11-10 Jakub Jelinek <jakub@redhat.com>
759 PR tree-optimization/82929
760 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
761 ops_swapped_p non-static data member.
762 (store_immediate_info::store_immediate_info): Clear it.
763 (imm_store_chain_info::coalesce_immediate_stores): If swapping
764 ops set ops_swapped_p.
765 (count_multiple_uses): Handle ops_swapped_p.
767 2017-11-10 Martin Liska <mliska@suse.cz>
769 * coverage.c (coverage_init): Stream information about
770 support of has_unexecuted_blocks.
771 * doc/gcov.texi: Document that.
772 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
773 * gcov.c (read_graph_file): Likewise.
774 (output_line_beginning): Fix a small issue with
777 2017-11-10 Bin Cheng <bin.cheng@arm.com>
779 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
780 reference of trivial component.
782 2017-11-10 Jakub Jelinek <jakub@redhat.com>
785 * gimple-ssa-store-merging.c
786 (pass_store_merging::terminate_all_aliasing_chains): For
787 gimple_store_p stmts also call refs_output_dependent_p.
789 PR rtl-optimization/82913
790 * compare-elim.c (try_merge_compare): Punt if def_insn is not
793 2017-11-09 Jeff Law <law@redhat.com>
795 * vr-values.h: New file with vr_values class.
796 * tree-vrp.c: Include vr-values.h
797 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
798 data objects into the vr_values class.
799 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
800 (get_value_range): Make it a member function within vr_values class.
801 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
802 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
803 (op_with_constant_singleton_value_range): Likewise.
804 (extract_range_for_var_from_comparison_expr): Likewise.
805 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
806 (extract_range_from_binary_expr): Likewise.
807 (extract_range_from_unary_expr): Likewise.
808 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
809 (check_for_binary_op_overflow, extract_range_basic): Likewise.
810 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
811 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
812 (compare_name_with_value, compare_names): Likewise.
813 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
814 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
815 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
816 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
817 (extract_range_from_phi_node): Likewise.
818 (simplify_truth_ops_using_ranges): Likewise.
819 (simplify_div_or_mod_using_ranges): Likewise.
820 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
821 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
822 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
823 (simplify_float_conversion_using_ranges): Likewise.
824 (simplify_internal_call_using_ranges): Likewise.
825 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
826 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
827 poitner via x_vr_values for calls into gimple folder.
828 (vrp_initialize_lattice): Make this the vr_values ctor.
829 (vrp_free_lattice): Make this the vr_values dtor.
830 (set_vr_value): New function.
831 (class vrp_prop): Add vr_values data member. Add various member
832 functions as well as member functions that delegate to vr_values.
833 (check_array_ref): Make a member function within vrp_prop class.
834 (search_for_addr_array, vrp_initialize): Likewise.
835 (vrp_finalize): Likewise. Revamp to avoid direct access to
836 vr_value, values_propagated, etc.
837 (check_array_bounds): Extract vrp_prop class instance pointer from
838 walk info structure. Use it to call member functions.
839 (check_all_array_refs): Make a member function within vrp_prop class.
840 Smuggle class instance pointer via walk info structure.
841 (x_vr_values): New local static.
842 (vrp_valueize): Use x_vr_values to get class instance.
843 (vr_valueize_1): Likewise.
844 (class vrp_folder): Add vr_values data member. Add various member
845 functions as well as member functions that delegate to vr_values.
846 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
847 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
848 class instance from vr_values. Use it to call member functions.
849 (vrp_dom_walker): Add vr_values data member.
850 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
851 instance via x_vr_values.
852 (identify_jump_threads): Accept vr_values as argument. Store
853 it into the walker structure.
854 (evrp_dom_walker): Add vr_values class data member. Add various
856 (evrp_dom_walker::try_find_new_range): Use vr_values data
857 member to access the memory allocator.
858 (evrp_dom_walker::before_dom_children): Store vr_values class
859 instance into the vrp_folder class.
860 (evrp_dom_walker::push_value_range): Rework to avoid direct
861 access to num_vr_values and vr_value.
862 (evrp_dom_walker::pop_value_range): Likewise.
863 (execute_early_vrp): Remove call to vrp_initialize_lattice.
864 Use vr_values to get to dump_all_value_ranges member function.
865 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
866 and simplify_cond_using_ranges_2 via vrp_prop class instance.
867 Pass vr_values class instance down to identify_jump_threads.
868 Remove call to vrp_free_lattice.
869 (debug_all_value_ranges): Remove.
871 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
872 (vrp_folder): Likewise.
874 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
875 Get it from the existing bitmap instead.
876 (vrp_intersect_ranges_1): Likewise.
878 2017-11-09 Jakub Jelinek <jakub@redhat.com>
880 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
882 (store_immediate_info::store_immediate_info): Add bitnotp argument,
884 (imm_store_chain_info::coalesce_immediate_stores): Break group
885 if bit_not_p is different.
886 (count_multiple_uses, split_group,
887 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
888 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
889 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
890 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
891 to store_immediate_info ctor.
893 2017-11-09 Jim Wilson <jimw@sifive.com>
895 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
896 (scan_prog_file): Likewise.
898 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
900 * bb-reorder.c (max_entry_frequency): Remove.
901 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
902 connect_traces, push_to_next_round_p): Remove prototypes.
903 (find_traces_1_round): Use counts only.
904 (push_to_next_round_p): Likewise.
905 (find_traces): Likewise.
906 (rotate_loop): Likewise.
907 (find_traces_1_round): Likewise.
908 (connect_traces): Likewise.
909 (edge_order): Likewise.
911 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
913 * config/arm/arm.c (output_return_instruction): Add comments to
914 indicate requirement for cmse_nonsecure_entry return to account
915 for the size of clearing instruction output here.
916 (thumb_exit): Likewise.
917 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
918 return in hardfloat mode.
920 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
922 * config/rs6000/rs6000.c (machine_function): Add a bool,
923 "toc_is_wrapped_separately".
924 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
925 if it wasn't explicitly set or unset, we are optimizing for speed, and
926 doing separate shrink-wrapping.
927 (rs6000_get_separate_components): Enable the TOC component if
928 saving the TOC register in the prologue.
929 (rs6000_components_for_bb): Handle the TOC component.
930 (rs6000_emit_prologue_components): Store the TOC register where needed.
931 (rs6000_set_handled_components): Mark TOC as handled, if handled.
932 (rs6000_emit_prologue): Don't save the TOC if that is already done.
934 2017-11-09 Martin Jambor <mjambor@suse.cz>
936 * ipa-param-manipulation.c: New file.
937 * ipa-param-manipulation.h: Likewise.
938 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
939 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
940 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
941 (ipa_parm_adjustment): Likewise.
942 (ipa_parm_adjustment_vec): Likewise.
943 (ipa_get_vector_of_formal_parms): Moved declaration to
944 ipa-param-manipulation.h.
945 (ipa_get_vector_of_formal_parm_types): Likewise.
946 (ipa_modify_formal_parameters): Likewise.
947 (ipa_modify_call_arguments): Likewise.
948 (ipa_combine_adjustments): Likewise.
949 (ipa_dump_param_adjustments): Likewise.
950 (ipa_modify_expr): Likewise.
951 (ipa_get_adjustment_candidate): Likewise.
952 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
953 ipa-param-manipulation.c.
954 (ipa_get_vector_of_formal_parm_types): Likewise.
955 (ipa_modify_formal_parameters): Likewise.
956 (ipa_modify_call_arguments): Likewise.
957 (ipa_modify_expr): Likewise.
958 (get_ssa_base_param): Likewise.
959 (ipa_get_adjustment_candidate): Likewise.
960 (index_in_adjustments_multiple_times_p): Likewise.
961 (ipa_combine_adjustments): Likewise.
962 (ipa_dump_param_adjustments): Likewise.
963 * tree-sra.c: Also include ipa-param-manipulation.h
964 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
967 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
968 Alan Hayward <alan.hayward@arm.com>
969 David Sherwood <david.sherwood@arm.com>
971 * doc/sourcebuild.texi (vect_masked_store): Document.
973 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
979 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
980 Alan Hayward <alan.hayward@arm.com>
981 David Sherwood <david.sherwood@arm.com>
983 * doc/sourcebuild.texi (vect_variable_length): Document.
985 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
986 Alan Hayward <alan.hayward@arm.com>
987 David Sherwood <david.sherwood@arm.com>
989 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
991 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
992 Alan Hayward <alan.hayward@arm.com>
993 David Sherwood <david.sherwood@arm.com>
995 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
997 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
998 Alan Hayward <alan.hayward@arm.com>
999 David Sherwood <david.sherwood@arm.com>
1001 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1002 previously undocumented selectors.
1003 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1005 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1007 * doc/rtl.texi (const_vector): Say that elements can be
1008 const_wide_ints too.
1009 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1010 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1011 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1012 * optabs.c (expand_vector_broadcast): Likewise.
1014 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1015 Alan Hayward <alan.hayward@arm.com>
1016 David Sherwood <david.sherwood@arm.com>
1018 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1019 scaled index even if the unscaled address was invalid.
1020 Don't increase the complexity of using a scale in that case.
1022 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1023 Alan Hayward <alan.hayward@arm.com>
1024 David Sherwood <david.sherwood@arm.com>
1026 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1027 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1028 UNITS_PER_WORD bytes.
1029 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1030 based on REGMODE_NATURAL_SIZE of the inner mode.
1031 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1032 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1033 * expmed.c (lowpart_bit_field_p): Divide the value up into
1034 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1035 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1036 whether something is likely to occupy more than one register.
1038 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1041 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1042 profile_count::adjust_for_ipa_scaling.
1043 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1044 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1046 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1048 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1050 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1051 (split_group): Add total_orig and total_new arguments, estimate the
1052 number of statements related to the store group without store merging
1053 and with store merging.
1054 (imm_store_chain_info::output_merged_store): Adjust split_group
1055 callers, punt if estimated number of statements with store merging
1056 is not smaller than estimated number of statements without it.
1058 (handled_load): Remove has_single_use checks.
1059 (pass_store_merging::process_store): Likewise.
1061 2017-11-09 Richard Biener <rguenther@suse.de>
1063 PR tree-optimization/82902
1064 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1066 2017-11-09 Martin Liska <mliska@suse.cz>
1069 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1072 2017-11-09 Martin Liska <mliska@suse.cz>
1074 PR tree-optimization/82669
1075 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1077 2017-11-09 Martin Liska <mliska@suse.cz>
1079 PR gcov-profile/48463
1080 * coverage.c (coverage_begin_function): Output also end locus
1081 of a function and information whether the function is
1083 * gcov-dump.c (tag_function): Parse and print the information.
1084 * gcov.c (INCLUDE_MAP): Add include.
1085 (INCLUDE_SET): Likewise.
1086 (struct line_info): Move earlier in the source file because
1087 of vector<line_info> in function_info structure.
1088 (line_info::line_info): Likewise.
1089 (line_info::has_block): Likewise.
1090 (struct source_info): Add new member index.
1091 (source_info::get_functions_at_location): New function.
1092 (function_info::group_line_p): New function.
1093 (output_intermediate_line): New function.
1094 (output_intermediate_file): Use the mentioned function.
1095 (struct function_start): New.
1096 (struct function_start_pair_hash): Likewise.
1097 (process_file): Add code that identifies group functions.
1098 Assign lines either to global or function scope.
1099 (generate_results): Skip artificial functions.
1100 (find_source): Assign index for each source file.
1101 (read_graph_file): Read new flag artificial and end_line.
1102 (add_line_counts): Assign it either to global of function scope.
1103 (accumulate_line_counts): Isolate core of the function to
1104 accumulate_line_info and call it for both function and global
1106 (accumulate_line_info): New function.
1107 (output_line_beginning): Fix GNU coding style.
1108 (print_source_line): New function.
1109 (output_line_details): Likewise.
1110 (output_function_details): Likewise.
1111 (output_lines): Iterate both source (global) scope and function
1113 (struct function_line_start_cmp): New class.
1114 * doc/gcov.texi: Reflect changes in documentation.
1116 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1119 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1120 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1121 and similarly for not instead of neg.
1123 2017-11-08 Andi Kleen <ak@linux.intel.com>
1125 * config/i386/i386.opt: Add -mforce-indirect-call.
1126 * config/i386/predicates.md: Check for flag_force_indirect_call.
1127 * doc/invoke.texi: Document -mforce-indirect-call
1129 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1131 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1133 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1134 config/riscv/riscv.c (predict.h): New include.
1135 (riscv_slow_unaligned_access_p): No longer static.
1136 (riscv_block_move_straight): Add require.
1137 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1139 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1142 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1143 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1144 nonimmediate_operand predicate for operand 1 instead of
1147 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1149 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1151 * config/aarch64/constraints.md (Uml): New constraint.
1152 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1155 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1157 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1158 of two vec_duplicates into a vec_concat.
1160 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1162 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1163 Simplify vec_merge of vec_duplicate and vec_concat.
1164 * config/aarch64/constraints.md (Utq): New constraint.
1165 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1168 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1170 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1171 Simplify vec_merge of vec_duplicate and const_vector.
1172 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1174 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1175 mode iterator. Update predicate on operand 1 to
1176 handle non-const_vec constants. Delete constraints.
1177 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1179 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1181 PR tree-optimization/78821
1182 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1184 (store_operand_info::store_operand_info): Initialize it to false.
1185 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1186 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1187 store in the group, and if chain_info is non-NULL, to ignore altogether
1189 (compatible_load_p): Fail if bit_not_p does not match.
1190 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1191 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1192 (pass_store_merging::process_store): Adjust
1193 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1194 call terminate_all_aliasing_chains newly when adding a store into
1195 a chain with non-NULL chain_info.
1197 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1199 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1201 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1203 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1205 (aarch64_layout_frame): Initialise emit_frame_chain.
1206 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1207 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1209 2017-11-08 Martin Liska <mliska@suse.cz>
1211 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1212 of gimple_call_internal_p.
1214 2017-11-07 Tom de Vries <tom@codesourcery.com>
1216 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1218 2017-11-07 Tom de Vries <tom@codesourcery.com>
1220 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1223 2017-11-08 Martin Liska <mliska@suse.cz>
1226 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1228 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1230 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1232 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1235 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1236 (*cmp<mode>_ccz_1): New insn with $k alternative.
1239 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1240 RTX_COMM_COMPARE as commutative as well.
1241 (ix86_binary_operator_ok): Formatting fix.
1242 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1243 *<code><mode>3<mask_name><round_saeonly_name>,
1244 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1245 *<s>mul<mode>3_highpart<mask_name>,
1246 *vec_widen_umult_even_v16si<mask_name>,
1247 *vec_widen_umult_even_v8si<mask_name>,
1248 *vec_widen_umult_even_v4si<mask_name>,
1249 *vec_widen_smult_even_v16si<mask_name>,
1250 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1251 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1252 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1253 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1254 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1255 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1256 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1257 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1258 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1259 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1260 ix86_binary_operator_ok. Formatting fixes.
1261 (*<plusminus_insn><mode>3<mask_name><round_name>,
1262 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1265 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1267 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1268 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1269 eq and ne if TARGET_ISEL.
1270 (cmp): New code_iterator.
1271 (UNS, UNSU_, UNSIK): New code_attrs.
1272 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1273 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1275 ("eq<mode>3"): ... this.
1276 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1278 ("ne<mode>3"): ... this.
1280 2017-11-07 Julia Koval <julia.koval@intel.com>
1283 * common/config/i386/i386-common.c
1284 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1285 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1286 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1287 * config/i386/i386.opt: Ditto.
1288 * config/i386/i386.c (ix86_target_string): Ditto.
1289 (ix86_option_override_internal): Ditto.
1290 (ix86_init_mpx_builtins): Move MPX to args2.
1291 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1292 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1293 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1294 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1295 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1296 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1297 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1299 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1302 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1303 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1304 (zero-extendsidi peephole2): Remove peephole.
1306 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1309 * stor-layout.c: Include attribs.h.
1310 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1311 explicit lookup of "aligned" attribute.
1313 2017-11-07 Andrew Waterman <andrew@sifive.com>
1315 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1316 (riscv_expand_block_move): Likewise.
1317 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1319 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1320 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1321 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1322 (riscv_adjust_block_mem): Likewise.
1323 (riscv_block_move_loop): Likewise.
1324 (riscv_expand_block_move): Likewise.
1325 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1327 2017-11-07 Michael Clark <michaeljclark@mac.com>
1329 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1330 (MUSL_DYNAMIC_LINKER): Likewise.
1332 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1334 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1335 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1338 2017-11-07 Richard Biener <rguenther@suse.de>
1340 * match.pd: Fix build.
1342 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1343 Jackson Woodruff <jackson.woodruff@arm.com>
1345 PR tree-optimization/71026
1346 * match.pd: Canonicalize negate in division.
1348 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1351 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1353 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1355 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1356 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1358 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1360 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1361 non-scalar integral types.
1362 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1363 (-(A-B), -(~A)): New transformations.
1365 2017-11-07 Tom de Vries <tom@codesourcery.com>
1367 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1368 semicolon after "do {} while (0)".
1369 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1370 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1371 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1372 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1373 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1375 2017-11-07 Tom de Vries <tom@codesourcery.com>
1377 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1378 after "do {} while (0)".
1379 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1380 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1381 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1382 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1383 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1385 2017-11-07 Tom de Vries <tom@codesourcery.com>
1388 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1390 (arm_rtx_costs_internal): Add missing semicolon after
1391 HANDLE_NARROW_SHIFT_ARITH call.
1393 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1395 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1396 disable isel if it was not set explicitly.
1398 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1400 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1402 (add_type_attribute) likewise.
1404 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1406 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1407 of struct ix86_frame.
1408 (ix86_initial_elimination_offset): Likewise.
1409 (ix86_expand_split_stack_prologue): Likewise.
1411 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1413 * tree-vrp.h (enum value_range_type): Update stale comment.
1415 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1416 Alan Hayward <alan.hayward@arm.com>
1417 David Sherwood <david.sherwood@arm.com>
1419 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1420 (aarch64_expand_vec_perm_const): Take the number of units too.
1421 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1422 (aarch64_expand_vec_perm_const): Likewise.
1423 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1424 (vec_perm<mode>): Update accordingly.
1426 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1427 Alan Hayward <alan.hayward@arm.com>
1428 David Sherwood <david.sherwood@arm.com>
1430 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1431 Take the number of units too.
1432 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1433 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1434 but check for a vector mode before rather than after the call.
1435 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1436 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1437 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1438 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1439 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1440 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1441 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1442 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1443 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1444 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1445 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1446 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1447 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1448 (aarch64_sqdmull2_n<mode>): Update accordingly.
1450 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1451 Alan Hayward <alan.hayward@arm.com>
1452 David Sherwood <david.sherwood@arm.com>
1454 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1455 the number of units too.
1456 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1457 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1458 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1459 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1460 (vec_store_lanesxi<mode>): Update accordingly.
1462 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1467 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1468 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1469 of units rather than the mode.
1470 * config/aarch64/iterators.md (nunits): New mode attribute.
1471 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1472 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1473 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1474 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1475 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1476 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1477 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1478 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1479 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1480 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1481 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1482 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1483 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1484 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1485 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1486 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1487 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1488 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1489 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1490 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1491 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1492 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1493 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1494 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1495 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1496 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1497 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1498 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1499 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1500 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1501 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1502 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1503 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1504 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1505 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1506 (aarch64_simd_vec_setv2di): Likewise.
1508 2017-11-06 Carl Love <cel@us.ibm.com>
1510 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1512 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1513 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1515 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1517 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1518 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1519 Add power 8 macro expansions.
1520 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1521 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1522 power 8 instructions. (VSX_XXBR): Add iterator.
1524 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1526 * config/arm/arm.md (predicable_short_it): Change default to "no",
1527 improve documentation, remove uses that are identical to the default.
1528 (enabled_for_depr_it): Rename to enabled_for_short_it.
1529 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1530 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1531 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1532 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1533 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1535 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1538 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1539 float128 helper macros, which are no longer used after deleting
1540 the old 'q' built-in functions, and moving the round to odd
1541 built-in functions to being special built-in functions.
1542 (BU_FLOAT128_2): Likewise.
1543 (BU_FLOAT128_1_HW): Likewise.
1544 (BU_FLOAT128_2_HW): Likewise.
1545 (BU_FLOAT128_3_HW): Likewise.
1546 (FABSQ): Delete old 'q' built-in functions.
1547 (COPYSIGNQ): Likewise.
1548 (SQRTF128_ODD): Move round to odd built-in functions to be
1549 special built-in functions, so that we can handle
1550 -mabi=ieeelongdouble.
1551 (TRUNCF128_ODD): Likewise.
1552 (ADDF128_ODD): Likewise.
1553 (SUBF128_ODD): Likewise.
1554 (MULF128_ODD): Likewise.
1555 (DIVF128_ODD): Likewise.
1556 (FMAF128_ODD): Likewise.
1557 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1558 built-in names to 'f128'.
1559 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1560 old 'q' built-in functions, as the machine independent code for
1561 'f128' built-in functions handles this.
1562 (rs6000_expand_builtin): Add expansion for float128 round to odd
1563 functions, keying off on -mabi=ieeelongdouble of whether to use
1564 the KFmode or TFmode variant.
1565 (rs6000_init_builtins): Initialize the _Float128 round to odd
1567 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1568 _Float128 'q' built-in functions are now mapped into the new
1569 'f128' built-in functions.
1571 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1573 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1574 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1576 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1579 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1580 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1581 * toplev.c: Include "ipa-fnsummary.h".
1582 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1584 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1586 PR tree-optimization/82838
1587 * gimple-ssa-store-merging.c
1588 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1589 on a separate gimple_seq which is then appended to seq.
1591 2017-11-06 Jeff Law <law@redhat.com>
1594 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1595 (get_probe_interval): New functions.
1596 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1597 (ix86_adjust_stack_and_probe): Likewise.
1598 (output_adjust_stack_and_probe): Likewise.
1599 (ix86_emit_probe_stack_range): Likewise.
1600 (ix86_expand_prologue): Likewise.
1602 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1604 PR tree-optimization/82816
1605 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1606 if the modes of the two types are the same.
1607 (convert_plusminus_to_widen): Likewise.
1609 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1611 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1613 (usadv16qi): New define_expand.
1614 (usadv8hi): New define_expand.
1616 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1619 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1621 (inline_transform): Likewise.
1622 * predict.c (counts_to_freqs): Remove useless conditional.
1623 * profile-count.h (profile_count::apply_scale): Move sanity check.
1624 * tree-inline.c (copy_bb): Always scale.
1625 (copy_cfg_body): Likewise.
1627 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1630 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1632 (*cmp_ite0): Add enabled_for_depr_it attribute.
1633 (*cmp_ite1): Likewise.
1635 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1637 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1640 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1642 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1643 return the value by pointer.
1644 (extract_range_from_multiplicative_op_1): Update accordingly.
1645 Return as soon as an operation fails.
1647 2017-11-05 Tom de Vries <tom@codesourcery.com>
1650 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1651 (DEF_SANITIZER_BUILTIN): ... here.
1652 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1653 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1655 2017-11-05 Tom de Vries <tom@codesourcery.com>
1658 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1660 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1661 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1662 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1664 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1665 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1666 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1668 2017-11-05 Tom de Vries <tom@codesourcery.com>
1671 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1674 2017-11-04 Michael Clark <michaeljclark@mac.com>
1676 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1677 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1679 (*addsi3_extended): Likewise.
1680 (*addsi3_extended2): Likewise.
1681 (<optab>si3): Likewise.
1682 (<optab>di3): Likewise.
1683 (<optab><mode>3): Likewise.
1684 (<*optabe>si3_internal): Likewise.
1685 (zero_extendqi<SUPERQI:mode>2): Likewise.
1686 (*add<mode>hi3): Likewise.
1687 (*xor<mode>hi3): Likewise.
1688 (<optab>di3): Likewise.
1689 (*<optab>si3_extend): Likewise.
1690 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1691 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1692 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1694 2017-11-04 Andrew Waterman <andrew@sifive.com>
1696 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1697 TARGET_STRICT_ALIGN based upon -mtune argument.
1699 2017-11-04 Andrew Waterman <andrew@sifive.com>
1701 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1703 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1705 config/i386/i386.c (choose_basereg): Use optional scratch
1706 register and add assertion.
1707 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1708 needed, and don't allocate stack.
1709 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1710 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1711 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1713 2017-11-03 Jeff Law <law@redhat.com>
1715 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1716 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1717 to probe at the start of a noreturn function.
1719 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1721 PR tree-optimization/78821
1722 * gimple-ssa-store-merging.c: Update the file comment.
1723 (MAX_STORE_ALIAS_CHECKS): Define.
1724 (struct store_operand_info): New type.
1725 (store_operand_info::store_operand_info): New constructor.
1726 (struct store_immediate_info): Add rhs_code and ops data members.
1727 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1728 and op1r arguments to the ctor, initialize corresponding data members.
1729 (struct merged_store_group): Add load_align_base and load_align
1731 (merged_store_group::merged_store_group): Initialize them.
1732 (merged_store_group::do_merge): Update them.
1733 (merged_store_group::apply_stores): Pick the constant for
1734 encode_tree_to_bitpos from one of the two operands, or skip
1735 encode_tree_to_bitpos if neither operand is a constant.
1736 (class pass_store_merging): Add process_store method decl. Remove
1737 bool argument from terminate_all_aliasing_chains method decl.
1738 (pass_store_merging::terminate_all_aliasing_chains): Remove
1739 var_offset_p argument and corresponding handling.
1740 (stmts_may_clobber_ref_p): New function.
1741 (compatible_load_p): New function.
1742 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1743 if there is overlap and rhs_code is not INTEGER_CST. For
1744 non-overlapping stores terminate group if rhs is not mergeable.
1745 (get_alias_type_for_stmts): Change first argument from
1746 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1747 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1748 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1750 (get_location_for_stmts): Change first argument from
1751 auto_vec<gimple *> & to vec<gimple *> &.
1752 (struct split_store): Remove orig_stmts data member, add orig_stores.
1753 (split_store::split_store): Create orig_stores rather than orig_stmts.
1754 (find_constituent_stmts): Renamed to ...
1755 (find_constituent_stores): ... this. Change second argument from
1756 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1757 to info structures rather than the statements.
1758 (split_group): Rename ALLOW_UNALIGNED argument to
1759 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1760 it. Adjust find_constituent_stores caller.
1761 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1762 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1763 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1764 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1765 (mem_valid_for_store_merging): New function.
1766 (handled_load): New function.
1767 (pass_store_merging::process_store): New method.
1768 (pass_store_merging::execute): Use process_store method. Adjust
1769 terminate_all_aliasing_chains caller.
1771 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1773 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1774 Return true for more constants, symbols and label references.
1775 (aarch64_valid_floating_const): Remove unused function.
1777 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1780 * config/aarch64/aarch64.c (aarch64_layout_frame):
1781 Undo forcing of LR at bottom of frame.
1783 2017-11-03 Jeff Law <law@redhat.com>
1786 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1787 for int_registers_saved.
1789 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1790 extracted from tree-ssa-dom.c.
1791 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1792 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1793 (record_equivalences_from_incoming_edge): Add additional argument
1794 to single_pred_edge_ignoring_loop_edges call.
1795 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1796 (uncprop_dom_walker::before_dom_children): Add additional argument
1797 to single_pred_edge_ignoring_loop_edges call.
1798 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1799 single_pred_edge_ignoring_loop_edges rather than open coding.
1800 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1802 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1804 * match.pd (-(-A)): Rewrite.
1806 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1808 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1809 (rs6000_emit_int_cmove): New declaration.
1810 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1811 (rs6000_emit_sISEL): Delete.
1812 (rs6000_emit_int_cmove): Make non-static.
1813 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1814 instead of rs6000_emit_sISEL.
1816 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1818 * asan.c (create_cond_insert_point): Maintain profile.
1819 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1821 * basic-block.h (struct basic_block_def): Remove frequency.
1822 (EDGE_FREQUENCY): Use to_frequency
1823 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1825 (find_traces): Update to use to_frequency.
1826 (find_traces_1_round): Likewise; use only IPA counts.
1827 (bb_to_key): Likewise.
1828 (connect_traces): Use IPA counts only.
1829 (copy_bb_p): Update to use to_frequency.
1830 (fix_up_crossing_landing_pad): Likewise.
1831 (sanitize_hot_paths): Likewise.
1832 * bt-load.c (basic_block_freq): Likewise.
1833 * cfg.c (init_flow): Set count_max to uninitialized.
1834 (check_bb_profile): Remove frequencies; check counts.
1835 (dump_bb_info): Do not dump frequencies.
1836 (update_bb_profile_for_threading): Update counts only.
1837 (scale_bbs_frequencies_int): Likewise.
1838 (MAX_SAFE_MULTIPLIER): Remove.
1839 (scale_bbs_frequencies_gcov_type): Update counts only.
1840 (scale_bbs_frequencies_profile_count): Update counts only.
1841 (scale_bbs_frequencies): Update counts only.
1842 * cfg.h (struct control_flow_graph): Add count-max.
1843 (update_bb_profile_for_threading): Update prototype.
1844 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1845 (find_many_sub_basic_blocks): Likewise.
1846 * cfgcleanup.c (try_forward_edges): Likewise.
1847 (try_crossjump_to_edge): Likewise.
1848 * cfgexpand.c (expand_gimple_cond): Likewise.
1849 (expand_gimple_tailcall): Likewise.
1850 (construct_init_block): Likewise.
1851 (construct_exit_block): Likewise.
1852 * cfghooks.c (verify_flow_info): Check consistency of counts.
1853 (dump_bb_for_graph): Do not dump frequencies.
1854 (split_block_1): Do not update frequencies.
1855 (split_edge): Do not update frequencies.
1856 (make_forwarder_block): Do not update frequencies.
1857 (duplicate_block): Do not update frequencies.
1858 (account_profile_record): Do not update frequencies.
1859 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1860 for global heuristics.
1861 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1862 (expected_loop_iterations_unbounded): Use counts only.
1863 * cfgloopmanip.c (scale_loop_profile): Simplify.
1864 (create_empty_loop_on_edge): Simplify
1866 (duplicate_loop_to_header_edge): Simplify
1867 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1868 (update_br_prob_note): Take care of removing note when profile
1870 (relink_block_chain): Do not dump frequency.
1871 (rtl_account_profile_record): Use to_frequency.
1872 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1873 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1874 (cgraph_update_edges_for_call_stmt_node): Likewise.
1875 (cgraph_edge::verify_count_and_frequency): Update.
1876 (cgraph_node::verify_node): Temporarily disable frequency verification.
1877 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1878 to_cgraph_frequency.
1879 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1880 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1882 (cgraph_node::expand_thunk): Update profile.
1883 * except.c (dw2_build_landing_pads): Do not update frequency.
1884 * final.c (compute_alignments): Use to_frequency.
1885 (dump_basic_block_info): Do not dump frequency.
1886 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1887 (dump_gimple_bb_header): Do not dump frequency.
1888 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1890 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1891 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1892 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1893 (init_before_recovery): Do not update frequency.
1894 (sched_create_recovery_edges): Do not update frequency.
1895 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1896 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1897 (ipa_cp_c_finalize): Set max_count to uninitialized.
1898 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1899 (param_change_prob): Use counts.
1900 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1902 * ipa-split.c (consider_split): Use to_frequency.
1903 (split_function): Use to_frequency.
1904 * ira-build.c (loop_compare_func): Likewise.
1905 (mark_loops_for_removal): Likewise.
1906 (mark_all_loops_for_removal): Likewise.
1907 * loop-doloop.c (doloop_modify): Do not update frequency.
1908 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1910 * lto-streamer-in.c (input_function): Update count_max.
1911 * omp-expand.c (expand_omp_taskreg): Update count_max.
1912 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1913 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1914 (maybe_hot_count_p): Use ipa counts only.
1915 (maybe_hot_bb_p): Simplify.
1916 (maybe_hot_edge_p): Simplify.
1917 (probably_never_executed): Do not take frequency argument.
1918 (probably_never_executed_bb_p): Do not pass frequency.
1919 (probably_never_executed_edge_p): Likewise.
1920 (combine_predictions_for_bb): Check that profile is nonzero.
1921 (propagate_freq): Do not set frequency.
1922 (drop_profile): Simplify.
1923 (counts_to_freqs): Simplify.
1924 (expensive_function_p): Use to_frequency.
1925 (propagate_unlikely_bbs_forward): Simplify.
1926 (determine_unlikely_bbs): Simplify.
1927 (estimate_bb_frequencies): Add hack to silence graphite issues.
1928 (compute_function_frequency): Use ipa counts.
1929 (pass_profile::execute): Update.
1930 (rebuild_frequencies): Use counts only.
1931 (force_edge_cold): Use counts only.
1932 * profile-count.c (profile_count::dump): Dump new count types.
1933 (profile_count::differs_from_p): Check compatiblity.
1934 (profile_count::to_frequency): New function.
1935 (profile_count::to_cgraph_frequency): New function.
1936 * profile-count.h (struct function): Declare.
1937 (enum profile_quality): Add profile_guessed_local and
1938 profile_guessed_global0.
1939 (class profile_proability): Decrease number of bits to 29;
1940 update from_reg_br_prob_note and to_reg_br_prob_note.
1941 (class profile_count: Update comment; decrease number of bits
1942 to 61. Check compatibility.
1943 (profile_count::compatible_p): New private member function.
1944 (profile_count::ipa_p): New member function.
1945 (profile_count::operator<): Handle global zero correctly.
1946 (profile_count::operator>): Handle global zero correctly.
1947 (profile_count::operator<=): Handle global zero correctly.
1948 (profile_count::operator>=): Handle global zero correctly.
1949 (profile_count::nonzero_p): New member function.
1950 (profile_count::force_nonzero): New member function.
1951 (profile_count::max): New member function.
1952 (profile_count::apply_scale): Handle IPA scalling.
1953 (profile_count::guessed_local): New member function.
1954 (profile_count::global0): New member function.
1955 (profile_count::ipa): New member function.
1956 (profile_count::to_frequency): Declare.
1957 (profile_count::to_cgraph_frequency): Declare.
1958 * profile.c (OVERLAP_BASE): Delete.
1959 (compute_frequency_overlap): Delete.
1960 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1961 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1962 * sched-ebb.c (rank): Use counts only.
1963 * shrink-wrap.c (handle_simple_exit): Use counts only.
1964 (try_shrink_wrapping): Use counts only.
1965 (place_prologue_for_one_component): Use counts only.
1966 * tracer.c (find_best_predecessor): Use to_frequency.
1967 (find_trace): Use to_frequency.
1968 (tail_duplicate): Use to_frequency.
1969 * trans-mem.c (expand_transaction): Do not update frequency.
1970 * tree-call-cdce.c: Do not update frequency.
1971 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1972 (gimple_merge_blocks): Likewise.
1973 (gimple_split_edge): Likewise.
1974 (gimple_duplicate_sese_region): Likewise.
1975 (gimple_duplicate_sese_tail): Likewise.
1976 (move_sese_region_to_fn): Likewise.
1977 (gimple_account_profile_record): Likewise.
1978 (insert_cond_bb): Likewise.
1979 * tree-complex.c (expand_complex_div_wide): Likewise.
1980 * tree-eh.c (lower_resx): Update profile.
1981 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1983 (initialize_cfun): Do not initialize frequencies
1984 (freqs_to_counts): Delete.
1985 (copy_cfg_body): Ignore count parameter.
1986 (copy_body): Update.
1987 (expand_call_inline): Update count_max.
1988 (optimize_inline_calls): Update count_max.
1989 (tree_function_versioning): Update count_max.
1990 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1991 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1993 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1994 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1995 (try_peel_loop): Likewise.
1996 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1998 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1999 (tree_transform_and_unroll_loop): Do not use frequencies
2000 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2001 Use reliable prediction only.
2002 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2003 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2004 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2005 probability scaling.
2006 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2007 not update frequency
2008 (any_remaining_duplicated_blocks): Likewise.
2009 (update_profile): Likewise.
2010 (estimated_freqs_path): Delete.
2011 (freqs_to_counts_path): Delete.
2012 (clear_counts_path): Delete.
2013 (ssa_fix_duplicate_block_edges): Likewise.
2014 (duplicate_thread_path): Likewise.
2015 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2016 * tree-tailcall.c (decrease_profile): Do not update frequency.
2017 (eliminate_tail_call): Likewise.
2018 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2019 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2020 (optimize_mask_stores): Likewise.
2021 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2022 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2023 (ubsan_expand_ptr_ifn): Update profile.
2024 * value-prof.c (gimple_ic): Simplify.
2025 * value-prof.h (gimple_ic): Update prototype.
2026 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2027 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2029 (want_inline_self_recursive_call_p): Likewise.
2030 (resolve_noninline_speculation): Only cummulate defined counts.
2031 (inline_small_functions): Use nonzero_p.
2032 (ipa_inline): Do not access freed node.
2034 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2036 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2037 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2039 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2041 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2042 non-legitimate address.
2044 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2046 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2047 (*lt0_<mode>di, *lt0_<mode>si): New.
2049 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2051 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2052 TARGET_PAIRED_FLOAT.
2054 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2055 Jim Wilson <jim.wilson@linaro.org>
2057 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2058 * config/aarch64/aarch64-tune.md: Regenerated.
2059 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2060 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2062 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2064 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2065 (arc_expand_prologue): Restore blink for millicode.
2066 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2068 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2071 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2072 gen_vec_duplicate after forcing the scalar into a register.
2074 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2076 * combine (try_combine): Print the insns input to try_combine to the
2079 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2082 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2083 Remove second argument from aarch64_process_target_attr call.
2084 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2086 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2088 (aarch64_handle_attr_arch): Remove second argument.
2089 (aarch64_handle_attr_cpu): Ditto.
2090 (aarch64_handle_attr_tune): Ditto.
2091 (aarch64_handle_attr_isa_flags): Ditto.
2092 (aarch64_process_one_target_attr): Ditto.
2093 (aarch64_process_target_attr): Ditto.
2094 (aarch64_option_valid_attribute_p): Remove second argument.
2095 on aarch64_process_target_attr call.
2097 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2099 * diagnostic.c: Include "selftest-diagnostic.h".
2100 (selftest::assert_location_text): New function.
2101 (selftest::test_diagnostic_get_location_text): New function.
2102 (selftest::diagnostic_c_tests): Call it.
2104 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2106 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2107 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2108 (class selftest::test_diagnostic_context): Move to...
2109 * selftest-diagnostic.c: New file.
2110 * selftest-diagnostic.h: New file.
2112 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2114 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2115 offset range for FT32B.
2116 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2117 * config/ft32/ft32.md: Add TARGET_NOPM.
2118 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2119 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2121 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2123 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2125 2017-11-02 Jeff Law <law@redhat.com>
2127 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2128 virtual keyword on FINAL OVERRIDE members.
2130 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2131 virtuals together. Add virtual destructor.
2132 (substitute_and_fold_engine): Similarly.
2134 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2136 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2138 2017-11-02 Richard Biener <rguenther@suse.de>
2140 PR tree-optimization/82795
2141 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2143 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2145 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2147 (gcc_GAS_CHECK_FEATURE): Remove.
2148 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2149 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2150 * configure: Regenerate.
2152 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2154 * config/arc/arc.c (hwloop_optimize): Account for empty
2157 2017-11-02 Richard Biener <rguenther@suse.de>
2160 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2161 Truncate ARRAY_REF index and element size.
2163 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2165 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2167 2017-11-01 Jeff Law <law@redhat.com>
2169 * tree-ssa-ccp.c (ccp_folder): New class derived from
2170 substitute_and_fold_engine.
2171 (ccp_folder::get_value): New member function.
2172 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2173 (ccp_fold_stmt): Remove prototype.
2174 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2175 * tree-ssa-copy.c (copy_folder): New class derived from
2176 substitute_and_fold_engine.
2177 (copy_folder::get_value): Renamed from get_value.
2178 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2179 * tree-vrp.c (vrp_folder): New class derived from
2180 substitute_and_fold_engine.
2181 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2182 (vrp_folder::get_value): New member function.
2183 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2184 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2185 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2186 provide a class interface to folder/substitute routines.
2187 (ssa_prop_fold_stmt_fn): Remove typedef.
2188 (ssa_prop_get_value_fn): Likewise.
2189 (subsitute_and_fold): Remove prototype.
2190 (replace_uses_in): Likewise.
2191 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2192 Renamed from replace_uses_in. Call the virtual member function
2193 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2194 (substitute_and_fold_dom_walker): Remove initialization of
2195 data member entries for calbacks. Add substitute_and_fold_engine
2196 member and initialize it.
2197 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2198 member functions for get_value, replace_phi_args_in c
2199 replace_uses_in, and fold_stmt calls.
2200 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2201 substitute_and_fold. Remove assert. Update ctor call.
2203 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2204 (ssa_prop_visit_phi_fn): Likewise.
2205 (class ssa_propagation_engine): New class to provide an interface
2207 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2209 (ssa_prop_visit_phi): Likewise.
2210 (ssa_propagation_engine::simulate_stmt): Moved into class.
2211 Call visit_phi/visit_stmt from the class rather than via
2212 file scoped static variables.
2213 (ssa_propagation_engine::simulate_block): Moved into class.
2214 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2215 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2216 set file scoped statics for the visit_stmt/visit_phi callbacks.
2217 * tree-complex.c (complex_propagate): New class derived from
2218 ssa_propagation_engine.
2219 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2220 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2221 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2223 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2224 ssa_propagation_engine.
2225 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2226 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2227 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2228 * tree-ssa-copy.c (copy_prop): New class derived from
2229 ssa_propagation_engine.
2230 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2231 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2232 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2233 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2234 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2235 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2236 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2238 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2240 PR rtl-optimization/82778
2241 PR rtl-optimization/82597
2242 * compare-elim.c (struct comparison): Add in_a_setter field.
2243 (find_comparison_dom_walker::before_dom_children): Remove killed
2244 bitmap and df_simulate_find_defs call, instead walk the defs.
2245 Compute last_setter and initialize in_a_setter. Merge definitions
2246 with first initialization for a few variables.
2247 (try_validate_parallel): Use insn_invalid_p instead of
2248 recog_memoized. Return insn rather than just the pattern.
2249 (try_merge_compare): Fix up comment. Don't uselessly test if
2250 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2252 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2255 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2256 Alan Hayward <alan.hayward@arm.com>
2257 David Sherwood <david.sherwood@arm.com>
2259 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2260 aarch64_hard_regno_nregs to get the number of registers
2263 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2264 Alan Hayward <alan.hayward@arm.com>
2265 David Sherwood <david.sherwood@arm.com>
2267 * config/aarch64/constraints.md (Upl): Rename to...
2269 * config/aarch64/aarch64.md
2270 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2278 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2281 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2282 Alan Hayward <alan.hayward@arm.com>
2283 David Sherwood <david.sherwood@arm.com>
2285 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2286 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2287 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2289 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2290 of a single element.
2291 * config/aarch64/iterators.md: Add a comment above the permute
2292 unspecs to say that they are generated directly by
2293 aarch64_expand_vec_perm_const.
2294 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2296 2017-11-01 Nathan Sidwell <nathan@acm.org>
2298 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2300 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2302 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2303 and medany code models, and describe what they do.
2305 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2307 Revert accidental duplicate:
2309 * combine.c (can_change_dest_mode): Reject changes in
2310 REGMODE_NATURAL_SIZE.
2312 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2314 PR rtl-optimization/64682
2315 PR rtl-optimization/69567
2316 PR rtl-optimization/69737
2317 PR rtl-optimization/82683
2318 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2319 register mentioned in the note, drop the note, unless it came from I3,
2320 in which case it should go to I3 again.
2322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2324 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2325 and return false if not.
2326 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2328 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2330 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2331 range is known to be empty.
2333 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2338 and CONST_VECTOR_NUNITS instead of computing the number of units from
2339 the byte sizes of the vector and element.
2340 (simplify_binary_operation_1): Likewise.
2341 (simplify_const_binary_operation): Likewise.
2342 (simplify_ternary_operation): Likewise.
2344 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2349 (int_mem_offset): ...this new function.
2350 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2351 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2352 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2355 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * lower-subreg.c (interesting_mode_p): New function.
2360 (compute_costs, find_decomposable_subregs, decompose_register)
2361 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2362 (resolve_clobber, dump_choices): Use it.
2364 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2365 Alan Hayward <alan.hayward@arm.com>
2366 David Sherwood <david.sherwood@arm.com>
2368 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2370 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2371 Alan Hayward <alan.hayward@arm.com>
2372 David Sherwood <david.sherwood@arm.com>
2374 * alias.c (find_base_value, find_base_term): Only process integer
2375 truncations. Check the precision rather than the size.
2377 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2378 Alan Hayward <alan.hayward@arm.com>
2379 David Sherwood <david.sherwood@arm.com>
2381 * machmode.h (is_narrower_int_mode): New function
2382 * optabs.c (expand_float, expand_fix): Use it.
2383 * dwarf2out.c (rotate_loc_descriptor): 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 * rtl.h (narrower_subreg_mode): New function.
2390 * ira-color.c (update_costs_from_allocno): Use it.
2392 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2393 Alan Hayward <alan.hayward@arm.com>
2394 David Sherwood <david.sherwood@arm.com>
2396 * optabs-query.h (convert_optab_p): New function, split out from...
2397 (convert_optab_handler): ...here.
2398 (widening_optab_handler): Delete.
2399 (find_widening_optab_handler): Remove permit_non_widening parameter.
2400 (find_widening_optab_handler_and_mode): Likewise. Provide an
2401 override that operates on mode class wrappers.
2402 * optabs-query.c (widening_optab_handler): Delete.
2403 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2404 parameter. Assert that the two modes are the same class and that
2405 the "from" mode is narrower than the "to" mode. Use
2406 convert_optab_handler instead of widening_optab_handler.
2407 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2408 instead of widening_optab_handler.
2409 * expr.c (expand_expr_real_2): Update calls to
2410 find_widening_optab_handler.
2411 * optabs.c (expand_widen_pattern_expr): Likewise.
2412 (expand_binop_directly): Take the insn_code as a parameter.
2413 (expand_binop): Only call find_widening_optab_handler for
2414 conversion optabs; use optab_handler otherwise. Update calls
2415 to find_widening_optab_handler and expand_binop_directly.
2416 Use convert_optab_handler instead of widening_optab_handler.
2417 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2418 find_widening_optab_handler and use scalar_mode rather than
2420 (convert_plusminus_to_widen): Likewise.
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 * machmode.h (fixed_size_mode): New class.
2427 * rtl.h (get_pool_mode): Return fixed_size_mode.
2428 * gengtype.c (main): Add fixed_size_mode.
2429 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2430 (get_raw_arg_mode): Likewise.
2431 * doc/tm.texi: Regenerate.
2432 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2433 * targhooks.c (default_get_reg_raw_mode): Likewise.
2434 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2435 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2436 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2437 (msp430_get_raw_result_mode): Likewise.
2438 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2439 * dbxout.c (dbxout_parms): Require fixed-size modes.
2440 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2441 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2442 * omp-low.c (lower_oacc_reductions): Likewise.
2443 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2444 (simplify_subreg): Update accordingly.
2445 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2446 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2447 that aren't fixed-size.
2448 (get_pool_mode): Return a fixed_size_mode.
2449 (output_constant_pool_2): Take a fixed_size_mode.
2451 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2452 Alan Hayward <alan.hayward@arm.com>
2453 David Sherwood <david.sherwood@arm.com>
2455 * doc/rtl.texi (vec_series): Document.
2456 (const): Say that the operand can be a vec_series.
2457 * rtl.def (VEC_SERIES): New rtx code.
2458 * rtl.h (const_vec_series_p_1): Declare.
2459 (const_vec_series_p): New function.
2460 * emit-rtl.h (gen_const_vec_series): Declare.
2461 (gen_vec_series): Likewise.
2462 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2463 (gen_vec_series): Likewise.
2464 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2465 * simplify-rtx.c (simplify_unary_operation): Handle negations
2467 (simplify_binary_operation_series): New function.
2468 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2469 (test_vector_ops_series): New function.
2470 (test_vector_ops): Call it.
2471 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2472 gen_const_vec_series.
2473 (altivec_lvsr): Likewise.
2474 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2476 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2477 Alan Hayward <alan.hayward@arm.com>
2478 David Sherwood <david.sherwood@arm.com>
2480 * doc/rtl.texi (const): Update description of address constants.
2481 Say that vector constants are allowed too.
2482 * common.md (E, F): Use CONSTANT_P instead of checking for
2484 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2485 checking for CONST_VECTOR.
2486 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2488 * expr.c (expand_expr_real_2): Check for vector modes instead
2489 of checking for CONST_VECTOR.
2490 * rtl.h (const_vec_p): New function.
2491 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2492 (unwrap_const_vec_duplicate): Handle them here too.
2494 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2495 David Malcolm <dmalcolm@redhat.com>
2496 Alan Hayward <alan.hayward@arm.com>
2497 David Sherwood <david.sherwood@arm.com>
2499 * rtl.h (vec_duplicate_p): New function.
2500 * selftest-rtl.c (assert_rtx_eq_at): New function.
2501 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2502 (assert_rtx_eq_at): Declare.
2503 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2504 * selftest-run-tests.c (selftest::run_tests): Call it.
2505 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2506 (simplify_unary_operation_1): Recursively handle vector duplicates.
2507 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2509 (simplify_subreg): Handle subregs of vector duplicates.
2510 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2511 (selftest::simplify_rtx_c_tests): New functions.
2513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2514 Alan Hayward <alan.hayward@arm.com>
2515 David Sherwood <david.sherwood@arm.com>
2517 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2518 (gen_vec_duplicate): Likewise.
2519 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2521 (gen_const_vector): ...here.
2522 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2523 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2524 whose elements are all equal.
2525 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2526 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2527 (simplify_relational_operation): Likewise.
2528 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2530 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2531 (aarch64_expand_vector_init): Likewise.
2532 * config/arm/arm.c (neon_vdup_constant): Likewise.
2533 (neon_expand_vector_init): Likewise.
2534 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2535 (arm_block_set_unaligned_vect): Likewise.
2536 (arm_block_set_aligned_vect): Likewise.
2537 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2538 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2539 (expand_vec_perm_even_odd_pack): Likewise.
2540 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2541 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2542 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2543 gen_const_vec_duplicate.
2544 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2545 * config/mips/mips.c (mips_gen_const_int_vector): Use
2546 gen_const_vec_duplicate.
2547 (mips_expand_vector_init): Use CONST0_RTX.
2548 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2549 (define_split): Use gen_const_vec_duplicate.
2550 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2551 (define_split): Use gen_const_vec_duplicate.
2552 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2553 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2554 * config/spu/spu.c (spu_const): Likewise.
2556 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2557 Alan Hayward <alan.hayward@arm.com>
2558 David Sherwood <david.sherwood@arm.com>
2560 * combine.c (can_change_dest_mode): Reject changes in
2561 REGMODE_NATURAL_SIZE.
2563 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2565 * configure.ac (--enable-libssp): New.
2566 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2567 trying to determine target-specific default. Adjust indentation.
2568 * configure: Regenerated.
2569 * doc/install.texi (Configuration): Expand --disable-libssp
2572 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2574 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2577 2017-10-31 Martin Jambor <mjambor@suse.cz>
2580 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2582 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2584 * auto-profile.c (autofdo_source_profile::read): Use
2585 UNKNOWN_LOCATION rather than 0.
2586 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2587 (warning_at): ...this overload.
2588 (warning_at_rich_loc_n): Rename to...
2589 (warning_n): ...this overload.
2590 (error_at_rich_loc): Rename to...
2591 (error_at): ...this overload.
2592 (pedwarn_at_rich_loc): Rename to...
2593 (pedwarn): ...this overload.
2594 (permerror_at_rich_loc): Rename to...
2595 (permerror): ...this overload.
2596 (inform_at_rich_loc): Rename to...
2597 (inform): ...this overload.
2598 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2599 (diagnostic_n_impl_richloc): Rename to...
2600 (diagnostic_n_impl): ...this rich_location *-based decl.
2601 (inform_at_rich_loc): Rename to...
2602 (inform): ...this, and add an assertion.
2603 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2604 (warning_at_rich_loc): Rename to...
2605 (warning_at): ...this, and add an assertion.
2606 (warning_at_rich_loc_n): Rename to...
2607 (warning_n): ...this, and add an assertion.
2608 (warning_n): Update location_t-based implementation for removal of
2609 location_t-based diagnostic_n_impl.
2610 (pedwarn_at_rich_loc): Rename to...
2611 (pedwarn): ...this, and add an assertion.
2612 (permerror_at_rich_loc): Rename to...
2613 (permerror): ...this, and add an assertion.
2614 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2615 (error_at_rich_loc): Rename to...
2616 (error_at): ...this, and add an assertion.
2617 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2618 (driver::do_spec_on_infiles): Likewise.
2619 * substring-locations.c (format_warning_va): Update for renaming
2620 of inform_at_rich_loc.
2622 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2624 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2625 _Float<N>X built-in functions so that the variant without the
2626 "__builtin_" prefix is only enabled for the GNU C and Objective C
2627 languages when they are in non-strict ANSI/ISO mode.
2628 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2629 * target.def (floatn_builtin_p): Add a target hook to control
2630 whether _Float<N> and _Float<N>X built-in functions without the
2631 "__builtin_" prefix are enabled, and return true for C and
2632 Objective C in the default hook. Include langhooks.h in
2634 * targhooks.h (default_floatn_builtin_p): Likewise.
2635 * targhooks.c (default_floatn_builtin_p): Likewise.
2636 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2637 floatn_builtin_p target hook.
2638 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2640 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2641 Eric Botcazou <ebotcazou@adacore.com>
2643 PR rtl-optimization/81803
2644 * lra-constraints.c (curr_insn_transform): Also reload the whole
2645 register for a strict subreg no wider than a word if this is for
2646 a WORD_REGISTER_OPERATIONS target.
2648 2017-10-31 Jason Merrill <jason@redhat.com>
2650 * gdbinit.in: Skip over inlines from timevar.h.
2652 2017-10-31 Martin Liska <mliska@suse.cz>
2654 * doc/gcov.texi: Document new option.
2655 * gcov.c (print_usage): Likewise print it.
2656 (process_args): Support the argument.
2657 (format_count): New function.
2658 (format_gcov): Use the function.
2660 2017-10-31 Martin Liska <mliska@suse.cz>
2662 * gcov.c (struct name_map): do not use typedef.
2663 Define operator== and operator<.
2664 (name_search): Remove.
2665 (name_sort): Remove.
2666 (main): Do not allocate names.
2667 (process_file): Add vertical space.
2668 (generate_results): Use std::find.
2669 (release_structures): Do not release memory.
2670 (find_source): Use std::find.
2672 2017-10-31 Martin Liska <mliska@suse.cz>
2674 * gcov.c (struct line_info): Remove it's typedef.
2675 (line_info::line_info): Add proper ctor.
2676 (line_info::has_block): Do not use a typedef.
2677 (struct source_info): Do not use typedef.
2678 (circuit): Likewise.
2679 (get_cycles_count): Likewise.
2680 (output_intermediate_file): Iterate via vector iterator.
2681 (add_line_counts): Use std::vector methods.
2682 (accumulate_line_counts): Likewise.
2683 (output_lines): Likewise.
2685 2017-10-31 Martin Liska <mliska@suse.cz>
2687 * gcov.c (struct source_info): Remove typedef.
2688 (source_info::source_info): Add proper ctor.
2689 (accumulate_line_counts): Use struct, not it's typedef.
2690 (output_gcov_file): Likewise.
2691 (output_lines): Likewise.
2692 (main): Do not allocate an array.
2693 (output_intermediate_file): Use size of vector container.
2694 (process_file): Resize the vector.
2695 (generate_results): Do not preallocate, use newly added vector
2697 (release_structures): Do not release sources.
2698 (find_source): Use vector methods.
2699 (add_line_counts): Do not use typedef.
2701 2017-10-31 Martin Liska <mliska@suse.cz>
2703 * doc/gcov.texi: Document that.
2704 * gcov.c (add_line_counts): Mark lines with a non-executed
2706 (output_line_beginning): Handle such lines.
2707 (output_lines): Pass new argument.
2708 (output_intermediate_file): Print it in intermediate format.
2710 2017-10-31 Martin Liska <mliska@suse.cz>
2712 * color-macros.h: New file.
2713 * diagnostic-color.c: Factor out color related to macros to
2715 * doc/gcov.texi: Document -k option.
2716 * gcov.c (INCLUDE_STRING): Include string.h.
2717 (print_usage): Add -k option.
2718 (process_args): Parse it.
2719 (pad_count_string): New function.
2720 (output_line_beginning): Likewise.
2721 (DEFAULT_LINE_START): New macro.
2722 (output_lines): Support color output.
2724 2017-10-31 Martin Liska <mliska@suse.cz>
2726 PR gcov-profile/82633
2727 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2728 their interaction with GCOV infrastructure.
2729 * configure.ac: Add -fkeep-{inline,static}-functions to
2731 * configure: Regenerate.
2733 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2736 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2738 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2741 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2742 into a register if it does not fit into an immediate offset field.
2744 2017-10-31 Olivier Hainque <hainque@adacore.com>
2746 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2748 2017-10-31 Julia Koval <julia.koval@intel.com>
2750 * config.gcc: Add gfniintrin.h.
2751 * config/i386/gfniintrin.h: New.
2752 * config/i386/i386-builtin-types.def
2753 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2754 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2755 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2756 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2757 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2758 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2759 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2760 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2761 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2762 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2763 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2764 * config/i386/immintrin.h: Include gfniintrin.h.
2765 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2767 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2769 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2771 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2773 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2774 (arm_ashldi3_1bit): Remove pattern.
2775 (ashrdi3): Remove shift by 1 expansion.
2776 (arm_ashrdi3_1bit): Remove pattern.
2777 (lshrdi3): Remove shift by 1 expansion.
2778 (arm_lshrdi3_1bit): Remove pattern.
2779 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2780 cost of ashldi3 by 1.
2781 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2782 (<shift>di3_neon): Likewise.
2784 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2786 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2787 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2788 and (*aarch64_simd_mov<VQ:mode>).
2789 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2790 pattern alternative.
2791 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2792 attributes to match pattern alternative.
2794 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2796 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2797 * config/rs6000/emmintrin.h: New file.
2798 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2800 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2802 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2803 * (movdi_vfp_cortexa8): Remove pattern.
2805 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2807 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2809 (Specific, alpha*-dec-osf5.1): Remove.
2810 (Specific, mips-sgi-irix5): Remove.
2811 (Specific, mips-sgi-irix6): Remove.
2813 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2816 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2817 arguments to clear_bit_region_be.
2819 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2821 * gimplify.c: Include memmodel.h.
2823 2017-10-30 Martin Jambor <mjambor@suse.cz>
2825 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2826 condition whether loop should be executed at all.
2828 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2831 gimple folding of vec_madd() intrinsics.
2832 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2833 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2834 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2836 2017-10-30 Richard Biener <rguenther@suse.de>
2838 PR tree-optimization/82762
2840 2017-10-23 Richard Biener <rguenther@suse.de>
2842 PR tree-optimization/82129
2844 2017-08-01 Richard Biener <rguenther@suse.de>
2846 PR tree-optimization/81181
2847 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2848 (compute_antic): ... end of iteration here.
2850 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2852 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2853 -std=iso9899:2017 and -std=gnu17.
2854 * doc/standards.texi (C Language): Document C17 support.
2855 * doc/cpp.texi (Overview): Mention -std=c17.
2856 (Standard Predefined Macros): Document C11 and C17 values of
2857 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2858 * doc/extend.texi (Inline): Do not list individual options for
2859 standards newer than C99.
2860 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2862 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2865 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2867 * asan.c (asan_finish_file): Align asan globals array by shadow
2870 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2873 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2874 (struct store_immediate_info): Add bitregion_start and bitregion_end
2876 (store_immediate_info::store_immediate_info): Add brs and bre
2877 arguments and initialize bitregion_{start,end} from those.
2878 (struct merged_store_group): Add bitregion_start, bitregion_end,
2879 align_base and mask fields. Drop unnecessary struct keyword from
2880 struct store_immediate_info. Add do_merge method.
2881 (clear_bit_region_be): Use memset instead of loop storing zeros.
2882 (merged_store_group::do_merge): New method.
2883 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2884 stores as long as the surrounding bitregions have no gaps.
2885 (merged_store_group::merge_overlapping): Use do_merge.
2886 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2887 is byte aligned, rather than requiring that start and width are
2888 byte aligned. Drop unnecessary struct keyword from
2889 struct store_immediate_info. Allocate and populate also mask array.
2890 Make start of the arrays relative to bitregion_start rather than
2891 start and size them according to bitregion_{end,start} difference.
2892 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2893 struct store_immediate_info.
2894 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2895 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2896 struct keyword from struct store_immediate_info.
2897 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2898 between stores as long as the surrounding bitregions have no gaps.
2900 (struct split_store): Add orig non-static data member.
2901 (split_store::split_store): Initialize orig to false.
2902 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2903 if there is exactly a single original stmt. Change stmts argument
2904 to pointer from reference, if NULL, don't push anything to it. Add
2905 first argument, use it to optimize skipping over orig stmts that
2906 are known to be before bitpos already. Simplify.
2907 (split_group): Return unsigned int count how many stores are or
2908 would be needed rather than a bool. Add allow_unaligned argument.
2909 Change split_stores argument from reference to pointer, if NULL,
2910 only do a dry run computing how many stores would be produced.
2911 Rewritten algorithm to use both alignment and misalign if
2912 !allow_unaligned and handle bitfield stores with gaps.
2913 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2914 from bitregion_start instead of start. Compute allow_unaligned
2915 here, if true, do 2 split_group dry runs to compute which one
2916 produces fewer stores and prefer aligned if equal. Punt if
2917 new count is bigger or equal than original before emitting any
2918 statements, rather than during that. Remove no longer needed
2919 new_ssa_names tracking. Replace num_stmts with
2920 split_stores.length (). Use 32-bit stack allocated entries
2921 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2922 if possible. Handle bitfields with gaps.
2923 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2924 Compute bitregion_{start,end} for the stores and construct
2925 store_immediate_info with that. Formatting fixes.
2927 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2930 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2931 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2933 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2935 * gimplify.c: Include tm_p.h.
2937 * common.opt (gcoff): Re-add as ignored option.
2938 (gcoff1, gcoff2, gcoff3): Likewise.
2940 * Makefile.in (OBJS): Delete sdbout.o.
2941 (GTFILES): Delete $(srcdir)/sdbout.c.
2942 * debug.h: Delete sdb_debug_hooks.
2943 * final.c: Delete sdbout.h include.
2944 (final_scan_insn): Delete SDB_DEBUG check.
2945 (rest_of_clean_state): Likewise.
2946 * output.h: Delete sdb_begin_function_line.
2949 * toplev.c: Delete sdbout.h include.
2950 (process_options): Delete SDB_DEBUG check.
2951 * tree-core.h (tree_type_common): Delete pointer field of
2953 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2954 TYPE_SYMTAB_POINTER.
2955 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2956 (TYPE_SYMTAB_IS_POINTER): Delete.
2957 (TYPE_SYMTAB_IS_DIE): Renumber.
2958 * xcoffout.c: Refer to former sdbout.c file.
2959 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2961 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2962 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2963 (Debugging Options): Delete -gcoff.
2964 (-gstabs): Delete SDB reference.
2966 (-gcoff@var{level}): Delete.
2967 * doc/passes.texi (Debugging information output): Delete SDB and
2968 sdbout.c references.
2969 * doc/tm.texi: Regenerate.
2970 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2971 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2973 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2974 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2975 (SDB_DEBUGGING_INFO): Delete.
2976 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2977 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2978 * target.def (output_source_filename): Delete COFF reference.
2980 * common.opt (gcoff): Delete.
2981 (gxcoff+): Update Negative chain.
2982 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2984 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2985 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2986 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2988 (expand_function_start): Change sdb reference to past tense.
2989 (expand_function_end): Change sdb reference to past tense.
2990 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2991 * opts.c (debug_type_names): Delete coff entry.
2992 (common_handle_option): Delete OPT_gcoff case.
2993 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2995 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2996 * config/cris/cris.h: Delete SDB reference in comment.
2997 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2998 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2999 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3000 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3002 (ix86_expand_prologue): Likewise.
3003 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3004 * config/ia64/ia64.h: Likewise.
3005 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3006 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3008 * config/mmix/mmix.h: Likewise.
3009 * config/nds32/nds32.c: Likewise.
3010 * config/stormy/storym16.h: Likewise.
3011 * config/visium/visium.h: Likewise.
3012 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3014 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3016 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3017 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3018 FRAME_POINTER_REGNUM point at high end of local var area.
3020 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3022 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3023 Move comment around. Do not reset best_edge for a copiable
3024 destination if the copy would cause a partition change.
3025 (better_edge_p): Remove redundant check.
3027 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3029 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3031 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3033 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3034 for math functions that have _Float<N> and _Float<N>X variants.
3035 (mathfn_built_in_2): Add support for math functions that have
3036 _Float<N> and _Float<N>X variants.
3037 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3038 (expand_builtin_mathfn_ternary): Add support for fma with
3039 _Float<N> and _Float<N>X variants.
3040 (expand_builtin): Likewise.
3041 (fold_builtin_3): Likewise.
3042 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3043 create math function _Float<N> and _Float<N>X variants as external
3045 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3046 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3047 the __builtin_ prefix and if not strict ansi, without the prefix.
3048 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3049 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3050 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3051 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3052 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3053 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3054 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3055 function signatures for fma _Float<N> and _Float<N>X variants.
3056 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3057 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3058 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3059 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3060 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3061 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3062 * gencfn-macros.c (print_case_cfn): Add support for math functions
3063 that have _Float<N> and _Float<N>X variants.
3064 (print_define_operator_list): Likewise.
3065 (fltfn_suffixes): Likewise.
3067 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3068 for math functions that have _Float<N> and _Float<N>X variants.
3069 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3070 and _Float<N>X variants.
3071 (COPYSIGN): Likewise.
3074 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3075 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3077 (integer_valued_read_call_p): Likewise.
3078 * fold-const-call.c (fold_const_call_ss): Likewise.
3079 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3080 _Float<N> and _Float<N>X variants.
3081 (fold_const_call_ssss): Add support for fma _Float<N> and
3082 _Float<N>X variants.
3083 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3084 support for copysign and fma _Float<N> and _Float<N>X variants.
3085 (backprop::process_builtin_call_use): Likewise.
3086 * tree-call-cdce.c (can_test_argument_range); Add support for
3087 sqrt _Float<N> and _Float<N>X variants.
3088 (edom_only_function): Likewise.
3089 (get_no_error_domain): Likewise.
3090 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3091 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3092 copysign _Float<N> and _Float<N>X variants.
3093 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3094 handled by machine independent code.
3095 (FMAF128): Likewise.
3096 * doc/cpp.texi (Common Predefined Macros): Document defining
3097 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3098 fma _Float<N> and _Float<N>X variants.
3100 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3103 * config/i386/i386-modes.def (CCFPU): Remove definition.
3104 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3105 (ix86_cc_modes_compatible): Ditto.
3106 (ix86_expand_carry_flag_compare): Ditto.
3107 (ix86_expand_int_movcc): Ditto.
3108 (ix86_expand_int_addcc): Ditto.
3109 (ix86_reverse_condition): Ditto.
3110 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3111 Return true/false for unordered/ordered fp comparisons.
3112 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3113 (ix86_prepare_fp_compare_args): Update for rename.
3114 (ix86_expand_fp_compare): Update for rename. Generate unordered
3115 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3116 (ix86_expand_sse_compare_and_jump): Ditto.
3117 * config/i386/predicates.md (fcmov_comparison_operator):
3118 Remove CCFPU mode handling.
3119 (ix86_comparison_operator): Ditto.
3120 (ix86_carry_flag_operator): Ditto.
3121 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3122 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3123 (*cmpu<mode>_cc_i387): Ditto.
3124 (FPCMP): Remove mode iterator.
3125 (unord): Remove mode attribute.
3126 (unord_subst): New define_subst transformation
3127 (unord): New define_subst attribute.
3129 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3130 (*cmpi<unord>xf_i387): Ditto.
3131 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3132 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3133 using unord_subst transformation.
3134 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3135 (round_saeonly): Also handle CCFP mode.
3136 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3137 Remove UNSPEC_SAHF unspec handling.
3139 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3141 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3143 2017-10-27 Jeff Law <law@redhat.com>
3145 * gimple-ssa-sprintf.c: Include domwalk.h.
3146 (class sprintf_dom_walker): New class, derived from dom_walker.
3147 (sprintf_dom_walker::before_dom_children): New function.
3148 (struct call_info): Moved into sprintf_dom_walker class
3149 (compute_formath_length, handle_gimple_call): Likewise.
3150 (sprintf_length::execute): Call the dominator walker rather
3151 than walking the statements.
3153 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3154 gimple statement locations.
3155 (check_array_bounds): Corresponding changes. Get the statement's
3156 location directly from wi->stmt.
3158 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3161 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3163 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3165 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3166 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3168 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3171 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3172 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3173 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3174 maybe_get_pool_constant.
3175 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3178 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3180 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3181 2.26 caveat. Update gas and gld versions.
3182 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3185 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3187 * cgraph.h (set_malloc_flag): Declare.
3188 * cgraph.c (set_malloc_flag_1): New function.
3189 (set_malloc_flag): Likewise.
3190 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3191 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3193 (read_ipa_call_summary): Add support for reading is_return_callee.
3194 (write_ipa_call_summary): Stream is_return_callee.
3195 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3196 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3197 ipa-prop.h, ipa-fnsummary.h.
3198 (pure_const_names): Change to static.
3199 (malloc_state_e): Define.
3200 (malloc_state_names): Define.
3201 (funct_state_d): Add field malloc_state.
3202 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3203 (check_retval_uses): New function.
3204 (malloc_candidate_p): Likewise.
3205 (analyze_function): Add support for malloc attribute.
3206 (pure_const_write_summary): Stream malloc_state.
3207 (pure_const_read_summary): Add support for reading malloc_state.
3208 (dump_malloc_lattice): New function.
3209 (propagate_malloc): New function.
3210 (warn_function_malloc): New function.
3211 (ipa_pure_const::execute): Call propagate_malloc and
3212 ipa_free_fn_summary.
3213 (pass_local_pure_const::execute): Add support for malloc attribute.
3214 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3215 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3217 2017-10-27 Martin Liska <mliska@suse.cz>
3219 PR gcov-profile/82457
3220 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3221 for fork-like functions to be properly instrumented.
3223 2017-10-27 Richard Biener <rguenther@suse.de>
3226 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3227 info when we redirected EH.
3229 2017-10-26 Michael Collison <michael.collison@arm.com>
3231 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3233 (<optab>_trunchf<GPI:mode>2: New pattern.
3234 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3235 * config/aarch64/iterators.md (wv): New mode attribute.
3236 (vf, VF): New mode attributes.
3237 (vgp, VGP): New mode attributes.
3238 (s): Update attribute with SImode and DImode prefixes.
3240 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3242 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3243 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3244 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3245 (nios2_option_overide): Initialize it. Don't allow R0-relative
3246 addressing with PIC.
3247 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3248 (nios2_symbolic_constant_p): Likewise.
3249 (nios2_legitimate_address_p): Likewise.
3250 (nios2_r0rel_section_name_p): New.
3251 (nios2_symbol_ref_in_r0rel_data_p): New.
3252 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3253 (r0rel_constant_p): New.
3254 (nios2_print_operand_address): Handle r0rel_constant_p.
3255 (nios2_cdx_narrow_form_p): Likewise.
3256 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3257 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3258 (Nios II Options): Document -mr0rel-sec.
3260 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3262 * config/nios2/nios2.c: Include xregex.h.
3263 (nios2_gprel_sec_regex): New.
3264 (nios2_option_overide): Initialize it. Don't allow GP-relative
3265 addressing with PIC.
3266 (nios2_small_section_name_p): Check for regex match.
3267 * config/nios2/nios2.opt (mgprel-sec=): New option.
3268 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3269 (Nios II Options): Document -mgprel-sec.
3271 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3273 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3275 2017-10-26 Tom de Vries <tom@codesourcery.com>
3277 PR tree-optimization/82707
3278 * gimple.c (gimple_copy): Fix unsharing of
3279 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3281 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3283 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3284 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3285 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3286 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3287 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3288 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3289 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3290 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3291 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3292 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3293 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3294 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3295 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3296 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3297 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3298 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3299 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3301 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3303 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3305 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3306 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3307 warning to rs6000.c. Remove the Undocumented flag, since it has
3309 (-mabi=ibmlongdouble): Likewise.
3310 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3311 already set, set the default format for long double.
3312 (rs6000_debug_reg_global): Print whether long double is IBM or
3314 (rs6000_option_override_internal): Rework setting long double
3315 format. Only warn if the user is changing the long double default
3316 and they did not use -Wno-psabi.
3317 * doc/invoke.texi (PowerPC options): Update the documentation for
3318 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3320 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3321 Alan Hayward <alan.hayward@arm.com>
3322 David Sherwood <david.sherwood@arm.com>
3324 * rtl.h (wider_subreg_mode): New function.
3325 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3326 rather than an unsigned int *.
3327 * ira-color.c (regno_max_ref_width): Replace with...
3328 (regno_max_ref_mode): ...this new variable.
3329 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3330 Use wider_subreg_mode.
3331 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3332 rather than an unsigned int *.
3333 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3334 (process_alt_operands): Likewise.
3335 (invariant_p): Likewise.
3336 * lra-spills.c (assign_mem_slot): Likewise.
3337 (add_pseudo_to_slot): Likewise.
3338 * lra.c (collect_non_operand_hard_regs): Likewise.
3339 (add_regs_to_insn_regno_info): Likewise.
3340 * reload1.c (regno_max_ref_width): Replace with...
3341 (regno_max_ref_mode): ...this new variable.
3342 (reload): Update accordingly. Update call to
3343 ira_sort_regnos_for_alter_reg.
3344 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3345 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3346 (scan_paradoxical_subregs): Likewise.
3348 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3350 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3351 (aarch64_frame): Add emit_frame_chain boolean.
3352 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3353 Move eh_return case to aarch64_layout_frame.
3354 (aarch64_layout_frame): Initialize emit_frame_chain.
3355 (aarch64_expand_prologue): Use emit_frame_chain.
3357 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3359 * config/aarch64/aarch64.c (aarch64_layout_frame):
3360 Ensure LR is always stored at the bottom of the callee-saves.
3361 Remove rarely used frame layout which saves callee-saves at top of
3362 frame, so the store of LR can be used as a valid probe in all cases.
3364 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3366 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3367 Improve unaligned TImode/TFmode base/offset split.
3369 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3370 Alan Hayward <alan.hayward@arm.com>
3371 David Sherwood <david.sherwood@arm.com>
3373 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3374 * combine.c (find_single_use_1): Likewise.
3375 (expand_field_assignment): Likewise.
3376 (move_deaths): Likewise.
3377 * lra-constraints.c (simplify_operand_subreg): Likewise.
3378 (curr_insn_transform): Likewise.
3379 * lra.c (collect_non_operand_hard_regs): Likewise.
3380 (add_regs_to_insn_regno_info): Likewise.
3381 * rtlanal.c (reg_referenced_p): Likewise.
3382 (covers_regno_no_parallel_p): Likewise.
3384 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3386 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3387 Don't print any bits outside the precision of the value.
3388 * wide-int.cc (test_printing): Add some new tests.
3390 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3392 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3393 supports -xbrace_comment option.
3394 * configure: Regenerate.
3395 * config.in: Regenerate.
3396 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3397 (ASM_CPU_SPEC): Use it.
3399 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3401 * target.def (static_rtx_alignment): New hook.
3402 * targhooks.h (default_static_rtx_alignment): Declare.
3403 * targhooks.c (default_static_rtx_alignment): New function.
3404 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3405 * doc/tm.texi: Regenerate.
3406 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3407 instead of targetm.constant_alignment. Remove call to
3409 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3410 (cris_preferred_mininum_alignment): New function, split out from...
3411 (cris_constant_alignment): ...here.
3412 (cris_static_rtx_alignment): New function.
3413 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3415 (ix86_constant_alignment): ...here.
3416 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3417 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3418 (mmix_static_rtx_alignment): New function.
3419 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3420 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3422 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3425 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3426 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3428 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3430 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3431 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3432 mark as a sign-extending load.
3433 (local_pic_load_u): Define.
3435 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3438 * fold-const.c (operand_equal_for_comparison_p): Also return true
3439 if ARG0 is a simple variant of ARG1 with narrower precision.
3440 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3442 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3444 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3446 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3447 scatter_static, scatter_per_elt.
3448 * x86-tune-costs.h: Add new cost entries.
3450 2017-10-25 Richard Biener <rguenther@suse.de>
3452 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3453 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3454 class pass_fre): Move to ...
3455 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3456 class pass_fre): ... here and adjust for statistics.
3458 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3461 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3462 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3463 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3464 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3467 2017-10-25 Richard Biener <rguenther@suse.de>
3469 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3470 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3471 eliminate_push_avail, eliminate_insert): Move inside...
3472 (class eliminate_dom_walker): ... this class in preparation
3474 (fini_eliminate): Remove by merging with ...
3475 (eliminate): ... this function. Adjust for class changes.
3476 (pass_pre::execute): Remove fini_eliminate call.
3477 (pass_fre::execute): Likewise.
3479 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3482 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3483 (VPERMI2, VPERMI2I): New mode iterators.
3484 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3485 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3487 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3488 mode iterator. Remove 3 old define_insn patterns.
3489 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3490 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3491 VPERMI2 mode iterator, remove the other two expanders.
3492 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3493 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3494 instructions, remove the other two patterns.
3495 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3496 mode iterator, remove the other two patterns.
3497 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3498 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3499 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3500 and adjust argument order accordingly.
3501 (ix86_expand_vec_perm): Adjust caller.
3502 (expand_vec_perm_1): Likewise.
3503 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3504 (expand_vec_perm_vpermt2_vpshub2): ... this.
3505 (ix86_expand_vec_perm_const_1): Adjust caller.
3506 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3509 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3510 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3511 (vec_shl_<mode>): Remove unused expander.
3512 (avx512bw_<shift_insn><mode>3): New define_insn.
3513 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3514 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3516 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3519 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3522 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3524 PR rtl-optimization/82396
3525 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3526 (autopref_multipass_init): Simplify initialization.
3527 (autopref_rank_data): Simplify sort order.
3528 * gcc/sched-int.h (autopref_multipass_data_): Remove
3529 multi_mem_insn_p, min_offset and max_offset.
3531 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3534 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3535 Check special value of flag_omit_frame_pointer.
3536 (aarch64_can_eliminate): Likewise.
3537 (aarch64_override_options_after_change_1): Simplify handling of
3538 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3540 2017-10-24 Richard Biener <rguenther@suse.de>
3542 PR tree-optimization/82697
3543 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3544 zero for conditional load and unconditional store.
3546 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3548 * doc/install.texi: Document bootstrap-cet.
3550 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3553 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3554 ENDBR instruction at function entrance if function is only
3557 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3560 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3561 patterns to better describe from which operation the CF is computed.
3562 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3563 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3564 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3565 is 0, use _0 suffixed expanders instead of emitting a comparison
3568 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3570 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3571 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3573 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3576 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3577 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3578 * loop-iv.c (iv_get_reaching_def): Likewise.
3579 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3580 variable is promoted and the partition contains undefined values.
3582 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3584 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3586 (nios2_address_cost): Define.
3587 (nios2_legitimize_address): Recognize (exp + constant) directly.
3588 (TARGET_ADDRESS_COST): Define.
3590 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3592 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3593 (nios2_symbolic_memory_operand_p): Declare.
3594 (nios2_split_large_constant): Declare.
3595 (nios2_split_symbolic_memory_operand): Declare.
3596 * config/nios2/nios2.c: Adjust includes.
3597 (nios2_symbolic_constant_allowed): New.
3598 (nios2_symbolic_constant_p): New.
3599 (nios2_plus_symbolic_constant_p): New.
3600 (nios2_valid_addr_expr_p): Recognize addresses involving
3602 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3603 (nios2_symbolic_memory_operand_p): New.
3604 (nios2_large_constant_p): New.
3605 (nios2_split_large_constant): New.
3606 (nios2_split_plus_large_constant): New.
3607 (nios2_split_symbolic_memory_operand): New.
3608 (nios2_legitimize_address): Code refactoring. Handle addresses
3609 involving symbolic constants.
3610 (nios2_emit_move_sequence): Likewise.
3611 (nios2_print_operand): Improve error output.
3612 (nios2_print_operand_address): Handle LO_SUM.
3613 (nios2_cdx_narrow_form_p): Likewise.
3614 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3615 operands involving symbolic constants.
3616 (movhi_internal, movsi_internal): Likewise.
3617 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3618 (extendhisi2, extendqi<mode>2): Likewise.
3620 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3622 * tree-pass.h (PROP_rtl_split_insns): Define.
3623 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3625 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3627 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3629 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3632 * target.def (const_not_ok_for_debug_p): Default to
3633 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3634 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3635 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3636 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3637 which targetm.const_not_ok_for_debug_p returned true.
3638 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3640 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3642 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3643 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3644 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3646 (ix86_const_not_ok_for_debug_p): New function.
3647 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3648 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3650 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3653 * system.h: Conditionally include "unique-ptr.h" if
3654 INCLUDE_UNIQUE_PTR is defined.
3655 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3656 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3658 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3660 * config/rl78/rl78.md: New define_expand "subdi3".
3662 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3665 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3666 DF_REF_INSN if DF_REF_INSN_INFO is false.
3668 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3670 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3671 xmm_move instead of sse_move.
3672 (sse_store_index): New function.
3673 (ix86_register_move_cost): Be more sensible about mismatch stall;
3674 model AVX moves correctly; make difference between sse->integer and
3676 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3677 moves; make difference between SSE and AVX.
3678 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3679 and zmm_move. Increase size of sse load and store tables;
3680 add unaligned load and store tables; add ssemmx_to_integer.
3681 * x86-tune-costs.h: Update all entries according to real
3682 move latencies from Agner Fog's manual and chip documentation.
3684 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3687 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3688 * config/i386/constraints.md (Wf): New constraint.
3689 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3690 (cmp<dwi>_doubleword): Removed.
3691 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3692 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3693 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3694 expand with cmp<dwi>_doubleword. For LTU and GEU use
3695 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3697 * common.opt (gcolumn-info): Enable by default.
3698 * doc/invoke.texi (gcolumn-info): Document new default.
3700 2017-10-23 Richard Biener <rguenther@suse.de>
3702 PR tree-optimization/82672
3703 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3704 Fold the stmt if we propagated into it.
3706 2017-10-23 Richard Biener <rguenther@suse.de>
3708 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3709 (bitmap_remove_expr_from_set): ... this. All callers call this
3710 for non-constant values.
3711 (bitmap_set_subtract): Rename to...
3712 (bitmap_set_subtract_expressions): ... this. Adjust and
3714 (bitmap_set_contains_value): Remove superfluous check.
3715 (bitmap_set_replace_value): Inline into single caller ...
3716 (bitmap_value_replace_in_set): ... here and simplify.
3717 (dependent_clean): Merge into ...
3718 (clean): ... this using an overload. Adjust.
3719 (prune_clobbered_mems): Adjust.
3720 (compute_antic_aux): Likewise.
3721 (compute_partial_antic_aux): Likewise.
3723 2017-10-23 Richard Biener <rguenther@suse.de>
3725 PR tree-optimization/82129
3727 2017-08-01 Richard Biener <rguenther@suse.de>
3729 PR tree-optimization/81181
3730 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3731 (compute_antic): ... end of iteration here.
3733 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3735 * target.def (starting_frame_offset): New hook.
3736 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3737 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3738 * doc/tm.texi.in: Regenerate.
3739 * hooks.h (hook_hwi_void_0): Declare.
3740 * hooks.c (hook_hwi_void_0): New function.
3741 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3742 STARTING_FRAME_OFFSET.
3743 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3744 * reload1.c (reload): Likewise.
3745 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3746 instead of STARTING_FRAME_OFFSET.
3747 * function.c (try_fit_stack_local): Likewise.
3748 (assign_stack_local_1): Likewise
3749 (instantiate_virtual_regs): Likewise.
3750 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3751 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3752 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3753 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3754 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3755 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3756 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3757 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3758 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3759 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3760 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3761 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3762 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3763 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3764 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3765 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3766 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3767 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3768 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3769 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3770 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3771 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3772 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3773 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3774 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3775 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3776 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3777 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3778 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3779 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3780 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3781 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3782 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3783 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3784 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3785 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3786 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3787 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3788 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3789 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3790 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3791 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3792 return a HOST_WIDE_INT.
3793 (avr_builtin_setjmp_frame_value): Use it instead of
3794 STARTING_FRAME_OFFSET.
3795 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3796 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3797 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3799 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3800 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3801 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3802 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3803 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3804 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3805 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3806 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3807 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3808 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3809 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3810 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3812 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3813 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3814 * config/mips/mips.c (mips_compute_frame_info): Refer to
3815 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3816 (mips_starting_frame_offset): New function.
3817 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3818 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3819 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3820 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3821 and return a HOST_WIDE_INT.
3822 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3823 (mmix_initial_elimination_offset): Refer to
3824 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3825 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3826 * config/pa/pa.c (pa_starting_frame_offset): New function.
3827 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3828 (pa_expand_prologue): Likewise.
3829 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3830 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3831 !FRAME_GROWS_DOWNWARD handling to...
3832 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3833 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3834 !FRAME_GROWS_DOWNWARD handling to...
3835 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3836 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3837 !FRAME_GROWS_DOWNWARD handling to...
3838 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3839 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3841 (rs6000_starting_frame_offset): New function.
3842 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3843 !FRAME_GROWS_DOWNWARD handling to...
3844 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3845 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3846 !FRAME_GROWS_DOWNWARD handling to...
3847 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3848 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3849 !FRAME_GROWS_DOWNWARD handling to...
3850 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3851 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3852 (rs6000_starting_frame_offset): New function.
3853 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3854 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3855 * config/vax/vax.c (vax_starting_frame_offset): New function.
3856 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3857 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3858 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3859 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3860 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3861 * system.h (STARTING_FRAME_OFFSET): Poison.
3863 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3865 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3866 SCALAR_TYPE_MODE instead of TYPE_MODE.
3868 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3869 Alan Hayward <alan.hayward@arm.com>
3870 David Sherwood <david.sherwood@arm.com>
3872 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3874 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3875 Alan Hayward <alan.hayward@arm.com>
3876 David Sherwood <david.sherwood@arm.com>
3878 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3880 2017-10-23 Richard Biener <rguenther@suse.de>
3882 PR tree-optimization/82129
3883 * tree-ssa-pre.c (bitmap_set_and): Remove.
3884 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3885 canonicalizing expressions in the set to those with lowest
3886 ID rather than taking that from the first edge.
3888 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3890 * combine.c (rtx_equal_for_field_assignment_p): Use
3891 byte_lowpart_offset.
3893 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3894 Alan Hayward <alan.hayward@arm.com>
3895 David Sherwood <david.sherwood@arm.com>
3897 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3898 to a SUBREG_PROMOTED_VAR.
3900 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3901 Alan Hayward <alan.hayward@arm.com>
3902 David Sherwood <david.sherwood@arm.com>
3904 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3905 (expand_debug_source_expr): Likewise.
3906 * combine.c (combine_simplify_rtx): Likewise.
3907 * cse.c (fold_rtx): Likewise.
3908 * optabs.c (expand_float): Likewise.
3909 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3910 (simplify_binary_operation_1): Likewise.
3912 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3913 Alan Hayward <alan.hayward@arm.com>
3914 David Sherwood <david.sherwood@arm.com>
3916 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3917 (record_promoted_value): Likewise.
3918 * expr.c (expand_expr_real_2): Likewise.
3919 * ree.c (update_reg_equal_equiv_notes): Likewise.
3920 (combine_set_extension): Likewise.
3921 * rtlanal.c (low_bitmask_len): Likewise.
3922 * simplify-rtx.c (neg_const_int): Likewise.
3923 (simplify_binary_operation_1): Likewise.
3925 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3926 Alan Hayward <alan.hayward@arm.com>
3927 David Sherwood <david.sherwood@arm.com>
3929 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3930 * regcprop.c (maybe_mode_change): Likewise.
3931 * reload1.c (alter_reg): Likewise.
3933 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3935 * inchash.h (inchash::hash::add_wide_int): New function.
3936 * lto-streamer-out.c (hash_tree): Use it.
3938 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3940 * inchash.h (inchash::hash::add_wide_int): Rename to...
3941 (inchash::hash::add_hwi): ...this.
3942 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3943 (polymorphic_call_target_hasher::hash): Likewise.
3944 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3945 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3946 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3947 * lto-streamer-out.c (hash_tree): Likewise.
3948 * optc-save-gen.awk: Likewise.
3949 * tree.c (add_expr): Likewise.
3951 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3954 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3955 for ordered inequality comparisons even with TARGET_IEEE_FP.
3957 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3960 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3961 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3962 Expand with cmp<dwi>_doubleword.
3964 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3966 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3967 List CET intrinsics.
3968 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3969 specific to -fcf-protection option.
3971 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3973 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3974 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3975 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3976 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3977 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3978 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3979 (extra_objs): Add cet.o for Linux/x86 targets.
3980 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3981 * config/i386/cet.c: New file.
3982 * config/i386/cetintrin.h: Likewise.
3983 * config/i386/t-cet: Likewise.
3984 * config/i386/cpuid.h (bit_SHSTK): New.
3985 (bit_IBT): Likewise.
3986 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3987 pass IBT and SHSTK bits.
3988 * config/i386/i386-builtin-types.def
3989 (VOID_FTYPE_UNSIGNED_PVOID): New.
3990 (VOID_FTYPE_UINT64_PVOID): Likewise.
3991 * config/i386/i386-builtin.def: Add CET intrinsics.
3992 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3993 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3994 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3995 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3997 * config/i386/i386.c (rest_of_insert_endbranch): New.
3998 (pass_data_insert_endbranch): Likewise.
3999 (pass_insert_endbranch): Likewise.
4000 (make_pass_insert_endbranch): Likewise.
4001 (ix86_notrack_prefixed_insn_p): Likewise.
4002 (ix86_target_string): Add -mibt, -mshstk flags.
4003 (ix86_option_override_internal): Add flag_cf_protection
4005 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4006 (ix86_print_operand): Add 'notrack' prefix output.
4007 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4008 (ix86_expand_builtin): Expand CET intrinsics.
4009 (x86_output_mi_thunk): Add 'endbranch' instruction.
4010 * config/i386/i386.h (TARGET_IBT): New.
4011 (TARGET_IBT_P): Likewise.
4012 (TARGET_SHSTK): Likewise.
4013 (TARGET_SHSTK_P): Likewise.
4014 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4015 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4016 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4017 (builtin_setjmp_setup): New pattern.
4018 (builtin_longjmp): Likewise.
4019 (rdssp<mode>): Likewise.
4020 (incssp<mode>): Likewise.
4021 (saveprevssp): Likewise.
4022 (rstorssp): Likewise.
4023 (wrss<mode>): Likewise.
4024 (wruss<mode>): Likewise.
4025 (setssbsy): Likewise.
4026 (clrssbsy): Likewise.
4027 (nop_endbr): Likewise.
4028 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4030 * config/i386/immintrin.h: Include <cetintrin.h>.
4031 * config/i386/linux-common.h
4032 (file_end_indicate_exec_stack_and_cet): New prototype.
4033 (TARGET_ASM_FILE_END): New.
4035 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4037 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4038 latencies instead of having separate table; make difference between
4039 integer and float costs.
4040 * i386.h (processor_costs): Remove scalar_stmt_cost,
4041 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4042 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4044 * x86-tune-costs.h: Remove entries which has been removed in
4045 procesor_costs from all tables; make cond_taken_branch_cost
4046 and cond_not_taken_branch_cost COST_N_INSNS based.
4048 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4050 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4052 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4054 * config/i386/i386.md (isa): Remove fma_avx512f.
4055 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4056 <avx512>_fmadd_<mode>_mask3<round_name>,
4057 <avx512>_fmsub_<mode>_mask<round_name>,
4058 <avx512>_fmsub_<mode>_mask3<round_name>,
4059 <avx512>_fnmadd_<mode>_mask<round_name>,
4060 <avx512>_fnmadd_<mode>_mask3<round_name>,
4061 <avx512>_fnmsub_<mode>_mask<round_name>,
4062 <avx512>_fnmsub_<mode>_mask3<round_name>,
4063 <avx512>_fmaddsub_<mode>_mask<round_name>,
4064 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4065 <avx512>_fmsubadd_<mode>_mask<round_name>,
4066 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4067 (*vec_widen_umult_even_v16si<mask_name>,
4068 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4069 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4071 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4073 * extend.texi: Add 'nocf_check' documentation.
4074 * gimple.texi: Add second parameter to
4075 gimple_build_call_from_tree.
4076 * invoke.texi: Add -fcf-protection documentation.
4077 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4079 2017-10-20 Richard Biener <rguenther@suse.de>
4081 PR tree-optimization/82473
4082 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4083 the largest input type.
4085 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4087 * c-attribs.c (handle_nocf_check_attribute): New function.
4088 (c_common_attribute_table): Add 'nocf_check' handling.
4089 * gimple-parser.c: Add second argument NULL to
4090 gimple_build_call_from_tree.
4091 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4092 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4094 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4095 * common.opt: Add fcf-protection flag.
4096 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4097 * flag-types.h: Add enum cf_protection_level.
4098 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4099 Add 'nocf_check' attribute propagation to gimple call.
4100 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4101 (gimple_build_call_from_tree): Update prototype.
4102 (gimple_call_nocf_check_p): New function.
4103 (gimple_call_set_nocf_check): Likewise.
4104 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4105 * ipa-icf.c: Add nocf_check attribute in statement hash.
4106 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4107 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4108 * toplev.c (process_options): Add flag_cf_protection handling.
4110 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4112 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4114 2017-10-20 Richard Biener <rguenther@suse.de>
4116 PR tree-optimization/82603
4117 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4118 remove false predicated stores.
4120 2017-10-20 Richard Biener <rguenther@suse.de>
4122 * graphite-isl-ast-to-gimple.c
4123 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4124 Remove return value and simplify, dump copied stmt after lhs
4126 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4127 Reduce dump verbosity.
4128 (gsi_insert_earliest): Likewise.
4129 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4130 * graphite.c (print_global_statistics): Adjust dumping.
4131 (print_graphite_scop_statistics): Likewise.
4132 (print_graphite_statistics): Do not dump loops here.
4133 (graphite_transform_loops): But here.
4135 2017-10-20 Nicolas Roche <roche@adacore.com>
4137 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4138 * configure: Regenerate.
4140 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4143 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4144 functions when optimizing replace GIMPLE_RETURN stmts with
4145 calls to __builtin_unreachable ().
4148 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4149 for -fsanitize=thread link of executables.
4150 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4151 link of executables.
4154 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4156 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4157 define_insns for logical vector shifts to use VI248_AVX512BW
4158 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4159 condition, useless isa and prefix attributes. Change the first
4160 2 of these define_insns to ...
4161 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4162 define_insn for avx512vl.
4163 (<shift_insn><mode>3): ... and this, new define_insn without
4164 masking for non-avx512vl.
4167 * config/i386/sse.md (*andnot<mode>3,
4168 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4169 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4170 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4171 not applied use empty suffix even for TARGET_AVX512VL.
4172 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4173 is applied, supply evex,evex or evex,evex,evex instead of just
4176 2017-10-20 Julia Koval <julia.koval@intel.com>
4178 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4179 (OPTION_MASK_ISA_GFNI_UNSET): New.
4180 (ix86_handle_option): Handle OPT_mgfni.
4181 * config/i386/cpuid.h (bit_GFNI): New.
4182 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4183 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4184 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4185 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4186 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4187 * config/i386/i386.opt: Add mgfni.
4189 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4191 * config/msp430/msp430.c (msp430_option_override): Disable
4192 -fdelete-null-pointer-checks.
4193 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4195 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4197 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4198 of x87 and SSE instructions.
4200 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4202 * asan.c (create_cond_insert_point): Do not update edge count.
4203 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4204 (afdo_propagate_circuit): Likewise.
4205 (afdo_calculate_branch_prob): Likewise.
4206 (afdo_annotate_cfg): Likewise.
4207 * basic-block.h (struct edge_def): Remove count.
4208 (edge_def::count): New accessor.
4209 * bb-reorder.c (rotate_loop): Update.
4210 (find_traces_1_round): Update.
4211 (connect_traces): Update.
4212 (sanitize_hot_paths): Update.
4213 * cfg.c (unchecked_make_edge): Update.
4214 (make_single_succ_edge): Update.
4215 (check_bb_profile): Update.
4216 (dump_edge_info): Update.
4217 (update_bb_profile_for_threading): Update.
4218 (scale_bbs_frequencies_int): Update.
4219 (scale_bbs_frequencies_gcov_type): Update.
4220 (scale_bbs_frequencies_profile_count): Update.
4221 (scale_bbs_frequencies): Update.
4222 * cfganal.c (connect_infinite_loops_to_exit): Update.
4223 * cfgbuild.c (compute_outgoing_frequencies): Update.
4224 (find_many_sub_basic_blocks): Update.
4225 * cfgcleanup.c (try_forward_edges): Update.
4226 (try_crossjump_to_edge): Update
4227 * cfgexpand.c (expand_gimple_cond): Update
4228 (expand_gimple_tailcall): Update
4229 (construct_exit_block): Update
4230 * cfghooks.c (verify_flow_info): Update
4231 (redirect_edge_succ_nodup): Update
4232 (split_edge): Update
4233 (make_forwarder_block): Update
4234 (duplicate_block): Update
4235 (account_profile_record): Update
4236 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4237 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4238 * cfgloopmanip.c (scale_loop_profile): Update.
4240 (lv_adjust_loop_entry_edge): Update.
4241 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4242 (force_nonfallthru_and_redirect): Update.
4243 (purge_dead_edges): Update.
4244 (rtl_flow_call_edges_add): Update.
4245 * cgraphunit.c (init_lowered_empty_function): Update.
4246 (cgraph_node::expand_thunk): Update.
4247 * gimple-pretty-print.c (dump_probability): Update.
4248 (dump_edge_probability): Update.
4249 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4250 * haifa-sched.c (sched_create_recovery_edges): Update.
4251 * hsa-gen.c (convert_switch_statements): Update.
4252 * ifcvt.c (dead_or_predicable): Update.
4253 * ipa-inline-transform.c (inline_transform): Update.
4254 * ipa-split.c (split_function): Update.
4255 * ipa-utils.c (ipa_merge_profiles): Update.
4256 * loop-doloop.c (add_test): Update.
4257 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4258 * lto-streamer-in.c (input_cfg): Update.
4259 (input_function): Update.
4260 * lto-streamer-out.c (output_cfg): Update.
4261 * modulo-sched.c (sms_schedule): Update.
4262 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4263 * predict.c (maybe_hot_edge_p): Update.
4264 (unlikely_executed_edge_p): Update.
4265 (probably_never_executed_edge_p): Update.
4266 (dump_prediction): Update.
4267 (drop_profile): Update.
4268 (propagate_unlikely_bbs_forward): Update.
4269 (determine_unlikely_bbs): Update.
4270 (force_edge_cold): Update.
4271 * profile.c (compute_branch_probabilities): Update.
4272 * reg-stack.c (better_edge): Update.
4273 * shrink-wrap.c (handle_simple_exit): Update.
4274 * tracer.c (better_p): Update.
4275 * trans-mem.c (expand_transaction): Update.
4276 (split_bb_make_tm_edge): Update.
4277 * tree-call-cdce.c: Update.
4278 * tree-cfg.c (gimple_find_sub_bbs): Update.
4279 (gimple_split_edge): Update.
4280 (gimple_duplicate_sese_region): Update.
4281 (gimple_duplicate_sese_tail): Update.
4282 (gimple_flow_call_edges_add): Update.
4283 (insert_cond_bb): Update.
4284 (execute_fixup_cfg): Update.
4285 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4286 * tree-complex.c (expand_complex_div_wide): Update.
4287 * tree-eh.c (lower_resx): Update.
4288 (unsplit_eh): Update.
4289 (cleanup_empty_eh_move_lp): Update.
4290 * tree-inline.c (copy_edges_for_bb): Update.
4291 (freqs_to_counts): Update.
4292 (copy_cfg_body): Update.
4293 * tree-ssa-dce.c (remove_dead_stmt): Update.
4294 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4295 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4296 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4297 (unloop_loops): Update.
4298 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4299 * tree-ssa-loop-split.c (connect_loops): Update.
4300 (split_loop): Update.
4301 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4302 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4303 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4304 * tree-ssa-reassoc.c (branch_fixup): Update.
4305 * tree-ssa-tail-merge.c (replace_block_by): Update.
4306 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4307 (compute_path_counts): Update.
4308 (update_profile): Update.
4309 (recompute_probabilities): Update.
4310 (update_joiner_offpath_counts): Update.
4311 (estimated_freqs_path): Update.
4312 (freqs_to_counts_path): Update.
4313 (clear_counts_path): Update.
4314 (ssa_fix_duplicate_block_edges): Update.
4315 (duplicate_thread_path): Update.
4316 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4317 (case_bit_test_cmp): Update.
4318 (collect_switch_conv_info): Update.
4319 (gen_inbound_check): Update.
4320 (do_jump_if_equal): Update.
4321 (emit_cmp_and_jump_insns): Update.
4322 * tree-tailcall.c (decrease_profile): Update.
4323 (eliminate_tail_call): Update.
4324 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4325 (vect_do_peeling): Update.
4326 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4327 * ubsan.c (ubsan_expand_null_ifn): Update.
4328 (ubsan_expand_ptr_ifn): Update.
4329 * value-prof.c (gimple_divmod_fixed_value): Update.
4330 (gimple_mod_pow2): Update.
4331 (gimple_mod_subtract): Update.
4332 (gimple_ic): Update.
4333 (gimple_stringop_fixed_value): Update.
4335 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4338 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4340 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4342 PR rtl-optimization/82395
4343 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4344 based on non_spilled_static_chain_regno_p.
4346 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4348 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4349 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4350 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4352 2017-10-19 Martin Sebor <msebor@redhat.com>
4354 PR tree-optimization/82596
4355 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4357 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4359 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4360 (get_mem_refs_of_builtin_call): Likewise.
4361 * builtins.c (expand_builtin_apply): Adjust call to
4362 allocate_dynamic_stack_space.
4363 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4364 the third argument to allocate_dynamic_stack_space, otherwise -1.
4365 (expand_builtin): Deal with all alloca variants.
4366 (is_inexpensive_builtin): Likewise.
4367 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4368 * calls.c (special_function_p): Deal with all alloca variants.
4369 (initialize_argument_information): Adjust call to
4370 allocate_dynamic_stack_space.
4371 (expand_call): Likewise.
4372 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4373 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4374 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4375 use it for the stack usage computation.
4376 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4377 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4378 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4379 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4380 (in_loop_p): Remove first argument and useless check.
4381 (pass_walloca::execute): Remove useless test and adjust call to above.
4382 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4383 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4384 (gimplify_call_expr): Deal with all alloca variants.
4385 * hsa-gen.c (gen_hsa_alloca): Likewise.
4386 (gen_hsa_insns_for_call): Likewise.
4387 * ipa-pure-const.c (special_builtin_state): Likewise.
4388 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4389 * tree-object-size.c (alloc_object_size): Likewise.
4390 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4391 (call_may_clobber_ref_p_1): Likewise.
4392 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4393 (ccp_fold_stmt): Likewise.
4394 (optimize_stack_restore): Likewise.
4395 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4396 (mark_all_reaching_defs_necessary_1): Likewise.
4397 (propagate_necessity): Likewise.
4398 (eliminate_unnecessary_stmts): Likewise.
4399 * tree.c (build_common_builtin_nodes): Build
4400 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4401 (build_alloca_call_expr): New function.
4402 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4403 (CASE_BUILT_IN_ALLOCA): Likewise.
4404 (build_alloca_call_expr): Declare.
4405 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4407 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4410 * dwarf2out.c (new_die_raw): New static inline function.
4411 (new_die): Use it to create the DIE.
4412 (add_AT_external_die_ref): Likewise.
4413 (clone_die): Likewise.
4414 (clone_as_declaration): Likewise.
4415 (dwarf2out_vms_debug_main_pointer): Likewise.
4416 (base_type_die): Likewise. Remove early return for corner cases.
4417 Do not call add_pubtype on the DIE here.
4418 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4419 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4420 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4421 native order exists for base types, attach the DIE manually and call
4422 add_pubtype on it. Do not equate a reverse order DIE to the type.
4424 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4426 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4427 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4429 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4431 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4433 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4434 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4435 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4436 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4437 * opts.c (sanitizer_opts): Add builtin.
4438 * ubsan.c (instrument_builtin): New function.
4439 (pass_ubsan::execute): Call it.
4440 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4441 * doc/invoke.texi: Document -fsanitize=builtin.
4443 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4444 builtins, store max (log2 (align), 0) into uchar field instead of
4445 align into uptr field.
4446 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4447 store uchar 0 field instead of uptr 0 field.
4448 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4449 instead of passing one address of struct with 2 locations pass
4450 two addresses of structs with 1 location each.
4451 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4452 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4453 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4454 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4455 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4456 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4457 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4458 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4460 2017-10-19 Martin Liska <mliska@suse.cz>
4463 * file-find.c (remove_prefix): Remove.
4464 * file-find.h (remove_prefix): Likewise.
4465 * gcc-ar.c: Remove smartness of lookup.
4467 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4469 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4470 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4471 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4473 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4476 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4477 (setcc + and to xor + setcc): New peephole2.
4479 2017-10-19 Tom de Vries <tom@codesourcery.com>
4481 * doc/sourcebuild.texi (Test Directives, Variants of
4482 dg-require-support): Add dg-require-stack-size.
4484 2017-10-19 Martin Liska <mliska@suse.cz>
4487 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4488 that have a large alignment.
4489 (gimplify_target_expr): Likewise.
4491 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4493 PR rtl-optimization/82602
4494 * ira.c (rtx_moveable_p): Return false for volatile asm.
4496 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4499 * config/i386/i386-modes.def (CCGZ): New CC mode.
4500 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4501 * config/i386/predicates.md (ix86_comparison_operator):
4503 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4504 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4505 with double-word subtraction.
4506 (put_condition_code): Handle CCGZmode.
4508 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4510 * wide-int.cc (debug (const wide_int &)): New.
4511 (debug (const wide_int *)): New.
4512 (debug (const widest_int &)): New.
4513 (debug (const widest_int *)): New.
4515 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4518 * lra-constraints.c (curr_insn_transform): Use non-input operand
4519 instead of output one for matched reload.
4521 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4523 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4524 (tree-ssa-loop-ivopts.h): New header file.
4525 (struct builtin_info): New fields.
4526 (classify_builtin_1): Compute and record base and offset parts for
4527 memset builtin partition by calling strip_offset.
4528 (offset_cmp, fuse_memset_builtins): New functions.
4529 (finalize_partitions): Fuse adjacent memset partitions by calling
4531 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4532 Expose the interface.
4533 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4535 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4537 PR tree-optimization/82574
4538 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4539 that data reference must be executed exactly once per iteration
4540 against the outermost loop in nest.
4541 (classify_partition): Update call to above function.
4543 2017-10-18 Richard Biener <rguenther@suse.de>
4545 PR tree-optimization/82591
4546 * graphite.c (graphite_transform_loops): Move code gen message
4548 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4549 Here. Handle scop_to_isl_ast failing.
4550 (scop_to_isl_ast): Limit the number of ISL operations.
4552 2017-10-18 Richard Biener <rguenther@suse.de>
4554 * graphite-isl-ast-to-gimple.c
4555 (translate_isl_ast_to_gimple::set_rename): Simplify.
4556 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4557 (graphite_copy_stmts_from_block): ... here.
4558 (copy_bb_and_scalar_dependences): Simplify.
4559 (add_parameters_to_ivs_params): Canonicalize.
4560 (generate_entry_out_of_ssa_copies): Simplify.
4561 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4562 by passing in ISL dimension.
4563 (parameter_index_in_region_1): Rename to ...
4564 (parameter_index_in_region): ... this.
4565 (extract_affine): Adjust assert, pass down parameter index.
4566 (add_param_constraints): Use range-info when available.
4567 (build_scop_context): Adjust.
4568 * sese.c (new_sese_info): Adjust.
4569 (free_sese_info): Likewise.
4570 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4571 Remove unused typedefs.
4572 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4574 2017-10-18 Martin Liska <mliska@suse.cz>
4576 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4578 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4580 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4581 (s390_sched_init): Do not reset s390_sched_state if we entered the
4582 current basic block via a fallthru edge and all others are unlikely.
4584 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4586 * config/s390/s390.c (NUM_SIDES): New variable.
4587 (LONGRUNNING_THRESHOLD): New variable.
4588 (LATENCY_FACTOR): New variable.
4589 (s390_sched_score): Decrease score for long-running instructions on
4591 (s390_sched_variable_issue): Perform bookkeeping for long-running
4594 2017-10-18 Richard Biener <rguenther@suse.de>
4596 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4597 Simplify with removal of the parameter rename map.
4598 (set_rename): Likewise.
4599 (should_copy_to_new_region): Likewise.
4600 (graphite_copy_stmts_from_block): Likewise.
4601 (copy_bb_and_scalar_dependences): Remove initialization of
4602 unused copied_bb_map.
4604 (copy_internal_parameters): Likewise.
4605 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4606 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4607 Use INTEGRAL_TYPE_P.
4608 (parameter_index_in_region_1): Rename to ...
4609 (assign_parameter_index_in_region): ... this. Assert we have
4610 a parameter we handle.
4611 (scan_tree_for_params): Adjust.
4612 * sese.h (parameter_rename_map_t): Remove.
4613 (struct sese_info_t): Remove unused parameter_rename_map and
4614 copied_bb_map members.
4615 * sese.c (new_sese_info): Adjust.
4616 (free_sese_info): Likewise.
4618 2017-10-18 Martin Liska <mliska@suse.cz>
4621 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4622 on an abnormal edge.
4624 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4626 * doc/invoke.texi (ffunction-sections and fdata-sections):
4629 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4631 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4632 the use statement can throw internally.
4634 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4636 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4637 any RTX present on the RHS of a SET.
4638 * compare-elim.c (try_eliminate_compare): Restore comment.
4640 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4642 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4643 may be also called on tcc_type nodes.
4644 * langhooks.c (lhd_tree_size): Likewise.
4646 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4648 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4649 format_warning_at_substring.
4650 (maybe_warn): Convert source_range * param to a location_t. Pass
4651 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4652 (format_directive): Remove code to extract source_ranges and
4653 source_range * in favor of just a location_t.
4654 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4656 * substring-locations.c (format_warning_va): Convert
4657 source_range * param to a location_t.
4658 (format_warning_at_substring): Likewise.
4659 * substring-locations.h (format_warning_va): Likewise.
4660 (format_warning_at_substring): Likewise.
4662 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4664 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4666 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4667 and scatter/gather ops.
4669 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4670 vec_gather_load and vec_scatter_store.
4671 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4672 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4673 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4674 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4675 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4676 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4678 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4680 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4681 Detect FTST instruction and handle its register pops. Only pop
4682 second operand if can_pop_second_op is true.
4683 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4684 set can_pop_second_op to false in the compare_for_stack_reg call.
4686 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4687 output_fp_compare for stack register operands.
4688 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4689 instructions here. Do not emit stack register pops here. Assert
4690 that FCOMPP pops next to top stack register. Rewrite function.
4692 2017-10-17 Nathan Sidwell <nathan@acm.org>
4695 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4696 use DECL_ASSEMBLER_NAME_RAW.
4699 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4702 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4703 Wilco Dijkstra <wilco.dijkstra@arm.com>
4705 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4706 converstion to Pmode from the buf_addr.
4708 2017-10-17 Richard Biener <rguenther@suse.de>
4710 * graphite-dependences.c (scop_get_reads_and_writes): Change
4711 output parameters to references.
4713 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4715 PR 71026/tree-optimization
4716 * fold-const.c (distribute_real_division): Removed.
4717 (fold_binary_loc): Remove calls to distribute_real_divison.
4719 2017-10-17 Richard Biener <rguenther@suse.de>
4721 * graphite-scop-detection.c
4722 (scop_detection::stmt_has_simple_data_refs_p): Always use
4723 the full nest as region.
4724 (try_generate_gimple_bb): Likewise.
4725 * sese.c (scalar_evolution_in_region): Simplify now that
4726 SCEV can handle instantiation in regions.
4727 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4728 in the non-loop part of a function if requested.
4730 2017-10-17 Richard Biener <rguenther@suse.de>
4732 PR tree-optimization/82563
4733 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4735 (graphite_regenerate_ast_isl): Call it.
4736 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4738 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4740 PR tree-optimization/82549
4741 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4742 Formatting fixes. Instead of calling make_bit_field_ref with negative
4745 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4747 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4748 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4749 _mm_maskz_reduce_ss): New.
4750 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4751 __builtin_ia32_reducess_mask): Ditto..
4752 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4753 * config/i386/sse.md (reduces<mode>): Renamed to ...
4754 (reduces<mode><mask_scalar_name>): ... this.
4755 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4757 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4758 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4760 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4762 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4763 * selftest-run-tests.c (selftest::run_tests): Call
4764 selftest::unique_ptr_tests_cc_tests.
4765 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4766 * unique-ptr-tests.cc: New file.
4768 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4771 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4773 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4774 (make_hard_regno_born, make_hard_regno_dead): Update
4775 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4777 2017-10-16 Jeff Law <law@redhat.com>
4779 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4781 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4783 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4785 2017-10-16 Olivier Hainque <hainque@adacore.com>
4787 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4788 with_cpu if we were configured for an e500v2 target cpu name.
4790 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4792 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4793 * doc/invoke.texi: Document +nodsp as a valid extension for
4796 2017-10-16 Martin Liska <mliska@suse.cz>
4798 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4799 (test_set_range): Likewise.
4800 (test_range_functions): Rename to ...
4801 (test_bit_in_range): ... this.
4802 (sbitmap_c_tests): Add new test.
4804 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4806 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4808 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4809 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4811 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4813 * config/aarch64/aarch64-builtins.c
4814 (aarch64_types_quadopu_lane_qualifiers): New.
4815 (TYPES_QUADOPU_LANE): New.
4816 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4817 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4818 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4819 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4820 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4821 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4822 (Vdottype, DOTPROD): New.
4823 (sur): Add SDOT and UDOT.
4825 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4827 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4828 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4829 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4831 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4832 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4833 Enable TARGET_DOTPROD.
4834 (cortex-a75.cortex-a55): Likewise.
4835 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4837 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4839 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4840 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4842 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4844 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4845 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4846 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4847 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4848 * config/arm/types.md (neon_dot, neon_dot_q): New.
4849 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4851 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4853 * config/arm/arm.h (TARGET_DOTPROD): New.
4854 * config/arm/arm.c (arm_arch_dotprod): New.
4855 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4856 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4857 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4858 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4859 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4860 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4861 * doc/invoke.texi (armv8.2-a): Document dotprod
4863 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4865 * i386.c (ix86_vec_cost): New function.
4866 (ix86_rtx_costs): Handle vector operations better.
4867 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4868 * x86-tune-costs.h: Add new costs to all tables.
4870 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4872 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4874 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4875 divsd, sqrtss and sqrtsd
4876 * x86-tune-costs.h: Add new entries to all costs.
4877 (znver1_cost): Fix to match real instruction latencies.
4879 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4880 Michael Collison <michael.collison@arm.com>
4882 * compare-elim.c: Include emit-rtl.h.
4883 (can_merge_compare_into_arith): New function.
4884 (try_validate_parallel): Likewise.
4885 (try_merge_compare): Likewise.
4886 (try_eliminate_compare): Call the above when no previous clobber
4888 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4891 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4895 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4896 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4900 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4901 to be any operand_equal_p operands. For & (B - 1) require
4902 B to be power of 2. Recognize
4903 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4905 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4908 * optabs.c (expand_memory_blockage): Fix call of
4909 targetm.have_memory_blockage.
4911 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4914 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4915 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4916 objects to extra_objs instead of overwriting it.
4918 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4920 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4921 Use any_fp_register_operand as operand[3] predicate. Simplify
4922 equality test for operands[2] and operands[4] memory location.
4923 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4924 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4925 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4926 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4927 any_fp_register_operand as operand[1] predicate. Simplify
4928 equality test for operands[0] and operands[3] memory location.
4929 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4930 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4931 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4933 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4935 * target-insns.def: Add memory_blockage.
4936 * optabs.c (expand_memory_blockage): New function.
4937 (expand_asm_memory_barrier): Rename ...
4938 (expand_asm_memory_blockage): ... to this.
4939 (expand_mem_thread_fence): Call expand_memory_blockage
4940 instead of expand_asm_memory_barrier.
4941 (expand_mem_singnal_fence): Ditto.
4942 (expand_atomic_load): Ditto.
4943 (expand_atomic_store): Ditto.
4944 * doc/md.texi (Standard Pattern Names For Generation):
4945 Document memory_blockage instruction pattern.
4947 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4949 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4950 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4951 * config/rl78/rl78.md: New define_expand "adddi3".
4953 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4955 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4958 2017-10-13 Jeff Law <law@redhat.com>
4960 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4962 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4965 * internal-fn.c (expand_mul_overflow): If both operands have
4966 the same highpart of -1 or 0 and the topmost bit of lowpart
4967 is different, overflow is if res <= 0 rather than res < 0.
4969 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4971 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4972 TARGET_P9_VECTOR code for unaligned_load case.
4974 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4976 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4978 2017-10-13 Nathan Sidwell <nathan@acm.org>
4980 * tree-core.h (tree_contains_struct): Make bool.
4981 * tree.c (tree_contains_struct): Likewise.
4982 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4983 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4984 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4985 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4986 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4988 2017-10-13 Richard Biener <rguenther@suse.de>
4990 * graphite-isl-ast-to-gimple.c
4991 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4992 parameters and dominance check.
4993 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4994 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4995 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4996 Do not update SSA form here or do intermediate IL verification.
4997 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4998 (graphite_initialize): Remove check on the number of loops in
4999 the function and inline into graphite_transform_loops.
5000 (graphite_finalize): Inline into graphite_transform_loops.
5001 (graphite_transform_loops): Perform SSA update and IL verification
5003 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5005 2017-10-13 Richard Biener <rguenther@suse.de>
5007 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5008 graphite_expression_type_precision): Avoid global constructor
5010 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5011 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5012 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5013 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5014 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5015 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5017 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5020 * config/i386/i386.h (ix86_red_zone_size): New.
5021 * config/i386/i386.md (push peephole2s): Replace
5022 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5024 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5025 Alan Hayward <alan.hayward@arm.com>
5026 David Sherwood <david.sherwood@arm.com>
5028 * combine.c (can_change_dest_mode): Reject changes in
5029 REGMODE_NATURAL_SIZE.
5031 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5032 Alan Hayward <alan.hayward@arm.com>
5033 David Sherwood <david.sherwood@arm.com>
5035 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5036 (expand_debug_source_expr): Likewise.
5037 * combine.c (combine_simplify_rtx): Likewise.
5038 * cse.c (fold_rtx): Likewise.
5039 * fwprop.c (canonicalize_address): Likewise.
5040 * targhooks.c (default_shift_truncation_mask): Likewise.
5042 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5043 Alan Hayward <alan.hayward@arm.com>
5044 David Sherwood <david.sherwood@arm.com>
5046 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5047 (widened_mode): Likewise.
5048 (expand_unop): Likewise.
5049 * ree.c (transform_ifelse): Likewise.
5050 (merge_def_and_ext): Likewise.
5051 (combine_reaching_defs): Likewise.
5052 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5054 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5055 Alan Hayward <alan.hayward@arm.com>
5056 David Sherwood <david.sherwood@arm.com>
5058 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5059 * combine.c (gen_lowpart_for_combine): Likewise.
5060 * dwarf2out.c (rtl_for_decl_location): Likewise.
5061 * final.c (alter_subreg): Likewise.
5062 * rtlhooks.c (gen_lowpart_general): Likewise.
5063 (gen_lowpart_if_possible): Likewise.
5065 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5066 Alan Hayward <alan.hayward@arm.com>
5067 David Sherwood <david.sherwood@arm.com>
5069 * calls.c (expand_call): Use subreg_lowpart_offset.
5070 * cse.c (cse_insn): Likewise.
5071 * regcprop.c (copy_value): Likewise.
5072 (copyprop_hardreg_forward_1): Likewise.
5074 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5077 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5078 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5079 =Q constraints to +Q and into insn condition add check
5080 that operands[0] and operands[1] are equal.
5081 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5082 =Q constraints to +Q and into insn condition add check
5083 that operands[0] is equal to either operands[1] or operands[2].
5086 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5087 instead of handling MINUS_EXPR twice (once for each argument),
5088 canonicalize operand order and handle just once, use rtype where
5089 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5092 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5093 any values of __C while still being pattern recognizable as a simple
5096 2017-10-13 Richard Biener <rguenther@suse.de>
5098 PR tree-optimization/82451
5100 2017-10-02 Richard Biener <rguenther@suse.de>
5102 PR tree-optimization/82355
5103 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5104 a mapping for the enclosing loop but avoid generating one for
5106 (copy_bb_and_scalar_dependences): Remove premature codegen
5107 error on PHIs in blocks duplicated into multiple places.
5108 * graphite-scop-detection.c
5109 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5110 in the region use it as loop and nest to analyze the DR in.
5111 (try_generate_gimple_bb): Likewise.
5112 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5113 (add_loop_constraints): For blocks in a loop not in the region
5114 create a dimension with a single iteration.
5115 * sese.h (gbb_loop_at_index): Remove assert.
5117 * cfgloop.c (loop_preheader_edge): For the loop tree root
5118 return the single successor of the entry block.
5119 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5120 Reset the SCEV hashtable and niters.
5121 * graphite-scop-detection.c
5122 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5123 assert that we only have POLYNOMIAL_CHREC that vary in loops
5124 contained in the region.
5125 (scop_detection::graphite_can_represent_expr): Adjust.
5126 (scop_detection::stmt_has_simple_data_refs_p): For loops
5127 not in the region set loop to NULL. The nest is now the
5128 entry edge to the region.
5129 (try_generate_gimple_bb): Likewise.
5130 * sese.c (scalar_evolution_in_region): Adjust for
5131 instantiate_scev change.
5132 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5133 Make nest parameter the edge into the region.
5134 (create_data_ref): Likewise.
5135 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5136 entry edge into a region and adjust instantiate_scev calls.
5137 (create_data_ref): Likewise.
5138 (graphite_find_data_references_in_stmt): Likewise.
5139 (find_data_references_in_stmt): Pass the loop preheader edge
5140 from the nest argument.
5141 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5142 parameter the edge into the region.
5143 (instantiate_parameters): Use the loop preheader edge as entry.
5144 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5146 (get_instantiated_value_entry): Make instantiate_below parameter
5147 the edge into the region.
5148 (instantiate_scev_name): Likewise. Adjust dominance checks,
5149 when we cannot use loop-based instantiation instantiate by
5150 walking use-def chains.
5151 (instantiate_scev_poly): Adjust.
5152 (instantiate_scev_binary): Likewise.
5153 (instantiate_scev_convert): Likewise.
5154 (instantiate_scev_not): Likewise.
5155 (instantiate_array_ref): Remove.
5156 (instantiate_scev_3): Likewise.
5157 (instantiate_scev_2): Likewise.
5158 (instantiate_scev_1): Likewise.
5159 (instantiate_scev_r): Do not blindly handle N-operand trees.
5160 Do not instantiate array-refs. Handle all constants and invariants.
5161 (instantiate_scev): Make instantiate_below parameter
5162 the edge into the region.
5163 (resolve_mixers): Use the loop preheader edge for the region
5164 parameter to instantiate_scev_r.
5165 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5167 2017-10-13 Richard Biener <rguenther@suse.de>
5169 PR tree-optimization/82525
5170 * graphite-isl-ast-to-gimple.c
5171 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5173 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5174 Fail code generation when we cannot represent the isl integer.
5175 (binary_op_to_tree): Elide modulo operations that are no-ops
5176 in the type we code generate. Remove now superfluous code
5179 2017-10-13 Richard Biener <rguenther@suse.de>
5181 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5182 (scop_detection::harmful_loop_in_region): Remove premature
5183 IV type restriction.
5184 (scop_detection::graphite_can_represent_scev): We can handle
5185 pointer IVs just fine.
5187 2017-10-13 Alan Modra <amodra@gmail.com>
5189 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5190 "Clobbers and Scratch Registers". Add paragraph on
5191 alternative to clobbers for scratch registers and OpenBLAS
5194 2017-10-13 Alan Modra <amodra@gmail.com>
5196 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5197 example of a memory input for a string of known length. Move
5198 commentary out of table. Add a number of new examples
5199 covering array memory inputs.
5201 2017-10-12 Martin Liska <mliska@suse.cz>
5203 PR tree-optimization/82493
5204 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5205 (test_range_functions): New function.
5206 (sbitmap_c_tests): Likewise.
5207 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5208 * selftest.h (sbitmap_c_tests): New function.
5210 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5212 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5214 * config/rs6000/amo.h: Fix spacing issue.
5216 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5219 * config/i386/i386.md (*ashl<mode>3_mask_1,
5220 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5221 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5224 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5226 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5227 (profile_probability): Set max_probability
5228 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5230 * profile-count.c (profile_probability::differs_from_p): Do not
5231 rely on max_probaiblity == 10000
5233 2017-10-12 Jeff Law <law@redhat.com>
5235 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5238 2017-10-12 Martin Sebor <msebor@redhat.com>
5242 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5243 (handle_alias_pairs): Call it.
5244 * common.opt (-Wattribute-alias): New option.
5245 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5246 * doc/invoke.texi (-Wattribute-alias): Document.
5248 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5251 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5253 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5255 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5256 (make_hard_regno_born, make_hard_regno_dead): Update
5257 bb_killed_pseudos and bb_gen_pseudos.
5259 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5261 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5263 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5265 * config/alpha/alpha.c (alpha_split_conditional_move):
5266 Use std::swap instead of manually swapping.
5267 (alpha_stdarg_optimize_hook): Ditto.
5268 (alpha_canonicalize_comparison): Ditto.
5270 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5272 * tree-loop-distribution.c (struct builtin_info): New struct.
5273 (struct partition): Refactor fields into struct builtin_info.
5274 (partition_free): Free struct builtin_info.
5275 (build_size_arg_loc, build_addr_arg_loc): Delete.
5276 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5277 information from struct builtin_info.
5278 (find_single_drs): New function refactored from classify_partition.
5279 Also moved builtin validity checks to this function.
5280 (compute_access_range, alloc_builtin): New functions.
5281 (classify_builtin_st, classify_builtin_ldst): New functions.
5282 (classify_partition): Refactor code into functions find_single_drs,
5283 classify_builtin_st and classify_builtin_ldst.
5284 (distribute_loop): Don't do runtime alias check when distributing
5286 (find_seed_stmts_for_distribution): New function.
5287 (pass_loop_distribution::execute): Refactor code finding seed
5288 stmts into above function. Support distribution for the innermost
5289 two-level loop nest. Adjust dump information.
5291 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5293 * tree-loop-distribution.c: Adjust the general comment.
5294 (NUM_PARTITION_THRESHOLD): New macro.
5295 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5296 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5297 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5298 in call to build_partition_graph.
5299 (finalize_partitions): New parameter. Make loop distribution more
5300 conservative by fusing more partitions.
5301 (distribute_loop): Don't do runtime alias check in case of loop nest
5303 (find_seed_stmts_for_distribution): New function.
5304 (prepare_perfect_loop_nest): New function.
5305 (pass_loop_distribution::execute): Refactor code finding seed stmts
5306 and loop nest into above functions. Support loop nest distribution.
5307 Adjust dump information accordingly.
5309 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5311 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5312 and set PTYPE_SEQUENTIAL for merged partition.
5314 2017-10-12 Richard Biener <rguenther@suse.de>
5316 PR tree-optimization/69728
5318 2017-09-19 Richard Biener <rguenther@suse.de>
5320 PR tree-optimization/69728
5321 * graphite-sese-to-poly.c (schedule_error): New global.
5322 (add_loop_schedule): Handle empty domain by failing the
5324 (build_original_schedule): Handle schedule_error.
5326 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5327 domain by returning an unchanged schedule.
5329 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5331 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5332 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5334 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5336 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5339 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5342 * expr.c (store_field): Don't optimize away bitsize == 0 store
5343 from CALL_EXPR with addressable return type.
5345 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5347 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5348 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5349 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5350 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5351 TARGET_ISEL instead of TARGET_ISEL<sel>.
5353 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5355 * config/rs6000/rs6000.c
5356 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5358 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5360 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5362 (reg_or_cint_operand): Fix comment.
5363 (reg_or_zero_operand): New predicate.
5364 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5365 * config/rs6000/rs6000.c (output_isel): Delete.
5366 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5367 instead of reg_or_cint_operand. Output instruction directly (not via
5369 (isel_unsigned_<mode>): Ditto.
5370 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5371 gpc_reg_operand. Add an instruction alternative for this. Output
5372 instruction directly.
5373 (*isel_reversed_unsigned_<mode>): Ditto.
5375 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5377 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5378 (TARGET_CANONICALIZE_COMPARISON): Define.
5380 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5383 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5384 Check whether the dest is REG before adding REG_EQUIV note.
5386 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5389 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5391 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5392 (make_hard_regno_born, make_hard_regno_dead): Update
5393 bb_killed_pseudos and bb_gen_pseudos.
5395 2017-10-11 Nathan Sidwell <nathan@acm.org>
5397 * incpath.h (enum incpath_kind): Name enum, prefix values.
5398 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5399 * incpath.c (heads, tails): Use INC_MAX.
5400 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5401 (merge_include_chains, split_quote_chain,
5402 register_include_chains): Update incpath_kind names.
5403 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5404 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5406 (add_framework_path, darwin_register_objc_includes): Likewise.
5407 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5409 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5411 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5412 Do not use float_operator operator predicate.
5413 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5414 * config/i386/predicates.md (float_operator): Remove predicate.
5416 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5418 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5419 (*jccxf_i387): Ditto.
5420 (*jcc<mode>_i387): Ditto.
5421 (*jccu<mode>_i387): Ditto.
5422 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5423 (*jcc_*_i387 splitters): Remove.
5424 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5425 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5426 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5429 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5431 * profile-count.h (slow_safe_scale_64bit): New function.
5432 (safe_scale_64bit): New inline.
5433 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5434 * profile-count.c: Include wide-int.h
5435 (slow_safe_scale_64bit): New.
5437 2017-10-11 Nathan Sidwell <nathan@acm.org>
5439 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5440 HAS_DECL_ASSEMBLER_NAME_P.
5441 * gimple-expr.c (gimple_decl_printable_name: Check
5442 HAS_DECL_ASSEMBLER_NAME_P too.
5443 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5444 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5445 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5446 * passes.c (rest_of_decl_compilation): Check
5447 HAS_DECL_ASSEMBLER_NAME_P too.
5448 * recog.c (verify_changes): Likewise.
5449 * tree-pretty-print.c (dump_decl_name): Likewise.
5450 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5452 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5453 (SET_DECL_ASSEMBLER_NAME): Use it.
5454 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5455 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5456 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5458 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5460 * config.gcc (i386, x86_64): Add extra objects.
5461 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5462 (ix86_min_insn_size): Declare.
5463 (ix86_issue_rate): Declare.
5464 (ix86_adjust_cost): Declare.
5465 (ia32_multipass_dfa_lookahead): Declare.
5466 (ix86_macro_fusion_p): Declare.
5467 (ix86_macro_fusion_pair_p): Declare.
5468 (ix86_bd_has_dispatch): Declare.
5469 (ix86_bd_do_dispatch): Declare.
5470 (ix86_core2i7_init_hooks): Declare.
5471 (ix86_atom_sched_reorder): Declare.
5472 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5473 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5474 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5475 (rip_relative_addr_p): Rename to ...
5476 (ix86_rip_relative_addr_p): ... this one; export.
5477 (memory_address_length): Update.
5478 (ix86_issue_rate): Move to x86-tune-sched.c.
5479 (ix86_flags_dependent): Move to x86-tune-sched.c.
5480 (ix86_agi_dependent): Move to x86-tune-sched.c.
5481 (exact_dependency_1): Move to x86-tune-sched.c.
5482 (exact_store_load_dependency): Move to x86-tune-sched.c.
5483 (ix86_adjust_cost): Move to x86-tune-sched.c.
5484 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5485 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5486 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5487 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5488 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5489 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5490 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5491 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5492 (min_insn_size): Rename to ...
5493 (ix86_min_insn_size): ... this one; export.
5494 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5495 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5496 (core2i7_first_cycle_multipass_backtrack): Move to
5497 x86-tune-sched-core.c.
5498 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5499 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5500 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5501 (ix86_avoid_jump_mispredicts): Update.
5502 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5503 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5504 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5505 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5506 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5507 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5508 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5509 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5510 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5511 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5512 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5513 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5514 (BIG): Move to ix86-tune-sched-bd.c.
5515 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5516 (enum insn_path): Move to ix86-tune-sched-bd.c.
5517 (get_mem_group): Move to ix86-tune-sched-bd.c.
5518 (is_cmp): Move to ix86-tune-sched-bd.c.
5519 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5520 (is_branch): Move to ix86-tune-sched-bd.c.
5521 (is_prefetch): Move to ix86-tune-sched-bd.c.
5522 (init_window): Move to ix86-tune-sched-bd.c.
5523 (allocate_window): Move to ix86-tune-sched-bd.c.
5524 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5525 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5526 (process_end_window): Move to ix86-tune-sched-bd.c.
5527 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5528 (find_constant): Move to ix86-tune-sched-bd.c.
5529 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5530 (has_immediate): Move to ix86-tune-sched-bd.c.
5531 (get_insn_path): Move to ix86-tune-sched-bd.c.
5532 (get_insn_group): Move to ix86-tune-sched-bd.c.
5533 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5534 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5535 (add_insn_window): Move to ix86-tune-sched-bd.c.
5536 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5537 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5538 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5539 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5540 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5541 (do_dispatch): Move to ix86-tune-sched-bd.c.
5542 (has_dispatch): Move to ix86-tune-sched-bd.c.
5543 * config/i386/t-i386: Add new object files.
5544 * config/i386/x86-tune-costs.h: New file.
5545 * config/i386/x86-tune-sched-atom.c: New file.
5546 * config/i386/x86-tune-sched-bd.c: New file.
5547 * config/i386/x86-tune-sched-core.c: New file.
5548 * config/i386/x86-tune-sched.c: New file.
5550 2017-10-11 Liu Hao <lh_mouse@126.com>
5552 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5553 the generic version below instead.
5554 (should_colorize): Recognize Windows consoles as terminals
5556 * pretty-print.c [__MINGW32__] (write_all): New function.
5557 [__MINGW32__] (find_esc_head): Likewise.
5558 [__MINGW32__] (find_esc_terminator): Likewise.
5559 [__MINGW32__] (eat_esc_sequence): Likewise.
5560 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5562 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5565 2017-10-11 Richard Biener <rguenther@suse.de>
5567 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5568 Properly call analyze_scalar_evolution with the loop of the stmt.
5570 2017-10-11 Richard Biener <rguenther@suse.de>
5572 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5573 * tree-core.h (tree_base): Add chrec_var union member.
5574 * tree.h (CHREC_VAR): Remove.
5575 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5576 * tree-chrec.h (build_polynomial_chrec): Adjust.
5577 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5578 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5580 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5582 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5583 * match.pd: ... here.
5584 ((T) X == (T) Y): Relax condition.
5586 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5588 PR tree-optimization/82472
5589 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5591 (break_alias_scc_partitions): Update postorder number.
5593 2017-10-11 Martin Liska <mliska@suse.cz>
5596 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5598 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5600 2017-10-11 Martin Liska <mliska@suse.cz>
5602 * print-rtl.c (print_insn): Move declaration of idbuf
5603 to same scope as name.
5605 2017-10-11 Martin Liska <mliska@suse.cz>
5610 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5613 2017-10-11 Martin Liska <mliska@suse.cz>
5616 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5619 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5621 * config.gcc (default_gnu_indirect_function): Default to yes for
5622 arm*-*-linux* with glibc.
5624 2017-10-11 Richard Biener <rguenther@suse.de>
5626 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5627 default-defs and types we do not want to analyze.
5628 (interpret_loop_phi): Replace unreachable code with an assert.
5629 (compute_scalar_evolution_in_loop): Remove and inline ...
5630 (analyze_scalar_evolution_1): ... here, replacing condition with
5631 what makes the intent clearer. Remove handling of cases
5632 get_scalar_evolution now handles.
5634 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5636 PR rtl-optimization/81434
5637 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5638 comment for main loop. In sched_group_found if, also add checks for
5639 pass and min_cost_group.
5641 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5643 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5644 (rs6000_insn_cost): New function.
5645 * config/rs6000/rs6000.md (cost): New attribute.
5647 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5648 H.J. Lu <hongjiu.lu@intel.com>
5652 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5653 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5654 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5655 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5656 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5657 OPTION_MASK_ISA_MMX and some other bit set require both
5658 mmx and the ISAs without the mmx bit.
5659 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5660 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5661 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5662 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5663 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5664 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5665 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5666 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5667 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5668 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5669 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5670 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5671 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5672 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5673 Add OPTION_MASK_ISA_MMX.
5675 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5677 * config.gcc (armv7*-*-freebsd*): New target.
5678 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5680 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5682 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5683 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5686 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5688 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5689 a template parameter.
5690 (WIDE_INT_REF_FOR): Update accordingly.
5691 * tree.h (wi::int_traits <const_tree>): Delete.
5692 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5693 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5694 (wi::tree_to_wide_ref): New typedef.
5695 (wi::to_wide): New function.
5696 * calls.c (get_size_range): Use wi::to_wide when operating on
5698 * cgraph.c (cgraph_node::create_thunk): Likewise.
5699 * config/i386/i386.c (ix86_data_alignment): Likewise.
5700 (ix86_local_alignment): Likewise.
5701 * dbxout.c (stabstr_O): Likewise.
5702 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5703 * expr.c (const_vector_from_tree): Likewise.
5704 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5705 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5706 (fold_negate_expr_1, int_const_binop_1, const_binop)
5707 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5708 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5709 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5710 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5711 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5712 (fold_not_const, round_up_loc): Likewise.
5713 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5714 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5715 (alloca_call_type): Likewise.
5716 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5717 * godump.c (go_output_typedef): Likewise.
5718 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5719 * internal-fn.c (get_min_precision): Likewise.
5720 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5721 * ipa-polymorphic-call.c
5722 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5723 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5724 (ipa_modify_call_arguments): Likewise.
5725 * match.pd: Likewise.
5726 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5727 * print-tree.c (print_node_brief, print_node): Likewise.
5728 * stmt.c (expand_case): Likewise.
5729 * stor-layout.c (layout_type): Likewise.
5730 * tree-affine.c (tree_to_aff_combination): Likewise.
5731 * tree-cfg.c (group_case_labels_stmt): Likewise.
5732 * tree-data-ref.c (dr_analyze_indices): Likewise.
5733 (prune_runtime_alias_test_list): Likewise.
5734 * tree-dump.c (dequeue_and_dump): Likewise.
5735 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5736 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5737 * tree-pretty-print.c (dump_generic_node): Likewise.
5738 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5739 (simple_iv_with_niters): Likewise.
5740 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5741 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5742 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5743 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5744 (refine_value_range_using_guard, number_of_iterations_ne_max)
5745 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5746 (get_cst_init_from_scev, record_nonwrapping_iv)
5747 (scev_var_range_cant_overflow): Likewise.
5748 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5749 * tree-ssa-pre.c (compute_avail): Likewise.
5750 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5751 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5752 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5753 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5754 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5755 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5756 (dump_case_nodes, try_switch_expansion): Likewise.
5757 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5758 (vect_do_peeling): Likewise.
5759 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5760 * tree-vect-stmts.c (vectorizable_load): Likewise.
5761 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5762 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5763 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5764 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5765 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5766 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5767 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5768 (evrp_dom_walker::before_dom_children): Likewise.
5769 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5770 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5771 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5772 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5773 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5774 (get_range_pos_neg): Likewise.
5775 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5776 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5777 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5778 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5779 * config/avr/avr.c (avr_fold_builtin): Likewise.
5780 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5781 * config/msp430/msp430.c (msp430_attr): Likewise.
5782 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5783 * config/powerpcspe/powerpcspe-c.c
5784 (altivec_resolve_overloaded_builtin): Likewise.
5785 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5786 (rs6000_expand_ternop_builtin): Likewise.
5787 * config/rs6000/rs6000-c.c
5788 (altivec_resolve_overloaded_builtin): Likewise.
5789 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5790 (rs6000_expand_ternop_builtin): Likewise.
5791 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5793 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5795 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5796 when copying loop nest with only one inner loop.
5798 2017-10-10 Richard Biener <rguenther@suse.de>
5800 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5801 blocks if SCEV is active.
5802 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5804 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5805 (scev_initialize): Assert we are not yet initialized.
5807 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5809 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5810 inner loop's exit stmt by making it always exit the loop, otherwise
5811 we would generate an infinite empty loop.
5813 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5815 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5816 renaming variables in new preheader if it's deleted.
5818 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5820 * tree-loop-distribution.c (struct partition): Remove unused field
5821 loops of the structure.
5822 (partition_alloc, partition_free): Ditto.
5823 (build_rdg_partition_for_vertex): Ditto.
5825 2017-10-09 Jeff Law <law@redhat.com>
5827 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5828 return type to match prototype and documentation.
5830 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5832 * config/rs6000/rs6000.c (processor_costs): Move to ...
5833 * config/rs6000/rs6000.h: ... here.
5834 (rs6000_cost): Declare.
5836 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5838 * except.c (setjmp_fn): New global variable.
5839 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5840 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5841 if DONT_USE_BUILTIN_SETJMP is defined.
5843 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5845 * target.def (insn_cost): New hook.
5846 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5847 * doc/tm.texi: Regenerate.
5848 * rtlanal.c (insn_cost): Use the new hook.
5850 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5852 * combine.c (combine_validate_cost): Compute the new insn_cost,
5853 not just pattern_cost.
5854 (try_combine): Adjust comment.
5856 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5858 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5860 * combine.c (uid_insn_cost): Adjust comment.
5861 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5863 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5864 * dse.c (find_shift_sequence): Ditto.
5865 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5866 (bb_valid_for_noce_process_p): Use pattern_cost.
5867 * rtl.h (insn_rtx_cost): Delete.
5868 (pattern_cost): New prototype.
5869 (insn_cost): New prototype.
5870 * rtlanal.c (insn_rtx_cost): Rename to...
5871 (pattern_cost): ... this.
5874 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5876 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5877 (*jcc<mode>_0_r_i387): Ditto.
5878 (*jccxf_r_i387): Ditto.
5879 (*jcc<mode>_r_i387): Ditto.
5880 (*jccu<mode>_r_i387): Ditto.
5881 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5882 (*jcc): Rename from *jcc_1.
5884 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5886 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5887 deferred rescans after the lvx/stvx recombination pre-pass.
5889 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5891 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5892 memory operation instruction support.
5893 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5894 (rs6000-ibm-aix[789]*): Likewise.
5895 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5896 Document new functions.
5898 2017-10-09 Richard Biener <rguenther@suse.de>
5900 PR tree-optimization/82397
5901 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5902 equality only for semantically equal trees.
5904 2017-10-09 Richard Biener <rguenther@suse.de>
5906 PR tree-optimization/82449
5907 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5908 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5909 allow constant addresses.
5910 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5913 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5915 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5918 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5921 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5924 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5927 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5929 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5932 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5933 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5935 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5937 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5938 (WI_BINARY_PREDICATE_RESULT): Likewise.
5939 (wi::binary_traits::operator_result): New type.
5940 (wi::binary_traits::predicate_result): Likewise.
5941 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5942 (generic_wide_int::operator==, generic_wide_int::operator!=)
5943 (generic_wide_int::operator&, generic_wide_int::and_not)
5944 (generic_wide_int::operator|, generic_wide_int::or_not)
5945 (generic_wide_int::operator^, generic_wide_int::operator+
5946 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5948 (operator~, unary operator-, operator==, operator!=, operator&)
5949 (operator|, operator^, operator+, binary operator-, operator*): New
5951 * expr.c (get_inner_reference): Use wi::bit_and_not.
5952 * fold-const.c (fold_binary_loc): Likewise.
5953 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5954 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5955 (bit_value_binop): Likewise.
5956 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5957 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5958 (extract_range_from_binary_expr_1): Likewise.
5959 (masked_increment): Likewise.
5960 (simplify_bit_ops_using_ranges): Likewise.
5962 2017-10-09 Martin Jambor <mjambor@suse.cz>
5965 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5966 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5967 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5969 (hsa_fixup_mov_insn_type): New function.
5970 (hsa_op_with_type::get_in_type): Use it.
5971 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5972 immediates in an assert.
5973 (hsa_op_with_type::extend_int_to_32bit): New method.
5974 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5975 types. Convert to dest type if necessary.
5976 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5977 (reg_for_gimple_ssa): Pass false as min32int to
5978 hsa_type_for_scalar_tree_type.
5979 (gen_hsa_addr): Fixup type when creating addresable temporary.
5980 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5981 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5982 necessary. Call hsa_fixup_mov_insn_type.
5983 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5984 extend operands and convert to dest type if necessary.
5985 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5986 to dest type if necessary.
5987 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5988 if conversion nt necessary and size matches.
5989 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5990 to dest type if necessary.
5991 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5992 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5994 (gen_hsa_clrsb): Likewise.
5995 (gen_hsa_ffs): Likewise.
5996 (gen_hsa_divmod): Extend operands and convert to dest type if
5998 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6000 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6002 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6003 Remove empty default arguments. Use a brace block as output
6005 (conditional return): Ditto.
6007 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6008 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6010 (group_ending_nop): Ditto.
6011 (doloop_end): Ditto.
6012 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6013 (splitters for those): Ditto.
6015 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6017 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6018 a conditional jump (and the compare for it) so that pc_rtx is the
6020 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6021 for the deleted and renamed ctr<mode>_internal[234] patterns.
6022 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6023 Delete second conditional return pattern.
6024 (ctr<mode>_internal2): Delete this second bdnz pattern.
6025 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6026 (ctr<mode>_internal4): Delete this second bdz pattern.
6028 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6030 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6031 (always_initialized_rtx_for_ssa_name_p): New predicate.
6032 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6033 (finish_out_of_ssa): Free new field of SA.
6034 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6035 * tree-ssa-coalesce.c: Include tree-ssa.h.
6036 (get_parm_default_def_partitions): Remove extern keyword.
6037 (get_undefined_value_partitions): New function.
6038 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6039 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6041 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6043 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6045 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6047 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6049 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6050 for targets that preffer 128bit.
6052 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6054 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6056 2017-10-08 Olivier Hainque <hainque@adacore.com>
6058 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6059 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6061 (thumb_set_return_address): Likewise.
6063 2017-10-08 Olivier Hainque <hainque@adacore.com>
6065 * common/config/arm/arm-common.c (arm_except_unwind_info):
6066 Handle DWARF2_UNWIND_INFO.
6068 2017-10-07 Michael Collison <michael.collison@arm.com>
6070 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6073 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6075 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6076 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6077 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6078 defined, force the creation of a new block for a dispatch label.
6080 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6082 * invoke.texi (Wsuggest-attribute=cold): Document.
6083 * common.opt (Wsuggest-attribute=cold): New
6084 * ipa-pure-const.c (warn_function_cold): New function.
6085 * predict.c (compute_function_frequency): Use it.
6086 * predict.h (warn_function_cold): Declare.
6088 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6090 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6093 2017-10-06 Martin Liska <mliska@suse.cz>
6095 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6096 keyword for member functions.
6097 (struct sanopt_tree_couple): New struct.
6098 (struct sanopt_tree_couple_hash): New function.
6099 (struct sanopt_ctx): Add new hash_map.
6100 (has_dominating_ubsan_ptr_check): New function.
6101 (record_ubsan_ptr_check_stmt): Likewise.
6102 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6103 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6104 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6106 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6109 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6110 aarch64_simd_valid_immediate on CONST_VECTORs.
6111 (aarch64_reg_or_bic_imm): Likewise.
6113 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6115 PR rtl-optimization/82396
6116 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6118 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6120 * graphite-dependences.c (scop_get_reads): Move code to...
6121 (scop_get_must_writes): Move code to...
6122 (scop_get_may_writes): Move code to...
6123 (scop_get_reads_and_writes): ... here.
6124 (scop_get_dependences): Call scop_get_reads_and_writes.
6126 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6128 PR tree-optimization/82434
6129 * fold-const.h (can_native_encode_type_p,
6130 can_native_encode_string_p): Remove.
6131 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6132 don't encode anything, just return what would be otherwise returned.
6133 (native_encode_fixed, native_encode_complex, native_encode_vector):
6135 (native_encode_string): Likewise. Inline by hand
6136 can_native_encode_string_p.
6137 (can_native_encode_type_p): Remove.
6138 (can_native_encode_string_p): Remove.
6139 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6140 STRING_CSTs using can_native_encode_string_p, test all
6141 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6142 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6143 argument from native_encode_expr.
6144 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6145 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6148 2017-10-06 Richard Biener <rguenther@suse.de>
6150 PR tree-optimization/82397
6151 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6152 operand_equal_p but rely on data_ref_compare_tree for detecting
6154 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6155 to match up with dr_group_sort_cmp.
6157 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6160 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6162 * config/s390/s390-builtin-types.def: Regenerate.
6164 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6167 * config/s390/s390-builtin-types.def: Regenerate.
6168 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6169 Change flag from B_VXE to B_VX.
6170 (s390_vec_min_dbl): Remove B_VXE flag.
6172 2017-10-06 Richard Biener <rguenther@suse.de>
6174 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6175 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6176 translate_isl_ast_to_gimple::is_valid_rename,
6177 translate_isl_ast_to_gimple::get_rename,
6178 translate_isl_ast_to_gimple::get_def_bb_for_const,
6179 translate_isl_ast_to_gimple::get_new_name,
6180 translate_isl_ast_to_gimple::collect_all_ssa_names,
6181 translate_isl_ast_to_gimple::copy_loop_phi_args,
6182 translate_isl_ast_to_gimple::collect_all_ssa_names,
6183 translate_isl_ast_to_gimple::copy_loop_phi_args,
6184 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6185 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6186 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6187 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6188 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6189 translate_isl_ast_to_gimple::copy_cond_phi_args,
6190 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6191 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6192 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6193 translate_isl_ast_to_gimple::rename_uses,
6194 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6195 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6196 (set_rename_for_each_def): Likewise.
6197 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6198 here. Handle rewriting SCEV analyzable uses here.
6199 (copy_bb_and_scalar_dependences): Generate code for PHI
6201 (graphite_regenerate_ast_isl): Adjust.
6202 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6203 (add_write, add_read): New functions.
6204 (build_cross_bb_scalars_def): Use it and simplify.
6205 (build_cross_bb_scalars_use): Likewise.
6206 (graphite_find_cross_bb_scalar_vars): Inline into...
6207 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6208 simulating out-of-SSA. Compute liveout and add dependencies.
6209 (build_scops): Force an empty entry block.
6210 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6212 (sese_build_liveouts): Declare.
6213 (sese_trivially_empty_bb_p): Likewise.
6214 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6215 compute liveout and debug_liveout.
6216 (sese_bad_liveouts_use): Remove.
6217 (sese_reset_debug_liveouts_bb): Likewise.
6218 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6219 (sese_build_liveouts): Build liveout and debug_liveout and store
6221 (new_sese_info): Adjust.
6222 (free_sese_info): Likewise.
6223 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6224 do not build liveout here.
6225 (move_sese_in_condition): Adjust region entry.
6226 (scev_analyzable_p): Match up with chrec_apply requirements.
6227 (sese_trivially_empty_bb_p): New.
6228 * tree-into-ssa.c (get_reaching_def): Properly support generating
6229 default-defs for incremental rewrite of anonymous names.
6231 2017-10-06 Richard Biener <rguenther@suse.de>
6233 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6234 precision do not perform modulo reduction.
6236 2017-10-06 Richard Biener <rguenther@suse.de>
6238 PR tree-optimization/82436
6239 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6240 conservatively choose the vectorization factor when checking
6241 whether we can perform the required load permutation.
6242 (vect_transform_slp_perm_load): Assert when we may not fail.
6244 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6246 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6247 message for incompatible -msdata=* and -mcall-* options.
6249 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6251 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6252 rate for post-reload scheduling.
6254 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6256 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6258 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6260 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6261 to improve monte carlo in scimark.
6263 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6265 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6266 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6267 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6268 pentium4_cost, nocona_cost): Set reassociation width to 1.
6269 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6270 width to 2 for fp operations and 1 otherwise.
6271 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6273 (atom_cost): Set reassociation width to 2.
6274 (slm_cost, generic_cost): Set fp reassociation width
6275 to 2 and 1 otherwise.
6276 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6277 (core_cost): Set fp reassociation width to 4 and vector to 2.
6278 (ix86_reassociation_width): Rewrite using cost table; special case
6279 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6280 and TARGET_AVX128_OPTIMAL.
6281 * config/i386/i386.h (processor_costs): Add
6282 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6283 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6284 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6285 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6286 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6287 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6289 2017-10-05 Nathan Sidwell <nathan@acm.org>
6291 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6293 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6295 * config/arm/arm.c (arm_test_fpu_data): New.
6296 (arm_run_selftests): Call arm_test_fpu_data.
6298 2017-10-04 Nathan Sidwell <nathan@acm.org>
6300 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6302 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6304 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6306 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6307 check type for aarch64_simd_valid_immediate.
6308 (aarch64_output_simd_mov_immediate): Update prototype.
6309 (aarch64_simd_valid_immediate): Update prototype.
6310 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6311 support for ORR-immediate.
6312 (and<mode>3): modified pattern to add support for BIC-immediate.
6313 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6314 now checks for valid immediate for BIC and ORR based on new enum
6316 (aarch64_output_simd_mov_immediate): Function now used to output
6317 BIC/ORR imm as well based on new enum argument.
6318 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6320 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6321 (aarch64_reg_or_bic_imm): Likewise.
6323 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6325 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6326 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6328 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6332 PR rtl-optimization/82396
6333 * haifa-sched.c (autopref_multipass_init): Simplify
6335 (autopref_rank_data): Simplify sort order.
6336 * sched-int.h (autopref_multipass_data_): Remove
6337 multi_mem_insn_p, min_offset and max_offset.
6339 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6341 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6343 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6345 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6348 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6350 * doc/sourcebuild.texi: Document vect_long_mult.
6352 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6354 PR tree-optimization/82413
6355 * fold-const.c (build_range_check): Use widest_int when comparing
6356 the maximum ETYPE value with HIGH.
6358 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6360 PR rtl-optimization/82396
6361 * haifa-sched.c (autopref_multipass_init): Simplify
6363 (autopref_rank_data): Simplify sort order.
6364 * sched-int.h (autopref_multipass_data_): Remove
6365 multi_mem_insn_p, min_offset and max_offset.
6367 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6369 PR tree-optimization/82381
6370 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6371 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6374 PR tree-optimization/82374
6375 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6376 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6377 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6378 current_function_decl to the new decl.
6380 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6382 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6383 helper macro for IEEE float128 hardware built-in functions.
6384 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6386 (TRUNCF128_ODD): Likewise.
6387 (ADDF128_ODD): Likewise.
6388 (SUBF128_ODD): Likewise.
6389 (MULF128_ODD): Likewise.
6390 (DIVF128_ODD): Likewise.
6391 (FMAF128_ODD): Likewise.
6392 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6393 UNSPEC_TRUNC_ROUND_TO_ODD.
6394 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6395 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6396 floating point round to odd instructions.
6397 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6398 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6399 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6400 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6401 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6402 (trunc<mode>sf2_hw): Change the truncate with round to odd
6403 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6404 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6405 to odd hardware instructions.
6406 (sub<mode>3_odd): Likewise.
6407 (mul<mode>3_odd): Likewise.
6408 (div<mode>3_odd): Likewise.
6409 (sqrt<mode>2_odd): Likewise.
6410 (fma<mode>4_odd): Likewise.
6411 (fms<mode>4_odd): Likewise.
6412 (nfma<mode>4_odd): Likewise.
6413 (nfms<mode>4_odd): Likewise.
6414 (trunc<mode>df2_odd): Change the truncate with round to odd
6415 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6417 * doc/extend.texi (PowerPC built-in functions): Update documentation
6418 for existing IEEE float128-bit built-in functions. Add built-in
6419 functions that generate the IEEE 128-bit floating point round to
6422 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6424 PR rtl-optimization/77729
6425 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6426 to (X&(C1&~C2))|C2 transformations.
6428 2017-10-03 Martin Jambor <mjambor@suse.cz>
6430 PR tree-optimization/82363
6431 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6432 mismatch, mark lacc written regardless of racc.
6434 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6436 PR tree-optimization/82381
6437 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6438 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6439 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6440 fallthrough into reassoc_stmt_dominates_stmt_p.
6443 * combine.c (combine_instructions): Don't combine in unreachable
6446 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6450 function to not use the have_cpu variable. Do not set cpu_index,
6451 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6453 (rs6000_valid_attribute_p): Remove duplicate initializations of
6454 old_optimize and func_optimize.
6455 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6456 (rs6000_activate_target_options): Make global.
6457 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6460 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6462 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6463 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6464 Set *poffset to 0 and *psize and *pmax_size to -1 if
6465 *poffset + *psize overflows in HOST_WIDE_INT.
6467 PR tree-optimization/82387
6468 PR tree-optimization/82388
6469 PR tree-optimization/82389
6470 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6471 instead of live_bytes non-NULL.
6473 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6476 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6479 2017-10-02 Richard Biener <rguenther@suse.de>
6481 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6482 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6483 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6485 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6487 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6488 requested precision matches the type's.
6489 * calls.c (alloc_max_size): Calculate the new candidate size as
6490 a widest_int and use wi::to_widest when comparing it with the
6491 current candidate size.
6492 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6493 zero rather than integer_zero_node.
6494 * match.pd: Check for a no-op conversion before using wi::add
6495 rather than after. Use tree_to_uhwi when summing small shift
6496 counts into an unsigned int.
6498 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6499 Alan Hayward <alan.hayward@arm.com>
6500 David Sherwood <david.sherwood@arm.com>
6503 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6504 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6505 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6506 POINTER_AND_FP_REGS.
6508 2017-10-02 Richard Biener <rguenther@suse.de>
6510 PR tree-optimization/82355
6511 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6512 a mapping for the enclosing loop but avoid generating one for
6514 (copy_bb_and_scalar_dependences): Remove premature codegen
6515 error on PHIs in blocks duplicated into multiple places.
6516 * graphite-scop-detection.c
6517 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6518 in the region use it as loop and nest to analyze the DR in.
6519 (try_generate_gimple_bb): Likewise.
6520 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6521 (add_loop_constraints): For blocks in a loop not in the region
6522 create a dimension with a single iteration.
6523 * sese.h (gbb_loop_at_index): Remove assert.
6525 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6527 * omp-expand.c (adjust_context_scope): New function.
6528 (expand_parallel_call): Call adjust_context_scope.
6530 2017-10-01 Jeff Law <law@redhat.com>
6532 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6533 dom_opt_dom_walker class with direct access to private members.
6534 Add comments. Call test_for_singularity.
6535 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6536 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6537 m_dummy_cond anymore.
6538 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6540 (pass_dominator:execute): Build the dummy_cond here and pass it
6541 to the dom_opt_dom_walker ctor.
6542 (test_for_singularity): New function.
6544 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6545 Maya Rashish <coypu@sdf.org>
6547 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6548 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6549 (alpha*-*-netbsd*) Use nbsd_tm_file.
6550 (arm*-*-netbsdelf*) Likewise.
6551 (i[34567]86-*-netbsdelf*) Likewise.
6552 (x86_64-*-netbsd*) Likewise.
6553 (mips*-*-netbsd*) Likewise.
6554 (powerpc-*-netbsd*) Likewise.
6555 (sh*-*-netbsd*) Likewise.
6556 (sparc-*-netbsdelf*) Likewise.
6557 (sparc64-*-netbsd*) Likewise.
6558 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6560 (vax-*-netbsdelf*) Likewise.
6561 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6562 (UINT_FAST8_TYPE) Likewise.
6563 (INT_FAST16_TYPE) Check CHAR_FAST16.
6564 (UINT_FAST16_TYPE) Likewise.
6566 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6569 * config/i386/i386.md
6570 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6571 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6572 *divmodsi4_zext_2): New define_insn_and_split.
6573 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6574 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6575 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6576 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6577 New define_insn_and_split.
6578 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6579 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6580 operands[1] having DImode when mode is SImode.
6582 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6583 always SImode for DIV and MOD in REG_EQUAL notes.
6585 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6588 * match.pd: Fix handling of NaNs in pattern.
6590 2017-09-29 Jeff Law <law@redhat.com>
6592 * sbitmap.c (bitmap_bit_in_range_p): New function.
6593 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6594 * tree-ssa-dse.c (live_bytes_read): New function.
6595 (dse_classify_store): Ignore reads of dead bytes.
6597 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6598 typos and whitespace errors.
6599 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6600 * config/s390/s390.c (s390_emit_prologue): Likewise.
6602 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6605 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6606 with a symbol for LRA.
6608 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6610 PR rtl-optimization/82338
6611 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6613 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6615 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6616 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6617 (qsort_chk): Declare.
6618 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6619 (qsort_chk): New function.
6621 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6623 PR tree-optimization/82337
6624 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6625 phi definition if the PHI result appears in an abnormal PHI.
6626 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6627 basis appears in an abnormal PHI.
6629 2017-09-29 Richard Biener <rguenther@suse.de>
6631 * graphite-isl-ast-to-gimple.c
6632 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6633 (binary_op_to_tree): Use it.
6634 (get_rename_from_scev): Likewise.
6635 (copy_loop_phi_nodes): Likewise.
6636 (copy_bb_and_scalar_dependences): Likewise.
6637 (translate_pending_phi_nodes): Likewise.
6639 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6642 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6643 for movabsq $(i32 << shift), r64.
6645 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6647 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6648 index when encoding %esp as %rsp to avoid 0x67 prefix.
6650 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6652 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6653 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6655 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6657 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6658 Extensions with more than 16 double VFP registers.
6659 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6660 to_clear_mask and all code related to it. Replace the remaining
6661 entry by a sbitmap and adapt code accordingly.
6663 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6665 * brig-builtins.def: Change pure attributes to const.
6667 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6669 * config.gcc (default_gnu_indirect_function): Default to yes for
6670 sparc*-*-linux* with glibc.
6672 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6674 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6675 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6676 when creating .init_array and .fini_array sections with priority
6679 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6682 * config/aarch64/aarch64.c
6683 (aarch64_builtin_support_vector_misalignment): Always return false
6684 when misalignment is unknown.
6686 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6688 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6689 this function to return false if the definition used by the swap
6690 instruction is artificial, or if the memory address from which the
6691 constant value is loaded is not represented by a base address held
6692 in a register or if the base address register is a frame or stack
6693 pointer. Additionally, return false if the base address of the
6694 loaded constant is a SYMBOL_REF but is not considered to be a
6696 (replace_swapped_load_constant): New function.
6697 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6698 loaded constant vector with a load of a swapped constant vector.
6700 2017-09-27 Carl Love <cel@us.ibm.com>
6702 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6703 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6704 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6707 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6709 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6710 first, always call autopref_rank_data otherwise.
6712 2017-09-27 Richard Biener <rguenther@suse.de>
6714 * graphite-scop-detection.c (find_scop_parameters): Move
6715 loop bound handling ...
6716 (gather_bbs::before_dom_children): ... here, avoiding the need
6717 to build scop_info->loop_nest.
6718 (record_loop_in_sese): Remove.
6719 * sese.h (sese_info_t::loop_nest): Remove.
6720 * sese.c (new_sese_info): Do not allocate loop_nest.
6721 (free_sese_info): Do not free loop_nest.
6723 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6726 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6727 lhs from calls if the lhs has addressable type.
6729 2017-09-27 Richard Biener <rguenther@suse.de>
6731 * graphite.h (scop::max_alias_set): New member.
6732 * graphite-scop-detection.c: Remove references to non-existing
6733 --param in comments.
6734 (build_alias_sets): Record the maximum alias set used for drs.
6735 (build_scops): Support zero as unlimited for
6736 --param graphite-max-arrays-per-scop.
6737 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6739 (build_poly_sr_1): ... here. Compute alias set based on the
6740 maximum alias set used for drs rather than
6741 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6743 2017-09-27 Richard Biener <rguenther@suse.de>
6745 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6746 --param loop-block-tile-size=0 to disable tiling.
6748 2017-09-27 Richard Biener <rguenther@suse.de>
6750 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6751 (graphite-max-nb-scop-params): Document special value zero.
6752 * domwalk.h (dom_walker::STOP): New symbolical constant.
6753 (dom_walker::dom_walker): Add optional parameter for bb to
6755 (dom_walker::~dom_walker): Declare.
6756 (dom_walker::before_dom_children): Document STOP return value.
6757 (dom_walker::m_user_bb_to_rpo): New member.
6758 (dom_walker::m_bb_to_rpo): Likewise.
6759 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6760 mapping here if not provided by the user.
6761 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6762 provided by the user.
6763 (dom_walker::STOP): Define.
6764 (dom_walker::walk): Do not compute bb to RPO mapping here.
6765 Support STOP return value from before_dom_children to stop
6767 * graphite-optimize-isl.c (optimize_isl): If the schedule
6768 is the same still generate code if -fgraphite-identity
6769 or -floop-parallelize-all are given.
6770 * graphite-scop-detection.c: Include cfganal.h.
6771 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6773 (gather_bbs::before_dom_children): Return STOP for BBs
6775 (build_scops): Compute bb to RPO mapping and pass it to
6776 the domwalk. Treat --param graphite-max-nb-scop-params=0
6777 as not limiting the number of params.
6778 * graphite.c (graphite_initialize): Remove limit on the
6779 number of basic-blocks in a function.
6780 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6781 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6782 default value of 10.
6784 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6786 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6787 Adjust code to eliminate needing to do the shift right 32-bits
6788 operation after XSCVDPSPN.
6790 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6792 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6793 ((X / Y) != 0 -> X >= Y): Likewise.
6795 2017-09-26 Carl Love <cel@us.ibm.com>
6797 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6798 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6799 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6800 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6801 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6802 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6803 definitions and overloading.
6804 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6805 statement for P9V_BUILTIN_XST_LEN_R.
6806 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6807 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6808 define_expand and define_insn for the instructions and builtins.
6809 * doc/extend.texi: Update the built-in documentation file for the new
6811 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6812 define_insn for the instructions
6814 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6817 * gcc/config/netbsd-protos.h: New file.
6818 * gcc/config/netbsd.c: New file.
6819 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6820 * gcc/config/t-netbsd: New file.
6821 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6822 (tmake_file) Add t-netbsd.
6823 (extra_objs) Add netbsd.o.
6825 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6829 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6831 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6833 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6834 sign extension from a vector register to a GPR by doing a 32-bit
6835 direct move and then an EXTSW.
6836 (extendsi<mode>2 splitter): Likewise.
6837 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6838 right or vector extract after doing XSCVDPSPN. Use
6839 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6841 (movdi_from_sf_zero_ext): Likewise.
6842 (reload_gpr_from_vsxsf): Likewise.
6843 (p8_mfvsrd_4_disf): Delete, no longer used.
6844 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6845 then needing to move the SFmode to a GPR to use the XSCVDPSP
6846 instruction instead of FRSP and XSCVDPSPN.
6847 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6848 it is adjacent to the other XSCVSPDP insns.
6849 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6850 SFmode to be in traditional Altivec registers.
6851 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6852 (vsx_xscvspdpn): Likewise.
6853 (vsx_xscvspdpn_scalar): Likewise.
6855 2017-09-26 Martin Jambor <mjambor@suse.cz>
6857 * tree-sra.c (compare_access_positions): Put integral types first,
6858 stabilize sorting of integral types, remove conditions putting
6859 non-full-precision integers last.
6860 (sort_and_splice_var_accesses): Disable scalarization if a
6861 non-integert would be represented by a non-full-precision integer.
6863 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6865 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6866 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6867 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6868 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6869 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6870 conditionals inside the function instead of around it. Call
6871 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6872 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6874 2017-09-26 Richard Biener <rguenther@suse.de>
6876 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6878 (scop_detection::build_scop_breadth): ... this. Removed.
6879 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6880 (scop_detection::harmful_stmt_in_bb): Likewise.
6881 (scop_detection::graphite_can_represent_stmt): Likewise.
6882 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6883 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6884 (scop_detection::can_represent_loop_1): ... this. Removed.
6885 (scop_detection::harmful_loop_in_region): Simplify after inlining
6886 the above and remove more quadraticness.
6887 (build_scops): Adjust.
6888 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6891 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6894 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6895 REGNO (base) == SP_REG if base is a REG.
6898 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6899 if it is different SSA_NAME.
6900 (optimize_range_tests_cmp_bitwise): New function.
6901 (optimize_range_tests): Call it.
6903 2017-09-26 Richard Biener <rguenther@suse.de>
6905 PR tree-optimization/82321
6906 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6907 for the def being inside the loop.
6909 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6911 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6913 * config/s390/s390-builtins.def: Fix constraint on op4.
6915 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6917 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6918 independent expanders.
6919 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6920 ("vec_ordered", "vec_unordered"): New expanders.
6922 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6924 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6927 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6929 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6930 vec_unpacks_lo_v16qi.
6931 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6933 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6935 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6936 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6937 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6939 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6941 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6943 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6945 ("*vec_slb<mode>"): New insn pattern.
6946 ("vec_shr_<mode>"): New expander.
6947 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6948 and force the shift count operand to V16QImode.
6949 ("vec_srb<mode>"): Set shift count mode to V16QI.
6951 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6953 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6954 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6955 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6957 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6960 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6962 2017-09-26 Richard Biener <rguenther@suse.de>
6964 PR tree-optimization/82320
6965 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6968 2017-09-25 Jeff Law <law@redhat.com>
6970 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6971 prototype for new argument.
6972 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6973 mostly extracted from rs6000_emit_allocate_stack.
6974 (rs6000_emit_probe_stack_range_stack_clash): New function.
6975 (rs6000_emit_allocate_stack): Call
6976 rs6000_emit_probe_stack_range_stack_clash as needed.
6977 (rs6000_emit_probe_stack_range): Add additional argument
6978 to call to gen_probe_stack_range{si,di}.
6979 (output_probe_stack_range): New.
6980 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6981 (output_probe_stack_range_stack_clash): New.
6982 (rs6000_emit_prologue): Emit notes into dump file as requested.
6983 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6984 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6985 Add additional operand and pass it to output_probe_stack_range.
6987 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6989 PR tree-optimization/82163
6990 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6991 (checking_verify_loop_closed_ssa): New parameter.
6992 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6993 (check_loop_closed_ssa_stmt): Delete.
6994 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6995 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6996 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6999 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7001 * brig-builtins.def: Treat HSAIL barrier builtins as
7002 setjmp/longjump style functions.
7004 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7006 * target.def (constant_alignment): New hook.
7007 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7008 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7009 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7010 * doc/tm.texi: Regenerate.
7011 * targhooks.h (default_constant_alignment): Declare.
7012 (constant_alignment_word_strings): Likewise.
7013 * targhooks.c (default_constant_alignment): New function.
7014 (constant_alignment_word_strings): Likewise.
7015 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7016 instead of CONSTANT_ALIGNMENT.
7017 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7018 (force_const_mem): Likewise.
7019 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7020 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7021 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7022 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7023 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7025 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7026 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7027 constant_alignment_word_strings.
7028 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7029 (CONSTANT_ALIGNMENT): Likewise.
7030 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7031 (arm_constant_alignment): New function.
7032 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7033 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7034 constant_alignment_word_strings.
7035 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7036 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7037 constant_alignment_word_strings.
7038 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7039 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7040 (cris_constant_alignment): New function.
7041 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7042 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7043 (epiphany_constant_alignment): New function.
7044 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7045 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7046 constant_alignment_word_strings.
7047 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7048 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7049 constant_alignment_word_strings.
7050 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7051 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7052 constant_alignment_word_strings.
7053 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7054 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7055 * config/i386/i386.c (ix86_constant_alignment): Make static.
7056 Use the same interface as the target hook.
7057 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7058 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7059 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7060 constant_alignment_word_strings.
7061 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7062 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7063 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7064 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7065 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7066 constant_alignment_word_strings.
7067 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7068 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7069 constant_alignment_word_strings.
7070 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7071 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7072 constant_alignment_word_strings.
7073 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7074 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7076 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7077 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7078 * config/mips/mips.c (mips_constant_alignment): New function.
7079 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7080 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7081 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7082 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7083 (mmix_constant_alignment): Make static. Use the same interface
7085 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7086 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7087 constant_alignment_word_strings.
7088 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7089 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7090 constant_alignment_word_strings.
7091 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7092 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7093 constant_alignment_word_strings.
7094 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7095 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7096 (rs6000_constant_alignment): New function.
7097 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7098 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7099 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7100 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7101 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7102 (rs6000_constant_alignment): New function.
7103 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7104 * config/s390/s390.c (s390_constant_alignment): New function.
7105 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7106 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7107 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7108 constant_alignment_word_strings.
7109 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7110 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7111 (sparc_constant_alignment): New function.
7112 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7113 * config/spu/spu.c (spu_constant_alignment): New function.
7114 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7115 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7116 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7117 constant_alignment_word_strings.
7118 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7119 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7120 constant_alignment_word_strings.
7121 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7122 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7123 constant_alignment_word_strings.
7124 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7125 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7126 (visium_constant_alignment): New function.
7127 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7128 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7129 (xtensa_constant_alignment): New function.
7130 * system.h (CONSTANT_ALIGNMENT): Poison.
7132 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7135 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7136 (rs6000_builtin_valid_without_lhs): New helper function.
7137 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7138 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7140 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7142 * target.h (vec_perm_indices): Use unsigned short rather than
7144 (auto_vec_perm_indices): Likewise.
7145 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7146 Use unsigned int rather than unsigned char.
7147 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7149 2017-09-25 Richard Biener <rguenther@suse.de>
7151 * cfgloop.h (sort_sibling_loops): Declare.
7152 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7153 (sort_sibling_loops): New function sorting the sibling loop list
7155 * graphite.c (graphite_transform_loops): Sort sibling loops.
7157 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7159 * target.def (vec_perm_const_ok): Change sel parameter to
7161 * optabs-query.c (can_vec_perm_p): Update accordingly.
7162 * doc/tm.texi: Regenerate.
7163 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7164 auto_vec_perm_indices and remove separate nelt field.
7165 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7166 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7167 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7168 (aarch64_expand_vec_perm_const): Update accordingly.
7169 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7170 to vec_perm_indices.
7171 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7172 auto_vec_perm_indices and remove separate nelt field.
7173 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7174 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7175 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7177 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7178 to vec_perm_indices.
7179 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7180 sel to vec_perm_indices.
7181 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7182 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7183 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7185 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7188 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7191 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7192 on the FUNCTION_DECL function context if it has a DIE that is a
7195 2017-09-25 Richard Biener <rguenther@suse.de>
7197 PR tree-optimization/82285
7198 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7201 2017-09-25 Tom de Vries <tom@codesourcery.com>
7205 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7208 2017-09-25 Richard Biener <rguenther@suse.de>
7210 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7211 ISL errors other than isl_error_quota happen. Dump if the
7212 schedule is the same.
7213 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7214 errors instead of aborting inside ISL.
7216 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7219 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7220 of libgcc_eh for m64.
7221 * config/i386/darwin64.h: Likewise.
7223 2017-09-25 Richard Biener <rguenther@suse.de>
7226 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7227 attribute for incomplete types nor twice for complete ones.
7229 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7232 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7233 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7236 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7239 * config/i386/i386.opt (mprefer-avx256): Use
7240 ix86_target_flags variable.
7241 * config/i386/i386.c (ix86_target_string): Move
7242 -mprefer-avx256 to flag2_opts.
7244 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7247 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7248 and x != -1 | y != -1 into (x & y) != -1.
7250 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7252 * config.gcc: Add new case statement to set
7253 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7254 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7255 s390x-*-linux* case statements. Added aarch64 to the list of
7256 supported architectures.
7258 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7260 PR tree-optimization/82289
7261 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7262 STMT_VINFO_RELEVANT_P.
7264 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7265 Alan Hayward <alan.hayward@arm.com>
7266 David Sherwood <david.sherwood@arm.com>
7268 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7269 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7270 (extract_range_from_binary_expr_1): Don't call
7271 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7273 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7274 Alan Hayward <alan.hayward@arm.com>
7275 David Sherwood <david.sherwood@arm.com>
7277 * target.def (preferred_vector_alignment): New hook.
7278 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7280 * doc/tm.texi: Regenerate.
7281 * targhooks.h (default_preferred_vector_alignment): Declare.
7282 * targhooks.c (default_preferred_vector_alignment): New function.
7283 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7285 (DR_TARGET_ALIGNMENT): New macro.
7286 (aligned_access_p): Update commentary.
7287 (vect_known_alignment_in_bytes): New function.
7288 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7290 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7291 Calculate the misalignment based on the target alignment rather than
7293 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7294 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7295 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7296 the target alignment, rather than masking the element misalignment
7297 with the number of elements in a vector. Also use the target
7298 alignment when calculating the maximum number of peels.
7299 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7300 instead of TYPE_ALIGN_UNIT.
7301 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7302 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7303 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7304 (vect_create_data_ref_ptr): Likewise.
7305 (vect_setup_realignment): Realign by ANDing with
7306 -DR_TARGET_MISALIGNMENT.
7307 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7308 the number of peels based on DR_TARGET_ALIGNMENT.
7309 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7310 with the guaranteed alignment boundary when deciding whether
7312 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7313 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7314 (ensure_base_align): Remove stmt_info parameter. Get the
7315 target base alignment from DR_TARGET_ALIGNMENT.
7316 (vectorizable_store): Update call accordingly. Interpret
7317 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7319 (vectorizable_load): Likewise.
7321 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7322 Alan Hayward <alan.hayward@arm.com>
7323 David Sherwood <david.sherwood@arm.com>
7325 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7326 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7327 (vect_enhance_data_refs_alignment): Likewise.
7329 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7331 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7332 error. Only quit immediately if parsing is complete.
7333 (BEGIN): Initialize fatal_err and parse_done.
7334 (begin fpu, end fpu): Check number of arguments.
7335 (begin arch, end arch): Likewise.
7336 (begin cpu, end cpu): Likewise.
7337 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7338 (optalias): Likewise.
7340 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7342 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7343 * config/arm/arm-isa.h: Delete. Move definitions to ...
7344 * arm-cpus.in: ... here. Use new feature and fgroup values.
7345 * config/arm/arm.c (arm_option_override): Use lower case for feature
7347 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7348 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7349 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7351 (print_isa_bits_for): New function.
7352 (gen_isa): New function.
7353 (gen_comm_data): Use print_isa_bits_for.
7354 (define feature): New keyword.
7355 (define fgroup): New keyword.
7356 * config/arm/t-arm (TM_H): Remove.
7357 (GTM_H): Add arm-isa.h.
7358 (arm-isa.h): Add rule to generate file.
7359 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7360 case for feature bit names.
7362 2017-09-22 Richard Biener <rguenther@suse.de>
7364 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7366 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7367 print of no dependency loops ...
7368 * graphite.c (graphite_transform_loops): ... here.
7369 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7371 (same_close_phi_node, remove_duplicate_close_phi,
7372 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7373 (canonicalize_loop_closed_ssa): ... here and simplify.
7374 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7375 (optimize_isl): Use dump_printf_loc to tell when we stopped
7376 optimizing because of an ISL timeout.
7378 2017-09-22 Richard Biener <rguenther@suse.de>
7380 PR tree-optimization/82291
7381 * tree-if-conv.c (predicate_mem_writes): Make sure to
7382 remove writes in blocks predicated with false.
7384 2017-09-22 Richard Biener <rguenther@suse.de>
7386 * sese.c: Include cfganal.h.
7387 (if_region_set_false_region): Remove.
7388 (create_if_region_on_edge): Likewise.
7389 (move_sese_in_condition): Re-implement without destroying
7392 2017-09-22 Richard Biener <rguenther@suse.de>
7394 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7395 Verify both BBs contain loop PHI nodes before dispatching to
7397 (graphite_regenerate_ast_isl): Do not recompute dominators,
7398 do not verify three times. Restructure for clarity.
7399 * graphite-scop-detection.c (same_close_phi_node,
7400 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7401 defined_in_loop_p, canonicalize_loop_closed_ssa,
7402 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7403 checking and SSA rewrite, move to ...
7404 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7405 (graphite_initialize): Do not pass in ctx, do not reset the
7406 SCEV cache, compute only dominators.
7407 (graphite_transform_loops): Allocate ISL ctx after
7408 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7409 Maintain post-dominators only around build_scops.
7410 * sese.c (if_region_set_false_region): Make static. Free
7411 and recompute dominators.
7412 (move_sese_in_condition): Assert we don't get called with
7413 post-dominators computed.
7414 * sese.h (if_region_set_false_region): Remove.
7416 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7418 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7419 mode attribute for TARGET_AVX512VL.
7421 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7423 * config/i386/i386.opt (mprefer-avx256): New option.
7424 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7426 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7427 for TARGET_PREFER_AVX256.
7428 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7430 2017-09-21 Jeff Law <law@redhat.com>
7432 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7433 Fix dump output if the only stack space is for pushed registers.
7435 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7437 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7440 2017-09-21 Martin Sebor <msebor@redhat.com>
7443 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7444 code (in C++) or code that triggers warnings.
7446 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7448 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7450 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7452 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7453 * sched-rgn.c: Ditto.
7454 * sel-sched-ir.c: Ditto.
7456 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7458 * toplev.h (set_random_seed): Adjust return type.
7459 * toplev.c (init_local_tick): Move eager initialization of random_seed
7460 to get_random_seed. Adjust comment.
7461 (init_random_seed): Inline to get_random_seed, delete.
7462 (get_random_seed): Initialize random_seed lazily.
7463 (set_random_seed): Do not return previous value.
7464 (print_switch_value): Do not call get_random_seed.
7466 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7468 * cgraph.c (delete_function_version): New, broken out from...
7469 (cgraph_node::delete_function_version): ...here. Rename to
7470 cgraph_node::delete_function_version_by_decl. Update all uses.
7471 (cgraph_node::remove): Call delete_function_version.
7473 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7476 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7477 VAR_DECLs to which addressable non-volatile parameters are mapped
7478 and for id->retvar after the return value assignment. Clear
7479 id->retval and id->retbnd after inlining.
7481 2017-09-21 Richard Biener <rguenther@suse.de>
7483 PR tree-optimization/82276
7484 PR tree-optimization/82244
7485 * tree-vrp.c (build_assert_expr_for): Set
7486 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7488 (remove_range_assertions): Revert earlier change.
7490 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7493 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7495 2017-09-21 Richard Biener <rguenther@suse.de>
7497 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7498 Restore valid IL after code generation errors.
7499 * graphite.c (graphite_transform_loops): Diagnose code
7500 generation issues as MSG_MISSED_OPTIMIZATION and continue
7501 with processing SCOPs.
7503 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7504 Alan Hayward <alan.hayward@arm.com>
7505 David Sherwood <david.sherwood@arm.com>
7507 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7508 rather than choosing between plus_constant and gen_rtx_<CODE>.
7509 * expr.c (emit_push_insn): Likewise.
7510 (expand_expr_real_2): Likewise.
7512 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7518 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7519 Alan Hayward <alan.hayward@arm.com>
7520 David Sherwood <david.sherwood@arm.com>
7522 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7523 calling tree_to_uhwi.
7525 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7526 Alan Hayward <alan.hayward@arm.com>
7527 David Sherwood <david.sherwood@arm.com>
7529 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7530 INTEGER_CST rather than a negative test for ADDR_EXPR.
7532 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7533 Alan Hayward <alan.hayward@arm.com>
7534 David Sherwood <david.sherwood@arm.com>
7536 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7537 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7539 2017-09-21 Richard Biener <rguenther@suse.de>
7541 PR tree-optimization/71351
7542 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7543 graphite_create_new_loop_guard): Remove, fold remaining parts
7545 (translate_isl_ast_node_for): ... here and simplify.
7547 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7550 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7551 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7552 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7553 alternative always use QI mode, for -Os imov (=R,R) alternative
7554 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7557 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7558 Jeff Law <law@redhat.com>
7560 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7561 (allocate_stack_space): New function, partially extracted from
7563 (s390_emit_prologue): Track offset to most recent stack probe.
7564 Code to allocate space moved into allocate_stack_space.
7565 Dump actions when no stack is allocated.
7566 (s390_prologue_plus_offset): New function.
7567 (s390_emit_stack_probe): Likewise.
7569 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7571 * common.opt (Wa, Wl, Wp, g, gz=): Add
7573 (gno-column-info): Remove.
7574 (gcolumn-info): Drop RejectNegative.
7576 (gno-record-gcc-switches): Remove.
7577 (grecord-gcc-switches): Drop RejectNegative.
7578 (gno-split-dwarf): Remove.
7579 (gsplit-dwarf): Drop RejectNegative.
7580 (gno-strict-dwarf): Remove.
7581 (gstrict-dwarf): Drop RejectNegative.
7582 * config/darwin.opt (gfull, gused): Add RejectNegative.
7583 * dwarf2out.c (gen_producer_string): Drop
7584 gno-record-gcc-switches handler.
7585 * optc-gen.awk: Add g to prefixes with negative forms.
7586 * opts-common.c (remapping_prefix_p): New.
7587 (find_opt): Check it.
7588 (generate_canonical_option): Test g prefix.
7589 (option_map): Add -gno- mapping.
7590 (add_misspelling_candidates): Check remapping_prefix_p.
7592 2017-09-20 Jeff Law <law@redhat.com>
7594 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7595 thinko in stack clash protection support.
7597 * explow.c (compute_stack_clash_protection_loop_data): Use
7598 CONST_INT_P instead of explicit test. Verify object is a
7599 CONST_INT_P before looking at INTVAL.
7600 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7601 instead of explicit test.
7603 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7606 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7607 address instead of to r1 and r11.
7609 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7611 * config.gcc: Support "knm".
7612 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7613 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7615 * config/i386/i386.c (m_KNM): Define.
7616 (processor_target_table): Add "knm".
7618 (ix86_option_override_internal): Add "knm".
7619 (ix86_issue_rate): Add PROCESSOR_KNM.
7620 (ix86_adjust_cost): Ditto.
7621 (ia32_multipass_dfa_lookahead): Ditto.
7622 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7623 (fold_builtin_cpu): Add M_INTEL_KNM.
7624 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7625 (processor_type): Add PROCESSOR_KNM.
7626 * config/i386/x86-tune.def: Add m_KNM.
7627 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7629 2017-09-20 Richard Biener <rguenther@suse.de>
7631 PR tree-optimization/80213
7632 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7633 are allowed in empty BBs as well.
7634 (canonicalize_loop_closed_ssa): Also look for other complex
7636 (scop_detection::get_sese): Include the loop-closed PHI block
7638 (scop_detection::merge_sese): Remove code adding extra blocks.
7639 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7640 (build_scops): Assert the final returned scop is invalid.
7642 2017-09-20 Richard Biener <rguenther@suse.de>
7644 PR tree-optimization/82264
7645 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7647 (vn_phi_lookup): Likewise.
7648 (vn_phi_insert): Likewise.
7650 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7652 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7653 that fits into uhwi or shwi, add DW_AT_const_value regardless
7654 of early_dwarf without going through RTL, using add_AT_unsigned
7657 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7658 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7659 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7660 (DEBUG_MACRO_SECTION): Likewise.
7661 (DEBUG_LTO_MACRO_SECTION): Likewise.
7662 (DEBUG_STR_DWO_SECTION): Likewise.
7663 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7664 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7665 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7666 (DEBUG_LTO_LINE_STR_SECTION): Define.
7667 (init_sections_and_labels): Initialize debug_line_str_section
7668 variable. Initialize debug_loc_section for -gdwarf-5 to
7669 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7671 2017-09-20 Richard Biener <rguenther@suse.de>
7673 * graphite-sese-to-poly.c (extract_affine): Properly handle
7674 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7676 2017-09-20 Richard Biener <rguenther@suse.de>
7678 PR tree-optimization/81373
7679 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7680 Force SESE live-out defs to be handled even if they are
7683 2017-09-19 Jeff Law <law@redhat.com>
7685 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7686 nothing for stack adjustments with REG_STACK_CHECK.
7687 * sched-deps.c (parse_add_or_inc): Reject insns with
7688 REG_STACK_CHECK from dependency breaking.
7689 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7690 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7691 * reg-notes.def (STACK_CHECK): New note.
7693 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7694 (ix86_expand_prologue): Dump stack clash info as needed.
7695 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7697 * function.c (dump_stack_clash_frame_info): New function.
7698 * function.h (dump_stack_clash_frame_info): Prototype.
7699 (enum stack_clash_probes): New enum.
7701 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7702 flag_stack_clash_protection.
7703 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7704 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7705 (arm_frame_pointer_required): Likewise.
7706 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7707 (ia64_expand_prologue): Likewise.
7708 * config/mips/mips.c (mips_expand_prologue): Likewise.
7709 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7710 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7711 (sparc_flat_expand_prologue): Likewise.
7712 * config/spu/spu.c (spu_expand_prologue): Likewise.
7714 * explow.c: Include "params.h".
7715 (anti_adjust_stack_and_probe_stack_clash): New function.
7716 (get_stack_check_protect): Likewise.
7717 (compute_stack_clash_protection_loop_data): Likewise.
7718 (emit_stack_clash_protection_loop_start): Likewise.
7719 (emit_stack_clash_protection_loop_end): Likewise.
7720 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7721 Use anti_adjust_stack_and_probe_stack_clash.
7722 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7723 (emit_stack_clash_protection_loop_start): Likewise.
7724 (emit_stack_clash_protection_loop_end): Likewise.
7725 * rtl.h (get_stack_check_protect): Prototype.
7726 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7727 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7728 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7730 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7732 * doc/tm.texi: Rebuilt.
7733 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7734 get_stack_check_protect.
7735 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7736 * config/arm/arm.c (arm_expand_prologue): Likewise.
7737 (arm_frame_pointer_required): Likewise.
7738 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7739 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7740 * config/mips/mips.c (mips_expand_prologue): Likewise.
7741 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7742 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7743 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7744 (sparc_flat_expand_prologue): Likewise.
7746 * common.opt (-fstack-clash-protection): New option.
7747 * flag-types.h (enum stack_check_type): Note difference between
7748 -fstack-check= and -fstack-clash-protection.
7749 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7750 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7751 * toplev.c (process_options): Issue warnings/errors for cases
7752 not handled with -fstack-clash-protection.
7753 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7754 (-fstack-check): Note additional problem with -fstack-check=generic.
7755 Note that -fstack-check is primarily for Ada and refer users
7756 to -fstack-clash-protection for stack-clash-protection.
7757 Document new params for stack clash protection.
7759 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7761 * config/i386/i386.c (ix86_split_long_move): Do not handle
7762 address used for LEA in a special way.
7764 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7766 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7768 2017-09-19 Martin Sebor <msebor@redhat.com>
7771 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7772 of incompatible types.
7774 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7776 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7777 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7778 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7779 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7781 2017-09-19 Richard Biener <rguenther@suse.de>
7783 PR tree-optimization/82244
7784 * tree-vrp.c (remove_range_assertions): Do not propagate
7785 a constant to abnormals but replace the assert with a copy.
7787 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7789 PR rtl-optimization/57878
7790 PR rtl-optimization/68988
7791 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7792 avoidance test involving non_reload_pseudos. Move frequency test
7793 below the general fragmentation avoidance test.
7795 2017-09-19 Richard Biener <rguenther@suse.de>
7797 PR tree-optimization/69728
7798 * graphite-sese-to-poly.c (schedule_error): New global.
7799 (add_loop_schedule): Handle empty domain by failing the
7801 (build_original_schedule): Handle schedule_error.
7803 2017-09-19 Richard Biener <rguenther@suse.de>
7805 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7806 Do not iterate to sibling loops but only to siblings of inner
7809 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7812 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7813 that the registers are different.
7815 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7817 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7818 to processor_model and "amdfam17h" to arch_names_table.
7819 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7821 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7824 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7827 Alan Hayward <alan.hayward@arm.com>
7828 David Sherwood <david.sherwood@arm.com>
7830 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7832 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7834 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7835 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7836 vect_schedule_slp_instance.
7837 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7838 Update call to vect_slp_analyze_node_operations. Simplify return
7840 (vect_slp_analyze_bb_1): Update call accordingly.
7841 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7842 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7843 (vect_schedule_slp): Update call accordingly.
7845 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7846 Alan Hayward <alan.hayward@arm.com>
7847 David Sherwood <david.sherwood@arm.com>
7849 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7850 with types that aren't in fact scalar.
7852 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7854 * tree-vect-slp.c (vect_record_max_nunits): New function,
7856 (vect_build_slp_tree_1): ...here.
7857 (vect_build_slp_tree_2): Call it for phis too.
7859 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7861 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7862 to vect_get_vec_def_for_operand when getting the mask operand.
7864 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7865 Alan Hayward <alan.hayward@arm.com>
7866 David Sherwood <david.sherwood@arm.com>
7868 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7871 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7872 Alan Hayward <alan.hayward@arm.com>
7873 David Sherwood <david.sherwood@arm.com>
7875 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7876 calculation for vector booleans.
7878 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7879 Alan Hayward <alan.hayward@arm.com>
7880 David Sherwood <david.sherwood@arm.com>
7882 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7884 (vect_transform_stmt): ...here.
7885 (vect_analyze_stmt): Use it instead of calling
7886 vectorizable_live_operation directly.
7888 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7890 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7891 non-SIMT targets in acc vector loops.
7893 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7895 * configure.ac: Add arc and check if assembler supports gdwarf2.
7896 * configure: Regenerate.
7898 2017-09-18 Richard Biener <rguenther@suse.de>
7900 PR tree-optimization/82220
7901 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7902 epilogue niters from the min_profitable_iters compute.
7904 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7907 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7908 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7909 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7911 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7914 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7915 switching to a new text section.
7917 2017-09-18 Richard Biener <rguenther@suse.de>
7919 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7921 (build_alias_set): Reject aliases with no access function.
7923 2017-09-18 Richard Biener <rguenther@suse.de>
7925 PR tree-optimization/79622
7926 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7928 (build_cross_bb_scalars_use): Likewise.
7930 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7932 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7934 2017-09-18 Alan Modra <amodra@gmail.com>
7937 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7938 stack_pointer_rtx for count 0. Update comments. Break up
7939 large rtl expression.
7941 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7943 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7944 Increase to 20 bytes.
7945 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7946 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7947 or avx version of the stub.
7949 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7952 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7953 compute the minimum stack alignment. Also update preferred stack
7954 boundary for leaf functions.
7956 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7958 PR tree-optimization/82228
7959 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7962 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7964 * common/config/nds32/nds32-common.c
7965 (nds32_option_optimization_table): Refine formatting.
7966 (nds32_option_optimization_table): Use -fsched-pressure and
7967 -fomit-frame-pointer for specific optimization level.
7969 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7971 * config/nds32/nds32.c: Refine formatting and comments.
7972 * config/nds32/nds32.h: Likewise.
7973 * config/nds32/nds32.md: Likewise.
7974 * config/nds32/nds32-cost.c: Likewise.
7975 * config/nds32/nds32-isr.c: Likewise.
7976 * config/nds32/nds32-md-auxiliary.c: Likewise.
7977 * config/nds32/nds32-multiple.md: Likewise.
7978 * config/nds32/nds32-predicates.c: Likewise.
7980 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7981 Jakub Jelinek <jakub@redhat.com>
7983 Add support for -std=c++2a.
7984 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7986 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7988 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7991 * doc/extend.texi (Common Function Attributes): Add
7992 references to ARM, AArch64, and S/390 specific attributes.
7993 (Function Specific Option Pragmas): Add AArch64 and S/390
7994 to list of back ends that support the target pragma.
7996 2017-09-15 Nathan Sidwell <nathan@acm.org>
7998 * doc/standards.texi: Fix C++17 description. Update URLs for
8001 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8003 * common.opt (Wcast-align=strict): New warning option.
8004 * doc/invoke.texi: Document -Wcast-align=strict.
8006 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8008 * cgraph.h (cgraph_thunk_info): Add comments.
8009 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8010 assert for VIRTUAL_* arguments stricter.
8012 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8014 PR tree-optimization/71026
8015 * match.pd: Move RDIV patterns from fold-const.c
8016 * fold-const.c (distribute_real_division): Removed.
8017 (fold_binary_loc): Remove calls to distribute_real_divison.
8019 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8021 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8022 c++1z and gnu++1z as deprecated. Change other references to
8023 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8024 Change -Wc++1z-compat to -Wc++17-compat.
8025 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8026 * dwarf2out.c (highest_c_language): Handle C++17.
8027 (gen_compile_unit_die): Likewise.
8029 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8031 PR rtl-optimization/82192
8032 * combine.c (make_extraction): Don't look through non-paradoxical
8033 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8036 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8037 Alan Hayward <alan.hayward@arm.com>
8038 David Sherwood <david.sherwood@arm.com>
8040 * target.def (function_arg_offset): New hook.
8041 * targhooks.h (default_function_arg_offset): Declare.
8042 * targhooks.c (default_function_arg_offset): New function.
8043 * function.c (locate_and_pad_parm): Use
8044 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8045 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8046 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8047 * doc/tm.texi: Regenerate.
8048 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8049 * config/spu/spu.c (spu_function_arg_offset): New function.
8050 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8051 * system.h (FUNCTION_ARG_OFFSET): Poison.
8053 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8054 Alan Hayard <alan.hayward@arm.com>
8055 David Sherwood <david.sherwood@arm.com>
8057 * target.def (truly_noop_truncation): New hook.
8058 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8059 than TRULY_NOOP_TRUNCATION.
8060 * hooks.h (hook_bool_uint_uint_true): Declare.
8061 * hooks.c (hook_bool_uint_uint_true): New function.
8062 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8063 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8064 * doc/tm.texi: Regenerate.
8065 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8066 rather than TRULY_NOOP_TRUNCATION in comments.
8067 (simplify_comparison): Likewise.
8068 (record_truncated_value): Likewise.
8069 * expmed.c (extract_bit_field_1): Likewise.
8070 (extract_split_bit_field): Likewise.
8071 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8072 instead of TRULY_NOOP_TRUNCATION.
8073 * function.c (assign_parm_setup_block): Likewise.
8074 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8075 * rtlhooks.c: Include target.h.
8076 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8077 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8078 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8079 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8080 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8081 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8082 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8083 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8084 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8085 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8086 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8087 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8088 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8089 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8090 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8091 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8092 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8093 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8094 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8095 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8096 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8097 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8098 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8099 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8100 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8101 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8102 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8103 rather than TRULY_NOOP_TRUNCATION in comments.
8104 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8105 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8106 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8107 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8108 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8109 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8110 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8111 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8112 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8113 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8114 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8115 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8116 rather than TRULY_NOOP_TRUNCATION in comments.
8117 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8118 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8119 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8120 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8121 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8122 TRULY_NOOP_TRUNCATION condition.
8123 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8124 (TRULY_NOOP_TRUNCATION): Delete.
8125 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8126 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8127 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8128 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8129 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8130 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8131 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8132 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8133 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8134 rather than TRULY_NOOP_TRUNCATION in comments.
8135 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8136 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8137 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8138 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8139 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8140 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8142 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8145 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8146 (*cmp_ior): Likewise.
8147 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8148 (*ior_scc_scc_cmp): Likewise.
8149 (*and_scc_scc): Likewise.
8150 (*and_scc_scc_cmp): Likewise.
8152 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8153 Alan Hayard <alan.hayward@arm.com>
8154 David Sherwood <david.sherwood@arm.com>
8156 * target.def (can_change_mode_class): New hook.
8157 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8158 (hard_regno_nregs): Likewise.
8159 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8160 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8161 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8162 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8163 (LOAD_EXTEND_OP): Update accordingly.
8164 * doc/tm.texi: Regenerate.
8165 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8166 CANNOT_CHANGE_MODE_CLASS.
8167 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8168 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8169 * combine.c (simplify_set): Update accordingly.
8170 * emit-rtl.c (validate_subreg): Likewise.
8171 * recog.c (general_operand): Likewise.
8172 * regcprop.c (mode_change_ok): Likewise.
8173 * reload1.c (choose_reload_regs): Likewise.
8174 (inherit_piecemeal_p): Likewise.
8175 * rtlanal.c (simplify_subreg_regno): Likewise.
8176 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8177 instead of CANNOT_CHANGE_MODE_CLASS.
8178 (reload_cse_simplify_operands): Likewise.
8179 * reload.c (push_reload): Use targetm.can_change_mode_class
8180 instead of CANNOT_CHANGE_MODE_CLASS.
8181 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8182 REG_CANNOT_CHANGE_MODE_P.
8183 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8184 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8185 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8186 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8187 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8188 (arm_can_change_mode_class): New function.
8189 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8190 than CANNOT_CHANGE_MODE_CLASS in comments.
8191 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8192 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8193 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8194 (ix86_can_change_mode_class): ...this new function, inverting the
8195 sense of the return value.
8196 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8197 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8198 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8199 (ia64_can_change_mode_class): New function.
8200 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8201 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8202 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8203 (m32c_can_change_mode_class): ...this new function, inverting the
8204 sense of the return value.
8205 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8206 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8207 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8208 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8209 (mips_can_change_mode_class): ...this new function, inverting the
8210 sense of the return value.
8211 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8212 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8213 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8214 (msp430_can_change_mode_class): New function.
8215 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8216 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8217 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8218 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8219 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8220 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8221 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8222 (pa_cannot_change_mode_class): Replace with...
8223 (pa_can_change_mode_class): ...this new function, inverting the
8224 sense of the return value.
8225 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8226 than CANNOT_CHANGE_MODE_CLASS in comments.
8227 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8228 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8229 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8230 (pdp11_cannot_change_mode_class): Replace with...
8231 (pdp11_can_change_mode_class): ...this new function, inverting the
8232 sense of the return value.
8233 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8234 * config/powerpcspe/powerpcspe-protos.h
8235 (rs6000_cannot_change_mode_class_ptr): Delete.
8236 * config/powerpcspe/powerpcspe.c
8237 (rs6000_cannot_change_mode_class_ptr): Delete.
8238 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8239 (rs6000_option_override_internal): Assign to
8240 targetm.can_change_mode_class instead of
8241 rs6000_cannot_change_mode_class_ptr.
8242 (rs6000_cannot_change_mode_class): Replace with...
8243 (rs6000_can_change_mode_class): ...this new function, inverting the
8244 sense of the return value.
8245 (rs6000_debug_cannot_change_mode_class): Replace with...
8246 (rs6000_debug_can_change_mode_class): ...this new function.
8247 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8248 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8249 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8250 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8251 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8253 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8254 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8255 (rs6000_option_override_internal): Assign to
8256 targetm.can_change_mode_class instead of
8257 rs6000_cannot_change_mode_class_ptr.
8258 (rs6000_cannot_change_mode_class): Replace with...
8259 (rs6000_can_change_mode_class): ...this new function, inverting the
8260 sense of the return value.
8261 (rs6000_debug_cannot_change_mode_class): Replace with...
8262 (rs6000_debug_can_change_mode_class): ...this new function.
8263 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8264 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8265 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8266 (s390_can_change_mode_class): ...this new function, inverting the
8267 sense of the return value.
8268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8269 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8270 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8271 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8272 (sh_cannot_change_mode_class): Replace with...
8273 (sh_can_change_mode_class): ...this new function, inverting the
8274 sense of the return value.
8275 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8276 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8277 (sparc_can_change_mode_class): New function.
8278 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8279 * config/spu/spu.c (spu_can_change_mode_class): New function.
8280 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8281 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8282 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8283 (visium_can_change_mode_class): New function.
8284 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8286 2017-09-15 Richard Biener <rguenther@suse.de>
8288 PR tree-optimization/82217
8289 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8290 but not undefined case.
8292 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8295 * postreload.c (reload_cse_simplify_operands): Skip
8296 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8298 2017-09-15 Richard Biener <rguenther@suse.de>
8300 PR tree-optimization/68823
8301 * graphite-scop-detection.c (build_alias_set): If we have a
8302 possible dependence check whether we can handle them by just
8303 looking at the DRs DR_ACCESS_FNs.
8304 (build_scops): If build_alias_set fails, fail the SCOP.
8306 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8308 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8309 to support float128 built-in functions that require the ISA 3.0
8311 (BU_FLOAT128_3_HW): Likewise.
8312 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8314 (FMAF128): Likewise.
8316 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8317 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8318 floating point instructions.
8319 (rs6000_invalid_builtin): Likewise.
8320 (rs6000_builtin_mask_names): Likewise.
8321 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8322 (RS6000_BTM_FLOAT128_HW): Likewise.
8323 (RS6000_BTM_COMMON): Likewise.
8324 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8326 * doc/extend.texi (RS/6000 built-in functions): Document the
8327 IEEE 128-bit floating point square root and fused multiply-add
8330 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8332 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8333 reg (r2) isn't in the set of registers defined in the prologue.
8335 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8336 Alan Hayward <alan.hayward@arm.com>
8337 David Sherwood <david.sherwood@arm.com>
8339 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8340 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8341 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8342 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8343 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8345 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8346 max_vectorization_factor.
8347 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8349 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8350 Alan Hayward <alan.hayward@arm.com>
8351 David Sherwood <david.sherwood@arm.com>
8353 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8354 (vect_worthwhile_without_simd_p): Declare.
8355 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8356 (vectorizable_reduction): Use it.
8357 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8358 (vectorizable_operation): Likewise.
8360 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8361 Alan Hayward <alan.hayward@arm.com>
8362 David Sherwood <david.sherwood@arm.com>
8364 * tree-vectorizer.h (vect_get_num_copies): New function.
8365 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8366 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8367 (vectorizable_induction): Likewise.
8368 (vectorizable_live_operation): Likewise.
8369 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8370 (vectorizable_bswap): Likewise.
8371 (vectorizable_call): Likewise.
8372 (vectorizable_conversion): Likewise.
8373 (vectorizable_assignment): Likewise.
8374 (vectorizable_shift): Likewise.
8375 (vectorizable_operation): Likewise.
8376 (vectorizable_store): Likewise.
8377 (vectorizable_load): Likewise.
8378 (vectorizable_condition): Likewise.
8379 (vectorizable_comparison): Likewise.
8380 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8382 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8383 Alan Hayward <alan.hayward@arm.com>
8384 David Sherwood <david.sherwood@arm.com>
8386 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8387 of vect_init_vector.
8389 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8390 Alan Hayward <alan.hayward@arm.com>
8391 David Sherwood <david.sherwood@arm.com>
8393 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8394 an inline wrapper that provides a location.
8395 (gimple_build_vector): Likewise.
8396 * gimple-fold.c (gimple_build_vector_from_val): New function.
8397 (gimple_build_vector): Likewise.
8398 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8399 functions to build the initial value. Always return a gimple value.
8400 (get_initial_defs_for_reduction): Likewise. Only compute
8402 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8403 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8404 (vectorizable_induction): Use gimple_build_vector rather than
8407 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8408 Alan Hayward <alan.hayward@arm.com>
8409 David Sherwood <david.sherwood@arm.com>
8411 * target.h (vec_perm_indices): New typedef.
8412 (auto_vec_perm_indices): Likewise.
8413 * optabs-query.h: Include target.h
8414 (can_vec_perm_p): Take a vec_perm_indices *.
8415 * optabs-query.c (can_vec_perm_p): Likewise.
8416 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8417 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8418 * tree-vect-generic.c (lower_vec_perm): Likewise.
8419 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8420 (vect_grouped_load_supported): Likewise.
8421 (vect_shift_permute_load_chain): Likewise.
8422 (vect_permute_store_chain): Use auto_vec_perm_indices.
8423 (vect_permute_load_chain): Likewise.
8424 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8425 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8426 Update uses of can_vec_perm_p.
8427 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8428 mode with a number of elements. Take a vec_perm_indices *.
8429 (vect_create_epilog_for_reduction): Update accordingly.
8430 Use auto_vec_perm_indices.
8431 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8432 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8433 (vect_transform_slp_perm_load): Likewise.
8434 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8435 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8436 (vect_gen_perm_mask_checked): Likewise.
8437 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8438 (vect_gen_perm_mask_checked): Likewise.
8439 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8440 (vectorizable_store): Likewise.
8441 (vectorizable_load): Likewise.
8442 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8443 (vectorizable_bswap): Likewise.
8445 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8446 Alan Hayward <alan.hayward@arm.com>
8447 David Sherwood <david.sherwood@arm.com>
8449 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8450 * tree.c (build_vector): Likewise.
8451 (build_vector_from_ctor): Update accordingly.
8452 (build_vector_from_val): Likewise.
8453 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8454 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8455 * tree-vect-generic.c (add_rshift): Likewise.
8456 (expand_vector_divmod): Likewise.
8457 (optimize_vector_constructor): Likewise.
8458 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8459 (vect_transform_slp_perm_load): Likewise.
8460 (vect_schedule_slp_instance): Likewise.
8461 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8462 (vectorizable_call): Likewise.
8463 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8464 * expmed.c (make_tree): Likewise.
8465 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8466 a vector passed to build_vector.
8467 (fold_convert_const): Likewise.
8468 (exact_inverse): Likewise.
8469 (fold_ternary_loc): Likewise.
8470 (fold_relational_const): Likewise.
8471 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8472 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8473 (const_unop): Likewise. Store the reduction accumulator in a
8474 variable rather than an array.
8475 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8476 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8477 the new vector, rather than constructing it after the input arrays.
8478 (native_interpret_vector): Use auto_vec<tree> when building
8479 a vector passed to build_vector. Add elements in order.
8480 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8481 auto_vec<tree> when building a vector passed to build_vector.
8482 (vect_create_epilog_for_reduction): Likewise.
8483 (vectorizable_induction): Likewise.
8484 (get_initial_def_for_reduction): Likewise. Fix indentation of
8486 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8488 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8489 passed to build_vector.
8491 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8492 Alan Hayward <alan.hayward@arm.com>
8493 David Sherwood <david.sherwood@arm.com>
8495 * tree-core.h (tree_base::u): Add an "nelts" field.
8496 (tree_vector): Use VECTOR_CST_NELTS as the length.
8497 * tree.c (tree_size): Likewise.
8498 (make_vector): Initialize VECTOR_CST_NELTS.
8499 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8500 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8501 TYPE_VECTOR_SUBPARTS.
8502 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8503 as the number of units, setting it from VECTOR_CST_NELTS.
8504 (const_vector_from_tree): Likewise.
8505 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8506 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8507 (fold_negate_expr_1): Likewise.
8508 (fold_convert_const): Likewise.
8509 (const_binop): Likewise. Differentiate the number of output and
8511 (const_unop): Likewise.
8512 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8513 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8514 in cases that did the opposite.
8516 2017-09-14 Richard Biener <rguenther@suse.de>
8518 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8521 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8523 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8525 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8528 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8529 if and where to split a bb, except for splitting before debug insn
8530 sequences followed by non-label real insn. Delete debug insns
8531 in between basic blocks.
8533 * combine.c (make_compound_operation_int): Formatting fixes.
8535 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8536 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8537 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8538 * config/sol2.h (LINK_EH_SPEC): Likewise.
8539 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8540 * config/s390/linux.h (LINK_SPEC): Likewise.
8541 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8542 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8543 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8544 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8545 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8546 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8547 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8548 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8550 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8552 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8554 (ENDFILE_LINUX_SPEC): Likewise.
8555 (LINK_EH_SPEC): Likewise.
8556 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8557 (LINK_OS_LINUX_SPEC32): Likewise.
8558 (LINK_OS_LINUX_SPEC64): Likewise.
8559 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8560 (LINK_OS_LINUX_SPEC): Likewise.
8562 2017-09-13 Martin Liska <mliska@suse.cz>
8565 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8566 CASE_HIGH is NULL_TREE.
8568 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * target.def (secondary_memory_needed): New hook.
8573 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8574 instead of SECONDARY_MEMORY_NEEDED.
8575 (secondary_memory_needed_mode): Likewise.
8576 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8577 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8578 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8579 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8580 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8581 * doc/tm.texi: Regenerate.
8582 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8583 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8584 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8585 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8586 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8587 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8588 mode argument first and change the reg_class arguments to reg_class_t.
8589 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8590 Make static. Update the call to inline_secondary_memory_needed.
8591 (ix86_register_move_cost): Update the call to
8592 inline_secondary_memory_needed.
8593 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8594 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8596 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8597 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8599 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8600 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8601 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8602 and match hook interface. Add comment from mips.h.
8603 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8604 * config/mmix/mmix.md (truncdfsf2): Refer to
8605 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8607 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8608 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8609 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8610 (pa_secondary_memory_needed): New function.
8611 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8612 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8613 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8614 (pdp11_secondary_memory_needed): Make static and match hook interface.
8615 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8616 * config/powerpcspe/powerpcspe-protos.h
8617 (rs6000_secondary_memory_needed_ptr): Delete.
8618 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8620 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8621 (rs6000_option_override_internal): Assign to
8622 targetm.secondary_memory_needed rather than
8623 rs6000_secondary_memory_needed_ptr.
8624 (rs6000_secondary_memory_needed): Match hook interface.
8625 (rs6000_debug_secondary_memory_needed): Likewise.
8626 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8627 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8628 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8629 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8630 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8631 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8633 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8634 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8635 (rs6000_option_override_internal): Assign to
8636 targetm.secondary_memory_needed rather than
8637 rs6000_secondary_memory_needed_ptr.
8638 (rs6000_secondary_memory_needed): Match hook interface.
8639 (rs6000_debug_secondary_memory_needed): Likewise.
8640 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8641 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8642 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8643 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8644 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8645 (sparc_secondary_memory_needed): New function.
8646 * lra-constraints.c (check_and_process_move): Refer to
8647 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8649 (curr_insn_transform): Likewise.
8650 (process_alt_operands): Use targetm.secondary_memory_needed
8651 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8652 (check_secondary_memory_needed_p): Likewise.
8653 (choose_split_class): Likewise.
8654 * reload.c: Unconditionally include code that was previously
8655 conditional on SECONDARY_MEMORY_NEEDED.
8656 (push_secondary_reload): Use targetm.secondary_memory_needed
8657 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8658 (push_reload): Likewise.
8659 * reload1.c: Unconditionally include code that was previously
8660 conditional on SECONDARY_MEMORY_NEEDED.
8661 (choose_reload_regs): Use targetm.secondary_memory_needed
8662 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8663 (gen_reload): Likewise.
8664 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8666 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8667 Alan Hayward <alan.hayward@arm.com>
8668 David Sherwood <david.sherwood@arm.com>
8670 * target.def (secondary_memory_needed_mode): New hook:
8671 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8672 * targhooks.h (default_secondary_memory_needed_mode): New function.
8673 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8674 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8675 * doc/tm.texi: Regenerate.
8676 * lra-constraints.c (check_and_process_move): Use
8677 targetm.secondary_memory_needed_mode instead of
8678 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8679 (curr_insn_transform): Likewise.
8680 * reload.c (get_secondary_mem): Likewise.
8681 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8682 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8684 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8685 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8686 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8687 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8688 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8690 * config/powerpcspe/powerpcspe-protos.h
8691 (rs6000_secondary_memory_needed_mode): Delete.
8692 * config/powerpcspe/powerpcspe.c
8693 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8694 (rs6000_secondary_memory_needed_mode): Make static.
8695 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8696 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8698 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8700 (rs6000_secondary_memory_needed_mode): Make static.
8701 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8702 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8703 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8704 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8705 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8707 (sparc_secondary_memory_needed_mode): New function.
8708 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8710 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8712 * config/aarch64/constraints.md (Umq): New constraint.
8713 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8715 (mov<mode>): Update condition.
8717 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8719 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8720 when bitposition is the same.
8722 2017-09-13 Richard Biener <rguenther@suse.de>
8724 * dwarf2out.c (output_die_symbol): Remove.
8725 (output_die): Do not output a DIEs symbol.
8727 2017-09-13 Richard Biener <rguenther@suse.de>
8730 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8731 default-def to avoid breaking iterator update with the weird
8732 interaction with cgraph_update_edges_for_call_stmt_node.
8734 2017-09-13 Richard Biener <rguenther@suse.de>
8736 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8737 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8738 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8739 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8741 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8743 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8744 Disable pc relative literal load irrespective of
8745 TARGET_FIX_ERR_A53_84341 for default.
8747 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8749 * config/sparc/sparc.c (output_return): Output the source location of
8750 the insn in the delay slot, if any.
8751 (output_sibcall): Likewise.
8753 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8756 * common.opt (-static-pie): New alias.
8757 (shared): Negate static-pie.
8758 (-no-pie): Update help text.
8760 (static-pie): New option.
8761 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8762 -static-pie support.
8763 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8764 (LINK_EH_SPEC): Likewise.
8765 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8766 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8767 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8768 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8769 (init_gcc_specs): Likewise.
8770 (init_spec): Likewise.
8771 (display_help): Update help message for -pie.
8772 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8775 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8777 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8778 (movdi_aarch64): Likewise.
8779 (movti_aarch64): Likewise.
8781 2017-09-12 Simon Wright <simon@pushface.org>
8784 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8785 calculation of the minor version, always output as 0.
8787 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8790 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8791 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8792 on it early, rather than manual conversion late. For
8793 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8794 instead of performing manual conversion.
8796 2017-09-12 Carl Love <cel@us.ibm.com>
8798 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8799 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8801 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8802 VMULOSW): Add definitions.
8803 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8804 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8805 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8807 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8809 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8811 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8813 (movti_aarch64): Likewise.
8814 (movdf_aarch64): Likewise.
8815 (movtf_aarch64): Likewise.
8816 (load_pairdi): Likewise.
8817 (store_pairdi): Likewise.
8818 (load_pairdf): Likewise.
8819 (store_pairdf): Likewise.
8820 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8821 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8822 (ldr_got_small_<mode>): Likewise.
8823 (ldr_got_small_28k_<mode>): Likewise.
8824 (ldr_got_tiny): Likewise.
8825 * config/aarch64/iterators.md (ldst_sz): New.
8826 (ldpstp_sz): Likewise.
8827 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8829 (thunderx_load): Split load_8 to load_16.
8830 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8832 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8833 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8834 (xgene1_store_pair): Split store_8 to store_16.
8835 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8836 (falkor_st_0_st_sd): Split store_8 to store_16.
8838 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8840 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8841 and store1/2/3/4 to store_4/8/12/16.
8842 * config/aarch64/aarch64.md: Update for rename.
8843 * config/arm/arm.md: Likewise.: Likewise.
8844 * config/arm/arm.c: Likewise.
8845 * config/arm/thumb1.md: Likewise.
8846 * config/arm/thumb2.md: Likewise.
8847 * config/arm/vfp.md: Likewise.
8848 * config/arm/arm-generic.md: Likewise.
8849 * config/arm/arm1020e.md: Likewise.
8850 * config/arm/arm1026ejs.md: Likewise.
8851 * config/arm/arm1136jfs.md: Likewise.
8852 * config/arm/arm926ejs.md: Likewise.
8853 * config/arm/cortex-a15.md: Likewise.
8854 * config/arm/cortex-a17.md: Likewise.
8855 * config/arm/cortex-a5.md: Likewise.
8856 * config/arm/cortex-a53.md: Likewise.
8857 * config/arm/cortex-a57.md: Likewise.
8858 * config/arm/cortex-a7.md: Likewise.
8859 * config/arm/cortex-a8.md: Likewise.
8860 * config/arm/cortex-a9.md: Likewise.
8861 * config/arm/cortex-m4.md: Likewise.
8862 * config/arm/cortex-m7.md: Likewise.
8863 * config/arm/cortex-r4.md: Likewise.
8864 * config/arm/exynos-m1.md: Likewise.
8865 * config/arm/fa526.md: Likewise.
8866 * config/arm/fa606te.md: Likewise.
8867 * config/arm/fa626te.md: Likewise.
8868 * config/arm/fa726te.md: Likewise.
8869 * config/arm/fmp626.md: Likewise.
8870 * config/arm/iwmmxt.md: Likewise.
8871 * config/arm/ldmstm.md: Likewise.
8872 * config/arm/marvell-pj4.md: Likewise.
8873 * config/arm/xgene1.md: Likewise.
8874 * config/aarch64/thunderx.md: Likewise.
8875 * config/aarch64/thunderx2t99.md: Likewise.
8876 * config/aarch64/falkor.md: Likewise.
8878 2017-09-12 Martin Liska <mliska@suse.cz>
8880 * attribs.c (private_lookup_attribute): New function.
8881 * attribs.h (private_lookup_attribute): Declared here.
8882 (lookup_attribute): Called from this place.
8884 2017-09-12 Richard Biener <rguenther@suse.de>
8886 PR tree-optimization/82157
8887 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8888 stmts with side-effects.
8890 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8891 Alan Hayward <alan.hayward@arm.com>
8892 David Sherwood <david.sherwood@arm.com>
8894 * target.def (hard_regno_nregs): New hook.
8895 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8896 * targhooks.h (default_hard_regno_nregs): Declare.
8897 * targhooks.c (default_hard_regno_nregs): New function.
8898 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8899 (TARGET_HARD_REGNO_NREGS): ...this hook.
8900 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8901 (CLASS_MAX_NREGS): Likewise.
8902 * doc/tm.texi: Regenerate.
8903 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8904 instead of HARD_REGNO_NREGS.
8905 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8906 HARD_REGNO_NREGS in the comment.
8907 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8908 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8909 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8910 Return an unsigned int.
8911 (TARGET_HARD_REGNO_NREGS): Redefine.
8912 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8913 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8914 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8915 (arc_hard_regno_nregs): New function.
8916 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8917 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8918 (arm_hard_regno_nregs): New function.
8919 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8920 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8921 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8922 (TARGET_HARD_REGNO_NREGS): Redefine.
8923 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8924 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8925 (HARD_REGNO_NREGS): Delete.
8926 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8927 (cr16_hard_regno_nregs): New function.
8928 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8929 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8930 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8931 (cris_hard_regno_nregs): New function.
8932 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8933 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8934 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8935 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8936 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8937 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8938 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8939 (frv_hard_regno_nregs): Make static. Take and return an
8941 (frv_class_max_nregs): Remove outdated copy of documentation.
8942 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8943 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8944 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8945 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8946 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8947 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8948 (TARGET_HARD_REGNO_NREGS): Redefine.
8949 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8950 (CLASS_MAX_NREGS): Update comment.
8951 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8952 (ia64_hard_regno_nregs): New function.
8953 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8954 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8955 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8956 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8957 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8959 (m32c_hard_regno_nregs): Likewise. Make static.
8960 (TARGET_HARD_REGNO_NREGS): Redefine.
8961 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8962 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8963 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8964 (m68k_hard_regno_nregs): New function.
8965 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8966 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8967 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8968 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8969 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8970 Take and return an unsigned int.
8971 (TARGET_HARD_REGNO_NREGS): Redefine.
8972 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8973 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8974 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8975 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8976 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8977 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8978 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8979 (msp430_hard_regno_nregs): Make static. Take and return an
8981 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8982 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8983 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8984 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8985 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8986 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8987 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8988 (TARGET_HARD_REGNO_NREGS): Redefine.
8989 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8990 (PA_HARD_REGNO_NREGS): ...this.
8991 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8992 (PA_HARD_REGNO_NREGS): ...this.
8993 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8994 (pa_hard_regno_nregs): New function.
8995 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8996 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8997 (pdp11_hard_regno_nregs): New function.
8998 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8999 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9000 (rs6000_hard_regno_nregs_hook): New function.
9001 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9002 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9003 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9004 Take and return an unsigned int. Move earlier in file.
9005 (TARGET_HARD_REGNO_NREGS): Redefine.
9006 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9007 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9008 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9009 (rl78_hard_regno_nregs): Make static. Take and return an
9011 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9012 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9013 (rs6000_hard_regno_nregs_hook): New function.
9014 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9015 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9016 (TARGET_HARD_REGNO_NREGS): Redefine.
9017 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9018 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9019 instead of HARD_REGNO_NREGS.
9020 (s390_hard_regno_nregs): New function.
9021 (s390_hard_regno_mode_ok): Add comment from s390.h.
9022 (TARGET_HARD_REGNO_NREGS): Redefine.
9023 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9024 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9025 (sh_hard_regno_nregs): New function.
9026 (sh_pass_in_reg_p): Use it.
9027 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9028 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9029 (sparc_hard_regno_nregs): New function.
9030 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9031 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9032 (spu_function_arg_advance): Use it, supplying a valid register number.
9033 (TARGET_HARD_REGNO_NREGS): Redefine.
9034 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9035 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9036 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9037 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9038 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9039 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9040 (CLASS_MAX_NREGS): Remove copy of old documentation.
9041 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9042 (visium_hard_regno_nregs): New function.
9043 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9044 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9045 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9046 xtensa_hard_regno_nregs): New function.
9047 * system.h (HARD_REGNO_NREGS): Poison.
9049 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9051 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9052 hard_regno_nregs instead of HARD_REGNO_NREGS.
9053 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9054 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9055 (c6x_expand_epilogue): Likewise.
9056 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9057 (frv_read_iacc_argument): Likewise.
9058 * config/sh/sh.c: Include regs.h.
9059 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9060 (regs_used): Likewise.
9061 (output_stack_adjust): Likewise.
9062 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9063 * expmed.c: Include regs.h.
9064 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9065 * ree.c: Include regs.h.
9066 (combine_reaching_defs): Use hard_regno_nregs instead of
9068 (add_removable_extension): Likewise.
9070 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9072 * regs.h (hard_regno_nregs): Turn into a function.
9073 (end_hard_regno): Update accordingly.
9074 * caller-save.c (setup_save_areas): Likewise.
9075 (save_call_clobbered_regs): Likewise.
9076 (replace_reg_with_saved_mem): Likewise.
9077 (insert_restore): Likewise.
9078 (insert_save): Likewise.
9079 * combine.c (can_change_dest_mode): Likewise.
9080 (move_deaths): Likewise.
9081 (distribute_notes): Likewise.
9082 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9083 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9084 (rs6000_split_multireg_move): Likewise.
9085 (rs6000_register_move_cost): Likewise.
9086 (rs6000_memory_move_cost): Likewise.
9087 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9088 (rs6000_split_multireg_move): Likewise.
9089 (rs6000_register_move_cost): Likewise.
9090 (rs6000_memory_move_cost): Likewise.
9091 * cselib.c (cselib_reset_table): Likewise.
9092 (cselib_lookup_1): Likewise.
9093 * emit-rtl.c (set_mode_and_regno): Likewise.
9094 * function.c (aggregate_value_p): Likewise.
9095 * ira-color.c (setup_profitable_hard_regs): Likewise.
9096 (check_hard_reg_p): Likewise.
9097 (calculate_saved_nregs): Likewise.
9098 (assign_hard_reg): Likewise.
9099 (improve_allocation): Likewise.
9100 (calculate_spill_cost): Likewise.
9101 * ira-emit.c (modify_move_list): Likewise.
9102 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9103 (ira_hard_reg_in_set_p): Likewise.
9104 * ira.c (setup_reg_mode_hard_regset): Likewise.
9105 (clarify_prohibited_class_mode_regs): Likewise.
9106 (check_allocation): Likewise.
9107 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9108 (lra_setup_reg_renumber): Likewise.
9109 (setup_try_hard_regno_pseudos): Likewise.
9110 (spill_for): Likewise.
9111 (assign_hard_regno): Likewise.
9112 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9113 * lra-constraints.c (in_class_p): Likewise.
9114 (lra_constraint_offset): Likewise.
9115 (simplify_operand_subreg): Likewise.
9116 (lra_constraints): Likewise.
9117 (split_reg): Likewise.
9118 (split_if_necessary): Likewise.
9119 (invariant_p): Likewise.
9120 (inherit_in_ebb): Likewise.
9121 * lra-lives.c (process_bb_lives): Likewise.
9122 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9123 (get_hard_regs): Likewise.
9124 (do_remat): Likewise.
9125 * lra-spills.c (assign_spill_hard_regs): Likewise.
9126 * mode-switching.c (create_pre_exit): Likewise.
9127 * postreload.c (reload_combine_recognize_pattern): Likewise.
9128 * recog.c (peep2_find_free_register): Likewise.
9129 * regcprop.c (kill_value_regno): Likewise.
9130 (set_value_regno): Likewise.
9131 (copy_value): Likewise.
9132 (maybe_mode_change): Likewise.
9133 (find_oldest_value_reg): Likewise.
9134 (copyprop_hardreg_forward_1): Likewise.
9135 * regrename.c (check_new_reg_p): Likewise.
9136 (regrename_do_replace): Likewise.
9137 * reload.c (push_reload): Likewise.
9138 (combine_reloads): Likewise.
9139 (find_dummy_reload): Likewise.
9140 (operands_match_p): Likewise.
9141 (find_reloads): Likewise.
9142 (find_equiv_reg): Likewise.
9143 (reload_adjust_reg_for_mode): Likewise.
9144 * reload1.c (count_pseudo): Likewise.
9145 (count_spilled_pseudo): Likewise.
9146 (find_reg): Likewise.
9147 (clear_reload_reg_in_use): Likewise.
9148 (free_for_value_p): Likewise.
9149 (allocate_reload_reg): Likewise.
9150 (choose_reload_regs): Likewise.
9151 (reload_adjust_reg_for_temp): Likewise.
9152 (emit_reload_insns): Likewise.
9153 (delete_output_reload): Likewise.
9154 * rtlanal.c (subreg_get_info): Likewise.
9155 * sched-deps.c (sched_analyze_reg): Likewise.
9156 * sel-sched.c (init_regs_for_mode): Likewise.
9157 (mark_unavailable_hard_regs): Likewise.
9158 (choose_best_reg_1): Likewise.
9159 (verify_target_availability): Likewise.
9160 * valtrack.c (dead_debug_insert_temp): Likewise.
9161 * var-tracking.c (track_loc_p): Likewise.
9162 (emit_note_insn_var_location): Likewise.
9163 * varasm.c (make_decl_rtl): Likewise.
9164 * reginfo.c (choose_hard_reg_mode): Likewise.
9165 (init_reg_modes_target): Refer directly to
9166 this_target_regs->x_hard_regno_nregs.
9168 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9170 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9171 instead of hard_regno_nregs.
9173 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9175 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9176 end_hard_regno instead of hard_regno_nregs.
9177 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9178 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9179 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9180 * ira-color.c (improve_allocation): Likewise.
9181 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9182 * lra-lives.c (mark_regno_live): Likewise.
9183 (mark_regno_dead): Likewise.
9184 * lra-remat.c (operand_to_remat): Likewise.
9185 * lra.c (collect_non_operand_hard_regs): Likewise.
9186 * postreload.c (reload_combine_note_store): Likewise.
9187 (move2add_valid_value_p): Likewise.
9188 * reload.c (regno_clobbered_p): Likewise.
9190 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9192 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9194 * config/v850/v850.c (v850_reorg): Likewise.
9195 * reload.c (refers_to_regno_for_reload_p): Likewise.
9196 (find_equiv_reg): Likewise.
9197 * reload1.c (reload_reg_reaches_end_p): Likewise.
9199 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9201 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9203 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9204 * config/arm/arm.c (output_move_neon): Likewise.
9205 (arm_attr_length_move_neon): Likewise.
9206 (neon_split_vcombine): Likewise.
9207 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9208 (c6x_mark_reg_written): Likewise.
9209 (c6x_dwarf_register_span): Likewise.
9210 * config/i386/i386.c (ix86_save_reg): Likewise.
9211 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9212 (rws_access_reg): Likewise.
9213 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9214 * mode-switching.c (create_pre_exit): Likewise.
9215 * ree.c (combine_reaching_defs): Likewise.
9216 (add_removable_extension): Likewise.
9217 * regcprop.c (find_oldest_value_reg): Likewise.
9218 (copyprop_hardreg_forward_1): Likewise.
9219 * reload.c (reload_inner_reg_of_subreg): Likewise.
9220 (push_reload): Likewise.
9221 (combine_reloads): Likewise.
9222 (find_dummy_reload): Likewise.
9223 (reload_adjust_reg_for_mode): Likewise.
9224 * reload1.c (find_reload_regs): Likewise.
9225 (forget_old_reloads_1): Likewise.
9226 (reload_reg_free_for_value_p): Likewise.
9227 (reload_adjust_reg_for_temp): Likewise.
9228 (emit_reload_insns): Likewise.
9229 (delete_output_reload): Likewise.
9230 * sel-sched.c (choose_best_reg_1): Likewise.
9231 (choose_best_pseudo_reg): Likewise.
9233 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9234 Alan Hayward <alan.hayward@arm.com>
9235 David Sherwood <david.sherwood@arm.com>
9237 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9238 * target.def (slow_unaligned_access): New hook.
9239 * targhooks.h (default_slow_unaligned_access): Declare.
9240 * targhooks.c (default_slow_unaligned_access): New function.
9241 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9242 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9243 * doc/tm.texi: Regenerate.
9244 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9245 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9246 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9248 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9249 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9251 (rs6000_slow_unaligned_access): New function.
9252 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9253 (expand_block_compare): Likewise.
9254 (expand_strn_compare): Likewise.
9255 (rs6000_rtx_costs): Likewise.
9256 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9257 (riscv_slow_unaligned_access): Likewise.
9258 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9259 (riscv_slow_unaligned_access_p): ...this and make static.
9260 (riscv_option_override): Update accordingly.
9261 (riscv_slow_unaligned_access): New function.
9262 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9263 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9264 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9265 (rs6000_slow_unaligned_access): New function.
9266 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9267 (rs6000_rtx_costs): Likewise.
9268 * config/rs6000/rs6000-string.c (expand_block_compare)
9269 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9270 of SLOW_UNALIGNED_ACCESS.
9271 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9272 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9273 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9274 of SLOW_UNALIGNED_ACCESS.
9275 * expmed.c (simple_mem_bitfield_p): Likewise.
9276 * expr.c (alignment_for_piecewise_move): Likewise.
9277 (emit_group_load_1): Likewise.
9278 (emit_group_store): Likewise.
9279 (copy_blkmode_from_reg): Likewise.
9280 (emit_push_insn): Likewise.
9281 (expand_assignment): Likewise.
9282 (store_field): Likewise.
9283 (expand_expr_real_1): Likewise.
9284 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9285 * lra-constraints.c (simplify_operand_subreg): Likewise.
9286 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9287 * gimple-ssa-store-merging.c: Likewise in block comment at start
9289 * tree-ssa-strlen.c: Include target.h.
9290 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9291 of SLOW_UNALIGNED_ACCESS.
9292 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9294 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9296 PR rtl-optimization/82185
9297 * expmed.c (emit_store_flag_int): Only test tem if it has been
9300 2017-09-12 Richard Biener <rguenther@suse.de>
9303 * match.pd ((FTYPE) N CMP CST): Fix typo.
9305 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9307 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9309 (mips_near_type_p): Add 'short_call' attribute as a synonym
9311 * doc/extend.texi (short_call): Document new function attribute.
9313 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9316 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9317 assertion check that in the condition.
9318 (get_atomic_generic_size): Likewise. Before testing if parameter
9319 has pointer type, if it has array type, call for C++
9320 default_conversion to perform array-to-pointer conversion.
9322 2017-09-12 Richard Biener <rguenther@suse.de>
9324 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9325 for operations we cannot scalarize.
9327 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9329 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9330 vectors heap vectors. Clean up comments.
9331 Make visited_bbs a reference.
9332 (profitable_jump_thread_path): Make GC
9333 vectors heap vectors. Clean up comments.
9335 (convert_and_register_jump_thread_path): Make GC vectors heap
9337 (check_subpath_and_update_thread_path): Same. Clean up comments.
9338 Make visited_bbs a reference.
9339 (handle_phi): Abstract common code to to
9340 register_jump_thread_path_if_profitable.
9341 Rename VAR_BB to DEF_BB.
9343 Make GC vectors heap vectors.
9344 Make visited_bbs a reference.
9345 (handle_assignment): Same.
9346 (register_jump_thread_path_if_profitable): New.
9347 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9349 Make GC vectors heap vectors. Clean up comments.
9350 Make visited_bbs a reference.
9351 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9352 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9355 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9358 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9359 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9361 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9363 Revert r251800 and r251799.
9365 2017-09-11 Martin Jambor <mjambor@suse.cz>
9368 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9369 arguments in advance.
9370 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9371 use it to find predecessor edges.
9372 (naive_outof_ssa): Collect vector of predecessors.
9374 2017-09-08 Jason Merrill <jason@redhat.com>
9376 PR c++/70029 - ICE with ref-qualifier and -flto
9377 * langhooks.h (struct lang_hooks_for_types): Add
9378 copy_lang_qualifiers.
9379 * attribs.c (build_type_attribute_qual_variant): Use it.
9380 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9382 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9383 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9385 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9388 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9389 (*mulsi3_sp64): New instruction.
9390 (mulsi3): New expander.
9392 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9394 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9396 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9398 * sancov.c: Include memmodel.h.
9400 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9403 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9406 2017-09-07 Carl Love <cel@us.ibm.com>
9408 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9409 the sldi instruction.
9411 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9413 * sancov.c: Include tm_p.h.
9415 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9418 * output.h (switch_to_other_text_partition): New declaration.
9419 * varasm.c (switch_to_other_text_partition): New function.
9420 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9421 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9422 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9423 to the other text partition before emitting LCL label and switch back
9424 after emitting the word after it.
9426 2017-09-07 Richard Biener <rguenther@suse.de>
9428 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9429 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9430 critical edges here, after loop init.
9431 (pass_data_pre): Remove PROP_no_crit_edges flags.
9432 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9433 for valueization of call args to avoid leaking VN_TOP.
9434 (visit_use): Assert we do not visit default defs.
9435 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9436 Use error_mark_node to more easily detect leaking VN_TOP.
9437 All default-defs are varying, not VN_TOP. Mark them visited.
9438 (run_scc_vn): Make code match comment.
9440 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9442 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9443 OPTION_MASK_FLOAT128_KEYWORD.
9444 (POWERPC_MASKS): Likewise.
9445 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9446 support for the -mfloat128-type option, and make -mfloat128
9447 default on PowerPC Linux systems. Define or undefine
9448 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9449 Define __float128 to be __ieee128 if IEEE 128-bit support is
9450 enabled, or undefine it.
9451 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9452 Delete defining __FLOAT128_TYPE__.
9453 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9454 -mfloat128-type option and make -mfloat128 default on PowerPC
9456 (TARGET_FLOAT128_TYPE): Likewise.
9457 (-mfloat128-type): Likewise.
9458 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9459 Delete the -mfloat128-type option and make -mfloat128 default on
9460 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9461 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9462 128-bit floating point is enabled. Change tests from using
9463 -mfloat128-type to -mfloat128.
9464 (rs6000_mangle_type): Use the correct mangling for the __float128
9465 type even if normal long double is restricted to 64-bits.
9466 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9468 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9469 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9470 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9472 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9473 documentation for -mfloat128.
9475 2017-09-06 Olivier Hainque <hainque@adacore.com>
9477 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9479 2017-09-06 Wish Wu <wishwu007@gmail.com>
9480 Jakub Jelinek <jakub@redhat.com>
9482 * asan.c (initialize_sanitizer_builtins): Add
9483 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9484 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9485 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9486 BT_FN_VOID_UINT64_PTR variables.
9487 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9488 (BT_FN_VOID_UINT16_UINT16): Likewise.
9489 (BT_FN_VOID_UINT32_UINT32): Likewise.
9490 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9491 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9492 (BT_FN_VOID_UINT64_PTR): Likewise.
9493 * common.opt (flag_sanitize_coverage): New variable.
9494 (fsanitize-coverage=trace-pc): Remove.
9495 (fsanitize-coverage=): Add.
9496 * flag-types.h (enum sanitize_coverage_code): New enum.
9497 * fold-const.c (fold_range_test): Disable non-short-circuit
9498 optimization if flag_sanitize_coverage.
9499 (fold_truth_andor): Likewise.
9500 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9501 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9502 (coverage_sanitizer_opts): New array.
9503 (get_closest_sanitizer_option): Add OPTS argument, handle also
9504 OPT_fsanitize_coverage_.
9505 (parse_sanitizer_options): Adjusted to also handle
9506 OPT_fsanitize_coverage_.
9507 (common_handle_option): Add OPT_fsanitize_coverage_.
9508 * sancov.c (instrument_comparison, instrument_switch): New function.
9509 (sancov_pass): Add trace-cmp support.
9510 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9511 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9512 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9513 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9514 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9515 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9516 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9517 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9518 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9519 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9521 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9523 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9524 error. Only quit immediately if parsing is complete.
9525 (BEGIN): Initialize fatal_err and parse_done.
9526 (begin fpu, end fpu): Check number of arguments.
9527 (begin arch, end arch): Likewise.
9528 (begin cpu, end cpu): Likewise.
9529 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9530 (optalias): Likewise.
9532 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9534 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9535 * config/arm/arm-isa.h: Delete. Move definitions to ...
9536 * arm-cpus.in: ... here. Use new feature and fgroup values.
9537 * config/arm/arm.c (arm_option_override): Use lower case for feature
9539 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9540 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9541 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9543 (print_isa_bits_for): New function.
9544 (gen_isa): New function.
9545 (gen_comm_data): Use print_isa_bits_for.
9546 (define feature): New keyword.
9547 (define fgroup): New keyword.
9548 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9549 (arm-isa.h): Add rule to generate file.
9550 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9551 case for feature bit names.
9553 2017-09-06 Richard Biener <rguenther@suse.de>
9555 * tree-ssa-pre.c (NECESSARY): Remove.
9556 (create_expression_by_pieces): Do not touch pass-local flags.
9557 (insert_into_preds_of_block): Likewise.
9558 (do_pre_regular_insertion): Likewise.
9559 (eliminate_insert): Likewise.
9560 (eliminate_dom_walker::before_dom_children): Likewise.
9561 (fini_eliminate): Do not look at inserted_exprs.
9562 (mark_operand_necessary): Remove.
9563 (remove_dead_inserted_code): Replace with simple work-list
9564 algorithm based on inserted_exprs and SSA uses.
9565 (pass_pre::execute): Re-order fini_eliminate and
9566 remove_dead_inserted_code.
9568 2017-09-06 Olivier Hainque <hainque@adacore.com>
9570 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9571 for VxWorks 7. Adjust surrounding comments.
9573 2017-09-06 Richard Biener <rguenther@suse.de>
9575 * gimple-ssa-strength-reduction.c
9576 (find_candidates_dom_walker::before_dom_children): Also allow
9579 2017-09-06 Richard Biener <rguenther@suse.de>
9581 PR tree-optimization/82108
9582 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9583 for gap in the non-permutation SLP case.
9585 2017-09-06 Martin Jambor <mjambor@suse.cz>
9587 PR tree-optimization/82078
9588 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9589 add_access_to_work_queue...
9590 (build_accesses_from_assign): ...here.
9591 (propagate_all_subaccesses): Make sure racc is the group
9592 representative, if there is one.
9594 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9597 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9600 2017-09-06 Richard Biener <rguenther@suse.de>
9602 * gimple-ssa-strength-reduction.c
9603 (find_candidates_dom_walker::before_doom_children): Use a
9604 type and not a mode check.
9606 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9609 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9611 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9613 2017-09-05 Jeff Law <law@redhat.com>
9615 PR tree-optimization/64910
9616 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9617 cases where we have 3 or more operands.
9619 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9622 * omp-low.c (lower_omp_for): Recompute tree invariant if
9623 gimple_omp_for_initial/final is ADDR_EXPR.
9626 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9627 into gimple val before gimplification fo the COND_EXPR.
9629 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9631 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9632 REGION_COPY argument.
9633 (thread_through_all_blocks): Remove unused argument to
9634 duplicate_thread_path.
9636 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9637 Alan Hayward <alan.hayward@arm.com>
9638 David Sherwood <david.sherwood@arm.com>
9640 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9641 Take a scalar_mode rather than a machine_mode.
9642 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9643 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9644 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9645 (aarch64_gen_adjusted_ldpstp): Likewise.
9646 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9648 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9649 Alan Hayward <alan.hayward@arm.com>
9650 David Sherwood <david.sherwood@arm.com>
9652 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9653 Take a scalar_int_mode instead of a machine_mode.
9654 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9655 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9656 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9657 (aarch64_simd_attr_length_rglist): Delete.
9658 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9659 a scalar_int_mode instead of a machine_mode.
9660 (aarch64_add_offset): Likewise.
9661 (aarch64_internal_mov_immediate): Likewise
9662 (aarch64_add_constant_internal): Likewise.
9663 (aarch64_add_constant): Likewise.
9664 (aarch64_movw_imm): Likewise.
9665 (aarch64_rtx_arith_op_extract_p): Likewise.
9666 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9667 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9668 Remove assert that the mode isn't a vector.
9669 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9670 (aarch64_expand_mov_immediate): Update calls after above changes.
9671 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9672 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9673 (aarch64_move_imm): Likewise.
9674 (aarch64_can_const_movi_rtx_p): Likewise.
9675 (aarch64_strip_extend): Likewise.
9676 (aarch64_extr_rtx_p): Likewise.
9677 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9678 a CONST_INT when the mode parameter is VOIDmode.
9679 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9681 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9683 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9684 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9685 (bitwise_type_for_mode): Update accordingly.
9687 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9689 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9690 * stor-layout.c (mode_for_size_tree): Likewise.
9691 (mode_for_array): Update accordingly.
9692 (layout_decl): Likewise.
9693 (compute_record_mode): Likewise. Only set the mode once.
9695 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9697 * target.def (get_mask_mode): Change return type to opt_mode.
9699 * doc/tm.texi: Regenerate.
9700 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9701 * targhooks.c (default_get_mask_mode): Likewise.
9702 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9703 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9704 targetm.get_mask_mode.
9705 * tree.c (build_truth_vector_type): Likewise.
9707 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9709 * machmode.h (mode_for_vector): Return an opt_mode.
9710 * stor-layout.c (mode_for_vector): Likewise.
9711 (mode_for_int_vector): Update accordingly.
9712 (layout_type): Likewise.
9713 * config/i386/i386.c (emit_memmov): Likewise.
9714 (ix86_expand_set_or_movmem): Likewise.
9715 (ix86_expand_vector_init): Likewise.
9716 (ix86_get_mask_mode): Likewise.
9717 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9719 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9720 * expmed.c (extract_bit_field_1): Likewise.
9721 * expr.c (expand_expr_real_2): Likewise.
9722 * optabs-query.c (can_vec_perm_p): Likewise.
9723 (can_vec_mask_load_store_p): Likewise.
9724 * optabs.c (expand_vec_perm): Likewise.
9725 * targhooks.c (default_get_mask_mode): Likewise.
9726 * tree-vect-stmts.c (vectorizable_store): Likewise.
9727 (vectorizable_load): Likewise.
9728 (get_vectype_for_scalar_type_and_size): Likewise.
9730 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9732 * machmode.h (mode_for_int_vector): New function.
9733 * stor-layout.c (mode_for_int_vector): Likewise.
9734 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9735 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9736 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9737 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9738 (s390_expand_vcond): Likewise.
9740 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9742 * machmode.h (opt_machine_mode): New type.
9743 (opt_mode<T>): Allow construction from anything that can be
9745 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9746 (mode_for_size): Return an opt_machine_mode.
9747 * stor-layout.c (mode_for_size): Likewise.
9748 (mode_for_size_tree): Update call accordingly.
9749 (bitwise_mode_for_mode): Likewise.
9750 (make_fract_type): Likewise.
9751 (make_accum_type): Likewise.
9752 * caller-save.c (replace_reg_with_saved_mem): Update call
9754 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9755 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9756 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9757 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9758 * expmed.c (extract_bit_field_1): Likewise.
9759 * reload.c (get_secondary_mem): Likewise.
9760 * varasm.c (assemble_integer): Likewise.
9761 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9764 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9766 * machmode.h (decimal_float_mode_for_size): New function.
9767 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9768 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9769 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9770 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9771 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9772 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9774 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9776 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9777 (get_builtin_sync_mode): Likewise.
9778 (expand_ifn_atomic_compare_exchange): Likewise.
9779 (expand_builtin_atomic_clear): Likewise.
9780 (expand_builtin_atomic_test_and_set): Likewise.
9781 (fold_builtin_atomic_always_lock_free): Likewise.
9782 * calls.c (compute_argument_addresses): Likewise.
9783 (emit_library_call_value_1): Likewise.
9784 (store_one_arg): Likewise.
9785 * combine.c (combine_instructions): Likewise.
9786 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9787 * config/arm/arm.c (arm_function_value): Likewise.
9788 (aapcs_allocate_return_reg): Likewise.
9789 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9790 * config/i386/i386.c (construct_container): Likewise.
9791 (ix86_gimplify_va_arg): Likewise.
9792 (ix86_expand_sse_cmp): Likewise.
9793 (emit_memmov): Likewise.
9794 (emit_memset): Likewise.
9795 (expand_small_movmem_or_setmem): Likewise.
9796 (ix86_expand_pextr): Likewise.
9797 (ix86_expand_pinsr): Likewise.
9798 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9799 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9801 * config/mips/mips.c (mips_function_value_1) Likewise.
9802 (mips_block_move_straight): Likewise.
9803 (mips_expand_ins_as_unaligned_store): Likewise.
9804 * config/powerpcspe/powerpcspe.c
9805 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9806 (rs6000_darwin64_record_arg_flush): Likewise.
9807 * config/rs6000/rs6000.c
9808 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9809 (rs6000_darwin64_record_arg_flush): Likewise.
9810 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9811 (sparc_function_value_1): Likewise.
9812 * config/spu/spu.c (adjust_operand): Likewise.
9813 (spu_emit_branch_or_set): Likewise.
9814 (arith_immediate_p): Likewise.
9815 * emit-rtl.c (gen_lowpart_common): Likewise.
9816 * expr.c (expand_expr_real_1): Likewise.
9817 * function.c (assign_parm_setup_block): Likewise.
9818 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9819 * reload1.c (alter_reg): Likewise.
9820 * stor-layout.c (mode_for_vector): Likewise.
9821 (layout_type): Likewise.
9823 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9825 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9826 (spu_convert_move): Likewise.
9827 * lower-subreg.c (resolve_simple_move): Likewise.
9829 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9832 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9833 define_insn to a define_expand.
9834 (altivec_vsum2sws_direct): New define_insn.
9835 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9837 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9839 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9842 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9847 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9848 realign stack if stack alignment needed is less than incoming
9851 2017-09-05 Marek Polacek <polacek@redhat.com>
9854 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9857 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9859 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9861 2017-09-05 Richard Biener <rguenther@suse.de>
9863 PR tree-optimization/82084
9864 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9866 2017-09-05 Richard Biener <rguenther@suse.de>
9868 PR tree-optimization/82102
9869 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9871 2017-09-05 Martin Liska <mliska@suse.cz>
9873 PR tree-optimization/82032
9874 * tree-cfg.c (generate_range_test): New function.
9875 * tree-cfg.h (generate_range_test): Declared here.
9876 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9877 (cleanup_control_expr_graph): Use it.
9878 * tree-switch-conversion.c (try_switch_expansion): Remove
9880 (emit_case_nodes): Use generate_range_test.
9882 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9885 * config/i386/i386.md (*<btsc><mode>_mask): Add
9886 TARGET_USE_BT to insn constraint.
9887 (*btr<mode>_mask): Ditto.
9889 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9891 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9892 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9894 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9897 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9898 TARGET_NEON and TARGET_IWMMXT.
9899 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9900 TARGET_NEON and TARGET_IWMMXT.
9901 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9903 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9905 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9906 (ix86_rewrite_tls_address): Ditto.
9907 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9908 (ix86_rewrite_tls_address_1): Ditto.
9909 (ix86_rewrite_tls_address): Ditto.
9910 * config/i386/predicates.md (tls_address_pattern): New predicate.
9911 * config/i386/i386.md (TLS address splitter): New splitter.
9913 2017-09-04 Richard Biener <rguenther@suse.de>
9915 PR tree-optimization/82084
9916 * fold-const.h (can_native_encode_string_p): Declare.
9917 * fold-const.c (can_native_encode_string_p): Factor out from ...
9918 (native_encode_string): ... here.
9919 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9920 vectorizing stores from constants we later cannot handle.
9922 2017-09-04 Marek Polacek <polacek@redhat.com>
9925 * doc/invoke.texi: Update -Wtautological-compare documentation.
9927 2017-09-04 Jeff Law <law@redhat.com>
9929 PR tree-optimization/64910
9930 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9931 swap the first and last operand if the last is a constant.
9933 2017-09-04 Marek Polacek <polacek@redhat.com>
9936 * convert.c (do_narrow): When sanitizing signed integer overflows,
9937 bail out for signed types.
9938 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9940 2017-09-04 Richard Biener <rguenther@suse.de>
9942 PR tree-optimization/82060
9943 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9944 Move devirtualization after stmt folding and before EH/AB/noreturn
9945 cleanup to get the stmt refs canonicalized. Use a bool instead
9946 of gimple_modified_p since that doesn't work for NOPs. Schedule
9947 NOPs generated by folding for removal.
9949 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * coretypes.h (pad_direction): New enum.
9954 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9955 (FUNCTION_ARG_PADDING): Likewise.
9956 * target.def (function_arg_padding): New hook.
9957 * targhooks.h (default_function_arg_padding): Declare.
9958 * targhooks.c (default_function_arg_padding): New function.
9959 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9960 (TARGET_FUNCTION_ARG_PADDING): ...this.
9961 * doc/tm.texi: Regenerate.
9962 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9963 instead of direction.
9964 (compute_argument_addresses): Likewise.
9965 (load_register_parameters): Likewise.
9966 (emit_library_call_value_1): Likewise.
9967 (store_one_arg): Use targetm.calls.function_arg_padding instead
9968 of FUNCTION_ARG_PADDING.
9969 (must_pass_in_stack_var_size_or_pad): Likewise.
9970 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9971 (emit_group_store): Likewise.
9972 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9973 instead of FUNCTION_ARG_PADDING.
9974 (emit_push_insn): Likewise, and propagate enum change throughout
9976 * function.h (direction): Delete.
9977 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9979 * function.c (assign_parm_find_stack_rtl): Likewise.
9980 (assign_parm_setup_block_p): Likewise.
9981 (assign_parm_setup_block): Likewise.
9982 (gimplify_parameters): Likewise.
9983 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9984 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9986 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9987 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9988 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9989 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9990 (aarch64_function_arg_padding): ...this new function.
9991 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9992 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9993 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9994 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9995 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9996 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9997 (arm_pad_arg_upward): Replace with...
9998 (arm_function_arg_padding): ...this new function.
9999 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10001 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10002 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10003 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10004 (ia64_hpux_function_arg_padding): Replace with...
10005 (ia64_function_arg_padding): ...this new function. Use pad_direction
10006 instead of direction. Check for TARGET_HPUX.
10007 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10008 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10009 (iq2000_function_arg_padding): New function.
10010 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10011 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10012 (mips_function_arg_padding): ...this new function.
10013 (mips_pad_reg_upward): Update accordingly.
10014 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10015 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10016 targetm.calls.function_arg_padding.
10017 (FUNCTION_ARG_PADDING): Delete.
10018 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10019 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10020 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10021 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10022 (nios2_block_reg_padding): Return pad_direction instead of direction.
10023 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10024 instead of direction.
10025 (nios2_function_arg_padding): Likewise. Make static.
10026 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10027 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10028 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10029 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10030 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10031 (pa_function_arg_padding): Make static. Return pad_direction instead
10033 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10034 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10035 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10036 instead of direction. Use targetm.calls.function_arg_padding.
10037 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10038 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10039 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10040 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10041 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10043 (function_arg_padding): Rename to...
10044 (rs6000_function_arg_padding): ...this. Make static. Return
10045 pad_direction instead of direction.
10046 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10047 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10048 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10049 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10050 instead of direction. Use targetm.calls.function_arg_padding.
10051 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10052 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10053 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10054 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10055 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10056 (function_arg_padding): Rename to...
10057 (rs6000_function_arg_padding): ...this. Make static. Return
10058 pad_direction instead of direction.
10059 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10060 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10061 * config/s390/s390.c (s390_function_arg_padding): New function.
10062 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10063 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10064 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10065 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10066 (function_arg_padding): Rename to...
10067 (sparc_function_arg_padding): ...this. Make static. Return
10068 pad_direction instead of direction.
10069 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10070 * config/spu/spu.c (spu_function_arg_padding): New function.
10071 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10072 * system.h (FUNCTION_ARG_PADDING): Poison.
10074 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10075 Alan Hayward <alan.hayward@arm.com>
10076 David Sherwood <david.sherwood@arm.com>
10078 * target.def (modes_tieable_p): New hook.
10079 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10080 (TARGET_MODES_TIEABLE_P): ...this.
10081 * doc/tm.texi.in: Regenerate.
10082 * hooks.h (hook_bool_mode_mode_true): Declare.
10083 * hooks.c (hook_bool_mode_mode_true): New function.
10084 * combine.c (subst): Use targetm.modes_tieable_p instead of
10086 * dse.c (find_shift_sequence): Likewise.
10087 * expmed.c (extract_low_bits): Likewise.
10088 * lower-subreg.c: Include target.h.
10089 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10091 * rtlanal.c (rtx_cost): Likewise.
10092 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10093 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10094 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10095 (TARGET_MODES_TIEABLE_P): Redefine.
10096 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10097 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10098 (TARGET_MODES_TIEABLE_P): Redefine.
10099 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10100 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10101 (arc_modes_tieable_p): New function.
10102 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10103 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10104 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10105 (arm_modes_tieable_p): Make static.
10106 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10107 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10108 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10109 (TARGET_MODES_TIEABLE_P): Redefine.
10110 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10111 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10112 (TARGET_MODES_TIEABLE_P): Redefine.
10113 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10114 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10115 (cr16_modes_tieable_p): New function.
10116 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10117 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10118 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10119 (TRULY_NOOP_TRUNCATION): Update comment.
10120 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10121 (TRULY_NOOP_TRUNCATION): Update comment.
10122 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10123 (frv_modes_tieable_p): New function.
10124 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10125 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10126 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10127 (TARGET_MODES_TIEABLE_P): Redefine.
10128 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10129 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10130 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10131 (TARGET_MODES_TIEABLE_P): Redefine.
10132 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10133 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10134 (ia64_modes_tieable_p): New function.
10135 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10136 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10137 (iq2000_modes_tieable_p): New function.
10138 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10139 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10140 (lm32_modes_tieable_p): New function.
10141 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10142 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10143 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10144 (TARGET_MODES_TIEABLE_P): Redefine.
10145 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10146 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10147 (m32r_modes_tieable_p): New function.
10148 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10149 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10150 (m68k_modes_tieable_p): New function.
10151 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10152 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10153 (mcore_modes_tieable_p): New function.
10154 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10155 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10157 (TARGET_MODES_TIEABLE_P): Redefine.
10158 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10159 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10160 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10161 (TARGET_MODES_TIEABLE_P): Redefine.
10162 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10163 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10164 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10165 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10166 (mn10300_modes_tieable_p): ...this and make static.
10167 (TARGET_MODES_TIEABLE_P): Redefine.
10168 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10169 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10170 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10171 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10172 (msp430_modes_tieable_p): Make static.
10173 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10174 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10175 (TARGET_MODES_TIEABLE_P): Redefine.
10176 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10177 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10178 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10179 (TARGET_MODES_TIEABLE_P): Redefine.
10180 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10181 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10182 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10183 (TARGET_MODES_TIEABLE_P): Redefine.
10184 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10185 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10186 (pdp11_modes_tieable_p): New function.
10187 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10188 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10189 (rs6000_modes_tieable_p): New function.
10190 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10191 * config/powerpcspe/powerpcspe.md: Update comment.
10192 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10193 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10194 (TARGET_MODES_TIEABLE_P): Redefine.
10195 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10196 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10197 (rl78_modes_tieable_p): New function.
10198 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10199 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10200 (rs6000_modes_tieable_p): New function.
10201 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10202 * config/rs6000/rs6000.md: Update comment.
10203 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10204 * config/rx/rx.c (rx_modes_tieable_p): New function.
10205 (TARGET_MODES_TIEABLE_P): Redefine.
10206 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10207 * config/s390/s390.c (s390_modes_tieable_p): New function.
10208 (TARGET_MODES_TIEABLE_P): Redefine.
10209 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10210 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10211 (sh_modes_tieable_p): New function.
10212 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10213 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10214 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10215 (sparc_modes_tieable_p): Make static.
10216 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10217 * config/spu/spu.c (spu_modes_tieable_p): New function.
10218 (TARGET_MODES_TIEABLE_P): Redefine.
10219 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10220 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10221 (TARGET_MODES_TIEABLE_P): Redefine.
10222 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10223 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10224 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10225 * config/v850/v850.c (v850_modes_tieable_p): New function.
10226 (TARGET_MODES_TIEABLE_P): Redefine.
10227 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10228 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10229 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10230 (visium_modes_tieable_p): New function.
10231 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10232 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10233 (xtensa_modes_tieable_p): New function.
10234 * system.h (MODES_TIEABLE_P): Poison.
10236 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10237 Alan Hayward <alan.hayward@arm.com>
10238 David Sherwood <david.sherwood@arm.com>
10240 * target.def (hard_regno_mode_ok): New hook.
10241 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10242 (TARGET_HARD_REGNO_MODE_OK): ...this.
10243 * doc/tm.texi.in: Regenerate.
10244 * hooks.h (hook_bool_uint_mode_true): Declare.
10245 * hooks.c (hook_bool_uint_mode_true): New function.
10246 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10247 HARD_REGNO_MODE_OK.
10248 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10249 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10250 instead of HARD_REGNO_MODE_OK.
10251 * caller-save.c: Include target.h.
10252 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10253 HARD_REGNO_MODE_OK.
10254 * combine.c (can_combine_p): Likewise.
10255 (combinable_i3pat): Likewise.
10256 (can_change_dest_mode): Likewise.
10257 * expr.c (init_expr_target): Likewise.
10258 (convert_move): Likewise.
10259 (convert_modes): Likewise.
10260 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10261 (setup_prohibited_mode_move_regs): Likewise.
10262 * ira.h (target_ira): Likewise.
10263 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10264 * lra-constraints.c (process_alt_operands): Likewise.
10265 (split_reg): Likewise.
10266 * recog.c (peep2_find_free_register): Likewise.
10267 * ree.c (combine_reaching_defs): Likewise.
10268 * regcprop.c (maybe_mode_change): Likewise.
10269 * reginfo.c (init_reg_sets_1): Likewise.
10270 (choose_hard_reg_mode): Likewise.
10271 (simplifiable_subregs): Likewise.
10272 * regrename.c (check_new_reg_p): Likewise.
10273 * reload.c (find_valid_class): Likewise.
10274 (find_valid_class_1): Likewise.
10275 (reload_inner_reg_of_subreg): Likewise.
10276 (push_reload): Likewise.
10277 (combine_reloads): Likewise.
10278 (find_dummy_reload): Likewise.
10279 (find_reloads): Likewise.
10280 * reload1.c (find_reg): Likewise.
10281 (set_reload_reg): Likewise.
10282 (allocate_reload_reg): Likewise.
10283 (choose_reload_regs): Likewise.
10284 (reload_adjust_reg_for_temp): Likewise.
10285 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10286 (simplify_subreg_regno): Likewise.
10287 * sel-sched.c (init_regs_for_mode): Likewise.
10288 * varasm.c (make_decl_rtl): Likewise.
10289 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10290 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10291 HARD_REGNO_MODE_OK.
10292 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10293 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10294 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10295 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10296 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10297 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10298 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10299 (arc_mode_class): Delete.
10300 (HARD_REGNO_MODE_OK): Delete.
10301 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10302 (arc_hard_regno_mode_ok): Rename old array to...
10303 (arc_hard_regno_mode_ok_modes): ...this.
10304 (arc_conditional_register_usage): Update accordingly.
10305 (arc_mode_class): Make static.
10306 (arc_hard_regno_mode_ok): New function.
10307 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10308 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10309 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10310 (arm_hard_regno_mode_ok): Make static.
10311 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10312 HARD_REGNO_MODE_OK.
10313 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10314 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10315 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10317 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10318 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10319 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10320 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10321 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10322 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10323 * config/bfin/predicates.md (valid_reg_operand): Use
10324 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10325 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10326 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10327 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10328 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10329 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10330 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10331 (cr16_hard_regno_mode_ok): Make static and return a bool.
10332 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10333 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10334 (cris_hard_regno_mode_ok): New function.
10335 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10336 (epiphany_mode_class): Delete.
10337 (HARD_REGNO_MODE_OK): Delete.
10338 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10339 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10340 (hard_regno_mode_ok): Rename to...
10341 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10342 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10343 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10344 HARD_REGNO_MODE_OK.
10345 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10346 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10347 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10348 (frv_hard_regno_mode_ok): Make static and return a bool.
10349 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10350 HARD_REGNO_MODE_OK.
10351 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10352 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10353 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10354 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10356 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10357 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10358 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10359 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10362 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10363 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10364 (ia64_hard_regno_mode_ok): New function.
10365 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10366 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10367 (iq2000_hard_regno_mode_ok): New function.
10368 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10369 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10370 (lm32_hard_regno_mode_ok): New function.
10371 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10372 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10373 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10374 instead of HARD_REGNO_MODE_OK.
10375 (m32c_hard_regno_ok): Rename to...
10376 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10377 (m32c_cannot_change_mode_class): Update accordingly.
10378 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10379 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10380 (m32r_mode_class): Delete.
10381 (HARD_REGNO_MODE_OK): Delete.
10382 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10383 (m32r_hard_regno_mode_ok): Rename to...
10384 (m32r_hard_regno_modes): ...this.
10385 (m32r_mode_class): Make static.
10386 (m32r_hard_regno_mode_ok): New function.
10387 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10388 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10389 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10390 (m68k_hard_regno_mode_ok): Make static.
10391 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10392 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10393 (mcore_hard_regno_mode_ok): New function.
10394 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10395 (HARD_REGNO_MODE_OK): Delete.
10396 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10398 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10399 (microblaze_hard_regno_mode_ok): New function.
10400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10401 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10402 (mips_hard_regno_mode_ok): Delete.
10403 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10404 (mips_hard_regno_mode_ok_p): ...this and make static.
10405 (mips_hard_regno_mode_ok_p): Rename to...
10406 (mips_hard_regno_mode_ok_uncached): ...this.
10407 (mips_hard_regno_mode_ok): New function.
10408 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10409 of HARD_REGNO_MODE_OK.
10410 (mips_option_override): Update after above name changes.
10411 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10412 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10413 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10414 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10415 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10416 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10417 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10418 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10419 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10420 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10421 (msp430_hard_regno_mode_ok): Make static and return a bool.
10422 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10423 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10424 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10426 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10427 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10428 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10429 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10430 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10431 (PA_HARD_REGNO_MODE_OK): ...this
10432 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10433 (PA_HARD_REGNO_MODE_OK): ...this.
10434 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10435 (pa_hard_regno_mode_ok): New function.
10436 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10437 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10438 (pdp11_hard_regno_mode_ok): New function.
10439 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10440 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10442 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10445 (rs6000_hard_regno_mode_ok): Rename to...
10446 (rs6000_hard_regno_mode_ok_uncached): ...this.
10447 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10448 (rs6000_hard_regno_mode_ok): New function.
10449 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10450 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10451 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10452 (riscv_hard_regno_mode_ok): ...this and make static.
10453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10454 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10455 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10456 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10457 (rl78_hard_regno_mode_ok): Make static and return bool.
10458 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10459 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10461 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10462 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10463 (rs6000_hard_regno_mode_ok): Rename to...
10464 (rs6000_hard_regno_mode_ok_uncached): ...this.
10465 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10466 (rs6000_hard_regno_mode_ok): New function.
10467 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10468 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10469 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10470 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10471 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10472 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10473 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10474 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10475 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10476 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10477 (sh_hard_regno_mode_ok): Make static.
10478 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10479 instead of HARD_REGNO_MODE_OK.
10480 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10481 (sparc_mode_class): Delete.
10482 (HARD_REGNO_MODE_OK): Delete.
10483 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10484 (hard_regno_mode_classes): Make static.
10485 (sparc_mode_class): Likewise.
10486 (sparc_hard_regno_mode_ok): New function.
10487 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10488 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10489 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10492 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10493 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10494 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10495 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10496 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10497 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10498 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10499 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10500 (visium_hard_regno_mode_ok): New function.
10501 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10502 instead of HARD_REGNO_MODE_OK.
10503 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10504 (HARD_REGNO_MODE_OK): Delete.
10505 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10506 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10507 (xtensa_option_override): Update accordingly.
10508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10509 (xtensa_hard_regno_mode_ok): New function.
10510 * system.h (HARD_REGNO_MODE_OK): Poison.
10512 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10513 Alan Hayward <alan.hayward@arm.com>
10514 David Sherwood <david.sherwood@arm.com>
10516 * target.def (hard_regno_call_part_clobbered): New hook.
10517 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10518 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10519 * doc/tm.texi: Regenerate.
10520 * hooks.h (hook_bool_uint_mode_false): Declare.
10521 * hooks.c (hook_bool_uint_mode_false): New function.
10522 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10523 * cselib.c (cselib_process_insn): Use
10524 targetm.hard_regno_call_part_clobbered instead of
10525 HARD_REGNO_CALL_PART_CLOBBERED.
10526 * ira-conflicts.c (ira_build_conflicts): Likewise.
10527 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10528 * lra-constraints.c (need_for_call_save_p): Likewise.
10529 * lra-lives.c: Include target.h.
10530 (check_pseudos_live_through_calls): Use
10531 targetm.hard_regno_call_part_clobbered instead of
10532 HARD_REGNO_CALL_PART_CLOBBERED.
10533 * regcprop.c: Include target.h.
10534 (copyprop_hardreg_forward_1): Use
10535 targetm.hard_regno_call_part_clobbered instead of
10536 HARD_REGNO_CALL_PART_CLOBBERED.
10537 * reginfo.c (choose_hard_reg_mode): Likewise.
10538 * regrename.c (check_new_reg_p): Likewise.
10539 * reload.c (find_equiv_reg): Likewise.
10540 * reload1.c (emit_reload_insns): Likewise.
10541 * sched-deps.c (deps_analyze_insn): Likewise.
10542 * sel-sched.c (init_regs_for_mode): Likewise.
10543 (mark_unavailable_hard_regs): Likewise.
10544 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10545 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10546 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10548 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10549 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10550 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10552 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10554 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10555 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10556 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10558 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10559 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10560 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10562 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10563 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10565 * config/powerpcspe/powerpcspe.c
10566 (rs6000_hard_regno_call_part_clobbered): New function.
10567 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10568 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10569 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10571 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10572 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10573 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10575 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10576 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10577 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10579 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10580 Alan Hayward <alan.hayward@arm.com>
10581 David Sherwood <david.sherwood@arm.com>
10583 * rtl.h (subreg_memory_offset): Declare.
10584 * emit-rtl.c (subreg_memory_offset): New function.
10585 * expmed.c (store_bit_field_1): Use it.
10586 * expr.c (undefined_operand_subword_p): Likewise.
10587 * simplify-rtx.c (simplify_subreg): Likewise.
10589 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10591 PR rtl-optimization/57448
10594 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10595 using atomic_load pattern.
10596 (expand_atomic_store): Likewise.
10598 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10601 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10602 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10603 handling. Use replace_call_with_value with NULL instead of
10604 gsi_replace, unlink_stmt_vdef and release_defs.
10606 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10609 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10611 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10614 * rtl.h (emit_library_call_value_1): Declare.
10615 (emit_library_call): Replace declaration with a series of overloads.
10616 Remove the parameter count argument.
10617 (emit_library_call_value): Likewise.
10618 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10619 with an "rtx_mode_t *".
10620 (emit_library_call_value): Delete.
10621 (emit_library_call): Likewise.
10622 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10623 (asan_emit_allocas_unpoison): Likewise.
10624 * builtins.c (expand_builtin_powi): Likewise.
10625 (expand_asan_emit_allocas_unpoison): Likewise.
10626 * cfgexpand.c (expand_main_function): Likewise.
10627 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10628 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10629 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10630 * config/arm/arm.c (arm_trampoline_init): Likewise.
10631 (arm_call_tls_get_addr): Likewise.
10632 (arm_expand_divmod_libfunc): Likewise.
10633 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10634 (smulsi3_highpart): Likewise.
10635 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10636 (c6x_expand_compare): Likewise.
10637 (c6x_expand_movmem): Likewise.
10638 * config/frv/frv.c (frv_trampoline_init): Likewise.
10639 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10640 (ix86_expand_divmod_libfunc): Likewise.
10641 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10642 (ia64_expand_compare): Likewise.
10643 (ia64_profile_hook): Likewise.
10644 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10645 (nonlocal_goto): Likewise.
10646 (restore_stack_nonlocal): Likewise.
10647 * config/m32r/m32r.c (block_move_call): Likewise.
10648 (m32r_trampoline_init): Likewise.
10649 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10650 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10651 (m68k_call_m68k_read_tp): Likewise.
10652 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10653 (microblaze_expand_divide): Likewise.
10654 * config/mips/mips.h (mips_args): Likewise.
10655 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10656 (MIPS_ICACHE_SYNC): Likewise.
10657 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10658 (nios2_trampoline_init): Likewise.
10659 * config/pa/pa.c (hppa_tls_call): Likewise.
10660 (pa_trampoline_init): Likewise.
10661 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10662 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10663 (expand_strn_compare): Likewise.
10664 (rs6000_generate_compare): Likewise.
10665 (rs6000_expand_float128_convert): Likewise.
10666 (output_profile_hook): Likewise.
10667 (rs6000_trampoline_init): Likewise.
10668 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10669 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10670 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10671 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10672 (rs6000_generate_compare): Likewise.
10673 (rs6000_expand_float128_convert): Likewise.
10674 (output_profile_hook): Likewise.
10675 (rs6000_trampoline_init): Likewise.
10676 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10677 * config/sh/sh.c (sh_trampoline_init): Likewise.
10678 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10679 (sparc_emit_float_lib_cmp): Likewise.
10680 (sparc32_initialize_trampoline): Likewise.
10681 (sparc64_initialize_trampoline): Likewise.
10682 (sparc_profile_hook): Likewise.
10683 * config/spu/spu.c (ea_load_store): Likewise.
10684 * config/spu/spu.md (floatunssidf2): Likewise.
10685 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10686 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10687 * config/visium/visium.c (expand_block_move_4): Likewise.
10688 (expand_block_move_2): Likewise.
10689 (expand_block_move_1): Likewise.
10690 (expand_block_set_4): Likewise.
10691 (expand_block_set_2): Likewise.
10692 (expand_block_set_1): Likewise.
10693 (visium_trampoline_init): Likewise.
10694 (visium_profile_hook): Likewise.
10695 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10696 (xtensa_setup_frame_addresses): Likewise.
10697 (xtensa_trampoline_init): Likewise.
10698 * except.c (sjlj_emit_function_enter): Likewise.
10699 (sjlj_emit_function_exit): Likewise.
10700 * explow.c (allocate_dynamic_stack_space): Likewise.
10701 (probe_stack_range): Likewise.
10702 * expr.c (convert_mode_scalar): Likewise.
10703 * optabs.c (expand_binop): Likewise.
10704 (expand_twoval_binop_libfunc): Likewise.
10705 (expand_unop): Likewise.
10706 (prepare_cmp_insn): Likewise.
10707 (prepare_float_lib_cmp): Likewise.
10708 (expand_float): Likewise.
10709 (expand_fix): Likewise.
10710 (expand_fixed_convert): Likewise.
10711 (maybe_emit_sync_lock_test_and_set): Likewise.
10712 (expand_atomic_compare_and_swap): Likewise.
10713 (expand_mem_thread_fence): Likewise.
10714 (expand_atomic_fetch_op): Likewise.
10716 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10718 * doc/generic.texi (OpenACC): Adjust URL.
10719 * doc/invoke.texi (C Dialect Options): Ditto.
10721 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10723 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10724 predicate for operand 1. Add (m,<S>) constraint.
10725 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10726 Prevent memory operand 1 with register operand 2.
10728 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10730 PR rtl-optimization/82024
10731 * combine.c (try_combine): If the combination result is a PARALLEL,
10732 and we only need to retain the SET in there that would be placed
10733 at I2, check that we can place that at I3 instead, before doing so.
10735 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10738 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10740 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10741 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10744 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10745 Jeff Law <law@redhat.com>
10747 * varasm.c (bss_initializer_p): Do not put constants into .bss
10748 (categorize_decl_for_section): Handle bss_initializer_p returning
10749 false when DECL_INITIAL is NULL.
10751 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10754 * config/s390/s390.c (s390_can_inline_p): New function.
10756 2017-09-01 Jeff Law <law@redhat.com>
10758 PR tree-optimization/82052
10759 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10760 Always initialize the returned slot after a hash table miss
10761 when INSERT is true.
10763 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10765 * config/s390/s390.md (mem_signal_fence): Remove.
10766 * doc/md.texi (mem_signal_fence): Remove.
10767 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10769 * target-insns.def (mem_signal_fence): Remove.
10771 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10774 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10777 * asan.c (create_odr_indicator): Strip name encoding from assembler
10778 name before appending it after __odr_asan_.
10780 2017-09-01 Martin Liska <mliska@suse.cz>
10782 PR tree-optimization/82059
10783 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10784 frequency only when an edge is redirected.
10786 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10788 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10789 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10790 (arc_conditional_register_usage): Remove ARC600 lp_count
10792 (arc_file_start): Emit Tag_ARC_CPU_variation.
10793 (arc_can_use_doloop_p): New conditions to use ZOLs.
10794 (hwloop_fail): New function.
10795 (hwloop_optimize): Likewise.
10796 (hwloop_pattern_reg): Likewise.
10797 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10798 (arc_reorg_loops): New function, calls reorg_loops.
10799 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10800 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10802 (arc_loop_hazard): Remove function, functionality moved into
10804 (arc_hazard): Remove arc_loop_hazard call.
10805 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10806 into hwloop_optimize.
10807 (arc_label_align): Remove ZOL handling.
10808 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10809 * config/arc/arc.md (doloop_begin): Remove pattern.
10810 (doloop_begin_i): Likewise.
10811 (doloop_end_i): Likewise.
10812 (doloop_fallback): Likewise.
10813 (doloop_fallback_m): Likewise.
10814 (doloop_end): Reimplement expand.
10815 (arc_lp): New pattern for LP instruction.
10816 (loop_end): New pattern.
10817 (loop_fail): Likewise.
10818 (decrement_and_branch_until_zero): Likewise.
10819 * config/arc/arc.opt (mlpc-width): New option.
10820 * doc/invoke.texi (mlpc-width): Document option.
10822 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10824 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10825 (arc_ccfsm_advance): Fix checking for delay slots.
10826 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10828 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10830 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10832 (movhi_insn): Update store instruction constraint which are saving
10833 6-bit short immediates.
10834 (movsi_insn): Consider also short scaled load operations.
10835 (zero_extendhisi2_i): Use Usd constraint instead of T.
10836 (extendhisi2_i): Add q constraint.
10837 (arc_clzsi2): Add type and length attributes.
10838 (arc_ctzsi2): Likewise.
10839 * config/arc/constraints.md (Usc): Update constraint, the
10840 assembler can parse two relocations for a single instruction.
10842 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10844 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10845 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10847 2017-08-31 Olivier Hainque <hainque@adacore.com>
10849 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10850 match as powerpc-wrs-vxworks*.
10852 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10854 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10855 register constraint for by-element operand.
10856 (aarch64_mls_elt_merge<mode>): Likewise.
10858 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10860 * config/arc/arc.c (arc_can_follow_jump): Check for short
10863 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10865 * config.gcc: Use g.opt for arc.
10866 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10867 functionality moved to ...
10868 (legitimate_scaled_address_p): New function, ...here.
10869 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10870 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10871 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10873 (arc_override_options): Handle G option.
10874 (arc_output_pic_addr_const): Correct function definition.
10875 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10876 (arc_decl_anon_ns_mem_p): Delete.
10877 (arc_in_small_data_p): Overhaul this function to take into
10878 consideration the value given via G option.
10879 (arc_rewrite_small_data_1): Renamed and corrected old
10880 arc_rewrite_small_data function.
10881 (arc_rewrite_small_data): New function.
10882 (small_data_pattern): Don't use pic_offset_table_rtx.
10883 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10884 * config/arc/simdext.md (movmisalignv2hi): Use
10885 prepare_move_operands function.
10887 (movmisalign*): Likewise.
10888 * doc/invoke.texi (ARC options): Document -G option.
10890 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10892 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10894 * config/arc/arc.c (arc_print_operand): Output scalled address for
10895 sdata whenever is possible.
10896 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10897 load/stores are available.
10898 (compact_sda_memory_operand): Check for the alignment required by
10899 code density instructions.
10900 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10902 * config/arc/constraints.md (Usd): Update constraint.
10903 (Us0): New constraint.
10904 (Usc): Update constraint.
10906 2017-08-31 Richard Biener <rguenther@suse.de>
10908 PR middle-end/82054
10909 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10910 function only once.
10912 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10914 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10915 Resize type_signature.
10917 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10918 Alan Hayward <alan.hayward@arm.com>
10919 David Sherwood <david.sherwood@arm.com>
10921 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10922 subregs whose inner modes can be stored in GPRs.
10923 (aarch64_classify_index): Likewise.
10925 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10926 Alan Hayward <alan.hayward@arm.com>
10927 David Sherwood <david.sherwood@arm.com>
10929 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10930 (V_INT_EQUIV): ...this.
10931 (v_cmp_result): Rename to...
10932 (v_int_equiv): ...this.
10933 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10934 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10935 (copysign<mode>3): Likewise.
10936 (aarch64_simd_bsl<mode>_internal): Likewise.
10937 (aarch64_simd_bsl<mode>): Likewise.
10938 (vec_cmp<mode><mode>): Likewise.
10939 (vcond<mode><mode>): Likewise.
10940 (vcond<v_cmp_mixed><mode>): Likewise.
10941 (vcondu<mode><v_cmp_mixed>): Likewise.
10942 (aarch64_cm<optab><mode>): Likewise.
10943 (aarch64_cmtst<mode>): Likewise.
10944 (aarch64_fac<optab><mode>): Likewise.
10945 (vec_perm_const<mode>): Likewise.
10946 (vcond_mask_<mode><v_cmp_result>): Rename to...
10947 (vcond_mask_<mode><v_int_equiv>): ...this.
10948 (vec_cmp<mode><v_cmp_result>): Rename to...
10949 (vec_cmp<mode><v_int_equiv>): ...this.
10951 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10952 Alan Hayward <alan.hayward@arm.com>
10953 David Sherwood <david.sherwood@arm.com>
10955 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10957 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10958 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10959 (UNSPEC_LD4_DREG): New unspecs.
10960 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10961 (aarch64_ld2<mode>_dreg_be): Replace with...
10962 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10964 (aarch64_ld3<mode>_dreg_le)
10965 (aarch64_ld3<mode>_dreg_be): Replace with...
10966 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10968 (aarch64_ld4<mode>_dreg_le)
10969 (aarch64_ld4<mode>_dreg_be): Replace with...
10970 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10973 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10975 PR tree-optimization/81987
10976 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10977 insert an initializer in a location not dominated by the stride
10980 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10982 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10983 on the entire header of the finally block in the fallthru case.
10985 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10987 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10989 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10991 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10992 rs6000_emit_move_from_cr and call renamed function.
10993 (rs6000_emit_prologue): Call renamed functions.
10994 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10995 movesi_from_cr, remove volatile CRs.
10997 2017-08-30 Jon Beniston <jon@beniston.com>
10998 Richard Biener <rguenther@suse.de>
11000 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11001 of VECTOR_MODE_P check.
11002 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11003 element vector types.
11005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11007 * df.h (df_read_modify_subreg_p): Remove in favor of...
11008 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11009 const_rtx instead of an rtx.
11010 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11011 * df-problems.c (df_word_lr_mark_ref): Likewise.
11012 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11013 (mark_pseudo_reg_dead): Likewise.
11014 (mark_ref_dead): Likewise.
11015 * reginfo.c (init_subregs_of_mode): Likewise.
11016 * sched-deps.c (sched_analyze_1): Likewise.
11017 * df-scan.c (df_def_record_1): Likewise.
11018 (df_uses_record): Likewise.
11019 (df_read_modify_subreg_p): Remove in favor of...
11020 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11021 const_rtx instead of an rtx.
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 * rtl.h (partial_subreg_p): New function.
11028 * caller-save.c (save_call_clobbered_regs): Use it.
11029 * calls.c (expand_call): Likewise.
11030 * combine.c (combinable_i3pat): Likewise.
11031 (simplify_set): Likewise.
11032 (make_extraction): Likewise.
11033 (make_compound_operation_int): Likewise.
11034 (gen_lowpart_or_truncate): Likewise.
11035 (force_to_mode): Likewise.
11036 (make_field_assignment): Likewise.
11037 (reg_truncated_to_mode): Likewise.
11038 (record_truncated_value): Likewise.
11039 (move_deaths): Likewise.
11040 * cse.c (record_jump_cond): Likewise.
11041 (cse_insn): Likewise.
11042 * cselib.c (cselib_lookup_1): Likewise.
11043 * expmed.c (extract_bit_field_using_extv): Likewise.
11044 * function.c (assign_parm_setup_reg): Likewise.
11045 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11046 * ira-build.c (create_insn_allocnos): Likewise.
11047 * lra-coalesce.c (merge_pseudos): Likewise.
11048 * lra-constraints.c (match_reload): Likewise.
11049 (simplify_operand_subreg): Likewise.
11050 (curr_insn_transform): Likewise.
11051 * lra-lives.c (process_bb_lives): Likewise.
11052 * lra.c (new_insn_reg): Likewise.
11053 (lra_substitute_pseudo): Likewise.
11054 * regcprop.c (mode_change_ok): Likewise.
11055 (maybe_mode_change): Likewise.
11056 (copyprop_hardreg_forward_1): Likewise.
11057 * reload.c (push_reload): Likewise.
11058 (find_reloads): Likewise.
11059 (find_reloads_subreg_address): Likewise.
11060 * reload1.c (alter_reg): Likewise.
11061 (eliminate_regs_1): Likewise.
11062 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11064 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11066 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11067 back to if statements, including unpack.
11069 2017-08-30 Martin Liska <mliska@suse.cz>
11071 PR inline-asm/82001
11072 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11074 (func_checker::compare_asm_inputs_outputs): ... this function.
11075 (func_checker::compare_gimple_asm): Use the function to compare
11076 also ASM constrains.
11077 * ipa-icf-gimple.h: Rename the function.
11079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11080 Alan Hayward <alan.hayward@arm.com>
11081 David Sherwood <david.sherwood@arm.com>
11083 * coretypes.h (complex_mode): New type.
11084 * gdbhooks.py (build_pretty_printer): Handle it.
11085 * machmode.h (complex_mode): New class.
11086 (complex_mode::includes_p): New function.
11087 (is_complex_int_mode): Likewise.
11088 (is_complex_float_mode): Likewise.
11089 * genmodes.c (get_mode_class): Handle complex mode classes.
11090 * function.c (expand_function_end): Use is_complex_int_mode.
11092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11093 Alan Hayward <alan.hayward@arm.com>
11094 David Sherwood <david.sherwood@arm.com>
11096 * coretypes.h (scalar_mode_pod): New typedef.
11097 * gdbhooks.py (build_pretty_printer): Handle it.
11098 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11099 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11100 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11101 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11102 as_a <scalar_mode>.
11104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11105 Alan Hayward <alan.hayward@arm.com>
11106 David Sherwood <david.sherwood@arm.com>
11108 * machmode.h (mode_for_vector): Take a scalar_mode instead
11110 * stor-layout.c (mode_for_vector): Likewise.
11111 * explow.c (promote_mode): Use as_a <scalar_mode>.
11112 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11115 Alan Hayward <alan.hayward@arm.com>
11116 David Sherwood <david.sherwood@arm.com>
11118 * target.def (preferred_simd_mode): Take a scalar_mode
11119 instead of a machine_mode.
11120 * targhooks.h (default_preferred_simd_mode): Likewise.
11121 * targhooks.c (default_preferred_simd_mode): Likewise.
11122 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11123 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11124 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11125 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11126 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11127 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11128 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11129 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11131 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11132 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11133 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11134 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11135 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11136 * doc/tm.texi: Regenerate.
11137 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11141 Alan Hayward <alan.hayward@arm.com>
11142 David Sherwood <david.sherwood@arm.com>
11144 * target.def (scalar_mode_supported_p): Take a scalar_mode
11145 instead of a machine_mode.
11146 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11147 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11148 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11149 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11150 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11151 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11152 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11153 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11154 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11155 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11156 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11157 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11158 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11159 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11161 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11162 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11163 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11164 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11165 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11167 * doc/tm.texi: Regenerate.
11169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11170 Alan Hayward <alan.hayward@arm.com>
11171 David Sherwood <david.sherwood@arm.com>
11173 * coretypes.h (opt_scalar_mode): New typedef.
11174 * gdbhooks.py (build_pretty_printers): Handle it.
11175 * machmode.h (mode_iterator::get_2xwider): Add overload for
11177 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11179 * expr.c (convert_mode_scalar): Likewise.
11180 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11181 * optabs.c (expand_float): Likewise.
11182 (expand_fix): Likewise.
11183 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * optabs.c (expand_float): Explicitly check for scalars before
11190 using a branching expansion.
11191 (expand_fix): 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 * expr.c (convert_mode): Split scalar handling out into...
11198 (convert_mode_scalar): ...this new function. Treat the modes
11201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11202 Alan Hayward <alan.hayward@arm.com>
11203 David Sherwood <david.sherwood@arm.com>
11205 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11207 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11210 Alan Hayward <alan.hayward@arm.com>
11211 David Sherwood <david.sherwood@arm.com>
11213 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11214 rather than a machine_mode.
11215 (fixed_from_string): Likewise.
11216 (fixed_convert): Likewise.
11217 (fixed_convert_from_int): Likewise.
11218 (fixed_convert_from_real): Likewise.
11219 (real_convert_from_fixed): Likewise.
11220 * fixed-value.c (fixed_from_double_int): Likewise.
11221 (fixed_from_string): Likewise.
11222 (fixed_convert): Likewise.
11223 (fixed_convert_from_int): Likewise.
11224 (fixed_convert_from_real): Likewise.
11225 (real_convert_from_fixed): Likewise.
11226 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11229 Alan Hayward <alan.hayward@arm.com>
11230 David Sherwood <david.sherwood@arm.com>
11232 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11233 of separate mode class checks. Do not allow vector modes here.
11234 (immed_wide_int_const): Use as_a <scalar_mode>.
11235 * explow.c (trunc_int_for_mode): Likewise.
11236 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11237 (wi::shwi): Likewise.
11238 (wi::min_value): Likewise.
11239 (wi::max_value): Likewise.
11240 * dwarf2out.c (loc_descriptor): Likewise.
11241 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11242 for CONST_WIDE_INT.
11244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11245 Alan Hayward <alan.hayward@arm.com>
11246 David Sherwood <david.sherwood@arm.com>
11248 * tree.h (SCALAR_TYPE_MODE): New macro.
11249 * expr.c (expand_expr_addr_expr_1): Use it.
11250 (expand_expr_real_2): Likewise.
11251 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11252 (fold_convert_const_fixed_from_int): Likewise.
11253 (fold_convert_const_fixed_from_real): Likewise.
11254 (native_encode_fixed): Likewise
11255 (native_encode_complex): Likewise
11256 (native_encode_vector): Likewise.
11257 (native_interpret_fixed): Likewise.
11258 (native_interpret_real): Likewise.
11259 (native_interpret_complex): Likewise.
11260 (native_interpret_vector): Likewise.
11261 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11262 (simd_clone_adjust_argument_types): Likewise.
11263 (simd_clone_init_simd_arrays): Likewise.
11264 (simd_clone_adjust): Likewise.
11265 * stor-layout.c (layout_type): Likewise.
11266 * tree.c (build_minus_one_cst): Likewise.
11267 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11268 * tree-inline.c (estimate_move_cost): Likewise.
11269 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11270 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11271 (vectorizable_reduction): Likewise.
11272 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11273 (vect_recog_mixed_size_cond_pattern): Likewise.
11274 (check_bool_pattern): Likewise.
11275 (adjust_bool_pattern): Likewise.
11276 (search_type_for_mask_1): Likewise.
11277 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11278 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11279 (vectorizable_load): Likewise.
11280 (vectorizable_store): Likewise.
11281 * ubsan.c (ubsan_encode_value): Likewise.
11282 * varasm.c (output_constant): Likewise.
11284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11285 Alan Hayward <alan.hayward@arm.com>
11286 David Sherwood <david.sherwood@arm.com>
11288 * coretypes.h (scalar_mode): New class.
11289 * machmode.h (scalar_mode): Likewise.
11290 (scalar_mode::includes_p): New function.
11291 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11292 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11293 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11294 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11295 * expmed.c (store_bit_field_1): Likewise.
11296 (extract_bit_field_1): Likewise.
11297 * expr.c (write_complex_part): Likewise.
11298 (read_complex_part): Likewise.
11299 (emit_move_complex_push): Likewise.
11300 (expand_expr_real_2): Likewise.
11301 * function.c (assign_parm_setup_reg): Likewise.
11302 (assign_parms_unsplit_complex): Likewise.
11303 * optabs.c (expand_binop): Likewise.
11304 * rtlanal.c (subreg_get_info): Likewise.
11305 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11306 * varasm.c (output_constant_pool_2): Likewise.
11308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11309 Alan Hayward <alan.hayward@arm.com>
11310 David Sherwood <david.sherwood@arm.com>
11312 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11314 (expmed_mult_highpart_optab): Likewise.
11315 (expmed_mult_highpart): Likewise.
11317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11318 Alan Hayward <alan.hayward@arm.com>
11319 David Sherwood <david.sherwood@arm.com>
11321 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11322 instead of a machine_mode.
11323 (builtin_memset_read_str): Likewise.
11324 * builtins.c (c_readstr): Likewise.
11325 (builtin_memcpy_read_str): Likewise.
11326 (builtin_strncpy_read_str): Likewise.
11327 (builtin_memset_read_str): Likewise.
11328 (builtin_memset_gen_str): Likewise.
11329 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11330 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11331 instead of a machine_mode.
11332 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11334 (make_extraction): Likewise.
11335 (try_widen_shift_mode): Take and return scalar_int_modes instead
11337 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11338 a scalar_int_mode instead of a machine_mode.
11339 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11340 (avr_addr_space_pointer_mode): Likewise.
11341 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11342 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11343 (msp430_unwind_word_mode): Likewise.
11344 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11345 (spu_addr_space_pointer_mode): Likewise.
11346 (spu_addr_space_address_mode): Likewise.
11347 (spu_libgcc_cmp_return_mode): Likewise.
11348 (spu_libgcc_shift_count_mode): Likewise.
11349 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11350 (rl78_addr_space_pointer_mode): Likewise.
11351 (fl78_unwind_word_mode): Likewise.
11352 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11354 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11355 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11356 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11357 (mips_valid_pointer_mode): Likewise.
11358 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11359 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11360 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11362 (ft32_addr_space_address_mode): Likewise.
11363 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11364 scalar_int_mode instead of a machine_mode.
11365 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11367 (m32c_addr_space_address_mode): Likewise.
11368 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11369 (rs6000_eh_return_filter_mode): Likewise.
11370 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11371 (rs6000_eh_return_filter_mode): Likewise.
11372 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11373 (s390_libgcc_shift_count_mode): Likewise.
11374 (s390_unwind_word_mode): Likewise.
11375 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11377 * target.def (mode_rep_extended): Likewise.
11378 (valid_pointer_mode): Likewise.
11379 (addr_space.valid_pointer_mode): Likewise.
11380 (eh_return_filter_mode): Return a scalar_int_mode rather than
11382 (libgcc_cmp_return_mode): Likewise.
11383 (libgcc_shift_count_mode): Likewise.
11384 (unwind_word_mode): Likewise.
11385 (addr_space.pointer_mode): Likewise.
11386 (addr_space.address_mode): Likewise.
11387 * doc/tm.texi: Regenerate.
11388 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11390 (do_jump): Use scalar_int_mode for local variables.
11391 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11392 rather than a machine_mode.
11393 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11394 (scompare_loc_descriptor_wide): Likewise.
11395 (scompare_loc_descriptor_narrow): Likewise.
11396 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11398 * except.c (sjlj_emit_dispatch_table): Likewise.
11399 (expand_builtin_eh_copy_values): Likewise.
11400 * explow.c (convert_memory_address_addr_space_1): Likewise.
11401 Take a scalar_int_mode rather than a machine_mode.
11402 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11403 than a machine_mode.
11404 (memory_address_addr_space): Use scalar_int_mode for local variables.
11405 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11406 rather than a machine_mode.
11407 * expmed.c (mask_rtx): Likewise.
11408 (init_expmed_one_conv): Likewise.
11409 (expand_mult_highpart_adjust): Likewise.
11410 (extract_high_half): Likewise.
11411 (expmed_mult_highpart_optab): Likewise.
11412 (expmed_mult_highpart): Likewise.
11413 (expand_smod_pow2): Likewise.
11414 (expand_sdiv_pow2): Likewise.
11415 (emit_store_flag_int): Likewise.
11416 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11418 (extract_low_bits): Likewise.
11419 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11421 * expr.c (pieces_addr::adjust): Likewise.
11422 (can_store_by_pieces): Likewise.
11423 (store_by_pieces): Likewise.
11424 (clear_by_pieces_1): Likewise.
11425 (expand_expr_addr_expr_1): Likewise.
11426 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11427 (expand_expr_real_1): Likewise.
11428 (try_casesi): Likewise.
11429 * final.c (shorten_branches): Likewise.
11430 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11431 type of "mode" to machine_mode.
11432 * internal-fn.c (expand_arith_overflow_result_store): Take a
11433 scalar_int_mode rather than a machine_mode.
11434 (expand_mul_overflow): Use scalar_int_mode for local variables.
11435 * loop-doloop.c (doloop_modify): Likewise.
11436 (doloop_optimize): Likewise.
11437 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11438 than a machine_mode.
11439 (expand_doubleword_shift_condmove): Likewise.
11440 (expand_doubleword_shift): Likewise.
11441 (expand_doubleword_clz): Likewise.
11442 (expand_doubleword_popcount): Likewise.
11443 (expand_doubleword_parity): Likewise.
11444 (expand_absneg_bit): Use scalar_int_mode for local variables.
11445 (prepare_float_lib_cmp): Likewise.
11446 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11447 rather than a machine_mode.
11448 (convert_memory_address_addr_space): Likewise.
11449 (get_mode_bounds): Likewise.
11450 (get_address_mode): Return a scalar_int_mode rather than a
11452 * rtlanal.c (get_address_mode): Likewise.
11453 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11454 than a machine_mode.
11455 * targhooks.c (default_mode_rep_extended): Likewise.
11456 (default_valid_pointer_mode): Likewise.
11457 (default_addr_space_valid_pointer_mode): Likewise.
11458 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11459 than a machine_mode.
11460 (default_libgcc_cmp_return_mode): Likewise.
11461 (default_libgcc_shift_count_mode): Likewise.
11462 (default_unwind_word_mode): Likewise.
11463 (default_addr_space_pointer_mode): Likewise.
11464 (default_addr_space_address_mode): Likewise.
11465 * targhooks.h (default_eh_return_filter_mode): Likewise.
11466 (default_libgcc_cmp_return_mode): Likewise.
11467 (default_libgcc_shift_count_mode): Likewise.
11468 (default_unwind_word_mode): Likewise.
11469 (default_addr_space_pointer_mode): Likewise.
11470 (default_addr_space_address_mode): Likewise.
11471 (default_mode_rep_extended): Take a scalar_int_mode rather than
11473 (default_valid_pointer_mode): Likewise.
11474 (default_addr_space_valid_pointer_mode): Likewise.
11475 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11477 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11478 rather than a machine_mode.
11479 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11480 for local variables.
11481 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11482 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11483 than a machine_mode.
11485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11486 Alan Hayward <alan.hayward@arm.com>
11487 David Sherwood <david.sherwood@arm.com>
11489 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11490 the mode argument to scalar_int_mode.
11491 (do_jump_by_parts_zero_rtx): Likewise.
11492 (do_jump_by_parts_equality_rtx): Likewise.
11493 (do_jump_by_parts_greater): Take a mode argument.
11494 (do_jump_by_parts_equality): Likewise.
11495 (do_jump_1): Update calls accordingly.
11497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11498 Alan Hayward <alan.hayward@arm.com>
11499 David Sherwood <david.sherwood@arm.com>
11501 * is-a.h (safe_dyn_cast): New function.
11502 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11503 (jump_table_for_label): Likewise.
11504 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11505 instead of an rtx_insn *.
11506 (shorten_branches): Use dyn_cast instead of LABEL_P and
11507 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11508 rtx_jump_table_data::get_data_mode.
11509 (final_scan_insn): Likewise.
11511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11512 Alan Hayward <alan.hayward@arm.com>
11513 David Sherwood <david.sherwood@arm.com>
11515 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11516 trying to combine a full-register integer set with a subreg
11519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11520 Alan Hayward <alan.hayward@arm.com>
11521 David Sherwood <david.sherwood@arm.com>
11523 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11524 that is always either address_mode or pointer_mode.
11526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11527 Alan Hayward <alan.hayward@arm.com>
11528 David Sherwood <david.sherwood@arm.com>
11530 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11531 when the two are known to be equal.
11533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11534 Alan Hayward <alan.hayward@arm.com>
11535 David Sherwood <david.sherwood@arm.com>
11537 * simplify-rtx.c (simplify_const_unary_operation): Use
11538 is_a <scalar_int_mode> instead of checking for a nonzero
11539 precision. Forcibly convert op_mode to a scalar_int_mode
11540 in that case. More clearly differentiate the operand and
11541 result modes and use the former when deciding what the value
11542 of a count-bits operation should be. Use is_int_mode instead
11543 of checking for a MODE_INT. Remove redundant check for whether
11544 this mode has a zero precision.
11546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11547 Alan Hayward <alan.hayward@arm.com>
11548 David Sherwood <david.sherwood@arm.com>
11550 * optabs.c (widen_leading): Change the type of the mode argument
11551 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11552 (widen_bswap): Likewise.
11553 (expand_parity): Likewise.
11554 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11555 (expand_ffs): Likewise.
11556 (epand_unop): Check for scalar integer modes before calling the
11559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11560 Alan Hayward <alan.hayward@arm.com>
11561 David Sherwood <david.sherwood@arm.com>
11563 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11565 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11572 value_mode arguments. Use scalar_int_mode internally.
11573 (store_bit_field_1): Rename the new integer mode from imode
11574 to op0_mode and use it instead of GET_MODE (op0). Update calls
11575 to store_split_bit_field, store_bit_field_using_insv and
11576 store_fixed_bit_field.
11577 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11578 Use scalar_int_mode internally. Use a bit count rather than a mode
11579 when calculating the largest bit size for get_best_mode.
11580 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11581 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11582 Remove assertion that OP0 has a scalar integer mode.
11583 (store_split_bit_field): Add op0_mode and value_mode arguments.
11584 Update calls to extract_fixed_bit_field.
11585 (extract_bit_field_using_extv): Add an op0_mode argument.
11586 Use scalar_int_mode internally.
11587 (extract_bit_field_1): Rename the new integer mode from imode to
11588 op0_mode and use it instead of GET_MODE (op0). Update calls to
11589 extract_split_bit_field, extract_bit_field_using_extv and
11590 extract_fixed_bit_field.
11591 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11592 to extract_split_bit_field and extract_fixed_bit_field_1.
11593 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11594 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11595 on the target mode.
11596 (extract_split_bit_field): Add an op0_mode argument. Update call
11597 to extract_fixed_bit_field.
11599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11600 Alan Hayward <alan.hayward@arm.com>
11601 David Sherwood <david.sherwood@arm.com>
11603 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11604 * explow.c (hard_function_value): Likewise.
11605 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11606 convert_to_mode call outside the loop.
11607 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11608 for the mode iterator. Require the mode specified by max_pieces
11610 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11612 (copy_blkmode_to_reg): Likewise.
11613 (set_storage_via_setmem): Likewise.
11614 * optabs.c (prepare_cmp_insn): Likewise.
11615 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11616 * stor-layout.c (finish_bitfield_representative): Likewise.
11618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11619 Alan Hayward <alan.hayward@arm.com>
11620 David Sherwood <david.sherwood@arm.com>
11622 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11623 * expr.c (convert_move): Use them.
11624 (convert_modes): Likewise.
11625 (store_expr_with_bounds): Likewise.
11627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11628 Alan Hayward <alan.hayward@arm.com>
11629 David Sherwood <david.sherwood@arm.com>
11631 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11632 parameter for the mode of "x". Remove the "known_x", "known_mode"
11633 and "known_ret" arguments. Change the type of the mode argument
11634 to scalar_int_mode.
11635 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11636 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11637 (reg_num_sign_bit_copies_for_combine): Likewise.
11638 * rtlanal.c (nonzero_bits1): Likewise.
11639 (num_sign_bit_copies1): Likewise.
11640 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11641 (reg_num_sign_bit_copies_general): Likewise.
11642 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11643 (reg_nonzero_bits_general): Likewise.
11645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11646 Alan Hayward <alan.hayward@arm.com>
11647 David Sherwood <david.sherwood@arm.com>
11649 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11650 than in subroutines. Return 1 for non-integer modes.
11651 (cached_num_sign_bit_copies): Change the type of the mode parameter
11652 to scalar_int_mode.
11653 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11654 classes. Handle CONST_INT_P first and then check whether X also
11655 has a scalar integer mode. Check the same thing for inner registers
11656 of a SUBREG and for values that are being extended or truncated.
11658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11659 Alan Hayward <alan.hayward@arm.com>
11660 David Sherwood <david.sherwood@arm.com>
11662 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11663 in subroutines. Return the mode mask for non-integer modes.
11664 (cached_nonzero_bits): Change the type of the mode parameter
11665 to scalar_int_mode.
11666 (nonzero_bits1): Likewise. Remove early exit for other mode
11667 classes. Handle CONST_INT_P first and then check whether X
11668 also has a scalar integer mode.
11670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11671 Alan Hayward <alan.hayward@arm.com>
11672 David Sherwood <david.sherwood@arm.com>
11674 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11675 Return a scalar_int_mode and assert that the size is greater than
11677 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11678 (op_by_pieces_d::op_by_pieces_d): Likewise.
11679 (op_by_pieces_d::run): Likewise.
11680 (can_store_by_pieces): Likewise.
11682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11683 Alan Hayward <alan.hayward@arm.com>
11684 David Sherwood <david.sherwood@arm.com>
11686 * combine.c (extract_left_shift): Add a mode argument and update
11688 (make_compound_operation_int): Change the type of the mode parameter
11689 to scalar_int_mode and update the call to extract_left_shift.
11691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11692 Alan Hayward <alan.hayward@arm.com>
11693 David Sherwood <david.sherwood@arm.com>
11695 * combine.c (simplify_and_const_int): Change the type of the mode
11696 parameter to scalar_int_mode.
11697 (simplify_and_const_int_1): Likewise. Update recursive call.
11699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11700 Alan Hayward <alan.hayward@arm.com>
11701 David Sherwood <david.sherwood@arm.com>
11703 * combine.c (simplify_compare_const): Check that the mode is a
11704 scalar_int_mode (rather than VOIDmode) before testing its
11706 (simplify_comparison): Move COMPARISON_P handling out of the
11707 loop and restrict the latter part of the loop to scalar_int_modes.
11708 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11709 and when considering SUBREG_REGs. Use is_int_mode instead of
11710 checking GET_MODE_CLASS against MODE_INT.
11712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11713 Alan Hayward <alan.hayward@arm.com>
11714 David Sherwood <david.sherwood@arm.com>
11716 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11717 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11718 shift_unit_mode and for modes involved in scalar shifts.
11720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11721 Alan Hayward <alan.hayward@arm.com>
11722 David Sherwood <david.sherwood@arm.com>
11724 * combine.c (force_int_to_mode): New function, split out from...
11725 (force_to_mode): ...here. Keep xmode up-to-date and use it
11726 instead of GET_MODE (x).
11728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11729 Alan Hayward <alan.hayward@arm.com>
11730 David Sherwood <david.sherwood@arm.com>
11732 * optabs-query.h (extraction_insn::struct_mode): Change type to
11733 opt_scalar_int_mode and update comment.
11734 (extraction_insn::field_mode): Change type to scalar_int_mode.
11735 (extraction_insn::pos_mode): Likewise.
11736 * combine.c (make_extraction): Update accordingly.
11737 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11738 (get_optab_extraction_insn): Likewise.
11739 * recog.c (simplify_while_replacing): Likewise.
11740 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11741 parameter to opt_scalar_int_mode.
11743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11744 Alan Hayward <alan.hayward@arm.com>
11745 David Sherwood <david.sherwood@arm.com>
11747 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11748 to a scalar_int_mode instead of a machine_mode.
11749 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11750 (get_best_mode): Return a boolean and use a pointer argument to store
11751 the selected mode. Replace the limit mode parameter with a bit limit.
11752 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11753 for the values returned by bit_field_mode_iterator::next_mode.
11754 (store_bit_field): Update call to get_best_mode.
11755 (store_fixed_bit_field): Likewise.
11756 (extract_fixed_bit_field): Likewise.
11757 * expr.c (optimize_bitfield_assignment_op): Likewise.
11758 * fold-const.c (optimize_bit_field_compare): Likewise.
11759 (fold_truth_andor_1): Likewise.
11760 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11761 Update for new type of m_mode.
11762 (get_best_mode): As above.
11764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11765 Alan Hayward <alan.hayward@arm.com>
11766 David Sherwood <david.sherwood@arm.com>
11768 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11769 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11770 (store_bit_field): Check is_a <scalar_int_mode> before calling
11771 strict_volatile_bitfield_p.
11772 (extract_bit_field): Likewise.
11774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11775 Alan Hayward <alan.hayward@arm.com>
11776 David Sherwood <david.sherwood@arm.com>
11778 * target.def (cstore_mode): Return a scalar_int_mode.
11779 * doc/tm.texi: Regenerate.
11780 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11781 * targhooks.h (default_cstore_mode): Likewise.
11782 * targhooks.c (default_cstore_mode): Likewise, using a forced
11784 * expmed.c (emit_cstore): Expect the target of the cstore to be
11787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11788 Alan Hayward <alan.hayward@arm.com>
11789 David Sherwood <david.sherwood@arm.com>
11791 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11793 (niter_desc): Likewise mode.
11794 (iv_analyze): Add a mode parameter.
11796 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11797 and change its type to scalar_int_mode.
11798 * loop-iv.c: Update commentary at head of file.
11799 (iv_constant): Pass the mode paraeter before the rtx it describes
11800 and change its type to scalar_int_mode. Remove VOIDmode handling.
11801 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11802 (iv_extend): Likewise.
11803 (shorten_into_mode): Likewise.
11804 (iv_add): Use scalar_int_mode.
11805 (iv_mult): Likewise.
11806 (iv_shift): Likewise.
11807 (canonicalize_iv_subregs): Likewise.
11808 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11809 it describes and change its mode to scalar_int_mode. Also change
11810 the type of the returned inner_mode to scalar_int_mode.
11811 (get_biv_step): Likewise, turning outer_mode from a pointer
11812 into a direct parameter. Update call to get_biv_step_1.
11813 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11814 iv_constant and get_biv_step.
11815 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11816 and change its type to scalar_int_mode. Don't initialise iv->mode
11817 to VOIDmode and remove later checks for its still being VOIDmode.
11818 Update calls to iv_analyze_op and iv_analyze_expr. Check
11819 is_a <scalar_int_mode> when changing the mode under consideration.
11820 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11821 Update call to iv_analyze_expr.
11822 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11823 inner register is not also a scalar_int_mode. Update call to
11825 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11826 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11827 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11828 separate mode class checks. Update calls to iv_analyze. Remove
11829 fix-up of VOIDmodes after iv_analyze_biv.
11830 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11831 don't have a scalar_int_mode. Update call to biv_p.
11833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11834 Alan Hayward <alan.hayward@arm.com>
11835 David Sherwood <david.sherwood@arm.com>
11837 * cfgexpand.c (convert_debug_memory_address): Use
11838 as_a <scalar_int_mode>.
11839 * combine.c (expand_compound_operation): Likewise.
11840 (make_extraction): Likewise.
11841 (change_zero_ext): Likewise.
11842 (simplify_comparison): Likewise.
11843 * cse.c (cse_insn): Likewise.
11844 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11845 (mem_loc_descriptor): Likewise.
11846 (loc_descriptor): Likewise.
11847 * expmed.c (init_expmed_one_mode): Likewise.
11848 (synth_mult): Likewise.
11849 (emit_store_flag_1): Likewise.
11850 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11851 of a comparison with size.
11852 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11853 (reduce_to_bit_field_precision): Likewise.
11854 * function.c (expand_function_end): Likewise.
11855 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11856 * loop-doloop.c (doloop_modify): Likewise.
11857 * optabs.c (expand_binop): Likewise.
11858 (expand_unop): Likewise.
11859 (expand_copysign_absneg): Likewise.
11860 (prepare_cmp_insn): Likewise.
11861 (maybe_legitimize_operand): Likewise.
11862 * recog.c (const_scalar_int_operand): Likewise.
11863 * rtlanal.c (get_address_mode): Likewise.
11864 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11865 (simplify_cond_clz_ctz): Likewise.
11866 * tree-nested.c (get_nl_goto_field): Likewise.
11867 * tree.c (build_vector_type_for_mode): Likewise.
11868 * var-tracking.c (use_narrower_mode): Likewise.
11870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11871 Alan Hayward <alan.hayward@arm.com>
11872 David Sherwood <david.sherwood@arm.com>
11874 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11875 * builtins.c (expand_builtin_signbit): Use it.
11876 * cfgexpand.c (expand_debug_expr): Likewise.
11877 * dojump.c (do_jump): Likewise.
11878 (do_compare_and_jump): Likewise.
11879 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11880 * expmed.c (make_tree): Likewise.
11881 * expr.c (expand_expr_real_2): Likewise.
11882 (expand_expr_real_1): Likewise.
11883 (try_casesi): Likewise.
11884 * fold-const-call.c (fold_const_call_ss): Likewise.
11885 * fold-const.c (unextend): Likewise.
11886 (extract_muldiv_1): Likewise.
11887 (fold_single_bit_test): Likewise.
11888 (native_encode_int): Likewise.
11889 (native_encode_string): Likewise.
11890 (native_interpret_int): Likewise.
11891 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11892 * internal-fn.c (expand_addsub_overflow): Likewise.
11893 (expand_neg_overflow): Likewise.
11894 (expand_mul_overflow): Likewise.
11895 (expand_arith_overflow): Likewise.
11896 * match.pd: Likewise.
11897 * stor-layout.c (layout_type): Likewise.
11898 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11899 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11900 * tree-ssanames.c (get_range_info): Likewise.
11901 * tree-switch-conversion.c (array_value_type) Likewise.
11902 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11903 (vect_recog_divmod_pattern): Likewise.
11904 (vect_recog_mixed_size_cond_pattern): Likewise.
11905 * tree-vrp.c (extract_range_basic): Likewise.
11906 (simplify_float_conversion_using_ranges): Likewise.
11907 * tree.c (int_fits_type_p): Likewise.
11908 * ubsan.c (instrument_bool_enum_load): Likewise.
11909 * varasm.c (mergeable_string_section): Likewise.
11910 (narrowing_initializer_constant_valid_p): Likewise.
11911 (output_constant): Likewise.
11913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11914 Alan Hayward <alan.hayward@arm.com>
11915 David Sherwood <david.sherwood@arm.com>
11917 * machmode.h (NARROWEST_INT_MODE): New macro.
11918 * expr.c (alignment_for_piecewise_move): Use it instead of
11919 GET_CLASS_NARROWEST_MODE (MODE_INT).
11920 (push_block): Likewise.
11921 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11923 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11926 Alan Hayward <alan.hayward@arm.com>
11927 David Sherwood <david.sherwood@arm.com>
11929 * postreload.c (move2add_valid_value_p): Change the type of the
11930 mode parameter to scalar_int_mode.
11931 (move2add_use_add2_insn): Add a mode parameter and use it instead
11933 (move2add_use_add3_insn): Likewise.
11934 (reload_cse_move2add): Update accordingly.
11936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11937 Alan Hayward <alan.hayward@arm.com>
11938 David Sherwood <david.sherwood@arm.com>
11940 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11942 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11943 * optabs.c (expand_unop): Likewise.
11945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11946 Alan Hayward <alan.hayward@arm.com>
11947 David Sherwood <david.sherwood@arm.com>
11949 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11950 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11951 (popcount_loc_descriptor): Likewise.
11952 (bswap_loc_descriptor): Likewise.
11953 (rotate_loc_descriptor): Likewise.
11954 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11955 calling the functions above.
11957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11958 Alan Hayward <alan.hayward@arm.com>
11959 David Sherwood <david.sherwood@arm.com>
11961 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11963 (try_combine): Likewise.
11964 (simplify_if_then_else): Likewise.
11965 * cse.c (cse_insn): Likewise.
11966 * dwarf2out.c (mem_loc_descriptor): Likewise.
11967 * emit-rtl.c (gen_lowpart_common): Likewise.
11968 * simplify-rtx.c (simplify_truncation): Likewise.
11969 (simplify_binary_operation_1): Likewise.
11970 (simplify_const_relational_operation): Likewise.
11971 (simplify_ternary_operation): Likewise.
11972 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11975 Alan Hayward <alan.hayward@arm.com>
11976 David Sherwood <david.sherwood@arm.com>
11978 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11979 * reload.c (push_reload): Likewise.
11980 (find_reloads): Likewise.
11982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11983 Alan Hayward <alan.hayward@arm.com>
11984 David Sherwood <david.sherwood@arm.com>
11986 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11987 (make_compound_operation_int): Likewise.
11988 (change_zero_ext): Likewise.
11989 * expr.c (convert_move): Likewise.
11990 (convert_modes): Likewise.
11991 * fwprop.c (forward_propagate_subreg): Likewise.
11992 * loop-iv.c (get_biv_step_1): Likewise.
11993 * optabs.c (widen_operand): Likewise.
11994 * postreload.c (move2add_valid_value_p): Likewise.
11995 * recog.c (simplify_while_replacing): Likewise.
11996 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11997 (simplify_binary_operation_1): Likewise. Remove redundant
11998 mode equality check.
12000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12001 Alan Hayward <alan.hayward@arm.com>
12002 David Sherwood <david.sherwood@arm.com>
12004 * combine.c (combine_simplify_rtx): Add checks for
12005 is_a <scalar_int_mode>.
12006 (simplify_if_then_else): Likewise.
12007 (make_field_assignment): Likewise.
12008 (simplify_comparison): Likewise.
12009 * ifcvt.c (noce_try_bitop): Likewise.
12010 * loop-invariant.c (canonicalize_address_mult): Likewise.
12011 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12014 Alan Hayward <alan.hayward@arm.com>
12015 David Sherwood <david.sherwood@arm.com>
12017 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12018 is_a <scalar_int_mode> instead of != BLKmode.
12020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12021 Alan Hayward <alan.hayward@arm.com>
12022 David Sherwood <david.sherwood@arm.com>
12024 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12025 instead of != VOIDmode.
12026 * combine.c (if_then_else_cond): Likewise.
12027 (change_zero_ext): Likewise.
12028 * dwarf2out.c (mem_loc_descriptor): Likewise.
12029 (loc_descriptor): Likewise.
12030 * rtlanal.c (canonicalize_condition): Likewise.
12031 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12034 Alan Hayward <alan.hayward@arm.com>
12035 David Sherwood <david.sherwood@arm.com>
12037 * simplify-rtx.c (simplify_binary_operation_1): Use
12038 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12041 Alan Hayward <alan.hayward@arm.com>
12042 David Sherwood <david.sherwood@arm.com>
12044 * wide-int.h (int_traits<unsigned char>) New class.
12045 (int_traits<unsigned short>) Likewise.
12046 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12047 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12049 * combine.c (set_nonzero_bits_and_sign_copies): Use
12050 is_a <scalar_int_mode>.
12051 (find_split_point): Likewise.
12052 (combine_simplify_rtx): Likewise.
12053 (simplify_logical): Likewise.
12054 (expand_compound_operation): Likewise.
12055 (expand_field_assignment): Likewise.
12056 (make_compound_operation): Likewise.
12057 (extended_count): Likewise.
12058 (change_zero_ext): Likewise.
12059 (simplify_comparison): Likewise.
12060 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12061 (ucompare_loc_descriptor): Likewise.
12062 (minmax_loc_descriptor): Likewise.
12063 (mem_loc_descriptor): Likewise.
12064 (loc_descriptor): Likewise.
12065 * expmed.c (init_expmed_one_mode): Likewise.
12066 * lra-constraints.c (lra_constraint_offset): Likewise.
12067 * optabs.c (prepare_libcall_arg): Likewise.
12068 * postreload.c (move2add_note_store): Likewise.
12069 * reload.c (operands_match_p): Likewise.
12070 * rtl.h (load_extend_op): Likewise.
12071 * rtlhooks.c (gen_lowpart_general): Likewise.
12072 * simplify-rtx.c (simplify_truncation): Likewise.
12073 (simplify_unary_operation_1): Likewise.
12074 (simplify_binary_operation_1): Likewise.
12075 (simplify_const_binary_operation): Likewise.
12076 (simplify_const_relational_operation): Likewise.
12077 (simplify_subreg): Likewise.
12078 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12079 * var-tracking.c (adjust_mems): Likewise.
12080 (prepare_call_arguments): Likewise.
12082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12083 Alan Hayward <alan.hayward@arm.com>
12084 David Sherwood <david.sherwood@arm.com>
12086 * machmode.h (is_int_mode): New fuction.
12087 * combine.c (find_split_point): Use it.
12088 (combine_simplify_rtx): Likewise.
12089 (simplify_if_then_else): Likewise.
12090 (simplify_set): Likewise.
12091 (simplify_shift_const_1): Likewise.
12092 (simplify_comparison): Likewise.
12093 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12094 * cse.c (notreg_cost): Likewise.
12095 (cse_insn): Likewise.
12096 * cselib.c (cselib_lookup_1): Likewise.
12097 * dojump.c (do_jump_1): Likewise.
12098 (do_compare_rtx_and_jump): Likewise.
12099 * dse.c (get_call_args): Likewise.
12100 * dwarf2out.c (rtl_for_decl_init): Likewise.
12101 (native_encode_initializer): Likewise.
12102 * expmed.c (emit_store_flag_1): Likewise.
12103 (emit_store_flag): Likewise.
12104 * expr.c (convert_modes): Likewise.
12105 (store_field): Likewise.
12106 (expand_expr_real_1): Likewise.
12107 * fold-const.c (fold_read_from_constant_string): Likewise.
12108 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12109 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12110 * optabs.c (expand_binop): Likewise.
12111 (expand_unop): Likewise.
12112 (expand_abs_nojump): Likewise.
12113 (expand_one_cmpl_abs_nojump): Likewise.
12114 * simplify-rtx.c (mode_signbit_p): Likewise.
12115 (val_signbit_p): Likewise.
12116 (val_signbit_known_set_p): Likewise.
12117 (val_signbit_known_clear_p): Likewise.
12118 (simplify_relational_operation_1): Likewise.
12119 * tree.c (vector_type_mode): Likewise.
12121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12122 Alan Hayward <alan.hayward@arm.com>
12123 David Sherwood <david.sherwood@arm.com>
12125 * machmode.h (smallest_mode_for_size): Fix formatting.
12126 (smallest_int_mode_for_size): New function.
12127 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12128 instead of smallest_mode_for_size.
12129 * combine.c (make_extraction): Likewise.
12130 * config/arc/arc.c (arc_expand_movmem): Likewise.
12131 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12132 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12133 * config/s390/s390.c (s390_expand_insv): Likewise.
12134 * config/sparc/sparc.c (assign_int_registers): Likewise.
12135 * config/spu/spu.c (spu_function_value): Likewise.
12136 (spu_function_arg): Likewise.
12137 * coverage.c (get_gcov_type): Likewise.
12138 (get_gcov_unsigned_t): Likewise.
12139 * dse.c (find_shift_sequence): Likewise.
12140 * expmed.c (store_bit_field_1): Likewise.
12141 * expr.c (convert_move): Likewise.
12142 (store_field): Likewise.
12143 * internal-fn.c (expand_arith_overflow): Likewise.
12144 * optabs-query.c (get_best_extraction_insn): Likewise.
12145 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12146 * stor-layout.c (layout_type): Likewise.
12147 (initialize_sizetypes): Likewise.
12148 * targhooks.c (default_get_mask_mode): Likewise.
12149 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12152 Alan Hayward <alan.hayward@arm.com>
12153 David Sherwood <david.sherwood@arm.com>
12155 * machmode.h (opt_mode::else_blk): New function.
12156 (int_mode_for_mode): Declare.
12157 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12158 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12160 * cfgexpand.c (expand_debug_expr): Likewise.
12161 * combine.c (gen_lowpart_or_truncate): Likewise.
12162 (gen_lowpart_for_combine): Likewise.
12163 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12164 * config/avr/avr.c (avr_to_int_mode): Likewise.
12165 (avr_out_plus_1): Likewise.
12166 (avr_out_plus): Likewise.
12167 (avr_out_round): Likewise.
12168 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12169 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12170 (s390_expand_vcond): Likewise.
12171 * config/spu/spu.c (spu_split_immediate): Likewise.
12172 (spu_expand_mov): Likewise.
12173 * dse.c (get_stored_val): Likewise.
12174 * expmed.c (store_bit_field_1): Likewise.
12175 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12177 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12178 (extract_low_bits): Likewise.
12179 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12180 handling rather than repeating the check.
12181 (emit_group_store): Likewise.
12182 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12183 * optabs.c (expand_absneg_bit): Likewise.
12184 (expand_copysign_absneg): Likewise.
12185 (expand_copysign_bit): Likewise.
12186 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12187 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12188 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12189 * var-tracking.c (prepare_call_arguments): Likewise.
12190 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12191 int_mode_for_mode instead of mode_for_size.
12192 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
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 * machmode.h (int_mode_for_size): New function.
12199 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12200 instead of mode_for_size.
12201 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12203 * combine.c (expand_field_assignment): Use int_mode_for_size
12204 instead of mode_for_size.
12205 (make_extraction): Likewise.
12206 (simplify_shift_const_1): Likewise.
12207 (simplify_comparison): Likewise.
12208 * dojump.c (do_jump): Likewise.
12209 * dwarf2out.c (mem_loc_descriptor): Likewise.
12210 * emit-rtl.c (init_derived_machine_modes): Likewise.
12211 * expmed.c (flip_storage_order): Likewise.
12212 (convert_extracted_bit_field): Likewise.
12213 * expr.c (copy_blkmode_from_reg): Likewise.
12214 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12215 * internal-fn.c (expand_mul_overflow): Likewise.
12216 * lower-subreg.c (simple_move): Likewise.
12217 * optabs-libfuncs.c (init_optabs): Likewise.
12218 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12219 * tree.c (vector_type_mode): Likewise.
12220 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12221 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12222 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12223 * tree-vect-stmts.c (vectorizable_load): Likewise.
12224 (vectorizable_store): Likewise.
12226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12227 Alan Hayward <alan.hayward@arm.com>
12228 David Sherwood <david.sherwood@arm.com>
12230 * coretypes.h (pod_mode): New type.
12231 (scalar_int_mode_pod): New typedef.
12232 * machmode.h (pod_mode): New class.
12233 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12234 * genmodes.c (emit_mode_int_n): Update accordingly.
12235 * lower-subreg.h (target_lower_subreg): Change type to
12236 scalar_int_mode_pod.
12237 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12238 scalar_int_mode_pod.
12240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12241 Alan Hayward <alan.hayward@arm.com>
12242 David Sherwood <david.sherwood@arm.com>
12244 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12245 machine_mode to scalar_int_mode.
12246 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12247 (rs6000_option_override_internal): Remove cast to int.
12248 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12249 machine_mode to scalar_int_mode.
12250 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12251 (rs6000_option_override_internal): Remove cast to int.
12252 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12253 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12255 * config/s390/s390.c (s390_expand_builtin): Likewise.
12256 * coretypes.h (scalar_int_mode): New type.
12257 (opt_scalar_int_mode): New typedef.
12258 * machmode.h (scalar_int_mode): New class.
12259 (scalar_int_mode::includes_p): New function.
12260 (byte_mode): Change type to scalar_int_mode.
12261 (word_mode): Likewise.
12262 (ptr_mode): Likewise.
12263 * emit-rtl.c (byte_mode): Likewise.
12264 (word_mode): Likewise.
12265 (ptr_mode): Likewise.
12266 (init_derived_machine_modes): Update accordingly.
12267 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12268 and MODE_PARTIAL_INT.
12269 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12270 opt_scalar_int_mode.
12272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12273 Alan Hayward <alan.hayward@arm.com>
12274 David Sherwood <david.sherwood@arm.com>
12276 * target.def (libgcc_floating_mode_supported_p): Take a
12278 * doc/tm.texi: Regenerate.
12279 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12281 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12282 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12286 Alan Hayward <alan.hayward@arm.com>
12287 David Sherwood <david.sherwood@arm.com>
12289 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12290 * doc/tm.texi: Regenerate.
12291 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12292 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12293 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12294 * targhooks.h (default_floatn_mode): Likewise.
12295 * targhooks.c (default_floatn_mode): Likewise.
12296 * tree.c (build_common_tree_nodes): Update accordingly.
12298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12299 Alan Hayward <alan.hayward@arm.com>
12300 David Sherwood <david.sherwood@arm.com>
12302 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12303 (mode_iterator::iterate_p): Likewise.
12304 (mode_iterator::get_wider): Likewise.
12305 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12308 Alan Hayward <alan.hayward@arm.com>
12309 David Sherwood <david.sherwood@arm.com>
12311 * coretypes.h (opt_scalar_float_mode): New typedef.
12312 * machmode.h (float_mode_for_size): New function.
12313 * emit-rtl.c (double_mode): Delete.
12314 (init_emit_once): Use float_mode_for_size.
12315 * stor-layout.c (layout_type): Likewise.
12316 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12319 Alan Hayward <alan.hayward@arm.com>
12320 David Sherwood <david.sherwood@arm.com>
12322 * output.h (assemble_real): Take a scalar_float_mode.
12323 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12324 * config/arm/arm.md (consttable_4): Likewise.
12325 (consttable_8): Likewise.
12326 (consttable_16): Likewise.
12327 * config/mips/mips.md (consttable_float): Likewise.
12328 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12329 * varasm.c (assemble_real): Take a scalar_float_mode.
12330 (output_constant_pool_2): Update accordingly.
12331 (output_constant): Likewise.
12333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12334 Alan Hayward <alan.hayward@arm.com>
12335 David Sherwood <david.sherwood@arm.com>
12337 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12338 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12339 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12340 (native_encode_real): Likewise.
12341 (native_interpret_real): Likewise.
12342 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12343 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12346 Alan Hayward <alan.hayward@arm.com>
12347 David Sherwood <david.sherwood@arm.com>
12349 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12350 <scalar_float_mode>. Simplify.
12351 (gen_extend_conv_libfunc): Likewise.
12353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12354 Alan Hayward <alan.hayward@arm.com>
12355 David Sherwood <david.sherwood@arm.com>
12357 * coretypes.h (scalar_float_mode): New type.
12358 * machmode.h (mode_traits::from_int): Use machine_mode if
12359 USE_ENUM_MODES is defined.
12360 (is_a): New function.
12362 (dyn_cast): Likewise.
12363 (scalar_float_mode): New class.
12364 (scalar_float_mode::includes_p): New function.
12365 (is_float_mode): Likewise.
12366 * gdbhooks.py (MachineModePrinter): New class.
12367 (build_pretty_printer): Use it for scalar_float_mode.
12368 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12369 (format_helper::format_helper): Turn into a template.
12370 * genmodes.c (get_mode_class): New function.
12371 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12372 or machine_mode if none.
12373 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12374 as_a <scalar_float_mode>.
12375 * dwarf2out.c (mem_loc_descriptor): Likewise.
12376 (insert_float): Likewise.
12377 (add_const_value_attribute): Likewise.
12378 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12379 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12380 (expand_unop): Update accordingly.
12381 (expand_abs_nojump): Likewise.
12382 (expand_copysign_absneg): Take a scalar_float_mode.
12383 (expand_copysign_bit): Likewise.
12384 (expand_copysign): Update accordingly.
12386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12387 Alan Hayward <alan.hayward@arm.com>
12388 David Sherwood <david.sherwood@arm.com>
12390 * coretypes.h (opt_mode): New class.
12391 * machmode.h (opt_mode): Likewise.
12392 (opt_mode::else_void): New function.
12393 (opt_mode::require): Likewise.
12394 (opt_mode::exists): Likewise.
12395 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12396 (GET_MODE_2XWIDER_MODE): Likewise.
12397 (mode_iterator::get_wider): Update accordingly.
12398 (mode_iterator::get_2xwider): Likewise.
12399 (mode_iterator::get_known_wider): Likewise, turning into a template.
12400 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12401 forcing a wider mode to exist.
12402 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12403 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12404 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12405 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12406 * lower-subreg.c (init_lower_subreg): Likewise.
12407 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12408 on the final iteration.
12409 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12410 a wider mode exists before asking for a move pattern.
12411 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12412 forcing a wider mode to exist.
12413 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12414 returning false if no such mode exists.
12415 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12416 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12417 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12418 Avoid checking for a MODE_INT if we already know the mode is not a
12420 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12421 forcing a wider mode to exist.
12422 (expmed_mult_highpart_optab): Likewise.
12423 (expmed_mult_highpart): Likewise.
12424 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12426 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12427 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12428 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12429 * internal-fn.c (expand_mul_overflow): Update use of
12430 GET_MODE_2XWIDER_MODE.
12431 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12432 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12433 GET_MODE_WIDER_MODE.
12434 (convert_plusminus_to_widen): Likewise.
12435 * tree-switch-conversion.c (array_value_type): Likewise.
12436 * var-tracking.c (emit_note_insn_var_location): Likewise.
12437 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12438 Return false inside rather than outside the loop if no wider mode
12440 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12441 and GET_MODE_2XWIDER_MODE
12442 (can_compare_p): Use else_void.
12443 * gdbhooks.py (OptMachineModePrinter): New class.
12444 (build_pretty_printer): Use it for opt_mode.
12446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12447 Alan Hayward <alan.hayward@arm.com>
12448 David Sherwood <david.sherwood@arm.com>
12450 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12451 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12454 Alan Hayward <alan.hayward@arm.com>
12455 David Sherwood <david.sherwood@arm.com>
12457 * machmode.h (mode_traits): New structure.
12458 (get_narrowest_mode): New function.
12459 (mode_iterator::start): Likewise.
12460 (mode_iterator::iterate_p): Likewise.
12461 (mode_iterator::get_wider): Likewise.
12462 (mode_iterator::get_known_wider): Likewise.
12463 (mode_iterator::get_2xwider): Likewise.
12464 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12465 (FOR_EACH_MODE): Likewise.
12466 (FOR_EACH_MODE_FROM): Likewise.
12467 (FOR_EACH_MODE_UNTIL): Likewise.
12468 (FOR_EACH_WIDER_MODE): Likewise.
12469 (FOR_EACH_2XWIDER_MODE): Likewise.
12470 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12471 * combine.c (simplify_comparison): Likewise
12472 * config/i386/i386.c (type_natural_mode): Likewise.
12473 * cse.c (cse_insn): Likewise.
12474 * dse.c (find_shift_sequence): Likewise.
12475 * emit-rtl.c (init_derived_machine_modes): Likewise.
12476 (init_emit_once): Likewise.
12477 * explow.c (hard_function_value): Likewise.
12478 * expmed.c (extract_fixed_bit_field_1): Likewise.
12479 (extract_bit_field_1): Likewise.
12480 (expand_divmod): Likewise.
12481 (emit_store_flag_1): Likewise.
12482 * expr.c (init_expr_target): Likewise.
12483 (convert_move): Likewise.
12484 (alignment_for_piecewise_move): Likewise.
12485 (widest_int_mode_for_size): Likewise.
12486 (emit_block_move_via_movmem): Likewise.
12487 (copy_blkmode_to_reg): Likewise.
12488 (set_storage_via_setmem): Likewise.
12489 (compress_float_constant): Likewise.
12490 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12491 * optabs-query.c (get_best_extraction_insn): Likewise.
12492 * optabs.c (expand_binop): Likewise.
12493 (expand_twoval_unop): Likewise.
12494 (expand_twoval_binop): Likewise.
12495 (widen_leading): Likewise.
12496 (widen_bswap): Likewise.
12497 (expand_parity): Likewise.
12498 (expand_unop): Likewise.
12499 (prepare_cmp_insn): Likewise.
12500 (prepare_float_lib_cmp): Likewise.
12501 (expand_float): Likewise.
12502 (expand_fix): Likewise.
12503 (expand_sfix_optab): Likewise.
12504 * postreload.c (move2add_use_add2_insn): Likewise.
12505 * reg-stack.c (reg_to_stack): Likewise.
12506 * reginfo.c (choose_hard_reg_mode): Likewise.
12507 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12508 * stor-layout.c (mode_for_size): Likewise.
12509 (smallest_mode_for_size): Likewise.
12510 (mode_for_vector): Likewise.
12511 (finish_bitfield_representative): Likewise.
12512 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12513 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12514 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12515 * var-tracking.c (prepare_call_arguments): Likewise.
12517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12518 Alan Hayward <alan.hayward@arm.com>
12519 David Sherwood <david.sherwood@arm.com>
12521 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12522 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12523 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12524 * machmode.h (mode_size): Move earlier in file.
12525 (mode_precision): Likewise.
12526 (mode_inner): Likewise.
12527 (mode_nunits): Likewise.
12528 (mode_unit_size): Likewise.
12529 (unit_unit_precision): Likewise.
12530 (mode_wider): Likewise.
12531 (mode_2xwider): Likewise.
12532 (machine_mode): New class.
12533 (mode_to_bytes): New function.
12534 (mode_to_bits): Likewise.
12535 (mode_to_precision): Likewise.
12536 (mode_to_inner): Likewise.
12537 (mode_to_unit_size): Likewise.
12538 (mode_to_unit_precision): Likewise.
12539 (mode_to_nunits): Likewise.
12540 (GET_MODE_SIZE): Use mode_to_bytes.
12541 (GET_MODE_BITSIZE): Use mode_to_bits.
12542 (GET_MODE_PRECISION): Use mode_to_precision.
12543 (GET_MODE_INNER): Use mode_to_inner.
12544 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12545 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12546 (GET_MODE_NUNITS): Use mode_to_nunits.
12547 * system.h (ALWAYS_INLINE): New macro.
12548 * config/powerpcspe/powerpcspe-c.c
12549 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12550 int for arg1_mode and arg2_mode.
12552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12553 Alan Hayward <alan.hayward@arm.com>
12554 David Sherwood <david.sherwood@arm.com>
12556 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12557 Prefix mode names with E_ in case statements.
12558 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12559 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12560 (aarch64_split_simd_move): Likewise.
12561 (aarch64_gen_storewb_pair): Likewise.
12562 (aarch64_gen_loadwb_pair): Likewise.
12563 (aarch64_gen_store_pair): Likewise.
12564 (aarch64_gen_load_pair): Likewise.
12565 (aarch64_get_condition_code_1): Likewise.
12566 (aarch64_constant_pool_reload_icode): Likewise.
12567 (get_rsqrte_type): Likewise.
12568 (get_rsqrts_type): Likewise.
12569 (get_recpe_type): Likewise.
12570 (get_recps_type): Likewise.
12571 (aarch64_gimplify_va_arg_expr): Likewise.
12572 (aarch64_simd_container_mode): Likewise.
12573 (aarch64_emit_load_exclusive): Likewise.
12574 (aarch64_emit_store_exclusive): Likewise.
12575 (aarch64_expand_compare_and_swap): Likewise.
12576 (aarch64_gen_atomic_cas): Likewise.
12577 (aarch64_emit_bic): Likewise.
12578 (aarch64_emit_atomic_swap): Likewise.
12579 (aarch64_emit_atomic_load_op): Likewise.
12580 (aarch64_evpc_trn): Likewise.
12581 (aarch64_evpc_uzp): Likewise.
12582 (aarch64_evpc_zip): Likewise.
12583 (aarch64_evpc_ext): Likewise.
12584 (aarch64_evpc_rev): Likewise.
12585 (aarch64_evpc_dup): Likewise.
12586 (aarch64_gen_ccmp_first): Likewise.
12587 (aarch64_gen_ccmp_next): Likewise.
12588 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12589 (alpha_emit_xfloating_libcall): Likewise.
12590 (emit_insxl): Likewise.
12591 (alpha_arg_type): Likewise.
12592 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12593 (arc_preferred_simd_mode): Likewise.
12594 (arc_secondary_reload): Likewise.
12595 (get_arc_condition_code): Likewise.
12596 (arc_print_operand): Likewise.
12597 (arc_legitimate_constant_p): Likewise.
12598 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12599 * config/arc/arc.md (casesi_load): Likewise.
12600 (casesi_compact_jump): Likewise.
12601 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12602 (cc_use_register): Likewise.
12603 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12604 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12605 (arm_init_iwmmxt_builtins): Likewise.
12606 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12607 (neon_expand_vector_init): Likewise.
12608 (arm_attr_length_move_neon): Likewise.
12609 (maybe_get_arm_condition_code): Likewise.
12610 (arm_emit_vector_const): Likewise.
12611 (arm_preferred_simd_mode): Likewise.
12612 (arm_output_iwmmxt_tinsr): Likewise.
12613 (thumb1_output_casesi): Likewise.
12614 (thumb2_output_casesi): Likewise.
12615 (arm_emit_load_exclusive): Likewise.
12616 (arm_emit_store_exclusive): Likewise.
12617 (arm_expand_compare_and_swap): Likewise.
12618 (arm_evpc_neon_vuzp): Likewise.
12619 (arm_evpc_neon_vzip): Likewise.
12620 (arm_evpc_neon_vrev): Likewise.
12621 (arm_evpc_neon_vtrn): Likewise.
12622 (arm_evpc_neon_vext): Likewise.
12623 (arm_validize_comparison): Likewise.
12624 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12625 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12626 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12627 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12628 (c6x_preferred_simd_mode): Likewise.
12629 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12630 (epiphany_rtx_costs): Likewise.
12631 * config/epiphany/predicates.md (proper_comparison_operator):
12633 * config/frv/frv.c (condexec_memory_operand): Likewise.
12634 (frv_emit_move): Likewise.
12635 (output_move_single): Likewise.
12636 (output_condmove_single): Likewise.
12637 (frv_hard_regno_mode_ok): Likewise.
12638 (frv_matching_accg_mode): Likewise.
12639 * config/h8300/h8300.c (split_adds_subs): Likewise.
12640 (h8300_rtx_costs): Likewise.
12641 (h8300_print_operand): Likewise.
12642 (compute_mov_length): Likewise.
12643 (output_logical_op): Likewise.
12644 (compute_logical_op_length): Likewise.
12645 (compute_logical_op_cc): Likewise.
12646 (h8300_shift_needs_scratch_p): Likewise.
12647 (output_a_shift): Likewise.
12648 (compute_a_shift_length): Likewise.
12649 (compute_a_shift_cc): Likewise.
12650 (expand_a_rotate): Likewise.
12651 (output_a_rotate): Likewise.
12652 * config/i386/i386.c (classify_argument): Likewise.
12653 (function_arg_advance_32): Likewise.
12654 (function_arg_32): Likewise.
12655 (function_arg_64): Likewise.
12656 (function_value_64): Likewise.
12657 (ix86_gimplify_va_arg): Likewise.
12658 (ix86_legitimate_constant_p): Likewise.
12659 (put_condition_code): Likewise.
12660 (split_double_mode): Likewise.
12661 (ix86_avx256_split_vector_move_misalign): Likewise.
12662 (ix86_expand_vector_logical_operator): Likewise.
12663 (ix86_split_idivmod): Likewise.
12664 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12665 (ix86_build_const_vector): Likewise.
12666 (ix86_build_signbit_mask): Likewise.
12667 (ix86_match_ccmode): Likewise.
12668 (ix86_cc_modes_compatible): Likewise.
12669 (ix86_expand_branch): Likewise.
12670 (ix86_expand_sse_cmp): Likewise.
12671 (ix86_expand_sse_movcc): Likewise.
12672 (ix86_expand_int_sse_cmp): Likewise.
12673 (ix86_expand_vec_perm_vpermi2): Likewise.
12674 (ix86_expand_vec_perm): Likewise.
12675 (ix86_expand_sse_unpack): Likewise.
12676 (ix86_expand_int_addcc): Likewise.
12677 (ix86_split_to_parts): Likewise.
12678 (ix86_vectorize_builtin_gather): Likewise.
12679 (ix86_vectorize_builtin_scatter): Likewise.
12680 (avx_vpermilp_parallel): Likewise.
12681 (inline_memory_move_cost): Likewise.
12682 (ix86_tieable_integer_mode_p): Likewise.
12683 (x86_maybe_negate_const_int): Likewise.
12684 (ix86_expand_vector_init_duplicate): Likewise.
12685 (ix86_expand_vector_init_one_nonzero): Likewise.
12686 (ix86_expand_vector_init_one_var): Likewise.
12687 (ix86_expand_vector_init_concat): Likewise.
12688 (ix86_expand_vector_init_interleave): Likewise.
12689 (ix86_expand_vector_init_general): Likewise.
12690 (ix86_expand_vector_set): Likewise.
12691 (ix86_expand_vector_extract): Likewise.
12692 (emit_reduc_half): Likewise.
12693 (ix86_emit_i387_round): Likewise.
12694 (ix86_mangle_type): Likewise.
12695 (ix86_expand_round_sse4): Likewise.
12696 (expand_vec_perm_blend): Likewise.
12697 (canonicalize_vector_int_perm): Likewise.
12698 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12699 (expand_vec_perm_1): Likewise.
12700 (expand_vec_perm_interleave3): Likewise.
12701 (expand_vec_perm_even_odd_pack): Likewise.
12702 (expand_vec_perm_even_odd_1): Likewise.
12703 (expand_vec_perm_broadcast_1): Likewise.
12704 (ix86_vectorize_vec_perm_const_ok): Likewise.
12705 (ix86_expand_vecop_qihi): Likewise.
12706 (ix86_expand_mul_widen_hilo): Likewise.
12707 (ix86_expand_sse2_abs): Likewise.
12708 (ix86_expand_pextr): Likewise.
12709 (ix86_expand_pinsr): Likewise.
12710 (ix86_preferred_simd_mode): Likewise.
12711 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12712 * config/i386/sse.md (*andnot<mode>3): Likewise.
12713 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12714 (*<code><mode>3): Likewise.
12715 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12716 (ia64_expand_atomic_op): Likewise.
12717 (ia64_arg_type): Likewise.
12718 (ia64_mode_to_int): Likewise.
12719 (ia64_scalar_mode_supported_p): Likewise.
12720 (ia64_vector_mode_supported_p): Likewise.
12721 (expand_vec_perm_broadcast): Likewise.
12722 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12723 (iq2000_function_arg_advance): Likewise.
12724 (iq2000_function_arg): Likewise.
12725 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12726 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12727 (m68k_libcall_value): Likewise.
12728 (m68k_function_value): Likewise.
12729 (sched_attr_op_type): Likewise.
12730 * config/mcore/mcore.c (mcore_output_move): Likewise.
12731 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12733 (microblaze_function_arg): Likewise.
12734 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12735 (mips_print_operand): Likewise.
12736 (mips_mode_ok_for_mov_fmt_p): Likewise.
12737 (mips_vector_mode_supported_p): Likewise.
12738 (mips_preferred_simd_mode): Likewise.
12739 (mips_expand_vpc_loongson_even_odd): Likewise.
12740 (mips_expand_vec_unpack): Likewise.
12741 (mips_expand_vi_broadcast): Likewise.
12742 (mips_expand_vector_init): Likewise.
12743 (mips_expand_vec_reduc): Likewise.
12744 (mips_expand_msa_cmp): Likewise.
12745 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12746 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12747 (cc_flags_for_mode): Likewise.
12748 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12749 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12750 (nds32_output_casesi_pc_relative): Likewise.
12751 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12752 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12753 (nvptx_gen_unpack): Likewise.
12754 (nvptx_gen_pack): Likewise.
12755 (nvptx_gen_shuffle): Likewise.
12756 (nvptx_gen_wcast): Likewise.
12757 (nvptx_preferred_simd_mode): Likewise.
12758 * config/pa/pa.c (pa_secondary_reload): Likewise.
12759 * config/pa/predicates.md (base14_operand): Likewise.
12760 * config/powerpcspe/powerpcspe-c.c
12761 (altivec_resolve_overloaded_builtin): Likewise.
12762 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12764 (rs6000_preferred_simd_mode): Likewise.
12765 (output_vec_const_move): Likewise.
12766 (rs6000_expand_vector_extract): Likewise.
12767 (rs6000_split_vec_extract_var): Likewise.
12768 (reg_offset_addressing_ok_p): Likewise.
12769 (rs6000_legitimate_offset_address_p): Likewise.
12770 (rs6000_legitimize_address): Likewise.
12771 (rs6000_emit_set_const): Likewise.
12772 (rs6000_const_vec): Likewise.
12773 (rs6000_emit_move): Likewise.
12774 (spe_build_register_parallel): Likewise.
12775 (rs6000_darwin64_record_arg_recurse): Likewise.
12776 (swap_selector_for_mode): Likewise.
12777 (spe_init_builtins): Likewise.
12778 (paired_init_builtins): Likewise.
12779 (altivec_init_builtins): Likewise.
12780 (do_load_for_compare): Likewise.
12781 (rs6000_generate_compare): Likewise.
12782 (rs6000_expand_float128_convert): Likewise.
12783 (emit_load_locked): Likewise.
12784 (emit_store_conditional): Likewise.
12785 (rs6000_output_function_epilogue): Likewise.
12786 (rs6000_handle_altivec_attribute): Likewise.
12787 (rs6000_function_value): Likewise.
12788 (emit_fusion_gpr_load): Likewise.
12789 (emit_fusion_p9_load): Likewise.
12790 (emit_fusion_p9_store): Likewise.
12791 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12792 (fusion_gpr_mem_load): Likewise.
12793 (fusion_addis_mem_combo_load): Likewise.
12794 (fusion_addis_mem_combo_store): Likewise.
12795 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12796 (fusion_gpr_mem_load): Likewise.
12797 (fusion_addis_mem_combo_load): Likewise.
12798 (fusion_addis_mem_combo_store): Likewise.
12799 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12801 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12802 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12803 (rs6000_preferred_simd_mode): Likewise.
12804 (output_vec_const_move): Likewise.
12805 (rs6000_expand_vector_extract): Likewise.
12806 (rs6000_split_vec_extract_var): Likewise.
12807 (reg_offset_addressing_ok_p): Likewise.
12808 (rs6000_legitimate_offset_address_p): Likewise.
12809 (rs6000_legitimize_address): Likewise.
12810 (rs6000_emit_set_const): Likewise.
12811 (rs6000_const_vec): Likewise.
12812 (rs6000_emit_move): Likewise.
12813 (rs6000_darwin64_record_arg_recurse): Likewise.
12814 (swap_selector_for_mode): Likewise.
12815 (paired_init_builtins): Likewise.
12816 (altivec_init_builtins): Likewise.
12817 (rs6000_expand_float128_convert): Likewise.
12818 (emit_load_locked): Likewise.
12819 (emit_store_conditional): Likewise.
12820 (rs6000_output_function_epilogue): Likewise.
12821 (rs6000_handle_altivec_attribute): Likewise.
12822 (rs6000_function_value): Likewise.
12823 (emit_fusion_gpr_load): Likewise.
12824 (emit_fusion_p9_load): Likewise.
12825 (emit_fusion_p9_store): Likewise.
12826 * config/rx/rx.c (rx_gen_move_template): Likewise.
12827 (flags_from_mode): Likewise.
12828 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12829 (s390_slb_comparison): Likewise.
12830 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12831 (s390_vector_mode_supported_p): Likewise.
12832 (s390_cc_modes_compatible): Likewise.
12833 (s390_match_ccmode_set): Likewise.
12834 (s390_canonicalize_comparison): Likewise.
12835 (s390_emit_compare_and_swap): Likewise.
12836 (s390_branch_condition_mask): Likewise.
12837 (s390_rtx_costs): Likewise.
12838 (s390_secondary_reload): Likewise.
12839 (__SECONDARY_RELOAD_CASE): Likewise.
12840 (s390_expand_cs): Likewise.
12841 (s390_preferred_simd_mode): Likewise.
12842 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12843 * config/sh/sh.c (sh_print_operand): Likewise.
12844 (dump_table): Likewise.
12845 (sh_secondary_reload): Likewise.
12846 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12847 * config/sh/sh.md (casesi_worker_1): Likewise.
12848 (casesi_worker_2): Likewise.
12849 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12850 (fcc_comparison_operator): Likewise.
12851 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12852 (emit_soft_tfmode_cvt): Likewise.
12853 (sparc_preferred_simd_mode): Likewise.
12854 (output_cbranch): Likewise.
12855 (sparc_print_operand): Likewise.
12856 (sparc_expand_vec_perm_bmask): Likewise.
12857 (vector_init_bshuffle): Likewise.
12858 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12859 (spu_vector_mode_supported_p): Likewise.
12860 (spu_expand_insv): Likewise.
12861 (spu_emit_branch_or_set): Likewise.
12862 (spu_handle_vector_attribute): Likewise.
12863 (spu_builtin_splats): Likewise.
12864 (spu_builtin_extract): Likewise.
12865 (spu_builtin_promote): Likewise.
12866 (spu_expand_sign_extend): Likewise.
12867 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12868 (tilegx_simd_int): Likewise.
12869 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12870 (tilepro_simd_int): Likewise.
12871 * config/v850/v850.c (const_double_split): Likewise.
12872 (v850_print_operand): Likewise.
12873 (ep_memory_offset): Likewise.
12874 * config/vax/vax.c (vax_rtx_costs): Likewise.
12875 (vax_output_int_move): Likewise.
12876 (vax_output_int_add): Likewise.
12877 (vax_output_int_subtract): Likewise.
12878 * config/visium/predicates.md (visium_branch_operator): Likewise.
12879 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12880 (visium_print_operand_address): Likewise.
12881 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12882 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12883 (xtensa_expand_conditional_branch): Likewise.
12884 (xtensa_copy_incoming_a7): Likewise.
12885 (xtensa_output_literal): Likewise.
12886 * dfp.c (decimal_real_maxval): Likewise.
12887 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12890 Alan Hayward <alan.hayward@arm.com>
12891 David Sherwood <david.sherwood@arm.com>
12893 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12894 (mode_nunits_inline): Likewise.
12895 (mode_inner_inline): Likewise.
12896 (mode_unit_size_inline): Likewise.
12897 (mode_unit_precision_inline): Likewise.
12898 (emit_insn_modes_h): Likewise. Also emit a #define of the
12900 (emit_mode_wider): Add an E_ prefix to mode names.
12901 (emit_mode_complex): Likewise.
12902 (emit_mode_inner): Likewise.
12903 (emit_mode_adjustments): Likewise.
12904 (emit_mode_int_n): Likewise.
12905 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12906 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12907 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12908 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12909 (CRC32_BUILTIN, ENTRY): Likewise.
12910 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12911 (aarch64_pop_regs): Likewise.
12912 (aarch64_process_components): Likewise.
12913 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12914 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12915 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12916 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12917 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12918 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12920 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12921 (pcmpistr): Likewise.
12922 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12923 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12924 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12926 * config/rl78/rl78.c (mduc_regs): Likewise.
12927 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12928 (htm_expand_builtin): Likewise.
12929 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12930 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12931 * config/xtensa/xtensa.c (print_operand): Likewise.
12932 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12933 (NUM_MODE_VECTOR_INT): Likewise.
12934 * genoutput.c (null_operand): Likewise.
12935 (output_operand_data): Likewise.
12936 * genrecog.c (print_parameter_value): Likewise.
12937 * lra.c (debug_operand_data): Likewise.
12939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12941 * dwarf2out.c (scompare_loc_descriptor_wide)
12942 (scompare_loc_descriptor_narrow): New functions, split out from...
12943 (scompare_loc_descriptor): ...here.
12944 * expmed.c (emit_store_flag_int): New function, split out from...
12945 (emit_store_flag): ...here.
12947 2017-08-30 Richard Biener <rguenther@suse.de>
12949 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12950 (dwarf2out_early_finish): Move setting of AT_pubnames from
12951 early debug output to early finish.
12953 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12955 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12956 and -mdata-region to the assembler.
12958 2017-08-30 Richard Biener <rguenther@suse.de>
12960 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12962 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12963 (dwarf2out_early_global_decl): Only generate a DIE for the
12964 abstract origin if it doesn't already exist or is a declaration DIE.
12965 (resolve_addr): Do not add the linkage name twice when
12966 generating a stub DIE for the DW_TAG_GNU_call_site target.
12968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12970 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12971 Use machine_mode rather than int for arg1_mode.
12973 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12976 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12977 that the second argument of the built-in functions to unpack
12978 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12979 switch statement instead a lot of if statements.
12980 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12981 Allow 64-bit values to be in Altivec registers as well as
12982 traditional floating point registers.
12983 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12985 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12987 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12988 MAX_REGS_PER_ADDRESS == 1.
12990 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12992 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12993 * config/i386/i386.c (ix86_option_override_internal): Simplify
12994 setting of opts->x_flag_entry.
12996 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12997 Jakub Jelinek <jakub@redhat.com>
12998 Richard Biener <rguenther@suse.de>
13000 PR tree-optimization/81503
13001 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13002 folded constant fits in the target type; reorder tests for clarity.
13004 2017-08-29 Martin Liska <mliska@suse.cz>
13006 * passes.def: Include pass_lower_switch.
13007 * stmt.c (dump_case_nodes): Remove and move to
13008 tree-switch-conversion.
13009 (case_values_threshold): Likewise.
13010 (expand_switch_as_decision_tree_p): Likewise.
13011 (emit_case_decision_tree): Likewise.
13012 (expand_case): Likewise.
13013 (balance_case_nodes): Likewise.
13014 (node_has_low_bound): Likewise.
13015 (node_has_high_bound): Likewise.
13016 (node_is_bounded): Likewise.
13017 (emit_case_nodes): Likewise.
13018 (struct simple_case_node): New struct.
13019 (add_case_node): Remove.
13020 (emit_case_dispatch_table): Use vector instead of case_list.
13021 (reset_out_edges_aux): Remove.
13022 (compute_cases_per_edge): Likewise.
13023 (expand_case): Build list of simple_case_node.
13024 (expand_sjlj_dispatch_table): Use it.
13025 * tree-switch-conversion.c (struct case_node): Moved from
13026 stmt.c and adjusted.
13027 (emit_case_nodes): Likewise.
13028 (node_has_low_bound): Likewise.
13029 (node_has_high_bound): Likewise.
13030 (node_is_bounded): Likewise.
13031 (case_values_threshold): Likewise.
13032 (reset_out_edges_aux): Likewise.
13033 (compute_cases_per_edge): Likewise.
13034 (add_case_node): Likewise.
13035 (dump_case_nodes): Likewise.
13036 (balance_case_nodes): Likewise.
13037 (expand_switch_as_decision_tree_p): Likewise.
13038 (emit_jump): Likewise.
13039 (emit_case_decision_tree): Likewise.
13040 (try_switch_expansion): Likewise.
13041 (do_jump_if_equal): Likewise.
13042 (emit_cmp_and_jump_insns): Likewise.
13043 (fix_phi_operands_for_edge): New function.
13044 (record_phi_operand_mapping): Likewise.
13045 (class pass_lower_switch): New pass.
13046 (pass_lower_switch::execute): New function.
13047 (make_pass_lower_switch): Likewise.
13048 (conditional_probability):
13049 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13050 * tree-pass.h: Add make_pass_lower_switch.
13052 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13055 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13058 2017-08-29 Richard Biener <rguenther@suse.de>
13060 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13061 we do not add a DW_AT_inline attribute twice.
13062 (gen_subprogram_die): Remove code setting DW_AT_inline on
13063 DECL_ABSTRACT_P nodes.
13065 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13067 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13068 calls to internal functions.
13069 (gimplify_modify_expr): Likewise.
13070 * tree-call-cdce.c (use_internal_fn): Likewise.
13071 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13072 (convert_to_divmod): Set the nothrow flag.
13073 * tree-if-conv.c (predicate_mem_writes): Likewise.
13074 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13075 (vectorizable_call): Likewise.
13076 (vectorizable_store): Likewise.
13077 (vectorizable_load): Likewise.
13078 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13079 (vect_recog_mask_conversion_pattern): Likewise.
13081 2017-08-29 Martin Liska <mliska@suse.cz>
13084 * gcc.c (driver_handle_option): Add new argument.
13085 * opts-common.c (handle_option): Pass
13086 target_option_override_hook.
13087 * opts-global.c (lang_handle_option): Add new option.
13088 (set_default_handlers): Add new argument.
13089 (decode_options): Likewise.
13090 * opts.c (target_handle_option): Likewise.
13091 (common_handle_option): Call target_option_override_hook.
13092 * opts.h (struct cl_option_handler_func): Add hook for
13093 target option override.
13094 (struct cl_option_handlers): Likewise.
13095 (set_default_handlers): Add new argument.
13096 (decode_options): Likewise.
13097 (common_handle_option): Likewise.
13098 (target_handle_option): Likewise.
13099 * toplev.c (toplev::main): Pass targetm.target_option.override
13102 2017-08-29 Richard Biener <rguenther@suse.de>
13103 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13105 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13106 life to the active subtree.
13108 2017-08-28 Jeff Law <law@redhat.com>
13110 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13111 derive_equivalences.
13112 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13113 Code moved into....
13114 (edge_info::derive_equivalences): New private member function
13116 * tree-ssa-dom.c (class edge_info): Changed from a struct
13117 to a class. Add ctor/dtor, methods and data members.
13118 (edge_info::edge_info): Renamed from allocate_edge_info.
13119 Initialize additional members.
13120 (edge_info::~edge_info): New.
13121 (free_dom_edge_info): Delete the edge info.
13122 (record_edge_info): Use new class & associated member functions.
13123 Tighten forms for testing for edge equivalences.
13124 (record_temporary_equivalences): Iterate over the simple
13125 equivalences rather than assuming there's only one per edge.
13126 (cprop_into_successor_phis): Iterate over the simple
13127 equivalences rather than assuming there's only one per edge.
13128 (optimize_stmt): Use operand_equal_p rather than pointer
13129 equality for mini-DSE code.
13131 2017-08-28 Nathan Sidwell <nathan@acm.org>
13133 * gcc.c (execute): Fold SIGPIPE handling into switch
13134 statement. Adjust internal error message.
13136 2017-08-28 Richard Biener <rguenther@suse.de>
13139 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13140 Do nothing for removed DIEs.
13142 2017-08-28 Richard Biener <rguenther@suse.de>
13144 PR tree-optimization/81977
13145 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13148 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13151 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13152 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13153 using targetm.gen_mem_thread_fence.
13155 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13158 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13159 predicate to register_operand. Reorder operands.
13160 (*btr<mode>): Ditto.
13161 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13162 (*btr<mode>_mask): Ditto.
13164 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13166 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13167 * config/rs6000/xmmintrin.h: New file.
13168 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13170 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13173 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13174 parameter and_insn and return it.
13175 (recombine_lvx_pattern): Insert a copy to ensure availability of
13176 the base register of the copied masking operation at the point of
13177 the instruction replacement.
13178 (recombine_stvx_pattern): Likewise.
13180 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13182 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13183 undocumented switches.
13184 (-mpower9-dform-vector): Likewise.
13185 (-mpower9-dform): Likewise.
13186 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13187 comments to delete references to -mpower9-dform* switches.
13188 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13189 Delete reference to -mpower9-dform* switches, test for
13190 -mpower9-vector instead.
13191 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13192 (OTHER_P9_VECTOR_MASKS): Likewise.
13193 (POWERPC_MASKS): Likewise.
13194 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13195 tests against -mpower9-dform* to -mpower9-vector. Delete code
13196 that checked for -mpower9-dform* consistancy with other options.
13197 Add test for -mpower9-misc to enable other power9 switches.
13198 (rs6000_init_hard_regno_mode_ok): Likewise.
13199 (rs6000_option_override_internal): Likewise.
13200 (rs6000_emit_prologue): Likewise.
13201 (rs6000_emit_epilogue): Likewise.
13202 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13203 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13204 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13206 (emit_fusion_p9_store): Likewise.
13207 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13208 resetting these macros if the assembler does not support ISA 3.0
13210 (TARGET_P9_DFORM_VECTOR): Likewise.
13211 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13212 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13214 2017-08-25 Alan Modra <amodra@gmail.com>
13216 PR rtl-optimization/81747
13217 * cse.c (cse_extended_basic_block): Don't attempt to record
13218 equivalences for degenerate conditional branches that branch
13219 to their fall-through.
13221 2017-08-24 Martin Sebor <msebor@redhat.com>
13223 PR middle-end/81908
13224 * gimple-fold.c (size_must_be_zero_p): New function.
13225 (gimple_fold_builtin_memory_op): Call it.
13227 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13229 * config/rs6000/mm_malloc.h: New file.
13231 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13233 PR tree-optimization/81913
13234 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13235 analysis when either IVs in condition can wrap.
13237 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13239 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13240 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13242 2017-08-24 Richard Biener <rguenther@suse.de>
13245 * targhooks.c (default_target_can_inline_p): Properly
13246 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13247 is present and always compare.
13248 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13249 infer -mfpmath=sse from TARGET_SSE_P.
13250 (ix86_can_inline_p): Properly use target_option_default_node when
13251 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13253 2017-08-24 Richard Biener <rguenther@suse.de>
13256 * dwarf2out.c (output_die): Handle flag_generate_offload like
13258 (output_comp_unit): Likewise.
13259 (gen_array_type_die): Likewise.
13260 (dwarf2out_early_finish): Likewise.
13261 (note_variable_value_in_expr): Likewise.
13262 (dwarf2out_finish): Likewise. Adjust assert.
13263 * cgraphunit.c (symbol_table::compile): Move setting of
13264 flag_generate_offload earlier ...
13265 (symbol_table::finalize_compilation_unit): ... here, before
13266 early debug finalization.
13268 2017-08-24 Richard Biener <rguenther@suse.de>
13270 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13271 and ipa-fnsummary.h.
13272 (ix86_can_inline_p): When ix86_fpmath flags do not match
13273 check whether the callee uses FP math at all.
13275 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13277 PR middle-end/81931
13278 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13279 instead of TYPE_PRECISION.
13281 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13282 Alan Hayward <alan.hayward@arm.com>
13283 David Sherwood <david.sherwood@arm.com>
13285 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13287 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13289 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13291 (ix86_frame::stack_realign_allocate): New field.
13292 (struct machine_frame_state): Modify comments.
13293 (machine_frame_state::sp_realigned_fp_end): New field.
13294 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13295 layout calculation.
13296 (sp_valid_at): Add assertion to assure no attempt to access invalid
13297 offset of a realigned stack.
13298 (fp_valid_at): Likewise.
13299 (choose_baseaddr): Modify comments.
13300 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13301 ix86_expand_prologue.
13302 (ix86_expand_prologue): Modify stack realignment and allocation.
13303 (ix86_expand_epilogue): Modify comments.
13304 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13305 avx2_runtime, avx512f, and avx512f_runtime.
13307 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13309 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13310 (mstackrealign): Do not init to -1.
13311 * config/i386/i386.c (ix86_option_override_internal):
13312 Check opts_set, not opts when setting default value of
13313 opts->x_ix86_force_align_arg_pointer.
13315 2017-08-23 Richard Biener <rguenther@suse.de>
13317 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13318 lang_hooks.decl_printable_name.
13319 * print-rtl-function.c (print_rtx_function): Likewise.
13320 * tree-pretty-print.c (dump_function_header): Likewise.
13322 2017-08-23 Richard Biener <rguenther@suse.de>
13325 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13326 -g0 at compile-time.
13328 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13330 PR middle-end/19706
13331 * doc/sourcebuild.texi (Other hardware attributes):
13334 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13336 PR middle-end/19706
13337 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13338 Fix single-use check.
13340 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13342 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13344 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13346 * doc/install.texi: Modify to add more details on running selected
13349 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13351 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13352 is combined with -mabi=ms.
13353 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13356 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13358 PR tree-optimization/81488
13359 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13360 and cached_basis fields.
13361 (MAX_SPREAD): New constant.
13362 (alloc_cand_and_find_basis): Initialize new fields.
13363 (clear_visited): New function.
13364 (create_phi_basis_1): Rename from create_phi_basis, set visited
13365 and cached_basis fields.
13366 (create_phi_basis): New wrapper function.
13367 (phi_add_costs_1): Rename from phi_add_costs, add spread
13368 parameter, set visited field, short-circuit when limits reached.
13369 (phi_add_costs): New wrapper function.
13370 (record_phi_increments_1): Rename from record_phi_increments, set
13372 (record_phi_increments): New wrapper function.
13373 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13374 (phi_incr_cost): New wrapper function.
13375 (all_phi_incrs_profitable_1): Rename from
13376 all_phi_incrs_profitable, set visited field.
13377 (all_phi_incrs_profitable): New wrapper function.
13379 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13380 Alan Hayward <alan.hayward@arm.com>
13381 David Sherwood <david.sherwood@arm.com>
13383 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13384 that takes the outer and inner modes.
13385 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13386 comparison as the canonical test for a paradoxical subreg.
13387 * combine.c (simplify_set): Use paradoxical_subreg_p.
13388 (make_extraction): Likewise.
13389 (force_to_mode): Likewise.
13390 (rtx_equal_for_field_assignment_p): Likewise.
13391 (gen_lowpart_for_combine): Likewise.
13392 (simplify_comparison): Likewise.
13393 * cse.c (equiv_constant): Likewise.
13394 * expmed.c (store_bit_field_1): Likewise.
13395 * final.c (alter_subreg): Likewise.
13396 * fwprop.c (propagate_rtx): Likewise.
13397 (forward_propagate_subreg): Likewise.
13398 * ira-conflicts.c (ira_build_conflicts): Likewise.
13399 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13400 * lra-constraints.c (curr_insn_transform): Likewise.
13401 (split_reg): Likewise.
13402 * lra-eliminations.c (move_plus_up): Likewise.
13403 (lra_eliminate_regs_1): Likewise.
13404 * recog.c (general_operand): Likewise.
13405 * ree.c (combine_reaching_defs): Likewise.
13406 * reload.c (push_reload): Likewise.
13407 (find_reloads): Likewise.
13408 * reload1.c (elimination_effects): Likewise.
13409 (compute_reload_subreg_offset): Likewise.
13410 (choose_reload_regs): Likewise.
13411 * rtlanal.c (subreg_lsb_1): Likewise.
13412 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13413 (simplify_subreg): Likewise.
13414 * var-tracking.c (track_loc_p): Likewise.
13415 * emit-rtl.c (byte_lowpart_offset): Likewise.
13416 (paradoxical_subreg_p): Delete out-of-line definition.
13418 2017-08-22 Jeff Law <law@redhat.com>
13420 PR tree-optimization/81741
13421 PR tree-optimization/71947
13422 * tree-ssa-dom.c: Include tree-inline.h.
13423 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13424 equivalences if one is more expensive to compute than the other.
13425 * tree-ssa-scopedtables.h (class const_or_copies): Make
13426 record_const_or_copy_raw method private.
13427 (class avail_exprs_stack): New method simplify_binary_operation.
13428 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13429 avail_exprs_stack::simplify_binary_operation as needed.
13430 (avail_exprs_stack::simplify_binary_operation): New function.
13432 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13434 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13435 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13436 (DOT_SYMBOLS): Likewise.
13437 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13438 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13439 (RS6000_ABI_NAME): Likewise.
13440 (TARGET_CMODEL): Likewise.
13441 (TOC_SECTION_ASM_OP): Likewise.
13442 (SET_CMODEL): New macro.
13443 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13445 2017-08-22 Richard Biener <rguenther@suse.de>
13447 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13448 to allow for free-lang-data replacements similar to verify_type_variant.
13450 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13451 Alan Hayward <alan.hayward@arm.com>
13452 David Sherwood <david.sherwood@arm.com>
13454 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13455 VOIDmode for the LABEL_REF.
13457 2017-08-22 Richard Biener <rguenther@suse.de>
13459 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13461 2017-08-22 Richard Biener <rguenther@suse.de>
13463 * common.opt (feliminate-dwarf2-dups): Ignore.
13464 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13465 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13466 same_die_p_wrap, compute_section_prefix,
13467 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13468 (comdat_symbol_id, comdat_symbol_number): Likewise.
13469 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13471 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13472 (output_die): Mark unreachable path unreachable.
13473 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13474 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13475 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13476 (dwarf2out_early_finish): Likewise.
13478 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13480 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13482 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13485 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13486 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13487 (avr_attribute_table) <io, io_low, address>: Initialize
13488 .decl_required with true.
13490 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13492 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13493 undocumented debugging options.
13494 (-mvsx-scalar-double): Likewise.
13495 (-mallow-df-permute): Likewise.
13496 (-mvectorize-builtins): Likewise.
13497 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13498 (rs6000_builtin_vectorized_function): Likewise.
13499 (rs6000_builtin_md_vectorized_function): Likewise.
13500 (rs6000_opt_vars): Likewise.
13502 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13505 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13506 (*btrq_imm): Rename from *btrq.
13507 (*btcq_imm): Rename from *btcq.
13508 (btsc): New code attribute.
13509 (*<btsc><mode>): New insn pattern.
13510 (*btr<mode>): Ditto.
13511 (*<btsc><mode>_mask): New insn_and_split pattern.
13512 (*btr<mode>_mask): Ditto.
13514 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13515 Alan Hayward <alan.hayward@arm.com>
13516 David Sherwood <david.sherwood@arm.com>
13518 * function.c (pad_below): Simplify padding calculation.
13520 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13521 Alan Hayward <alan.hayward@arm.com>
13522 David Sherwood <david.sherwood@arm.com>
13524 * target.def (function_prologue): Remove frame size argument.
13525 (function_epilogue): Likewise.
13526 * doc/tm.texi: Regenerate.
13527 * final.c (final_start_function): Update call to function_prologue.
13528 (final_end_function): Update call to function_epilogue.
13529 (default_function_pro_epilogue): Remove frame size argument.
13530 * output.h (default_function_pro_epilogue): Likewise.
13531 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13532 (arm_output_function_prologue): Likewise.
13533 * config/frv/frv.c (frv_function_prologue): Likewise.
13534 (frv_function_epilogue): Likewise.
13535 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13536 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13537 (ia64_output_function_epilogue): Likewise.
13538 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13539 (m32r_output_function_epilogue): Likewise.
13540 * config/microblaze/microblaze.c (microblaze_function_prologue)
13541 (microblaze_function_epilogue): Likewise.
13542 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13543 (mips_output_function_epilogue): Likewise.
13544 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13545 (mmix_target_asm_function_epilogue): Likewise.
13546 * config/msp430/msp430.c (msp430_start_function): Likewise.
13547 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13548 (nds32_asm_function_epilogue): Likewise.
13549 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13550 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13551 (pa_output_function_epilogue): Likewise.
13552 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13553 (rs6000_output_function_epilogue): Likewise.
13554 * config/rl78/rl78.c (rl78_start_function): Likewise.
13555 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13556 (rs6000_output_function_epilogue): Likewise.
13557 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13558 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13559 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13560 (sparc_asm_function_epilogue): Likewise.
13562 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13564 * tree.h (type_has_mode_precision_p): New function.
13565 * convert.c (convert_to_integer_1): Use it.
13566 * expr.c (expand_expr_real_2): Likewise.
13567 (expand_expr_real_1): Likewise.
13568 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13569 * match.pd: Likewise.
13570 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13571 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13572 * tree-tailcall.c (process_assignment): Likewise.
13573 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13574 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13575 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13576 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13577 (vectorizable_assignment): Likewise.
13578 (vectorizable_shift): Likewise.
13579 (vectorizable_operation): Likewise.
13580 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13582 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13584 * match.pd: Add pow (C, x) simplification.
13586 2017-08-21 Richard Biener <rguenther@suse.de>
13588 PR tree-optimization/81900
13589 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13590 for blocks with abnormal predecessors.
13591 (compute_antic): Do not set visited flag prematurely.
13593 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13596 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13598 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13600 * stor-layout.h (vector_type_mode): Move to...
13601 * tree.h (vector_type_mode): ...here.
13602 * stor-layout.c (vector_type_mode): Move to...
13603 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13605 2017-08-21 Richard Biener <rguenther@suse.de>
13607 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13608 register_external_die hooks.
13609 (debug_false_tree_charstarstar_uhwistar): Declare.
13610 (debug_nothing_tree_charstar_uhwi): Likewise.
13611 * debug.c (do_nothing_debug_hooks): Adjust.
13612 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13613 (debug_nothing_tree_charstar_uhwi): Likewise.
13614 * dbxout.c (dbx_debug_hooks): Adjust.
13615 (xcoff_debug_hooks): Likewise.
13616 * sdbout.c (sdb_debug_hooks): Likewise.
13617 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13618 * dwarf2out.c (macinfo_label_base): New global.
13619 (dwarf2out_register_external_die): New function for the
13620 register_external_die hook.
13621 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13622 (dwarf2_debug_hooks): Use them.
13623 (dwarf2_lineno_debug_hooks): Adjust.
13624 (struct die_struct): Add with_offset flag.
13625 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13626 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13627 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13628 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13629 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13630 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13631 defining section names for the early LTO debug variants.
13632 (reset_indirect_string): New helper.
13633 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13634 (print_dw_val): Add support for offsetted symbol references.
13635 (get_ultimate_context): Split out from is_cxx.
13636 (is_cxx): Use get_ultimate_context.
13637 (is_fortran): Add decl overload.
13638 (compute_comp_unit_symbol): Split out worker from
13639 compute_section_prefix.
13640 (compute_section_prefix): Call compute_comp_unit_symbol and
13641 set comdat_type_p here.
13642 (output_die): Skip DIE symbol output for the LTO added one.
13643 Handle DIE symbol references with offset.
13644 (output_comp_unit): Guard section name mangling properly.
13645 For LTO debug sections emit a symbol at the section beginning
13646 which we use to refer to its DIEs.
13647 (add_abstract_origin_attribute): For DIEs registered via
13648 dwarf2out_register_external_die directly refer to the early
13649 DIE rather than indirectly through the shadow one we created.
13650 Remove obsolete call to dwarf2out_abstract_function for
13651 non-function/block origins.
13652 (gen_array_type_die): When generating early LTO debug do
13653 not emit DW_AT_string_length.
13654 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13655 late when in LTO. As suggested place a gcc_unreachable for
13656 the DECL_ABSTRACT_P case.
13657 (gen_subprogram_die): Avoid another specification DIE
13658 for early built declarations/definitions for the late LTO case.
13659 (gen_variable_die): Add type references for late duplicated VLA dies
13661 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13662 we have the abstract instance already.
13663 (process_scope_var): Adjust decl DIE contexts in LTO which
13664 first puts them in limbo.
13665 (gen_decl_die): Do not generate type DIEs late apart from
13666 types for VLAs or for decls we do not yet have a DIE. Do not
13667 call dwarf2out_abstract_function late.
13668 (dwarf2out_early_global_decl): Make sure to create DIEs
13669 for abstract instances of a decl first.
13670 (dwarf2out_late_global_decl): Adjust comment.
13671 (output_macinfo_op): With multiple macro sections use
13672 macinfo_label_base to distinguish labels.
13673 (output_macinfo): Likewise. Update macinfo_label_base.
13674 Pass in the line info label.
13675 (note_variable_value_in_expr): When generating LTO resolve
13676 all variable values here by generating DIEs as needed.
13677 (init_sections_and_labels): Add early LTO debug flag parameter
13678 and generate different sections and names if set. Add generation
13679 counter for the labels so we can have multiple of them.
13680 (reset_dies): Helper to allow DIEs to be output multiple times.
13681 (dwarf2out_finish): When outputting DIEs to the fat part of an
13682 LTO object first reset DIEs.
13683 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13684 (modified_type_die): Check for decl_ultimate_origin being self
13686 (gen_type_die_with_usage): Likewise.
13687 (gen_typedef_die): Allow decl_ultimate_origin being self.
13688 (set_decl_abstract_flags): Remove.
13689 (set_block_abstract_flags): Likewise.
13690 (dwarf2out_abstract_function): Treat the early generated DIEs
13691 as the abstract copy and only add DW_AT_inline and
13692 DW_AT_artificial here and call set_decl_origin_self.
13693 If the DIE has an abstract origin don't do anything.
13694 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13695 if we have none yet (Go fails to build one, PR78628).
13696 (variably_modified_type_p): Prevent endless recursion for Ada
13697 cyclic pointer types.
13698 * lto-streamer-in.c: Include debug.h.
13699 (dref_queue): New global.
13700 (lto_read_tree_1): Stream in DIE references.
13701 (lto_input_tree): Register DIE references.
13702 (input_function): Stream DECL_DEBUG_ARGS.
13703 * lto-streamer-out.c: Include debug.h.
13704 (lto_write_tree_1): Output DIE references.
13705 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13706 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13707 (output_function): Stream DECL_DEBUG_ARGS.
13708 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13709 Stream DECL_ABSTRACT_ORIGIN.
13710 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13711 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13712 DECL_CONTEXT for file-scope decls.
13713 * lto-streamer.h (struct dref_entry): Declare.
13714 (dref_queue): Likewise.
13715 * cfgexpand.c (pass_expand::execute): Do not call the
13716 outlining_inline_function hook here.
13717 * lto-wrapper.c (debug_obj): New global.
13718 (tool_cleanup): Unlink it if required.
13719 (debug_objcopy): New function.
13720 (run_gcc): Handle early debug sections in the IL files by
13721 extracting them to separate files, partially linkin them and
13722 feeding the result back as result to the linker.
13723 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13724 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13725 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13726 sections into a separate segment.
13727 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13729 (darwin_asm_dwarf_section): Likewise.
13730 (darwin_asm_output_dwarf_offset): Likewise.
13731 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13733 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13734 Alan Hayward <alan.hayward@arm.com>
13735 David Sherwood <david.sherwood@arm.com>
13737 * read-md.h (md_reader::record_potential_iterator_use): Replace
13738 pointer argument with an rtx and an index.
13739 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13740 (apply_mode_iterator): Likewise.
13741 (apply_code_iterator): Likewise.
13742 (apply_int_iterator): Likewise.
13743 (apply_subst_iterator): Likewise.
13744 (record_iterator_use): Likewise.
13745 (record_attribute_use): Likewise.
13746 (md_reader::record_potential_iterator_use): Likewise. Update calls
13747 to record_iterator_use and apply_iterator.
13748 (iterator_use): Replace ptr with x and index.
13749 (attribute_use): Likewise.
13750 (apply_attribute_uses): Update calls to apply_iterator.
13751 (apply_iterators): Likewise. Update initialization of iterator_use.
13752 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13753 and record_potential_iterator_use.
13754 (rtx_reader::read_rtx_operand): Likewise.
13756 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13757 Alan Hayward <alan.hayward@arm.com>
13758 David Sherwood <david.sherwood@arm.com>
13760 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13763 2017-08-21 Richard Biener <rguenther@suse.de>
13765 PR middle-end/81884
13766 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13767 at struct end conservatively when comparing common bases.
13769 2017-08-21 Richard Biener <rguenther@suse.de>
13771 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13772 (mem_ref_in_stmt): Remove.
13773 (determine_max_movement): Use ref index to get at the reference.
13774 (invariantness_dom_walker::before_dom_children): Deal with
13775 lim data already initialized.
13776 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13778 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13780 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13781 (TARGET_ISA_ROUND): Ditto.
13782 (TARGET_ROUND): Ditto.
13783 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13784 * config/i386/i386.md: Ditto.
13785 * config/i386/sse.md: Ditto.
13786 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13787 with OPTION_MASK_ISA_SSE4_1.
13789 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13792 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13793 __builtin_ia32_lzcnt_u16.
13795 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13798 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13799 (rs6000_set_current_function): Rewrite function to use it.
13801 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13804 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13805 and TYPE_WARN_IF_NOT_ALIGN.
13806 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13807 (handle_warn_if_not_align): New.
13808 (place_union_field): Call handle_warn_if_not_align.
13809 (place_field): Call handle_warn_if_not_align.
13810 Copy TYPE_WARN_IF_NOT_ALIGN.
13811 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13812 (layout_type): Likewise.
13813 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13815 (tree_decl_common): Add warn_if_not_align.
13816 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13817 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13818 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13819 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13820 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13821 * doc/extend.texi: Document warn_if_not_aligned attribute.
13822 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13824 2017-08-17 Martin Liska <mliska@suse.cz>
13827 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13828 (get_data_dependence): Use it as pointer type.
13829 (distribute_loop): Likewise.
13831 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13833 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13834 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13835 (p8_vmrgow_<mode>_direct): New define_insn.
13836 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13837 handle endianness for vmrgew and vmrgow permute patterns.
13839 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13841 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13842 * config/rs6000/rs6000-cpus.def: Remove comment.
13843 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13844 (POWERPC_MASKS): Likewise.
13845 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13846 use of TARGET_VSX_TIMODE.
13847 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13848 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13849 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13850 (rs6000_option_override_internal): Remove dead code.
13851 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13852 (rs6000_legitimize_reload_address): Likewise.
13853 (rs6000_legitimate_address_p): Likewise.
13854 (rs6000_opt_masks): Delete "vsx-timode".
13855 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13856 from function comment.
13857 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13858 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13859 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13861 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13862 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13863 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13864 (VSX_TI): Likewise.
13866 (define_peephole2): Likewise.
13868 2017-08-17 Martin Sebor <msebor@redhat.com>
13871 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13872 past the end of an array.
13873 (test_pp_format): Add test cases.
13875 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13877 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13878 missing ECF_NOTHROW flags.
13880 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13883 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13884 operands residing in integer registers.
13885 (*vsx_le_perm_load_<mode>): Likewise.
13886 (*vsx_le_perm_store_<mode>): Likewise.
13887 (define_peephole2): Add peepholes to optimize the above.
13889 2017-08-17 Marek Polacek <polacek@redhat.com>
13891 PR middle-end/81814
13892 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13893 to mimic what shorten_compare did. Change the return type to bool.
13894 (fold_cond_expr_with_comparison): Update call to
13895 operand_equal_for_comparison_p.
13896 (fold_ternary_loc): Likewise.
13898 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13900 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13902 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13904 2017-08-17 Richard Biener <rguenther@suse.de>
13906 * tree-ssa-structalias.c (solve_graph): When propagating
13907 to successors update the graphs succ edges and avoid duplicate work.
13909 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13912 * config/i386/i386.c (ix86_option_override_internal): Save target
13913 specific options after ix86_stack_protector_guard_reg was changed.
13915 2017-08-17 Richard Biener <rguenther@suse.de>
13917 PR tree-optimization/81827
13918 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13919 (new_var_info): Initialize it conservatively.
13920 (get_call_vi): Mark register vars.
13921 (new_scalar_tmp_constraint_exp): Likewise.
13922 (handle_rhs_call): Likewise.
13923 (handle_const_call): Likewise.
13924 (create_function_info_for): Likewise.
13925 (solve_constraints): Sort varinfos to separate register from
13926 non-register vars to pack points-to solution bitmaps during
13929 2017-08-17 Marek Polacek <polacek@redhat.com>
13931 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13933 2017-08-17 Richard Biener <rguenther@suse.de>
13935 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13936 to true when overflow is undefined and we saturated the result.
13938 2017-08-17 Alan Modra <amodra@gmail.com>
13941 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13942 Don't use store multiple if only one reg needs saving.
13943 (interesting_frame_related_regno): New function.
13944 (rs6000_frame_related): Don't emit frame info for regs that
13946 (rs6000_emit_epilogue): Likewise.
13948 2017-08-16 Nathan Sidwell <nathan@acm.org>
13950 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13951 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13952 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13953 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13954 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13955 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13956 (verify_type): Adjust for TYPE_BINFO move.
13957 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13958 process TYPE_BINFO directly.
13959 (hash_tree): Likewise.
13960 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13962 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13965 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13967 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13969 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13972 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13973 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13974 Add ix86_binary_operator_ok to insn constraint.
13975 (*iordi_1_bts): Ditto.
13976 (*xordi_1_btc): Ditto.
13977 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13978 Update corresponding peephole2 pattern.
13982 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13984 PR tree-optimization/81832
13985 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13986 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13988 2017-08-16 Marek Polacek <polacek@redhat.com>
13991 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13992 regarding size_zero_node.
13994 2017-08-16 Martin Liska <mliska@suse.cz>
13997 * config.gcc: Respect previously set extra_objs in case
14000 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14002 PR tree-optimization/81835
14003 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14004 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14005 not depend on the phi.
14007 2017-08-16 Alan Modra <amodra@gmail.com>
14009 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14012 2017-08-16 Alan Modra <amodra@gmail.com>
14014 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14015 (save_reg_p): ..into this. Update all callers.
14016 (first_reg_to_save): Simplify.
14018 2017-08-16 Alan Modra <amodra@gmail.com>
14020 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14023 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14027 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14028 constant count if that count is less than 32.
14030 2017-08-15 Nathan Sidwell <nathan@acm.org>
14032 * gcc.c (execute): Emit friendlier message if inferior is killed
14033 by an external cause.
14035 2017-08-15 Richard Biener <rguenther@suse.de>
14037 PR tree-optimization/81790
14038 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14039 CONSTRUCTORs from simplifying and VN.
14041 2017-08-14 Martin Sebor <msebor@redhat.com>
14043 * builtin-attrs.def: Add comments.
14045 2017-08-14 Martin Sebor <msebor@redhat.com>
14048 * doc/extend.texi (attribute nonstring): Document new attribute.
14050 2017-08-14 Martin Sebor <msebor@redhat.com>
14053 * tree-diagnostic.c (default_tree_printer): Handle %G.
14054 * gimple-pretty-print.h (percent_G_format): Declare new function.
14055 * gimple-pretty-print.c (percent_G_format): Define.
14056 * tree-pretty-print.c (percent_K_format): Add argument.
14058 2017-08-14 Martin Sebor <msebor@redhat.com>
14060 PR translation/79998
14061 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14062 Remove a stray space.
14064 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14067 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14068 (*iordi_1_bts): Ditto.
14069 (*xordi_1_btc): Ditto.
14071 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14074 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14075 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14077 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14078 quoted strings, and make more translator-friendly.
14079 (darwin_rs6000_override_options): Likewise.
14080 (rs6000_option_override_internal): Likewise.
14081 (rs6000_return_in_memory): Fix overlong line.
14082 (init_cmulative_args): Use quoted strings, and make more
14083 translator-friendly.
14084 (rs6000_pass_by_reference): Fix overlong line.
14085 (def_builtin): Use quoted strings.
14086 (altivec_expand_predicate_builtin): Use quoted strings, and make
14087 more translator-friendly.
14088 (htm_expand_builtin): Use quoted strings.
14089 (cpu_expand_builtin): Use quoted strings, and make more
14090 translator-friendly.
14091 (altivec_expand_builtin): Likewise.
14092 (paired_expand_predicate_builtin): Likewise.
14093 (rs6000_invalid_builtin): Likewise.
14094 (builtin_function_type): Use quoted strings.
14095 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14096 more translator-friendly.
14097 (rs6000_trampoline_init): Likewise.
14098 (rs6000_handle_altivec_attribute): Likewise.
14099 (rs6000_inner_target_options): Use quoted strings.
14100 (rs6000_disable_incompatible_switches): Likewise.
14101 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14102 strings, and make more translator-friendly.
14103 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14105 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14107 PR tree-optimization/81799
14108 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14109 cond_expr to simple gimple operand.
14111 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14113 PR middle-end/46932
14114 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14116 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14120 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14121 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14122 TARGET_GASISR_PROLOGUES.
14123 * config/avr/avr.c (avr_option_override): Same.
14124 (avr_pass_pre_proep::execute): Same.
14126 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14129 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14130 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14132 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14134 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14135 define_insn_and_split. Split to a memory load from 0 in
14136 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14137 using PTR mode iterator.
14138 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14139 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14140 (*add_tp_<mode>): Redefine as define_insn_and_split.
14141 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14142 address space. Merge with *add_tp_x32 using PTR mode iterator.
14143 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14144 Split to an add with a memory load from 0 in
14145 DEFAULT_TLS_SEG_REG address space.
14147 2017-08-12 Andrew Pinski <apinski@cavium.com>
14149 * config/aarch64/aarch64-option-extensions.def (rdma):
14150 Fix feature string to what Linux prints out in /proc/cpuinfo.
14152 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14155 * dwarf2out.c (modified_type_die): For C typedef types that have
14156 an ultimate origin, process the ultimate origin instead of the
14158 (gen_typedef_die): Assert that input DECLs have no ultimate
14160 (gen_type_die_with_usage): For typedef variants that have an
14161 ultimate origin, just call gen_decl_die on the original DECL.
14162 (process_scope_var): Avoid creating DIEs for local typedefs and
14163 concrete static variables.
14165 2017-08-12 Alan Modra <amodra@gmail.com>
14169 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14170 match gnu-user.h startfile.
14171 (ENDFILE_LINUX_SPEC): Similarly.
14173 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14176 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14178 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14180 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14181 * config/aarch64/aarch64.md (mov<mode>): Change.
14182 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14183 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14184 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14186 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14188 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14189 for storage order barriers.
14191 2017-08-11 Martin Liska <mliska@suse.cz>
14194 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14195 variables of void type.
14197 2017-08-11 Martin Liska <mliska@suse.cz>
14199 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14200 TARGET_SUPPORTS_ALIASES.
14201 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14202 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14203 (optimize_weakref): Likewise.
14204 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14205 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14206 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14208 2017-08-11 Martin Liska <mliska@suse.cz>
14211 * config/i386/i386.c (make_resolver_func): Do complete
14212 refactoring of the function.
14214 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14217 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14218 * config/i386/i386.c (ix86_stack_protect_guard): Use
14219 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14220 * doc/invoke.texi (x86 Options): Document
14221 -mstack-protector-guard-symbol= option.
14223 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14225 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14226 * config/i386/i386.c (ix86_split_stack_guard): New function.
14227 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14228 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14229 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14230 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14231 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14232 (split_stack_space_check): Call ix86_split_stack_guard.
14234 2017-08-10 Martin Sebor <msebor@redhat.com>
14236 * print-tree.c (print_node): Print location using the established
14238 Replace spaces with colons.
14239 (debug_raw, debug): Ditto.
14241 2017-08-10 Martin Sebor <msebor@redhat.com>
14244 * pretty-print.c (pp_format): Correct the handling of %s precision.
14246 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14249 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14251 (ix86_finalize_stack_frame_flags): This. Also clear
14252 frame_pointer_needed if -fno-omit-frame-pointer is used without
14254 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14255 with ix86_finalize_stack_frame_flags.
14256 (ix86_expand_epilogue): Likewise.
14257 (ix86_expand_split_stack_prologue): Likewise.
14258 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14260 2017-08-10 Martin Liska <mliska@suse.cz>
14263 * c-attribs.c (handle_target_attribute):
14264 Report warning for an empty string argument of target attribute.
14266 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14269 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14271 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14272 or DECL_NONLOCAL labels.
14273 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14274 or DECL_NONLOCAL labels here.
14276 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14279 to indicate when early gimple folding has been disabled.
14280 (rs6000_gimple_fold_builtin): Add debug content.
14281 (rs6000_invalid_builtin): Fix whitespace.
14282 (rs6000_expand_builtin): Fix whitespace.
14283 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14285 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14288 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14289 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14291 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14293 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14294 (qdf24xx): Likewise.
14295 * config/aarch64/aarch64.md: Include falkor.md.
14296 * config/aarch64/falkor.md: New.
14298 2017-08-09 Marek Polacek <polacek@redhat.com>
14301 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14302 * diagnostic.c (emit_diagnostic): Add a comment.
14303 (emit_diagnostic_valist): New function.
14305 2017-08-09 Marek Polacek <polacek@redhat.com>
14308 * input.c (make_location): New overload.
14309 * input.h (make_location): Declare.
14311 2017-08-08 Alan Modra <amodra@gmail.com>
14312 H.J. Lu <hongjiu.lu@intel.com>
14315 * gcc.c (NO_PIE_SPEC): Delete.
14316 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14318 (LINK_PIE_SPEC): ..to here.
14319 (LINK_COMMAND_SPEC): Support -no-pie.
14320 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14321 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14323 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14324 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14325 (ENDFILE_CRTEND_SPEC): Similarly.
14327 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14330 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14331 (mstack-protector-guard-offset=): Ditto.
14332 * config/i386/i386.c (ix86_option_override): Handle
14333 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14335 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14336 ix86_stack_protect_guard_offset variables.
14337 (TARGET_STACK_PROTECT_GUARD): Always define.
14338 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14339 and -mstack-protector-guard-offset= options.
14341 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14343 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14344 boundary case for the last candidate.
14346 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14348 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14349 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14351 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14353 PR middle-end/19706
14354 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14355 * config/aarch64/aarch64-builtins.c
14356 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14357 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14358 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14360 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14361 Andrew Pinski <pinskia@gmail.com>
14363 PR middle-end/19706
14364 * internal-fn.def (XORSIGN): New.
14365 * optabs.def (xorsign_optab): New.
14366 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14367 (convert_expand_mult_copysign): New.
14368 (pass_optimize_widening_mul::execute): Call
14369 convert_expand_mult_copysign.
14371 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14373 PR tree-optimization/81354
14374 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14375 Insert on edges rather than explicitly creating landing pads.
14376 (analyze_candidates_and_replace): Commit edge inserts.
14378 2017-08-08 Richard Biener <rguenther@suse.de>
14380 PR middle-end/81719
14381 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14382 (expand_simple_operations): Also look through ADDR_EXPRs with
14383 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14385 2017-08-08 Richard Biener <rguenther@suse.de>
14387 PR tree-optimization/81723
14388 * tree-vect-slp.c (struct bst_traits): New hash traits.
14389 (bst_fail): New global.
14390 (vect_build_slp_tree_2): New worker, split out from ...
14391 (vect_build_slp_tree): ... this now wrapping it with using
14392 bst_fail set to cache SLP tree build fails. Properly handle
14394 (vect_analyze_slp_instance): Allocate and free bst_fail.
14396 2017-08-08 Martin Liska <mliska@suse.cz>
14399 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14400 LABEL_DECLs that can be from a different function.
14402 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14404 PR tree-optimization/81744
14405 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14406 loop's number of iterations.
14408 2017-08-08 Martin Liska <mliska@suse.cz>
14410 * asan.c: Include header files.
14411 * attribs.c (build_decl_attribute_variant): New function moved
14413 (build_type_attribute_qual_variant): Likewise.
14414 (cmp_attrib_identifiers): Likewise.
14415 (simple_cst_list_equal): Likewise.
14416 (omp_declare_simd_clauses_equal): Likewise.
14417 (attribute_value_equal): Likewise.
14418 (comp_type_attributes): Likewise.
14419 (build_type_attribute_variant): Likewise.
14420 (lookup_ident_attribute): Likewise.
14421 (remove_attribute): Likewise.
14422 (merge_attributes): Likewise.
14423 (merge_type_attributes): Likewise.
14424 (merge_decl_attributes): Likewise.
14425 (merge_dllimport_decl_attributes): Likewise.
14426 (handle_dll_attribute): Likewise.
14427 (attribute_list_equal): Likewise.
14428 (attribute_list_contained): Likewise.
14429 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14430 (lookup_attribute_by_prefix): Likewise.
14431 * bb-reorder.c: Include header files.
14432 * builtins.c: Likewise.
14433 * calls.c: Likewise.
14434 * cfgexpand.c: Likewise.
14435 * cgraph.c: Likewise.
14436 * cgraphunit.c: Likewise.
14437 * convert.c: Likewise.
14438 * dwarf2out.c: Likewise.
14439 * final.c: Likewise.
14440 * fold-const.c: Likewise.
14441 * function.c: Likewise.
14442 * gimple-expr.c: Likewise.
14443 * gimple-fold.c: Likewise.
14444 * gimple-pretty-print.c: Likewise.
14445 * gimple.c: Likewise.
14446 * gimplify.c: Likewise.
14447 * hsa-common.c: Likewise.
14448 * hsa-gen.c: Likewise.
14449 * internal-fn.c: Likewise.
14450 * ipa-chkp.c: Likewise.
14451 * ipa-cp.c: Likewise.
14452 * ipa-devirt.c: Likewise.
14453 * ipa-fnsummary.c: Likewise.
14454 * ipa-inline.c: Likewise.
14455 * ipa-visibility.c: Likewise.
14457 * lto-cgraph.c: Likewise.
14458 * omp-expand.c: Likewise.
14459 * omp-general.c: Likewise.
14460 * omp-low.c: Likewise.
14461 * omp-offload.c: Likewise.
14462 * omp-simd-clone.c: Likewise.
14463 * opts-global.c: Likewise.
14464 * passes.c: Likewise.
14465 * predict.c: Likewise.
14466 * sancov.c: Likewise.
14467 * sanopt.c: Likewise.
14468 * symtab.c: Likewise.
14469 * toplev.c: Likewise.
14470 * trans-mem.c: Likewise.
14471 * tree-chkp.c: Likewise.
14472 * tree-eh.c: Likewise.
14473 * tree-into-ssa.c: Likewise.
14474 * tree-object-size.c: Likewise.
14475 * tree-parloops.c: Likewise.
14476 * tree-profile.c: Likewise.
14477 * tree-ssa-ccp.c: Likewise.
14478 * tree-ssa-live.c: Likewise.
14479 * tree-ssa-loop.c: Likewise.
14480 * tree-ssa-sccvn.c: Likewise.
14481 * tree-ssa-structalias.c: Likewise.
14482 * tree-ssa.c: Likewise.
14483 * tree-streamer-in.c: Likewise.
14484 * tree-vectorizer.c: Likewise.
14485 * tree-vrp.c: Likewise.
14486 * tsan.c: Likewise.
14487 * ubsan.c: Likewise.
14488 * varasm.c: Likewise.
14489 * varpool.c: Likewise.
14490 * tree.c: Remove functions moved to attribs.[ch].
14491 * tree.h: Likewise.
14492 * config/aarch64/aarch64.c: Add attrs.h header file.
14493 * config/alpha/alpha.c: Likewise.
14494 * config/arc/arc.c: Likewise.
14495 * config/arm/arm.c: Likewise.
14496 * config/avr/avr.c: Likewise.
14497 * config/bfin/bfin.c: Likewise.
14498 * config/c6x/c6x.c: Likewise.
14499 * config/cr16/cr16.c: Likewise.
14500 * config/cris/cris.c: Likewise.
14501 * config/darwin.c: Likewise.
14502 * config/epiphany/epiphany.c: Likewise.
14503 * config/fr30/fr30.c: Likewise.
14504 * config/frv/frv.c: Likewise.
14505 * config/ft32/ft32.c: Likewise.
14506 * config/h8300/h8300.c: Likewise.
14507 * config/i386/winnt.c: Likewise.
14508 * config/ia64/ia64.c: Likewise.
14509 * config/iq2000/iq2000.c: Likewise.
14510 * config/lm32/lm32.c: Likewise.
14511 * config/m32c/m32c.c: Likewise.
14512 * config/m32r/m32r.c: Likewise.
14513 * config/m68k/m68k.c: Likewise.
14514 * config/mcore/mcore.c: Likewise.
14515 * config/microblaze/microblaze.c: Likewise.
14516 * config/mips/mips.c: Likewise.
14517 * config/mmix/mmix.c: Likewise.
14518 * config/mn10300/mn10300.c: Likewise.
14519 * config/moxie/moxie.c: Likewise.
14520 * config/msp430/msp430.c: Likewise.
14521 * config/nds32/nds32-isr.c: Likewise.
14522 * config/nds32/nds32.c: Likewise.
14523 * config/nios2/nios2.c: Likewise.
14524 * config/nvptx/nvptx.c: Likewise.
14525 * config/pa/pa.c: Likewise.
14526 * config/pdp11/pdp11.c: Likewise.
14527 * config/powerpcspe/powerpcspe.c: Likewise.
14528 * config/riscv/riscv.c: Likewise.
14529 * config/rl78/rl78.c: Likewise.
14530 * config/rx/rx.c: Likewise.
14531 * config/s390/s390.c: Likewise.
14532 * config/sh/sh.c: Likewise.
14533 * config/sol2.c: Likewise.
14534 * config/sparc/sparc.c: Likewise.
14535 * config/spu/spu.c: Likewise.
14536 * config/stormy16/stormy16.c: Likewise.
14537 * config/tilegx/tilegx.c: Likewise.
14538 * config/tilepro/tilepro.c: Likewise.
14539 * config/v850/v850.c: Likewise.
14540 * config/vax/vax.c: Likewise.
14541 * config/visium/visium.c: Likewise.
14542 * config/xtensa/xtensa.c: Likewise.
14544 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14547 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14548 constraints since the -mupper-regs-* switches have been
14550 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14551 into a vector from a double word element that was extracted from
14552 another vector, and eliminate extra XXPERMDI instructions.
14553 (vsx_concat_<mode>_2): Likewise.
14554 (vsx_concat_<mode>_3): Likewise.
14555 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14556 concat to allow optimizing inserts from previous extracts.
14558 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14560 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14561 memory reference to a SSP offset in TLS address space.
14562 (ix86_print_operand) <case '@'>: Remove.
14563 (ix86_print_operand_punct_valid_p): Remove '@' code.
14564 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14565 UNSPEC_SP_TLS_TEST.
14566 (stack_tls_protect_set_<mode>): Remove.
14567 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14568 (stack_tls_protect_test_<mode>): Remove.
14569 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14571 2017-08-07 Olivier Hainque <hainque@adacore.com>
14574 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14576 2017-08-07 Douglas Rupp <rupp@adacore.com>
14578 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14579 variable was referenced as multidir in command.
14581 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14584 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14587 2017-08-07 Martin Liska <mliska@suse.cz>
14589 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14590 * config/rl78/rl78.c: Add include of attribs.h.
14591 * config/sh/sh.c: Likewise.
14592 * config/v850/v850.c: Likewise.
14594 2017-08-07 Tom de Vries <tom@codesourcery.com>
14596 PR middle-end/78266
14597 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14599 2017-08-07 Martin Liska <mliska@suse.cz>
14601 * config/mips/mips.c: Include attribs.h.
14603 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14606 * doc/invoke.texi: Document change in behvaior for -Ofast for
14609 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14611 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14613 (aarch64_pop_regs): Likewise.
14614 (aarch64_gen_load_pair): Likewise.
14615 (aarch64_save_callee_saves): Likewise.
14616 (aarch64_restore_callee_saves): Likewise.
14618 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14620 * config/i386/i386.c: Revert the last change.
14622 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14625 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14627 (ix86_finalize_stack_frame_flags): This. Also clear
14628 frame_pointer_needed if -fno-omit-frame-pointer is used without
14630 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14631 with ix86_finalize_stack_frame_flags.
14632 (ix86_expand_epilogue): Likewise.
14633 (ix86_expand_split_stack_prologue): Likewise.
14635 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14638 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14639 to P_AES for Westmere.
14641 2017-08-07 Jonathan Yong <10walls@gmail.com>
14643 * config/i386/mingw.opt (fset-stack-executable): Removed.
14644 * config/i386/cygming.opt (fset-stack-executable): Moved
14646 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14648 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14650 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14652 2017-08-07 Marek Polacek <polacek@redhat.com>
14654 PR middle-end/81737
14655 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14657 2017-08-07 Martin Liska <mliska@suse.cz>
14659 * attribs.h (canonicalize_attr_name): New function.
14660 (cmp_attribs): Move from c-format.c and adjusted.
14661 (is_attribute_p): Moved from tree.h.
14662 * tree-inline.c: Add new includes.
14663 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14664 (private_is_attribute_p): Remove.
14665 (private_lookup_attribute): Likewise.
14666 (private_lookup_attribute_by_prefix): Simplify.
14667 (remove_attribute): Use is_attribute_p.
14668 * tree.h: Remove removed declarations.
14670 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14672 PR middle-end/81698
14673 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14674 instead of computing it in the function. Formatting fix.
14675 (expand_case): Don't rely on default_edge being the first edge,
14676 clear it if removing it, pass default_edge to
14677 emit_case_dispatch_table.
14678 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14681 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14683 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14684 insn in the function, emit NOP after the insn.
14686 2017-08-06 Tom de Vries <tom@codesourcery.com>
14688 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14691 2017-08-06 Tom de Vries <tom@codesourcery.com>
14693 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14696 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14698 PR tree-optimization/57371
14699 * match.pd: New pattern.
14701 2017-08-04 Marek Polacek <polacek@redhat.com>
14703 PR middle-end/81695
14704 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14705 perform the computation in offset_int.
14707 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14709 PR tree-optimization/81136
14710 * tree-vectorizer.h: Include tree-hash-traits.h.
14711 (vec_base_alignments): New typedef.
14712 (vec_info): Add a base_alignments field.
14713 (vect_record_base_alignments): Declare.
14714 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14716 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14717 (create_data_ref): Add an is_conditional_in_stmt argument.
14718 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14719 the is_conditional_in_stmt field.
14720 (data_ref_loc): Add an is_conditional_in_stmt field.
14721 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14722 (find_data_references_in_stmt): Update call to create_data_ref.
14723 (graphite_find_data_references_in_stmt): Likewise.
14724 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14725 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14726 (vect_record_base_alignment): New function.
14727 (vect_record_base_alignments): Likewise.
14728 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14729 for nested statements even if we fail to compute a misalignment.
14730 Use pooled base alignments for unconditional references.
14731 (vect_find_same_alignment_drs): Compare base addresses instead
14733 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14734 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14736 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14738 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14739 Add an explicit name for the enum. Use auto_vec for slp_instances
14740 and grouped_stores.
14741 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14743 (_bb_vec_info): Add a constructor and destructor.
14744 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14745 (destroy_loop_vec_info): Delete.
14746 (vect_destroy_datarefs): Likewise.
14747 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14748 (vec_info::vec_info): New function.
14749 (vec_info::~vec_info): Likewise.
14750 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14751 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14752 destroy_loop_vec_info.
14753 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14754 (_loop_vec_info::_loop_vec_info): ...this.
14755 (destroy_loop_vec_info): Replace with...
14756 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14757 the stmt_vec_infos. Leave handling of vec_info information to its
14758 destructor. Remove explicit vector releases.
14759 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14760 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14761 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14762 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14763 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14764 (destroy_bb_vec_info): Replace with...
14765 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14766 information to its destructor.
14767 (vect_slp_analyze_bb_1): Use new and delete instead of
14768 new_bb_vec_info and destroy_bb_vec_info.
14769 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14772 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14774 * tree-data-ref.h (subscript): Add access_fn field.
14775 (data_dependence_relation): Add could_be_independent_p.
14776 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14777 (same_access_functions): Move to tree-data-ref.c.
14778 * tree-data-ref.c (ref_contains_union_access_p): New function.
14779 (access_fn_component_p): Likewise.
14780 (access_fn_components_comparable_p): Likewise.
14781 (dr_analyze_indices): Add a reference to access_fn_component_p.
14782 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14784 (constant_access_functions): Likewise.
14785 (add_other_self_distances): Likewise.
14786 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14787 (initialize_data_dependence_relation): Use XCNEW and remove
14788 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14789 of access functions that have the same type. Allow the
14790 subsequence to end with different bases in some circumstances.
14791 Record the chosen access functions in SUB_ACCESS_FN.
14792 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14793 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14794 (subscript_dependence_tester_1): Likewise dra and drb.
14795 (build_classic_dist_vector): Update calls accordingly.
14796 (subscript_dependence_tester): Likewise.
14797 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14798 DDR_COULD_BE_INDEPENDENT_P.
14799 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14800 comp_alias_ddrs instead of may_alias_ddrs.
14801 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14803 (vect_analyze_data_ref_dependence): Use it if
14804 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14805 distance vectors if that fails.
14806 (dependence_distance_ge_vf): New function.
14807 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14808 LOOP_VINFO_MAY_ALIAS_DDRS.
14810 2017-08-04 Richard Biener <rguenther@suse.de>
14812 PR middle-end/81705
14813 * fold-const.c (fold_binary_loc): Properly restrict
14814 minus_var0 && minus_var1 case when associating undefined overflow
14817 2017-08-04 Tom de Vries <tom@codesourcery.com>
14819 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14821 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14823 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14824 Don't start diagnostic messages with a capital letter.
14825 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14827 (rs6000_invalid_builtin): Likewise.
14828 (rs6000_trampoline_init): Likewise.
14830 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14833 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14834 after setting changeable df flags.
14836 2017-08-03 Richard Biener <rguenther@suse.de>
14838 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14839 up if the use is in USE - X.
14841 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14843 * toplev.c (dumpfile.h): New include.
14844 (internal_error_reentered): New static function. Use it...
14845 (internal_error_function): ...here to handle reentered internal_error.
14847 2017-08-03 Richard Biener <rguenther@suse.de>
14849 PR middle-end/81148
14850 * fold-const.c (split_tree): Add minus_var and minus_con
14851 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14852 here but always use minus_*.
14853 (associate_trees): Assert we never associate with MINUS_EXPR
14854 and NULL first operand. Do not recurse for PLUS_EXPR operands
14855 when associating as MINUS_EXPR either.
14856 (fold_binary_loc): Track minus_var and minus_con.
14858 2017-08-03 Tom de Vries <tom@codesourcery.com>
14861 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14862 ACCEL_COMPILER, apply finish_options on
14863 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14865 2017-08-03 Tom de Vries <tom@codesourcery.com>
14868 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14869 function_entry_patch_area_size > 0.
14871 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14874 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14875 instead of 10??LU, perform unit multiplication in wide_int,
14876 don't change alloc_object_size_limit if the limit is larger
14879 PR tree-optimization/81655
14880 PR tree-optimization/81588
14881 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14882 the case when ranges[i].low and high are 1 for unsigned type with
14885 PR middle-end/81052
14886 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14887 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14889 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14891 * tree-vrp.h: Add include guard.
14893 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14896 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14897 (ud2): New insn pattern.
14898 * config/i386/i386.c (ix86_expand_epilogue):
14899 For naked functions, generate ud2 instead of trap insn.
14901 2017-08-02 Marek Polacek <polacek@redhat.com>
14904 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14906 2017-08-02 Tom de Vries <tom@codesourcery.com>
14907 Cesar Philippidis <cesar@codesourcery.com>
14909 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14910 Add missing edge probabilities.
14912 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14914 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14915 Correct endianness.
14917 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14919 PR middle-end/79499
14920 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14921 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14922 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14923 prologue_seq sequences - if any.
14925 2017-08-02 Richard Biener <rguenther@suse.de>
14927 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14928 via vectors if supported, integer extracts via punning if supported
14929 or otherwise vector extracts.
14931 2017-08-02 Richard Biener <rguenther@suse.de>
14933 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14935 (bitmap_insert_into_set): ... this.
14937 2017-08-02 Richard Biener <rguenther@suse.de>
14939 PR tree-optimization/81633
14941 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14943 PR tree-optimization/71752
14944 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14946 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14948 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14949 (machine_function::call_ms2sysv_pad_out): Remove field.
14950 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14951 (ix86_compute_frame_layout): Likewise.
14953 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14956 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14957 attribute with interrupt attribute.
14959 2017-08-01 Andrew Pinski <apinski@cavium.com>
14961 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14962 BIT_INSERT_EXPR's operand 1
14963 to see if the types precision matches.
14965 2017-08-01 Martin Liska <mliska@suse.cz>
14967 PR middle-end/70140
14968 * builtins.c (expand_builtin_memcpy_args): Remove.
14969 (expand_builtin_memcpy): Call newly added function
14970 expand_builtin_memory_copy_args.
14971 (expand_builtin_memcpy_with_bounds): Likewise.
14972 (expand_builtin_mempcpy): Remove last argument.
14973 (expand_builtin_mempcpy_with_bounds): Likewise.
14974 (expand_builtin_memory_copy_args): New function created from
14975 expand_builtin_mempcpy_args with small modifications.
14976 (expand_builtin_mempcpy_args): Remove.
14977 (expand_builtin_stpcpy): Remove unused argument.
14978 (expand_builtin): Likewise.
14979 (expand_builtin_with_bounds): Likewise.
14981 2017-08-01 Martin Liska <mliska@suse.cz>
14984 Make mempcpy more optimal (PR middle-end/70140).
14986 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14989 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14990 __builtin_vec_cmpne verify both arguments are compatible vectors
14991 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14992 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14993 move computation of aligned to after checking the argument types.
14997 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14998 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15001 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15002 Doug Rupp <rupp@adacore.com>
15003 Olivier Hainque <hainque@adacore.com>
15005 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15006 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15008 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15009 for TARGET_OS_CPP_BUILTIN.
15010 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15011 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15013 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15014 passing required abi options to the assembler for EABI configurations.
15015 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15016 of .text.hot and .text.unlikely sections for kernel modules when
15017 using ARM style exceptions.
15018 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15019 options. Add EXTRA_CC1_SPEC.
15020 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15022 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15024 (ARM_TARGET2_DWARF_FORMAT): Define.
15025 * config/arm/t-vxworks: Adjust multilib control to removal of the
15026 Diab command line options.
15028 2017-08-01 Martin Liska <mliska@suse.cz>
15030 PR gcov-profile/81561
15031 * gcov.c (unblock): Make unblocking safe as we need to preserve
15032 index correspondence of blocks and block_lists.
15034 2017-08-01 Richard Biener <rguenther@suse.de>
15036 PR tree-optimization/81181
15037 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15038 (compute_antic): ... end of iteration here.
15040 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15042 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15043 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15044 (ftree-slp-vectorize): Likewise.
15045 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15046 can no longer be set independent of flag_tree_loop_vectorize.
15047 * omp-general.c (emp_max_vf): Likewise.
15048 * opts.c (enable_fdo_optimizations): Remove references to
15049 flag_tree_vectorize, these are now implicit.
15050 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15051 and leave it for the options machinery.
15053 2017-08-01 Martin Liska <mliska@suse.cz>
15055 PR middle-end/70140
15056 * builtins.c (expand_builtin_memcpy_args): Remove.
15057 (expand_builtin_memcpy): Call newly added function
15058 expand_builtin_memory_copy_args.
15059 (expand_builtin_memcpy_with_bounds): Likewise.
15060 (expand_builtin_mempcpy): Remove last argument.
15061 (expand_builtin_mempcpy_with_bounds): Likewise.
15062 (expand_builtin_memory_copy_args): New function created from
15063 expand_builtin_mempcpy_args with small modifications.
15064 (expand_builtin_mempcpy_args): Remove.
15065 (expand_builtin_stpcpy): Remove unused argument.
15066 (expand_builtin): Likewise.
15067 (expand_builtin_with_bounds): Likewise.
15069 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15072 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15073 print "ds:" only for immediates in generic address space.
15075 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15078 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15079 (ix86_function_ok_for_sibcall): Return false for naked functions.
15081 2017-08-01 Richard Biener <rguenther@suse.de>
15083 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15084 (compute_antic): Seed worklist with exit block predecessors.
15085 * cfganal.c (dfs_find_deadend): For a cycle return the source
15086 of the edge closing it.
15088 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15090 * config/aarch64/aarch64.c
15091 (aarch64_can_const_movi_rtx_p): Move 0 check.
15093 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15095 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15096 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15098 * match.pd: Ditto in address comparison pattern.
15100 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15102 PR tree-optimization/81627
15103 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15104 closed ssa form for store-store chain.
15106 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15108 PR tree-optimization/81620
15109 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15110 for store-store chain.
15112 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15114 PR tree-optimization/81588
15115 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15116 ranges[i].in_p, invert comparison code ccode. For >/>=,
15117 swap rhs1 and rhs2 and comparison code unconditionally,
15118 for </<= don't do that. Don't swap rhs1/rhs2 again if
15119 ranges[i].in_p, instead invert comparison code ccode if
15120 opcode or oe->rank is BIT_IOR_EXPR.
15123 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15124 a direct optab to conversion optab.
15125 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15126 with GET_MODE_INNER as last argument instead of optab_handler.
15127 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15128 vector extraction if possible and optab is available.
15129 * expr.c (store_constructor): Use convert_optab_handler instead
15130 of optab_handler. Use vector initialization from smaller
15131 vectors if possible and optab is available.
15132 * tree-vect-stmts.c (vectorizable_load): Likewise.
15133 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15134 now have two modes.
15135 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15136 of vec_init from half-sized vectors with the same element mode.
15137 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15138 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15139 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15140 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15141 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15142 after mode in gen_vec_extract* calls.
15143 (vec_extract<mode>): Renamed to ...
15144 (vec_extract<mode><ssescalarmodelower>): ... this.
15145 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15146 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15147 element mode after mode in gen_vec_init* calls.
15148 (VEC_INIT_HALF_MODE): New mode iterator.
15149 (vec_init<mode>): Renamed to ...
15150 (vec_init<mode><ssescalarmodelower>): ... this.
15151 (vec_init<mode><ssehalfvecmodelower>): New expander.
15152 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15153 (vec_extractv2sfsf): ... this.
15154 (vec_initv2sf): Renamed to ...
15155 (vec_initv2sfsf): ... this.
15156 (vec_extractv2si): Renamed to ...
15157 (vec_extractv2sisi): ... this.
15158 (vec_initv2si): Renamed to ...
15159 (vec_initv2sisi): ... this.
15160 (vec_extractv4hi): Renamed to ...
15161 (vec_extractv4hihi): ... this.
15162 (vec_initv4hi): Renamed to ...
15163 (vec_initv4hihi): ... this.
15164 (vec_extractv8qi): Renamed to ...
15165 (vec_extractv8qiqi): ... this.
15166 (vec_initv8qi): Renamed to ...
15167 (vec_initv8qiqi): ... this.
15168 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15169 (vec_init<mode>): Renamed to ...
15170 (vec_init<mode><VEC_base_l>): ... this.
15171 (vec_extract<mode>): Renamed to ...
15172 (vec_extract<mode><VEC_base_l>): ... this.
15173 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15174 (vec_initv2sfsf): ... this.
15175 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15176 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15177 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15178 element mode after mode in gen_vec_init* calls.
15179 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15180 (vec_init<mode><Vel>): ... this.
15181 (vec_extract<mode>): Renamed to ...
15182 (vec_extract<mode><Vel>): ... this.
15183 * config/aarch64/iterators.md (Vel): New mode attribute.
15184 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15185 Add element mode after mode in gen_vec_extract* calls.
15186 * config/s390/vector.md (non_vec_l): New mode attribute.
15187 (vec_extract<mode>): Renamed to ...
15188 (vec_extract<mode><non_vec_l>): ... this.
15189 (vec_init<mode>): Renamed to ...
15190 (vec_init<mode><non_vec_l>): ... this.
15191 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15192 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15194 * config/arm/iterators.md (V_elem_l): New mode attribute.
15195 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15196 (vec_extract<mode><V_elem_l>): ... this.
15197 (vec_extractv2di): Renamed to ...
15198 (vec_extractv2didi): ... this.
15199 (vec_init<mode>): Renamed to ...
15200 (vec_init<mode><V_elem_l>): ... this.
15201 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15202 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15203 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15204 Add element mode after gen_vec_extract* calls.
15205 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15206 (vec_init<mode><unitmode>): ... this.
15207 (vec_extract<mode>): Renamed to ...
15208 (vec_extract<mode><unitmode>): ... this.
15209 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15210 (vec_init<mode><unitmode>): ... this.
15211 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15212 (vec_initv2sfsf): ... this.
15213 (vec_extractv2sf): Renamed to ...
15214 (vec_extractv2sfsf): ... this.
15215 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15216 Add element mode after gen_vec_extract* calls.
15217 * config/mips/mips.md (unitmode): New mode iterator.
15218 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15219 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15220 * config/spu/spu.md (inner_l): New mode attribute.
15221 (vec_init<mode>): Renamed to ...
15222 (vec_init<mode><inner_l>): ... this.
15223 (vec_extract<mode>): Renamed to ...
15224 (vec_extract<mode><inner_l>): ... this.
15225 * config/sparc/sparc.md (veltmode): New mode iterator.
15226 (vec_init<VMALL:mode>): Renamed to ...
15227 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15228 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15229 (vec_initv2sisi): ... this.
15230 (vec_initv2sf): Renamed to ...
15231 (vec_initv2sfsf): ... this.
15232 (vec_extractv2sf): Renamed to ...
15233 (vec_extractv2sfsf): ... this.
15234 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15235 (vec_init<mode>): Renamed to ...
15236 (vec_init<mode><VEC_base_l>): ... this.
15237 (vec_extract<mode>): Renamed to ...
15238 (vec_extract<mode><VEC_base_l>): ... this.
15239 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15240 (vec_initv2sfsf): ... this.
15241 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15242 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15243 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15244 gen_vec_init* calls.
15246 2017-08-01 Richard Biener <rguenther@suse.de>
15248 PR tree-optimization/81297
15249 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15250 TREE_OVERFLOW from INTEGER_CSTs.
15252 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15254 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15256 2017-07-31 Carl Love <cel@us.ibm.com>
15258 * config/rs6000/rs6000-c: Add support for built-in functions
15259 vector signed char vec_xl_be (signed long long, signed char *);
15260 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15261 vector signed int vec_xl_be (signed long long, signed int *);
15262 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15263 vector signed long long vec_xl_be (signed long long, signed long long *);
15264 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15265 vector signed short vec_xl_be (signed long long, signed short *);
15266 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15267 vector double vec_xl_be (signed long long, double *);
15268 vector float vec_xl_be (signed long long, float *);
15269 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15270 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15271 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15273 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15274 (altivec_expand_builtin): Add switch statement to call
15275 altivec_expand_xl_be for each builtin.
15276 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15277 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15278 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15279 __builtin_vsx_le_be_v16qi.
15280 * doc/extend.texi: Update the built-in documentation file for the
15281 new built-in functions.
15283 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15286 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15288 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15290 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15292 * config.gcc: Add z14.
15293 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15294 CPU model numbers for z13s and z14.
15295 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15297 * config/s390/s390-opts.h (enum processor_type): Rename
15298 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15299 * config/s390/s390.c (processor_table): Add field for CPU name to
15300 be passed to Binutils.
15301 (s390_asm_output_machine_for_arch): Use the new field in
15302 processor_table for Binutils.
15303 (s390_expand_builtin): Replace arch12 with z14.
15304 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15305 (s390_get_sched_attrmask): Likewise.
15306 (s390_get_unit_mask): Likewise.
15307 * config/s390/s390.opt: Add z14 to processor_type enum.
15309 2017-07-31 Martin Jambor <mjambor@suse.cz>
15312 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15313 regardless of optimization level.
15315 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15316 Martin Liska <mliska@suse.cz>
15318 * predict.def: Remove old comment and adjust probability.
15319 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15320 PREDICT statements.
15322 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15325 * config/i386/i386.c (ix86_function_naked): New function.
15326 (ix86_can_use_return_insn_p): Return false for naked functions.
15327 (ix86_expand_prologue): Skip prologue for naked functions.
15328 (ix86_expand_epilogue): Skip epilogue for naked functions
15329 and emit trap instruction.
15330 (ix86_warn_func_return): New function.
15331 (ix86_attribute_table): Add "naked" attribute specification.
15332 (TARGET_WARN_FUNC_RETURN): Define.
15333 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15335 2017-07-31 Martin Liska <mliska@suse.cz>
15337 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15338 (dump_gimple_bb_header): Always dump BB info.
15339 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15341 2017-07-31 Martin Liska <mliska@suse.cz>
15344 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15345 also with current_function_decl non-null equality.
15347 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15350 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15351 change type to the element type, instead add eltype variable and
15352 use it where we are interested in the element type.
15354 PR tree-optimization/81603
15355 * ipa-polymorphic-call.c
15356 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15357 offset arithmetic in offset_int, bail out if the resulting bit offset
15358 doesn't fit into shwi.
15360 2017-07-31 Martin Liska <mliska@suse.cz>
15362 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15363 (gimplify_save_expr): Fix comment.
15365 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15368 * config/i386/i386.c (ix86_function_arg): Update arguments for
15370 (ix86_compute_frame_layout): Set the initial stack offset to
15371 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15372 INCOMING_FRAME_SP_OFFSET.
15373 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15374 stack before exception handler returns.
15375 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15376 the 'ERROR_CODE' for exception handler.
15378 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15380 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15381 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15382 (ASM_OUTPUT_REG_POP): Ditto.
15383 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15384 instead of asm_fprintf to output pure string.
15386 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15388 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15389 to imported_module_or_decl hook.
15390 (debug_nothing_tree_tree_tree_bool): Remove.
15391 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15392 * debug.c (do_nothing_debug_hooks): Use
15393 debug_nothing_tree_tree_tree_bool_bool instead of
15394 debug_nothing_tree_tree_tree_bool.
15395 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15396 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15397 * sdbout.c (sdb_debug_hooks): Likewise.
15398 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15399 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15401 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15402 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15403 attribute, don't add anything.
15405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15407 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15408 (fold_build2_stat_loc): Likewise.
15409 (fold_build3_stat_loc): Likewise.
15410 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15411 (fold_build1_loc): Remove macro.
15412 (fold_build2_loc): Likewise.
15413 (fold_build3_loc): Likewise.
15415 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15417 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15418 (gimple_build_debug_bind_source_stat): Likewise.
15419 * gimple.h (gimple_build_debug_bind): Remove macro.
15420 (gimple_build_debug_bind_source): Likewise.
15422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15424 * bitmap.c (bitmap_alloc): Adjust.
15425 (bitmap_gc_alloc): Likewise.
15426 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15428 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15430 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15431 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15432 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15433 (bitmap_gc_alloc_stat): Likewise.
15434 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15436 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15438 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15439 * rtl.h (shallow_copy_rtx): Remove macro.
15441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15443 * emit-rtl.c (gen_raw_REG): Adjust.
15444 * gengenrtl.c (gendef): Likewise.
15445 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15446 * rtl.h (rtx_alloc): Remove macro.
15448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15450 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15451 (build_tree_list_stat): Likewise.
15452 * tree.h (build_tree_list): Remove macro.
15453 (build_tree_list_vec): Likewise.
15455 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15457 * tree.c (make_vector_stat): Remove _stat from name.
15458 (build_vector_stat): Likewise.
15459 * tree.h (make_vector_stat): Remove macro.
15460 (build_vector_stat): Likewise.
15462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15464 * tree.h (build_var_debug_value): Remove prototype.
15466 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15468 * tree.c (tree_cons_stat): Remove _stat from name.
15469 * tree.h (tree_cons): Remove macro.
15471 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15473 * tree.c (build_vl_exp_stat): Remove _stat from name.
15474 * tree.h (build_vl_exp): Remove macro.
15476 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15478 * tree.c (build_decl_stat): Remove _stat from name.
15479 * tree.h (build_decl): Remove macro.
15481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15483 * gimple.c (gimple_build_with_ops_stat): Adjust.
15484 (gimple_alloc_stat): Remove _stat from name.
15485 * gimple.h (gimple_alloc): Remove macro.
15487 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15489 * tree.c (make_tree_vec_stat): Remove _stat from name.
15490 (grow_tree_vec_stat): Likewise.
15491 * tree.h (make_tree_vec_stat): Adjust prototype.
15492 (grow_tree_vec_stat): Likewise.
15493 (make_tree_vec): Remove macro.
15494 (grow_tree_vec): Likewise.
15496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15498 * fold-const.c (fold_build1_stat_loc): Adjust.
15499 (fold_build2_stat_loc): Likewise.
15500 (fold_build3_stat_loc): Likewise.
15501 * tree.c (build0_stat): Remove _stat from name.
15502 (build1_stat): Likewise.
15503 (build2_stat): Likewise.
15504 (build3_stat): Likewise.
15505 (build4_stat): Likewise.
15506 (build5_stat): Likewise.
15507 * tree.h (build1_loc): Remove macro, and rename _stat function
15509 (build2_loc): Likewise.
15510 (build3_loc): Likewise.
15511 (build4_loc): Likewise.
15512 (build5_loc): Likewise.
15514 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15516 * tree.c (make_int_cst_stat): Remove _stat from name.
15517 * tree.h (make_int_cst_stat): Adjust prototype.
15518 (make_int_cst): Remove macro.
15520 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15522 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15523 * tree.h (make_tree_binfo_stat): Adjust prototype.
15524 (make_tree_binfo): Remove.
15526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15528 * tree.c (copy_node_stat): Rename to copy_node.
15529 (build_distinct_type_copy): Adjust.
15530 * tree.h (copy_node_stat): Adjust prototype.
15531 (copy_node): Remove macro.
15533 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15535 * tree.c (make_node_stat): rename to make_node.
15536 (build_tree_list_stat): Adjust.
15537 (build0_stat): Likewise.
15538 (build2_stat): Likewise.
15539 (build3_stat): Likewise.
15540 (build4_stat): Likewise.
15541 (build5_stat): Likewise.
15542 (build_decl_stat): Likewise.
15543 * tree.h (make_node_stat): Adjust prototype.
15544 (make_node): remove macro.
15546 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15548 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15549 (PPC_FEATURE2_SCV): Likewise.
15550 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15552 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15554 * config/aarch64/aarch64.c
15555 (aarch64_internal_mov_immediate): Add new special pattern.
15556 * config/aarch64/aarch64.md (*movdi_aarch64):
15557 Add reg/32bit const mov case.
15559 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15560 Richard Sandiford <richard.sandiford@linaro.org>
15562 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15563 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15564 Add integer and movi cases.
15565 (movi-split-hf-df-sf split, fp16): New.
15566 (enabled): Added TARGET_FP_F16INST.
15567 * config/aarch64/iterators.md (GPF_HF): New.
15568 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15570 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15572 * config/aarch64/aarch64.c
15573 (aarch64_simd_container_mode): Add prototype.
15574 (aarch64_expand_mov_immediate): Add HI support.
15575 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15576 (aarch64_can_const_movi_rtx_p): New.
15577 (aarch64_preferred_reload_class):
15578 Remove restrictions of using FP registers for certain SIMD operations.
15579 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15580 (aarch64_valid_floating_const): Add integer move validation.
15581 (aarch64_simd_imm_scalar_p): Remove.
15582 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15583 (aarch64_legitimate_constant_p): Expand list of supported cases.
15584 * config/aarch64/aarch64-protos.h
15585 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15586 (aarch64_reinterpret_float_as_int): New.
15587 (aarch64_simd_imm_scalar_p): Remove.
15588 * config/aarch64/constraints.md (Uvi): New.
15589 (Dd): Split into Ds and new Dd.
15590 * config/aarch64/aarch64.md (*movsi_aarch64):
15592 (*movdi_aarch64): Add SIMD mov case.
15594 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15596 * tree-predcom.c: (struct chain): Handle store-store chain in which
15597 stores for elimination only store loop invariant values.
15598 (execute_pred_commoning_chain): Ditto.
15599 (prepare_initializers_chain_store_elim): Ditto.
15600 (prepare_finalizers): Ditto.
15601 (is_inv_store_elimination_chain): New function.
15602 (initialize_root_vars_store_elim_1): New function.
15604 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15606 * tree-predcom.c: Revise general description of the pass.
15607 (enum chain_type): New enum type for store elimination.
15608 (struct chain): New field supporting store elimination.
15609 (struct component): Ditto.
15610 (dump_chain): Dump store-stores chain.
15611 (release_chain): Release resources.
15612 (split_data_refs_to_components): Compute and create component
15613 contains only stores for elimination.
15614 (get_chain_last_ref_at): New function.
15615 (make_invariant_chain): Initialization.
15616 (make_rooted_chain): Specify chain type in parameter and record it.
15617 (add_looparound_copies): Skip for store-stores chain.
15618 (determine_roots_comp): Compute type of chain and pass it to
15620 (initialize_root_vars_store_elim_2): New function.
15621 (finalize_eliminated_stores): New function.
15622 (remove_stmt): Handle store for elimination.
15623 (execute_pred_commoning_chain): Execute predictive commoning on
15624 store-store chains.
15625 (determine_unroll_factor): Skip unroll for store-stores chain.
15626 (prepare_initializers_chain_store_elim): New function.
15627 (prepare_initializers_chain): Hanlde store-store chain.
15628 (prepare_finalizers_chain, prepare_finalizers): New function.
15629 (tree_predictive_commoning_loop): Return integer value indicating
15630 if loop is unrolled or lcssa form is corrupted.
15631 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15633 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15635 * tree-predcom.c (initialize_root): Delete.
15636 (execute_pred_commoning_chain): Initialize root vars and replace
15637 reference of non-combined chain directly, rather than call above
15640 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15642 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15643 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15645 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15647 * tree-predcom.c (struct chain): New field init_seq.
15648 (release_chain): Release init_seq.
15649 (prepare_initializers_chain): Record intialization stmts in above
15651 (insert_init_seqs): New function.
15652 (tree_predictive_commoning_loop): Call insert_init_seqs.
15654 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15656 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15658 2017-07-28 Richard Biener <rguenther@suse.de>
15660 * match.pd: Remove superfluous :c.
15661 * genmatch.c (simplify::id): Add member.
15662 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15664 (current_id): New global.
15665 (dt_node::parent): Move from ...
15666 (dt_operand::parent): ... here. Add for_id member.
15667 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15668 (decision_tree::find_node): Relax order requirement when
15669 merging DT_TRUE nodes to ones inbetween the current simplify
15670 and the one we try to merge with. Add diagnostic whenever
15671 we need to enforce pattern order by not merging.
15672 (decision_tree::insert): Set current_id.
15673 (decision_tree::print_node): Dump parent node and for_id.
15674 (parser::last_id): Add member.
15675 (parser::push_simplify): Assign unique id.
15676 (parser::parser): Initialize last_id.
15678 2017-07-28 Martin Liska <mliska@suse.cz>
15681 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15682 gimple_build_debug_bind.
15684 2017-07-28 Richard Biener <rguenther@suse.de>
15686 PR tree-optimization/81502
15687 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15689 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15691 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15692 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15693 for BIT_FIELD_REF args.
15694 * fold-const.c (make_bit_field_ref): Likewise.
15695 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15697 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15700 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15701 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15702 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15703 Or it into SANITIZER_UNDEFINED.
15704 * ubsan.c: Include gimple-fold.h and varasm.h.
15705 (ubsan_expand_ptr_ifn): New function.
15706 (instrument_pointer_overflow): New function.
15707 (maybe_instrument_pointer_overflow): New function.
15708 (instrument_object_size): Formatting fix.
15709 (pass_ubsan::execute): Call instrument_pointer_overflow
15710 and maybe_instrument_pointer_overflow.
15711 * internal-fn.c (expand_UBSAN_PTR): New function.
15712 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15713 * sanitizer.def (__ubsan_handle_pointer_overflow,
15714 __ubsan_handle_pointer_overflow_abort): New builtins.
15715 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15716 * internal-fn.def (UBSAN_PTR): New internal function.
15717 * opts.c (sanitizer_opts): Add pointer-overflow.
15718 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15719 * fold-const.c (build_range_check): Compute pointer range check in
15720 integral type if pointer arithmetics would be needed. Formatting
15723 2017-07-28 Martin Liska <mliska@suse.cz>
15726 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15727 parameters that are of a variable-length.
15729 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15731 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15733 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15734 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15735 (CRT_CALL_STATIC_FUNCTION): Likewise.
15736 (ASM_DEFAULT_SPEC): New define.
15737 (ASM_SPEC32): Likewise.
15738 (ASM_SPEC64): Likewise.
15739 (ASM_SPEC_COMMON): Likewise.
15740 (ASM_SPEC): Likewise.
15741 (INVALID_64BIT): Likewise.
15742 (LINK_OS_DEFAULT_SPEC): Likewise.
15743 (LINK_OS_SPEC32): Likewise.
15744 (LINK_OS_SPEC64): Likewise.
15745 (POWERPC_LINUX): Likewise.
15746 (PTRDIFF_TYPE): Likewise.
15747 (RESTORE_FP_PREFIX): Likewise.
15748 (RESTORE_FP_SUFFIX): Likewise.
15749 (SAVE_FP_PREFIX): Likewise.
15750 (SAVE_FP_SUFFIX): Likewise.
15751 (SIZE_TYPE): Likewise.
15752 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15753 (TARGET_64BIT): Likewise.
15754 (TARGET_64BIT): Likewise.
15755 (TARGET_AIX): Likewise.
15756 (WCHAR_TYPE_SIZE): Likewise.
15757 (WCHAR_TYPE): Undefine.
15758 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15759 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15760 (CPP_OS_RTEMS_SPEC): Delete.
15761 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15762 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15764 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15766 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15768 PR tree-optimization/81578
15769 * tree-parloops.c (build_new_reduction): Bail out if
15770 reduction_code isn't one of the standard OpenMP reductions.
15771 Move the details printing after that decision.
15773 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15775 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15776 related to reload_in_progress.
15777 (splat_input_operand): Likewise.
15778 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15780 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15782 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15783 (TARGET_INSTANTIATE_DECLS): Likewise.
15784 (legitimate_indexed_address_p): Delete reload_in_progress code.
15785 (rs6000_debug_legitimate_address_p): Likewise.
15786 (rs6000_eliminate_indexed_memrefs): Likewise.
15787 (rs6000_emit_le_vsx_store): Likewise.
15788 (rs6000_emit_move_si_sf_subreg): Likewise.
15789 (rs6000_emit_move): Likewise.
15790 (register_to_reg_type): Likewise.
15791 (rs6000_pre_atomic_barrier): Likewise.
15792 (rs6000_machopic_legitimize_pic_address): Likewise.
15793 (rs6000_allocate_stack_temp): Likewise.
15794 (rs6000_address_for_fpconvert): Likewise.
15795 (rs6000_address_for_altivec): Likewise.
15796 (rs6000_secondary_memory_needed_rtx): Delete function.
15797 (rs6000_check_sdmode): Likewise.
15798 (rs6000_alloc_sdmode_stack_slot): Likewise.
15799 (rs6000_instantiate_decls): Likewise.
15800 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15801 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15802 Delete reload_in_progress.
15803 (*vec_reload_and_plus_<mptrsize>): Likewise.
15804 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15805 (vsx_div_v2di): Likewise.
15806 (vsx_udiv_v2di): Likewise.
15808 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15810 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15811 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15812 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15813 (rs6000_debug_reg_global): Delete print of LRA status.
15814 (rs6000_option_override_internal): Delete dead LRA related code.
15815 (rs6000_lra_p): Delete function.
15816 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15818 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15820 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15821 * config/riscv/rtems.h: New file.
15823 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15824 Sudakshina Das <sudi.das@arm.com>
15826 * config/aarch64/aarch64.md
15827 (define_split for and<mode>3nr_compare): Move
15828 non aarch64_logical_operand to a register.
15829 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15830 register immediate operand to a register.
15831 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15833 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15835 PR middle-end/81564
15836 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15838 2017-07-27 Richard Biener <rguenther@suse.de>
15840 PR tree-optimization/81573
15841 PR tree-optimization/81494
15842 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15843 multi defuse cycle case.
15845 2017-07-27 Richard Biener <rguenther@suse.de>
15847 PR tree-optimization/81571
15848 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15851 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15853 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15854 earlier and only if MASK_FPU is set. Adjust formatting.
15856 2017-07-27 Martin Liska <mliska@suse.cz>
15858 * opt-functions.awk: Add validation of value of Init.
15859 * optc-gen.awk: Pass new argument.
15861 2017-07-27 Martin Liska <mliska@suse.cz>
15863 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15864 Fix wrong condition.
15866 2017-07-27 Martin Liska <mliska@suse.cz>
15868 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15869 BBs and edges seen by autoFDO.
15871 2017-07-27 Richard Biener <rguenther@suse.de>
15873 PR tree-optimization/81502
15874 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15875 with incompatible but same sized type.
15876 (execute_update_addresses_taken): Likewise.
15878 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15880 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15881 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15883 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15886 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15887 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15888 Change s_operand to memory_operand.
15890 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15892 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15893 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15894 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15895 Emit instructions rather than returning an expression. Handle TFmode
15896 and KFmode by casting to TImode.
15897 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15898 (rs6000_emit_le_vsx_store): Likewise.
15899 * config/rs6000/vsx.md (VSX_TI): New iterator.
15900 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15901 (*vsx_le_undo_permute_<mode>): Likewise.
15902 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15903 emit the split sequence.
15904 (*vsx_le_perm_store_<mode>): Likewise.
15906 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15908 PR tree-optimization/81555
15909 PR tree-optimization/81556
15910 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15911 if true, force CHANGED for the recursive invocation.
15912 (reassociate_bb): Remember original length of ops array, pass
15913 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15915 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15916 attributes for noipa attribute. For naked attribute use
15917 lookup_attribute first before lookup_attribute_spec.
15918 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15920 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15921 for functions with noipa attribute.
15922 (cgraph_externally_visible_p): Return true for functions with noipa
15924 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15925 for functions with noipa attribute.
15926 * doc/extend.texi: Document noipa function attribute.
15927 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15928 also for functions with noipa attribute.
15929 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15931 2017-07-26 Andrew Pinski <apinski@cavium.com>
15933 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15934 vec_unalign_load_cost and vec_unalign_store_cost.
15936 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15938 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15939 -mvsx-small-integer option.
15940 (ISA_3_0_MASKS_IEEE): Likewise.
15941 (OTHER_VSX_VECTOR_MASKS): Likewise.
15942 (POWERPC_MASKS): Likewise.
15943 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15944 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15945 code, only testing for DImode being allowed in non-VSX floating
15947 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15948 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15950 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15951 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15952 TARGET_P8_VECTOR test.
15953 (rs6000_secondary_reload_simple_move): Likewise.
15954 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15955 since TARGET_P9_VECTOR was already tested.
15956 (rs6000_opt_masks): Remove -mvsx-small-integer.
15957 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15958 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15960 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15961 test for TARGET_VEXTRACTUB was used, and that uses
15963 (p9 extract splitter): Likewise.
15964 (vsx_extract_<mode>_di_p9): Likewise.
15965 (vsx_extract_<mode>_store_p9): Likewise.
15966 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15967 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15968 the elimination of TARGET_VSX_SMALL_INTEGER.
15969 (vsx_extract_<mode>_p8): Likewise.
15970 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15971 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15972 (vsx_set_<mode>_p9): Likewise.
15973 (vsx_set_v4sf_p9): Likewise.
15974 (vsx_set_v4sf_p9_zero): Likewise.
15975 (vsx_insert_extract_v4sf_p9): Likewise.
15976 (vsx_insert_extract_v4sf_p9_2): Likewise.
15977 * config/rs6000/rs6000.md (sign extend splitter): Change
15978 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15979 (floatsi<mode>2_lfiwax_mem): Likewise.
15980 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15981 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15982 since a test for TARGET_P9_VECTOR was used.
15983 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15984 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15985 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15986 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15987 TARGET_P8_VECTOR test.
15988 (fix_trunc<mode>si2_stfiwx): Likewise.
15989 (fix_trunc<mode>si2_internal): Likewise.
15990 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15991 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15993 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15994 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15995 TARGET_P8_VECTOR test.
15996 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15997 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15998 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16000 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16001 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16002 since a test for TARGET_P9_VECTOR was used.
16003 (splitter for loading small constants): Likewise.
16005 2017-07-26 Andrew Pinski <apinski@cavium.com>
16007 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16010 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16013 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16014 (fp_valid_at): Likewise.
16016 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16018 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16019 (qdf24xx_addrcost_table): Likewise.
16020 (cortexa57_tunings): Update to use generic_branch_cost.
16021 (cortexa72_tunings): Likewise.
16022 (cortexa73_tunings): Likewise.
16023 (qdf24xx_tunings): Likewise.
16025 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16027 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16028 (thunderx2t99_branch_cost): Likewise.
16029 (cortexa35_tunings): Update to use generic_branch_cost.
16030 (cortexa53_tunings): Likewise.
16031 (cortexa57_tunings): Likewise.
16032 (cortexa72_tunings): Likewise.
16033 (cortexa73_tunings): Likewise.
16034 (thunderx2t99_tunings): Likewise.
16036 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16038 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16039 (sparc_option_override): Honour MASK_FSMULD.
16040 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16041 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16042 * config/sparc/sparc.opt (mfsmuld): New option.
16043 * doc/invoke.texi (mfsmuld): Document option.
16045 2017-07-26 Marek Polacek <polacek@redhat.com>
16047 PR middle-end/70992
16048 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16050 2017-07-26 Richard Biener <rguenther@suse.de>
16052 * gimple-match-head.c (do_valueize): Return OP if valueize
16054 (get_def): New helper to get at the def stmt of a SSA name
16055 if valueize allows.
16056 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16057 do_valueize to get at the def stmt.
16058 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16060 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16062 PR middle-end/46932
16063 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16065 2017-07-26 Martin Liska <mliska@suse.cz>
16068 * function.c (expand_function_start): Make expansion of
16069 nonlocal_goto_save_area after parm_birth_insn.
16071 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16073 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16074 from all CPU target flags enable members.
16076 2017-07-26 Richard Biener <rguenther@suse.de>
16078 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16079 (decision_tree::gen): Make 'type' const.
16080 (write_predicate): Likewise.
16082 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16084 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16085 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16086 (rs6000_option_override_internal): Likewise.
16087 (rs6000_expand_vector_set): Likewise.
16088 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16089 (TARGET_UPPER_REGS_SF): Likewise.
16090 (TARGET_UPPER_REGS_DI): Likewise.
16091 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16092 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16093 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16094 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16095 (Splitters for DI constants in Altivec registers): Likewise.
16096 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16097 (vsx_set_v4sf_p9): Likewise.
16098 (vsx_set_v4sf_p9_zero): Likewise.
16099 (vsx_insert_extract_v4sf_p9): Likewise.
16100 (vsx_insert_extract_v4sf_p9_2): Likewise.
16102 2017-07-25 Carl Love <cel@us.ibm.com>
16104 * doc/extend.texi: Update the built-in documentation file for the
16105 existing built-in functions
16106 vector signed char vec_cnttz (vector signed char);
16107 vector unsigned char vec_cnttz (vector unsigned char);
16108 vector signed short vec_cnttz (vector signed short);
16109 vector unsigned short vec_cnttz (vector unsigned short);
16110 vector signed int vec_cnttz (vector signed int);
16111 vector unsigned int vec_cnttz (vector unsigned int);
16112 vector signed long long vec_cnttz (vector signed long long);
16113 vector unsigned long long vec_cnttz (vector unsigned long long);
16115 2017-07-25 Andrew Pinski <apinski@cavium.com>
16117 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16118 accesses where the use is for the first operand of a BIT_INSERT.
16120 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16123 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16124 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16126 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16128 * config/i386/gstabs.h: Delete.
16129 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16131 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16133 * config/i386/i386.c (ix86_decompose_address): Do not check for
16134 register RTX when looking at index_reg or base_reg.
16135 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16137 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16139 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16140 to update EH info here.
16142 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16144 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16146 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16148 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16150 2017-07-25 Torsten Duwe <duwe@suse.de>
16152 * common.opt: Introduce -fpatchable-function-entry
16153 command line option, and its variables function_entry_patch_area_size
16154 and function_entry_patch_area_start.
16155 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16156 including a two-value parser.
16157 * target.def (print_patchable_function_entry): New target hook.
16158 * targhooks.h (default_print_patchable_function_entry): New function.
16159 * targhooks.c (default_print_patchable_function_entry): Likewise.
16160 * toplev.c (process_options): Switch off IPA-RA if
16161 patchable function entries are being generated.
16162 * varasm.c (assemble_start_function): Look at the
16163 patchable-function-entry command line switch and current
16164 function attributes and maybe generate NOP instructions by
16165 calling the print_patchable_function_entry hook.
16166 * doc/extend.texi: Document patchable_function_entry attribute.
16167 * doc/invoke.texi: Document -fpatchable_function_entry
16168 command line option.
16169 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16171 * doc/tm.texi: Re-generate.
16173 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16176 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16177 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16179 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16181 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16183 2017-07-25 Richard Biener <rguenther@suse.de>
16185 PR tree-optimization/81455
16186 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16187 not walk in cycles when looking for guards.
16189 2017-07-25 Richard Biener <rguenther@suse.de>
16191 PR tree-optimization/81529
16192 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16193 when optimizing backedge uses.
16195 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16197 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16199 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16200 to dl_section_ref. On AIX, append an expression to subtract
16201 the size of the section length to dl_section_ref.
16203 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16205 * configure.ac: If any of the config.* scripts fail, exit 1.
16206 * configure: Regenerate.
16208 2017-07-25 Richard Biener <rguenther@suse.de>
16210 PR middle-end/81546
16211 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16212 of immediate uses, be more verbose on errors.
16214 2017-07-25 Richard Biener <rguenther@suse.de>
16216 PR tree-optimization/81510
16217 * tree-vect-loop.c (vect_is_simple_reduction): When the
16218 reduction stmt is not inside the loop bail out.
16220 2017-07-25 Richard Biener <rguenther@suse.de>
16222 PR tree-optimization/81303
16223 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16224 profitability check against LOOP_VINFO_NITERSM1.
16226 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16228 * domwalk.c (cmp_bb_postorder): Simplify.
16229 (sort_bbs_postorder): New function. Use it...
16230 (dom_walker::walk): ...here to optimize common cases.
16232 2017-07-25 Martin Liska <mliska@suse.cz>
16235 * ipa-visibility.c (function_and_variable_visibility): Make the
16236 redirection just on target that supports aliasing.
16237 Fix GNU coding style.
16239 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16242 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16244 * config/arm/rtems.h: Likewise.
16245 * config/bfin/rtems.h: Likewise.
16246 * config/i386/rtemself.h: Likewise.
16247 * config/lm32/rtems.h: Likewise.
16248 * config/m32c/rtems.h: Likewise.
16249 * config/m68k/rtemself.h: Likewise.
16250 * config/microblaze/rtems.h: Likewise.
16251 * config/mips/rtems.h: Likewise.
16252 * config/moxie/rtems.h: Likewise.
16253 * config/nios2/rtems.h: Likewise.
16254 * config/powerpcspe/rtems.h: Likewise.
16255 * config/rs6000/rtems.h: Likewise.
16256 * config/rtems.h: Likewise.
16257 * config/sh/rtems.h: Likewise.
16258 * config/sh/rtemself.h: Likewise.
16259 * config/sparc/rtemself.h: Likewise.
16261 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16264 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16265 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16266 * tree-ssa-structalias.c (alias_get_name): Same.
16268 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16271 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16272 instructions if no du chain is found.
16274 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16276 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16278 2017-07-25 Richard Biener <rguenther@suse.de>
16280 PR middle-end/81505
16281 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16284 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16286 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16287 upper-regs options.
16288 (ISA_2_7_MASKS_SERVER): Likewise.
16289 (ISA_3_0_MASKS_IEEE): Likewise.
16290 (OTHER_P8_VECTOR_MASKS): Likewise.
16291 (OTHER_VSX_VECTOR_MASKS): Likewise.
16292 (POWERPC_MASKS): Likewise.
16293 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16294 duplicate list of options.
16295 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16296 explicit -mupper-regs options.
16297 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16298 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16299 alias for -mupper-regs-df.
16300 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16301 (rs6000_init_hard_regno_mode_ok): Likewise.
16302 (rs6000_option_override_internal): Likewise.
16303 (rs6000_opt_masks): Likewise.
16304 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16305 options in terms of whether -mvsx or -mpower8-vector was used.
16306 (TARGET_UPPER_REGS_DI): Likewise.
16307 (TARGET_UPPER_REGS_SF): Likewise.
16308 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16309 -mupper-regs-* options.
16311 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16313 * passes.c (emergency_dump_function): Print some empty lines and a
16314 header before the RTL dump.
16316 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16318 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16320 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16323 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16324 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16326 2017-07-24 Carl Love <cel@us.ibm.com>
16328 * config/rs6000/rs6000-c.c: Add support for built-in functions
16329 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16330 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16331 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16332 vec_extract_fp_from_shortl): Add defines for the two builtins.
16333 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16334 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16336 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16337 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16338 * doc/extend.texi: Update the built-in documentation file for the
16339 new built-in function.
16341 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16344 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16346 * doc/generic.texi: Likewise.
16347 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16348 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16350 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16352 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16353 (aarch64_mls_elt_merge<mode>): Likewise.
16355 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16357 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16358 having __cxa_atexit.
16360 2017-07-23 Michael Collison <michael.collison@arm.com>
16362 * config/arm/arm.c (arm_option_override): Deprecate
16363 use of -mstructure-size-boundary.
16364 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16365 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16367 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16370 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16371 Reduce cost estimate for direct moves.
16373 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16376 * config/i386/i386.c (ix86_option_override_internal): Disable
16377 BMI, BMI2 and TBM instructions for -m16.
16379 2017-07-21 Carl Love <cel@us.ibm.com>
16381 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16382 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16383 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16384 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16385 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16386 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16387 VMULOSW): New enum "unspec" values.
16388 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16389 altivec_vmulosw): New patterns.
16390 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16391 VMULOSW): Add definitions.
16393 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16395 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16396 (qdf24xx): Likewise.
16397 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16398 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16399 (AARCH64_FL_V8_1): Renumber.
16400 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16401 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16402 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16403 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16404 rdma to feature modifiers list.
16406 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16408 PR middle-end/56727
16409 * ipa-visibility (function_and_variable_visibility): Convert
16410 recursive PLT call to direct call if appropriate.
16412 2017-07-21 Andrew Pinski <apinski@cavium.com>
16414 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16415 operand 1 to see if the types precision matches.
16416 * fold-const.c (operand_equal_p): Likewise.
16418 2017-07-21 Richard Biener <rguenther@suse.de>
16420 PR tree-optimization/81303
16421 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16422 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16423 (vect_peeling_hash_get_lowest_cost): Adjust.
16424 (vect_enhance_data_refs_alignment): Likewise. Use
16425 vect_get_peeling_costs_all_drs to compute the penalty for no
16426 peeling to match up costs.
16428 2017-07-21 Richard Biener <rguenther@suse.de>
16430 PR tree-optimization/81500
16431 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16432 we didn't identify a reduction path.
16434 2017-07-21 Tom de Vries <tom@codesourcery.com>
16435 Cesar Philippidis <cesar@codesourcery.com>
16437 PR gcov-profile/81442
16438 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16441 2017-07-21 Tom de Vries <tom@codesourcery.com>
16444 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16446 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16447 nvptx_override_options_after_change.
16449 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16451 * dwarf2out.c (output_file_names): Avoid double testing for
16452 dwarf_version >= 5.
16454 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16456 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16458 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16460 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16462 (try_crossjump_to_edge): Do not punt on partitioned functions.
16464 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16466 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16467 Put all BBs reachable only via paths crossing cold region to cold
16469 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16471 2016-07-21 Richard Biener <rguenther@suse.de>
16473 PR tree-optimization/81303
16474 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16475 into account prologue and epilogue iterations when raising
16476 min_profitable_iters to sth at least covering one vector iteration.
16478 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16480 * config/arm/arm.c (arm_test_cpu_arch_dat):
16483 2017-07-20 Nathan Sidwell <nathan@acm.org>
16485 Remove TYPE_METHODS.
16486 * tree.h (TYPE_METHODS): Delete.
16487 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16488 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16489 (dbxout_type_methods): Scan TYPE_FIELDS.
16490 (dbxout_type): Don't check TYPE_METHODS here.
16491 * function.c (use_register_for_decl): Always ignore register for
16492 class types when not optimizing.
16493 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16494 * tree.c (free_lang_data_in_type): Stitch out member functions and
16495 templates from TYPE_FIELDS.
16496 (build_distinct_type_copy, verify_type_variant,
16497 verify_type): Member fns are on TYPE_FIELDS.
16498 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16499 * tree-pretty-print.c (dump_generic_node): Likewise.
16501 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16504 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16505 V2TImode and V4TImode.
16506 (ix86_expand_vector_extract): Likewise.
16507 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16508 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16509 (ssescalarmode): Handle V4TImode and V2TImode.
16510 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16511 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16512 (VEXTRACTI128_MODE): New mode iterator.
16513 (splitter for *vec_extractv?ti first element): New.
16514 (VEC_INIT_MODE): New mode iterator.
16515 (vec_init<mode>): Consolidate 3 expanders into one using
16516 VEC_INIT_MODE mode iterator.
16518 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16520 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16521 non_spilled_static_chain_regno_p.
16523 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16525 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16527 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16529 * bb-reorder.c (connect_traces): Allow copying of blocks within
16532 2017-07-20 Richard Biener <rguenther@suse.de>
16534 * gimple.h (gimple_phi_result): Add gphi * overload.
16535 (gimple_phi_result_ptr): Likewise.
16536 (gimple_phi_arg): Likewise. Adjust index assert to only
16537 allow actual argument accesses rather than all slots available
16539 (gimple_phi_arg_def): Add gphi * overload.
16540 * tree-phinodes.c (make_phi_node): Initialize only actual
16542 (resize_phi_node): Clear memory not covered by old node,
16543 do not initialize excess argument slots.
16544 (reserve_phi_args_for_new_edge): Initialize new argument slot
16547 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16549 PR tree-optimization/81388
16551 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16553 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16554 by removing computation of may_be_zero.
16556 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16557 Tom de Vries <tom@codesourcery.com>
16559 PR middle-end/81030
16560 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16561 when gimple level profile disagrees with what RTL expander did.
16563 2017-07-20 Richard Biener <rguenther@suse.de>
16565 PR tree-optimization/61171
16566 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16567 (vect_analyze_stmt): Add slp instance parameter.
16568 (vectorizable_reduction): Likewise.
16569 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16570 (vect_is_simple_reduction): Deal with chains not detected
16571 as SLP reduction chain, specifically not properly associated
16572 chains containing a mix of plus/minus.
16573 (get_reduction_op): Remove.
16574 (get_initial_defs_for_reduction): Simplify, pass in whether
16575 this is a reduction chain, pass in the SLP node for the PHIs.
16576 (vect_create_epilog_for_reduction): Get the SLP instance as
16578 (vectorizable_reduction): Get the SLP instance as arg.
16579 During analysis remember the SLP node with the PHIs in the
16580 instance. Simplify getting at the vectorized reduction PHIs.
16581 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16582 through SLP instance.
16583 (vect_slp_analyze_operations): Likewise.
16584 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16585 (vect_transform_stmt): Likewise.
16587 2017-07-20 Tom de Vries <tom@codesourcery.com>
16589 PR tree-optimization/81489
16590 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16591 read of phi arg location to before loop that modifies phi.
16593 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16595 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16598 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16600 PR middle-end/81331
16601 * except.c (execute): Fix ordering issue.
16603 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16605 PR rtl-optimization/81423
16606 * combine.c (make_compound_operation_int): Don't try to optimize
16607 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16609 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16611 PR rtl-optimization/81423
16612 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16613 with a constant that is -1 in the truncated to mode.
16615 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16617 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16618 (determine_unlikely_bbs): ... here.
16619 * predict.h (propagate_unlikely_bbs_forward): Declare.
16620 * cfgexpand.c (pass_expand::execute): Use it.
16621 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16623 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16624 propagate_unlikely_bbs_forward.
16626 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16628 PR middle-end/81331
16629 * except.c (maybe_add_nop_after_section_switch): New function.
16632 2017-07-19 Tom de Vries <tom@codesourcery.com>
16634 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16636 2017-07-19 Tom de Vries <tom@codesourcery.com>
16638 * gimple.h (gimple_phi_arg): Make assert more strict.
16640 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16642 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16643 * config/rs6000/mmintrin.h: New file.
16644 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16646 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16648 PR tree-optimization/81346
16649 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16651 2017-07-19 Tom de Vries <tom@codesourcery.com>
16653 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16655 2017-07-19 Tom de Vries <tom@codesourcery.com>
16657 * config/nvptx/nvptx-modes.def: Add V2DImode.
16658 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16659 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16660 (nvptx_output_mov_insn): Handle lack of mov.b128.
16661 (nvptx_print_operand): Handle 'H' and 'L' codes.
16662 (nvptx_vector_mode_supported): Allow V2DImode.
16663 (nvptx_preferred_simd_mode): New function.
16664 (nvptx_data_alignment): New function.
16665 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16666 nvptx_preferred_simd_mode.
16667 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16669 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16671 2017-07-19 Tom de Vries <tom@codesourcery.com>
16673 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16674 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16675 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16676 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16677 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16678 (mov<VECIM>_insn): New define_insn.
16679 (define_expand "mov<VECIM>): New define_expand.
16681 2017-07-19 Tom de Vries <tom@codesourcery.com>
16683 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16685 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16687 PR tree-optimization/81346
16688 * fold-const.h (fold_div_compare, range_check_type): Declare.
16689 * fold-const.c (range_check_type): New function.
16690 (build_range_check): Use range_check_type.
16691 (fold_div_compare): No longer static, rewritten into
16692 a match.pd helper function.
16693 (fold_comparison): Don't call fold_div_compare here.
16694 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16695 as helper function.
16697 2017-07-19 Nathan Sidwell <nathan@acm.org>
16699 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16700 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16701 * tree.c (find_decls_types_r, verify_type): Use
16702 TYPE_{MIN,MAX}_VALUE_RAW.
16703 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16704 (hash_tree): Likewise.
16705 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16707 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16710 2017-07-18 Tom de Vries <tom@codesourcery.com>
16712 PR middle-end/81464
16713 * omp-expand.c (expand_omp_for_static_chunk): Handle
16714 equal-argument loop exit phi.
16716 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16719 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16720 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16721 operand 2 predicate.
16722 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16723 operand 2 predicate.
16724 (ror,rol -> rorx splitters): Use const_int_operand as
16725 operand 2 predicate.
16727 2017-06-18 Richard Biener <rguenther@suse.de>
16729 PR tree-optimization/81410
16730 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16731 the gap in the ! slp_perm SLP case after each group.
16733 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16735 PR middle-end/81463
16736 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16739 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16741 PR middle-end/81462
16742 * predict.c (set_even_probabilities): Cleanup; do not affect
16743 probabilities that are already known.
16744 (combine_predictions_for_bb): Call even when count is set.
16746 2017-07-18 Nathan Sidwell <nathan@acm.org>
16748 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16751 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16754 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16755 optimization for loop niter analysis.
16757 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16760 * config/avr/avr.c (avr_optimize_casesi): Don't use
16761 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16763 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16765 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16766 body_cost_vec from _vect_peel_extended_info.
16767 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16768 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16771 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16773 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16775 2017-07-18 Richard Biener <rguenther@suse.de>
16777 PR tree-optimization/80620
16778 PR tree-optimization/81403
16779 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16780 info when re-using a VN table entry.
16782 2017-07-18 Richard Biener <rguenther@suse.de>
16784 PR tree-optimization/81418
16785 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16786 vectype_in. Verify that with lane-reducing reduction operations
16787 we have a single def-use cycle.
16789 2017-07-17 Carl Love <cel@us.ibm.com>
16791 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16793 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16794 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16795 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16796 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16797 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16798 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16799 VMULOSW): New enum "unspec" values.
16800 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16801 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16802 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16803 altivec_vmulosw): New patterns.
16804 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16805 VMULOSW): Add definitions.
16806 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16808 * config/alpha/alpha.c: Include predict.h.
16810 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16812 * tree-vrp.c (compare_assert_loc): Fix comparison function
16813 to return predictable results.
16815 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16817 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16819 (subdi3): Likewise.
16820 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16821 * doc/invoke.texi (mexpand-adddi): Update text.
16823 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16825 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16826 that also clobbers the CC register. The old expand code is moved
16828 (*arc_clzsi2): ... here.
16829 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16830 the CC register. The old expand code is moved to ...
16831 (arc_ctzsi2): ... here.
16833 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16835 * config/arc/arc.opt (mindexed-loads): Use initial value
16836 TARGET_INDEXED_LOADS_DEFAULT.
16837 (mauto-modify-reg): Use initial value
16838 TARGET_AUTO_MODIFY_REG_DEFAULT.
16839 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16840 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16841 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16842 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16844 2017-07-17 Martin Liska <mliska@suse.cz>
16847 * opts.c (finish_options): Do not allow -fgnu-tm
16848 w/ -fsanitize={kernel-,}address. Say sorry.
16850 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16853 * tree-loop-distribution.c (classify_partition): Only assert on
16854 numer of iterations.
16855 (merge_dep_scc_partitions): Delete prameter. Update function call.
16856 (distribute_loop): Remove code handling loop with unknown niters.
16857 (pass_loop_distribution::execute): Skip loop with unknown niters.
16859 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16862 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16863 function sort_partitions_by_post_order.
16865 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16867 PR tree-optimization/81374
16868 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16869 the max index of basic blocks, rather than number of basic blocks.
16871 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16873 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16875 (arc_legitimate_pic_operand_p): Likewise.
16876 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16878 (arc_needs_pcl_p): Likewise.
16879 (arc_legitimate_pc_offset_p): Likewise.
16880 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16881 function is also used in constrains.md.
16882 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16883 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16884 PLUS. Only return true/false in known cases, otherwise assert.
16885 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16886 is already called in arc_legitimate_constant_p.
16887 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16889 (LEGITIMATE_PIC_OPERAND_P): Use
16890 arc_raw_symbolic_reference_mentioned_p function.
16891 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16896 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16897 Andrew Burgess <andrew.burgess@embecosm.com>
16899 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16900 (arc_return_address_register): New function.
16901 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16902 (arc_handle_fndecl_attribute): Add naked attribute.
16903 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16904 (TARGET_WARN_FUNC_RETURN): Likewise.
16905 (arc_allocate_stack_slots_for_args): New function.
16906 (arc_warn_func_return): Likewise.
16907 (machine_function): Change type fn_type.
16908 (arc_compute_function_type): Consider new naked function type,
16909 change function return type.
16910 (arc_must_save_register): Adapt to handle new
16911 arc_compute_function_type's return type.
16912 (arc_expand_prologue): Likewise.
16913 (arc_expand_epilogue): Likewise.
16914 (arc_return_address_regs): Delete.
16915 (arc_return_address_register): New function.
16916 (arc_epilogue_uses): Use above function.
16917 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16918 (arc_function_type): Change encoding, add naked type.
16919 (ARC_INTERRUPT_P): Change to handle the new encoding.
16920 (ARC_FAST_INTERRUPT_P): Likewise.
16921 (ARC_NORMAL_P): Define.
16922 (ARC_NAKED_P): Likewise.
16923 (arc_compute_function_type): Delete prototype.
16924 * config/arc/arc.md (in_ret_delay_slot): Use
16925 arc_return_address_register function.
16926 (simple_return): Likewise.
16927 (p_return_i): Likewise.
16929 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16931 PR tree-optimization/81428
16932 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16933 can't be built for those types.
16935 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16937 Remove stuff dead since r239246.
16939 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16940 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16941 (avr_inform_devices): Remove dead stuff.
16943 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16945 * config/arm/arm_neon.h: Fix softp typo.
16947 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16949 PR tree-optimization/81365
16950 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16951 aggregate moves onto bb predecessor edges, make sure there are no
16952 loads that could alias the lhs in between the start of bb and the
16955 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16958 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16959 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16960 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16962 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16964 PR tree-optimization/81396
16965 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16966 (init_symbolic_number): Initialize it to 1.
16967 (perform_symbolic_merge): Add n_ops from both operands into the new
16969 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16970 without base_addr as useless if they need more than one operation.
16971 (bswap_replace): Handle !bswap case for NULL base_addr.
16973 2017-07-17 Tom de Vries <tom@codesourcery.com>
16976 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16979 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16981 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16982 conditional builtin define __FIX_LEON3FT_B2BST.
16984 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16986 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16987 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16990 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16992 PR rtl-optimization/81424
16993 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16994 to remove potential trapping from operands if -fnon-call-exceptions.
16996 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16998 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16999 profile_proability for scalling.
17000 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17002 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17004 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17006 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17008 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17009 fixpoint arithmetics.
17011 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17013 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17014 fixpoint arithmetics.
17016 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17018 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17019 fixpoint arithmetics.
17021 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17023 * profile-count.h (profile_probability::from_reg_br_prob_note,
17024 profile_probability::to_reg_br_prob_note): New functions.
17025 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17026 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17027 * predict.c (probability_reliable_p): Update.
17028 (edge_probability_reliable_p): Update.
17029 (br_prob_note_reliable_p): Update.
17030 (invert_br_probabilities): Update.
17031 (add_reg_br_prob_note): New function.
17032 (combine_predictions_for_insn): Update.
17033 * asan.c (asan_clear_shadow): Update.
17034 * cfgbuild.c (compute_outgoing_frequencies): Update.
17035 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17036 (update_br_prob_note): Update.
17037 (rtl_verify_edges): Update.
17038 (purge_dead_edges): Update.
17039 (fixup_reorder_chain): Update.
17040 * emit-rtl.c (try_split): Update.
17041 * ifcvt.c (cond_exec_process_insns): Update.
17042 (cond_exec_process_if_block): Update.
17043 (dead_or_predicable): Update.
17044 * internal-fn.c (expand_addsub_overflow): Update.
17045 (expand_neg_overflow): Update.
17046 (expand_mul_overflow): Update.
17047 * loop-doloop.c (doloop_modify): Update.
17048 * loop-unroll.c (compare_and_jump_seq): Update.
17049 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17050 * predict.h: Update.
17051 * reorg.c (mostly_true_jump): Update.
17053 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17054 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17055 * config/arc/arc.c: (emit_unlikely_jump): Update.
17056 * config/arm/arm.c: (emit_unlikely_jump): Update.
17057 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17058 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17059 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17060 (ix86_print_operand): Update.
17061 (ix86_split_fp_branch): Update.
17062 (predict_jump): Update.
17063 * config/ia64/ia64.c (ia64_print_operand): Update.
17064 * config/mmix/mmix.c (mmix_print_operand): Update.
17065 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17066 (rs6000_expand_split_stack_prologue): Update.
17067 * config/rs6000/rs6000.c: Update.
17068 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17069 (s390_expand_vec_movstr): Update.
17070 (s390_expand_cs_tdsi): Update.
17071 (s390_expand_split_stack_prologue): Update.
17072 * config/sh/sh.c (sh_print_operand): Update.
17073 (expand_cbranchsi4): Update.
17074 (expand_cbranchdi4): Update.
17075 * config/sparc/sparc.c (output_v9branch): Update.
17076 * config/spu/spu.c (get_branch_target): Update.
17077 (ea_load_store_inline): Update.
17078 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17079 * config/tilepro/tilepro.c: Update.
17081 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17083 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17084 (gimplify_save_expr): Likewise.
17086 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17088 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17090 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17092 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17094 * timevar.def (TV_IPA_FNSUMMARY): Define.
17096 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17098 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17099 to back store errata sensitive sequence from being generated.
17100 (sqrtdf2_fix): Likewise.
17102 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17104 * tree-ssa-threadupdate.c (compute_path_counts,
17105 update_joiner_offpath_counts): Use profile_probability.
17107 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17110 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17112 * config/arm/arm-c.c (arm_cpu_builtins): Define
17113 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17115 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17117 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17118 array entries to represent __ieee128 versions of the
17119 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17120 scalar_extract_sig, and scalar_insert_exp built-in functions.
17121 (altivec_resolve_overloaded_builtin): Add special case handling
17122 for the __builtin_scalar_insert_exp function, as represented by
17123 the P9V_BUILTIN_VEC_VSIEDP constant.
17124 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17125 exponent support for __ieee128 argument.
17126 (VSESQP): Add scalar extract signature support for __ieee128
17128 (VSTDCNQP): Add scalar test negative support for __ieee128
17130 (VSIEQP): Add scalar insert exponent support for __int128 argument
17131 with __ieee128 result.
17132 (VSIEQPF): Add scalar insert exponent support for __ieee128
17133 argument with __ieee128 result.
17134 (VSTDCQP): Add scalar test data class support for __ieee128
17136 (VSTDCNQP): Add overload support for scalar test negative with
17137 __ieee128 argument.
17138 (VSTDCQP): Add overload support for scalar test data class
17139 __ieee128 argument.
17140 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17141 UNSPEC_VSX_SXSIGDP.
17142 (UNSPEC_VSX_SIEXPQP): New constant.
17143 (xsxexpqp): New insn for VSX scalar extract exponent quad
17145 (xsxsigqp): New insn for VSX scalar extract significand quad
17147 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17148 precision with floating point argument.
17149 (xststdcqp): New expand for VSX scalar test data class quad
17151 (xststdcnegqp): New expand for VSX scalar test negative quad
17153 (xststdcqp): New insn to match expansions for VSX scalar test data
17154 class quad precision and VSX scalar test negative quad precision.
17155 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17156 special case operand checking to enforce that second operand of
17157 VSX scalar test data class with quad precision argument is a 7-bit
17159 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17160 prototypes and descriptions of __ieee128 versions of
17161 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17162 scalar_test_data_class, and scalar_test_neg built-in functions.
17164 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17166 PR tree-optimization/81162
17167 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17168 replace a negate with an add.
17170 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17172 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17174 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17176 * config/arm/arm-c.c (arm_cpu_builtins): Define
17177 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17179 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17181 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17182 (armv8-r): Set ARM Cortex-R52 as default CPU.
17183 * config/arm/arm-tables.opt: Regenerate.
17184 * config/arm/arm-tune.md: Regenerate.
17185 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17187 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17188 extension for -mcpu=cortex-r52.
17190 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17192 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17193 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17194 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17195 (fp-armv8): Define it as FP_ARMv8 only.
17196 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17197 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17199 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17200 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17201 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17202 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17203 than TARGET_FPU_ARMV8.
17204 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17205 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17206 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17208 * config/arm/neon.md (neon_vrint): Likewise.
17209 (neon_vcvt): Likewise.
17210 (neon_<fmaxmin_op><mode>): Likewise.
17211 (<fmaxmin><mode>3): Likewise.
17212 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17213 * config/arm/predicates.md (arm_cond_move_operator): Check against
17214 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17216 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17218 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17219 to top of function.
17221 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17223 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17224 loop in comment with memset.
17226 2017-07-14 Martin Liska <mliska@suse.cz>
17228 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17229 * dwarf2out.c (is_java): Remove the function.
17230 (output_pubname): Remove usage of the function.
17231 (lower_bound_default): Remove usage of DW_LANG_Java.
17232 (gen_compile_unit_die): Likewise.
17233 * gcc.c: Remove compiler defaults for .java and .zip files.
17234 * gimple-expr.c (remove_suffix): Change as there's no longer
17235 extension than 4-letter one.
17236 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17237 (gimplify_save_expr): Likewise.
17238 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17239 as it's possible even for other languages than Java.
17240 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17241 * lto-opts.c (lto_write_options): Remove reference to Java.
17242 * opts.c (strip_off_ending): Update file extension handling.
17243 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17244 * tree-eh.c (lower_resx): Likewise.
17245 * tree.c (free_lang_data_in_type): Remove dead code.
17246 (find_decls_types_r): Likewise.
17247 (build_common_builtin_nodes): Remove Java from a comment.
17248 (verify_type): Remove dead code.
17249 * varasm.c (assemble_external): Remove Java from a comment.
17251 2017-07-14 Martin Liska <mliska@suse.cz>
17253 * opts.c (finish_options): Add quotes.
17254 (common_handle_option): Likewise.
17256 2017-07-14 Martin Liska <mliska@suse.cz>
17258 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17259 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17260 Remove N_SO_PASCAL.
17261 * dwarf2out.c (lower_bound_default): Do not handle
17263 (gen_compile_unit_die): Likewise.
17264 * gcc.c: Remove default extension binding for GNU Pascal.
17265 * stmt.c: Remove Pascal language from a comment.
17266 * xcoffout.c: Likewise.
17268 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17271 * diagnostic-show-locus.c (fixit_cmp): New function.
17272 (layout::layout): Sort m_fixit_hints.
17273 (column_range::column_range): Assert that the values are valid.
17274 (struct char_span): New struct.
17275 (correction::overwrite): New method.
17276 (struct source_line): New struct.
17277 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17278 calls in terms of classes source_line and char_span, and
17279 correction::overwrite.
17280 (selftest::test_overlapped_fixit_printing_2): New function.
17281 (selftest::diagnostic_show_locus_c_tests): Call it.
17283 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17285 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17286 early if there is no lhs.
17288 2017-07-13 Martin Liska <mliska@suse.cz>
17290 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17291 (gen_reference_type_die): Likewise.
17292 * stor-layout.c: Remove Pascal-related comment.
17294 2017-07-13 Martin Liska <mliska@suse.cz>
17296 * opts.c (finish_options): Add quotes to error messages.
17297 (parse_sanitizer_options): Likewise.
17299 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17301 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17303 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17305 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17307 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17309 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17311 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17313 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17316 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17317 provides the hardware capability bits, define the macro
17318 __BUILTIN_CPU_SUPPORTS__.
17319 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17320 if GLIBC does not provide the hardware capability bits. Add a
17321 gcc_unreachable call if the built-in cpu function is neither
17322 __builtin_cpu_is nor __builtin_cpu_supports.
17323 (rs6000_get_function_versions_dispatcher): Change the warning
17324 that an old GLIBC is used which does not export the capability
17325 bits to be an error.
17326 * doc/extend.texi (target_clones attribute): Document the
17327 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17328 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17329 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17330 the macros defined by GCC if the newer GLIBC is available.
17332 2017-07-12 Jeff Law <law@redhat.com>
17334 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17335 remaining includes slightly.
17336 * config/riscv/riscv-builtins.c: Include profile-count.h.
17338 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17341 * config/avr/avr.c (avr_set_current_function): In diagnostic
17342 messages: Quote keywords and (parts of) identifiers.
17343 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17346 2017-07-12 Carl Love <cel@us.ibm.com>
17348 * config/rs6000/rs6000-c.c: Add support for built-in functions
17349 vector bool char vec_revb (vector bool char);
17350 vector bool short vec_revb (vector short char);
17351 vector bool int vec_revb (vector bool int);
17352 vector bool long long vec_revb (vector bool long long);
17353 * doc/extend.texi: Update the built-in documentation file for the
17354 new built-in functions.
17356 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17358 * config/s390/s390.md: Remove movcc splitter.
17360 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17362 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17363 load/store on condition.
17365 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17368 * config/avr/avr.c (avr_encode_section_info)
17369 [progmem && !TREE_READONLY]: Error if progmem object needs
17372 2017-07-11 Michael Collison <michael.collison@arm.com>
17374 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17377 2017-07-11 Carl Love <cel@us.ibm.com>
17379 * config/rs6000/rs6000-c.c: Add support for builtins
17380 vector unsigned int vec_parity_lsbb (vector signed int);
17381 vector unsigned int vec_parity_lsbb (vector unsigned int);
17382 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17383 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17384 vector unsigned long long vec_parity_lsbb (vector signed long long);
17385 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17386 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17387 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17388 * doc/extend.texi: Update the built-in documentation file for the
17389 new built-in functions.
17391 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17393 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17394 (layout::m_primary_loc): New field.
17395 (layout::layout): Initialize new field. Move location filtering
17396 logic from here to...
17397 (layout::maybe_add_location_range): ...this new method. Add
17398 support for filtering to just the lines already specified by other
17400 (layout::will_show_line_p): New method.
17401 (gcc_rich_location::add_location_if_nearby): New method.
17402 (selftest::test_add_location_if_nearby): New test function.
17403 (selftest::diagnostic_show_locus_c_tests): Call it.
17404 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17407 2017-07-11 Tom de Vries <tom@codesourcery.com>
17409 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17410 (bb_first_real_insn): New function.
17411 (nvptx_single): Add extra initialization of broadcasted condition
17414 2017-07-11 Nathan Sidwell <nathan@acm.org>
17416 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17418 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17420 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17421 Binutils doc as TEXI will mess them up.
17422 * doc/invoke.texi (AVR Options): Same here.
17424 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17426 * config/sparc/sparc.opt (mfix-ut700): New option.
17427 (mfix-gr712rc): Likewise.
17428 (sparc_fix_b2bst): New variable.
17429 * doc/invoke.texi (SPARC options): Document them.
17430 (ARM options): Fix warnings.
17431 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17432 instructions to prevent sequences that can trigger the store-store
17433 errata for certain LEON3FT processors.
17434 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17435 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17436 * config/sparc/sparc.md (fix_b2bst): New attribute.
17437 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17439 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17442 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17444 (*rsqrtsf2_sse): Ditto.
17446 (div<mode>3): Macroize insn from divdf3 and divsf3
17447 using MODEF mode iterator.
17449 2017-07-10 Martin Sebor <msebor@redhat.com>
17451 PR tree-optimization/80397
17452 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17453 instead of testing for equality to INTEGER_TYPE.
17455 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17457 * config.gcc: Remove uclibc from arc target spec.
17459 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17461 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17463 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17466 * lto-wrapper.c (remove_option): New function.
17467 (merge_and_complain): Merge PIC/PIE options more realistically.
17469 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17471 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17475 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17476 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17477 * config.in: Regenerate.
17478 * configure: Regenerate.
17479 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17480 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17481 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17482 (TARGET_GASISR_PROLOGUES): ...target mask.
17483 * common/config/avr/avr-common.c
17484 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17485 Set -mgas-isr-prologues.
17486 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17487 INSERT_PASS_BEFORE for it.
17488 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17489 * config/avr/avr.c (avr_option_override)
17490 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17491 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17492 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17493 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17494 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17496 (make_avr_pass_pre_proep): New function.
17497 (emit_push_sfr) <treg>: Add argument to function and use it
17498 instead of TMP_REG.
17499 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17500 and set machine->gasisr.yes.
17501 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17502 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17503 __gcc_isr.n_pushed to .L__stack_usage.
17504 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17505 (avr_asm_final_postscan_insn): ...this new static function.
17506 * config/avr/avr.h (machine_function)
17507 <is_no_gccisr, use_L__stack_usage>: New fields.
17508 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17509 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17510 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17511 (gasisr, *gasisr): New expander and insn.
17512 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17513 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17514 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17516 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17518 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17521 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17523 Move jump-tables out of .text again.
17526 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17527 (ASM_OUTPUT_ADDR_VEC): New function.
17528 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17529 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17530 INSN_ADDRESSes as asm comment.
17531 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17532 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17533 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17534 * config/avr/avr.md (*tablejump): Adjust comment.
17535 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17536 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17538 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17539 (avr_output_addr_vec): New proto.
17540 (avr_log_t) <insn_addresses>: New field.
17542 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17545 * config/i386/i386.c (ix86_function_arg_advance): Set
17546 outgoing_args_on_stack to true if there are outgoing arguments
17548 (ix86_function_arg): Likewise.
17549 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17550 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17551 * config/i386/i386.h (machine_function): Add
17552 outgoing_args_on_stack.
17554 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17556 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17557 supporting pthreds.
17558 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17560 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17562 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17563 (REAL_H): Remove $(MACHMODE_H).
17564 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17566 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17567 $(MACHMODE_H) and double-int.h.
17568 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17570 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17573 2017-07-07 Andrew Pinski <apinski@cavium.com>
17575 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17576 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17578 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17580 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17581 Add warning if GCC was not configured to link against a GLIBC that
17582 exports the hardware capability bits.
17583 (make_resolver_func): Make resolver function private and not a
17584 COMDAT function. Create the name with clone_function_name instead
17585 of make_unique_name.
17588 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17589 correct operand in doing the split.
17591 2017-07-07 Carl Love <cel@us.ibm.com>
17593 * config/rs6000/rs6000-c: Add support for built-in function
17594 vector unsigned short vec_pack_to_short_fp32 (vector float,
17596 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17597 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17598 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17599 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17600 (convert_4f32_8i16): Add define_expand.
17601 * doc/extend.texi: Update the built-in documentation file for the
17602 new built-in function.
17604 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17606 * config/sparc/m8.md: New file.
17607 * config/sparc/sparc.md: Include m8.md.
17609 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17611 * config/sparc/sparc.opt: New option -mvis4b.
17612 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17613 (sparc_option_override): Handle VIS4B.
17614 (enum sparc_builtins): Define
17615 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17616 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17617 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17618 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17619 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17620 (check_constant_argument): New function.
17621 (sparc_vis_init_builtins): Define builtins
17622 __builtin_vis_dictunpack{8,16,32},
17623 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17624 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17625 __builtin_vis_fpcmpde{8,16,32}shl and
17626 __builtin_vis_fpcmpur{8,16,32}shl.
17627 (sparc_expand_builtin): Check that the constant operands to
17628 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17629 constant and in range.
17630 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17632 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17633 (SPARC_IMM5_P): Likewise.
17634 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17635 (enabled): Handle vis4b.
17636 (UNSPEC_DICTUNPACK): New unspec.
17637 (UNSPEC_FPCMPSHL): Likewise.
17638 (UNSPEC_FPUCMPSHL): Likewise.
17639 (UNSPEC_FPCMPDESHL): Likewise.
17640 (UNSPEC_FPCMPURSHL): Likewise.
17641 (cpu_feature): New CPU feature `vis4b'.
17642 (dictunpack{8,16,32}): New insns.
17643 (FPCSMODE): New mode iterator.
17644 (fpcscond): New code iterator.
17645 (fpcsucond): Likewise.
17646 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17647 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17648 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17649 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17650 * config/sparc/constraints.md: Define constraints `q' for unsigned
17651 2-bit integer constants and `t' for unsigned 5-bit integer
17653 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17655 (imm5_operand_dictunpack16): Likewise.
17656 (imm5_operand_dictunpack32): Likewise.
17657 (imm2_operand): Likewise.
17658 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17659 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17660 ditunpack* and fpcmp*shl builtins.
17662 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17664 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17665 * config.in: Add HAVE_AS_SPARC6 define.
17666 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17668 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17670 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17671 (CPP_CPU_SPEC): Handle m8.
17672 (ASM_CPU_SPEC): Likewise.
17673 * config/sparc/sparc-opts.h (enum processor_type): Add
17675 * config/sparc/sparc.c (m8_costs): New struct.
17676 (sparc_option_override): Handle TARGET_CPU_m8.
17677 (sparc32_initialize_trampoline): Likewise.
17678 (sparc64_initialize_trampoline): Likewise.
17679 (sparc_issue_rate): Likewise.
17680 (sparc_register_move_cost): Likewise.
17681 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17682 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17683 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17684 (CPP_CPU_SPEC): Handle M8.
17685 (ASM_CPU_SPEC): Likewise.
17686 (AS_M8_FLAG): Define.
17687 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17688 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17689 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17691 * configure: Regenerate.
17692 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17695 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17697 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17699 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17700 ("*movdi_insn_sp32"): Do not set v3pipe.
17701 ("*movsi_insn"): Likewise.
17702 ("*movdi_insn_sp64"): Likewise.
17703 ("*movsf_insn"): Likewise.
17704 ("*movdf_insn_sp32"): Likewise.
17705 ("*movdf_insn_sp64"): Likewise.
17706 ("*zero_extendsidi2_insn_sp64"): Likewise.
17707 ("*sign_extendsidi2_insn"): Likewise.
17708 ("*mov<VM32:mode>_insn"): Likewise.
17709 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17710 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17711 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17712 ("<vlop:code><VL:mode>3"): Likewise.
17713 ("*not_<vlop:code><VL:mode>3"): Likewise.
17714 ("*nand<VL:mode>_vis"): Likewise.
17715 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17716 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17717 ("one_cmpl<VL:mode>2"): Likewise.
17718 ("faligndata<VM64:mode>_vis"): Likewise.
17719 ("alignaddrsi_vis"): Likewise.
17720 ("alignaddrdi_vis"): Likweise.
17721 ("alignaddrlsi_vis"): Likewise.
17722 ("alignaddrldi_vis"): Likewise.
17723 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17724 ("bmaskdi_vis"): Likewise.
17725 ("bmasksi_vis"): Likewise.
17726 ("bshuffle<VM64:mode>_vis"): Likewise.
17727 ("cmask8<P:mode>_vis"): Likewise.
17728 ("cmask16<P:mode>_vis"): Likewise.
17729 ("cmask32<P:mode>_vis"): Likewise.
17730 ("pdistn<P:mode>_vis"): Likewise.
17731 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17733 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17735 * config/sparc/sparc.md ("subtype"): New insn attribute.
17736 ("*wrgsr_sp64"): Set insn subtype.
17737 ("*rdgsr_sp64"): Likewise.
17738 ("alignaddrsi_vis"): Likewise.
17739 ("alignaddrdi_vis"): Likewise.
17740 ("alignaddrlsi_vis"): Likewise.
17741 ("alignaddrldi_vis"): Likewise.
17742 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17743 ("fexpand_vis"): Likewise.
17744 ("fpmerge_vis"): Likewise.
17745 ("faligndata<VM64:mode>_vis"): Likewise.
17746 ("bshuffle<VM64:mode>_vis"): Likewise.
17747 ("cmask8<P:mode>_vis"): Likewise.
17748 ("cmask16<P:mode>_vis"): Likewise.
17749 ("cmask32<P:mode>_vis"): Likewise.
17750 ("fchksm16_vis"): Likewise.
17751 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17752 ("fmean16_vis"): Likewise.
17753 ("fp<plusminus_insn>64_vis"): Likewise.
17754 ("<plusminus_insn>v8qi3"): Likewise.
17755 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17756 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17757 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17758 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17759 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17760 ("*movqi_insn"): Likewise.
17761 ("*movhi_insn"): Likewise.
17762 ("*movsi_insn"): Likewise.
17763 ("movsi_pic_gotdata_op"): Likewise.
17764 ("*movdi_insn_sp32"): Likewise.
17765 ("*movdi_insn_sp64"): Likewise.
17766 ("movdi_pic_gotdata_op"): Likewise.
17767 ("*movsf_insn"): Likewise.
17768 ("*movdf_insn_sp32"): Likewise.
17769 ("*movdf_insn_sp64"): Likewise.
17770 ("*zero_extendhisi2_insn"): Likewise.
17771 ("*zero_extendqihi2_insn"): Likewise.
17772 ("*zero_extendqisi2_insn"): Likewise.
17773 ("*zero_extendqidi2_insn"): Likewise.
17774 ("*zero_extendhidi2_insn"): Likewise.
17775 ("*zero_extendsidi2_insn_sp64"): Likewise.
17776 ("ldfsr"): Likewise.
17777 ("prefetch_64"): Likewise.
17778 ("prefetch_32"): Likewise.
17779 ("tie_ld32"): Likewise.
17780 ("tie_ld64"): Likewise.
17781 ("*tldo_ldub_sp32"): Likewise.
17782 ("*tldo_ldub1_sp32"): Likewise.
17783 ("*tldo_ldub2_sp32"): Likewise.
17784 ("*tldo_ldub_sp64"): Likewise.
17785 ("*tldo_ldub1_sp64"): Likewise.
17786 ("*tldo_ldub2_sp64"): Likewise.
17787 ("*tldo_ldub3_sp64"): Likewise.
17788 ("*tldo_lduh_sp32"): Likewise.
17789 ("*tldo_lduh1_sp32"): Likewise.
17790 ("*tldo_lduh_sp64"): Likewise.
17791 ("*tldo_lduh1_sp64"): Likewise.
17792 ("*tldo_lduh2_sp64"): Likewise.
17793 ("*tldo_lduw_sp32"): Likewise.
17794 ("*tldo_lduw_sp64"): Likewise.
17795 ("*tldo_lduw1_sp64"): Likewise.
17796 ("*tldo_ldx_sp64"): Likewise.
17797 ("*mov<VM32:mode>_insn"): Likewise.
17798 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17799 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17801 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17803 * config/sparc/sparc.md ("type"): New insn type viscmp.
17804 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17806 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17807 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17808 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17809 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17811 ("n7_vis_logical_11cycle"): Likewise.
17812 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17813 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17814 * config/sparc/niagara.md ("niag_vis"): Likewise.
17815 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17816 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17820 * config/sparc/sparc.md: New instruction type `bmask'.
17821 (bmaskdi_vis): Use the `bmask' type.
17822 (bmasksi_vis): Likewise.
17823 * config/sparc/ultra3.md (us3_array): Likewise.
17824 * config/sparc/niagara7.md (n7_array): Likewise.
17825 * config/sparc/niagara4.md (n4_array): Likewise.
17826 * config/sparc/niagara2.md (niag2_vis): Likewise.
17827 (niag3_vis): Likewise.
17828 * config/sparc/niagara.md (niag_vis): Likewise.
17830 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17832 * ipa-comdats.c: Remove optimize check from gate.
17833 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17834 for functions not optimized.
17835 (ipa_fn_summary_read): Skip optimize check.
17836 (ipa_fn_summary_write): Likewise.
17837 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17839 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17841 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17842 (check_callers): Give up on caller which is not optimized.
17843 (inline_small_functions): Likewise.
17844 (ipa_inline): Do not give up when not optimizing.
17845 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17846 away unoptimizes cdtors.
17847 (whole_program_function_and_variable_visibility): Do
17848 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17849 * ipa.c (process_references): Do not check optimize.
17850 (symbol_table::remove_unreachable_nodes): Update optimize check.
17851 (set_writeonly_bit): Update optimize check.
17852 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17853 (pass_ipa_single_use::gate): Remove.
17855 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17857 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17858 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17859 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17860 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17861 permute_load, permute_store, adjust_extract, adjust_splat,
17862 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17863 replace_swap_with_copy, dump_swap_insn_table,
17864 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17865 recombine_lvx_pattern, recombine_stvx_pattern,
17866 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17867 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17868 to file rs6000-p8swap.c.
17869 * config/rs6000/rs6000-p8swap.c: New file.
17870 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17871 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17872 and rs6000*-*-* targets.
17874 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17876 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17878 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17880 * lto-wrapper.c (merge_and_complain): Do not merge
17881 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17882 fsigned_zeros, ftrapping_math, fwrapv.
17883 (append_compiler_options): Do not track these options.
17884 (append_linker_options): Likewie
17886 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17888 * cgraphunit.c (cgraph_node::finalize_function): When
17889 !flag_toplevel_reorde set no_reorder flag.
17890 (varpool_node::finalize_decl): Likewise.
17891 (symbol_table::compile): Drop no toplevel reorder path.
17893 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17895 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17896 edges; zero probability is not better than uninitialized.
17898 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17900 * asan.h (asan_sanitize_allocas_p): Declare.
17901 * asan.c (asan_sanitize_allocas_p): New function.
17902 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17903 (handle_builtin_alloca): Likewise.
17904 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17905 if !asan_sanitize_allocas_p.
17906 * params.def (asan-instrument-allocas): Add new option.
17907 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17908 * opts.c (common_handle_option): Disable allocas sanitization for
17911 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17913 * asan.c: Include gimple-fold.h.
17914 (get_last_alloca_addr): New function.
17915 (handle_builtin_stackrestore): Likewise.
17916 (handle_builtin_alloca): Likewise.
17917 (asan_emit_allocas_unpoison): Likewise.
17918 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17919 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17920 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17921 (instrument_builtin_call): Pass gimple iterator to
17922 get_mem_refs_of_builtin_call.
17923 (last_alloca_addr): New global.
17924 * asan.h (asan_emit_allocas_unpoison): Declare.
17925 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17926 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17927 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17928 if function calls alloca.
17929 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17930 * gimple-fold.h (replace_call_with_value): Declare.
17931 * internal-fn.c: Include asan.h.
17932 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17933 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17935 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17937 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17938 (C_SELFTEST_FLAGS): New.
17939 (CPP_SELFTEST_FLAGS): New.
17940 (SELFTEST_DEPS): New, from deps of s-selftest.
17941 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17942 (CPP_SELFTEST_DEPS): New.
17943 (selftest): Add dependency on s-selftest-c++.
17944 (s-selftest): Rename to...
17945 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17946 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17947 than SELFTEST_FLAGS.
17948 (selftest-gdb): Rename to...
17949 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17951 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17952 (selftest-valgrind): Rename to...
17953 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17955 (selftest-valgrind): Reintroduce as an alias for
17956 selftest-c-valgrind.
17957 (s-selftest-c++): New.
17958 (selftest-c++-gdb): New.
17959 (selftest-c++-valgrind): New.
17961 2017-07-06 Olivier Hainque <hainque@adacore.com>
17963 * gcc.c (process_command): When deciding if undefined variables
17964 should be ignored when processing specs, accept "gcc -v" as well.
17966 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17968 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17969 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17971 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17973 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17974 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17975 * config/arm/arm-tables.opt: Regenerate.
17976 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17978 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17980 2017-07-06 Carl Love <cel@us.ibm.com>
17982 * ChangeLog: Clean up from mid air collision
17984 2017-07-06 Carl Love <cel@us.ibm.com>
17986 * config/rs6000/rs6000-c.c: Add support for built-in functions
17987 vector signed int vec_subc (vector signed int, vector signed int);
17988 vector signed __int128 vec_subc (vector signed __int128,
17989 vector signed __int128);
17990 vector unsigned __int128 vec_subc (vector unsigned __int128,
17991 vector unsigned __int128);
17992 vector signed int vec_sube (vector signed int, vector signed int,
17993 vector signed int);
17994 vector unsigned int vec_sube (vector unsigned int,
17995 vector unsigned int,
17996 vector unsigned int);
17997 vector signed __int128 vec_sube (vector signed __int128,
17998 vector signed __int128,
17999 vector signed__int128);
18000 vector unsigned __int128 vec_sube (vector unsigned __int128,
18001 vector unsigned __int128,
18002 vector unsigned __int128);
18003 vector signed int vec_subec (vector signed int, vector signed int,
18004 vector signed int);
18005 vector unsigned int vec_subec (vector unsigned int,
18006 vector unsigned int,
18007 vector unsigned int);
18008 vector signed __int128 vec_subec (vector signed __int128,
18009 vector signed __int128,
18010 vector signed__int128);
18011 vector unsigned __int128 vec_subec (vector unsigned __int128,
18012 vector unsigned __int128,
18013 vector unsigned __int128);
18014 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18015 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18016 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18017 BU_ALTIVEC_OVERLOAD_X definitions.
18018 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18019 * doc/extend.texi: Update the built-in documentation file for the new
18020 built-in functions.
18022 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18025 * diagnostic-show-locus.c (layout::layout): Use start and finish
18026 spelling location for the start and finish of each range.
18027 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18028 Add unused aspect param.
18029 * input.c (expand_location_1): Add "aspect" param, and use it
18030 to access the correct part of the location.
18031 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18033 (expand_location_to_spelling_point): Likewise.
18034 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18035 param, and pass it to expand_location_1.
18037 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18039 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18040 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18041 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18042 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18043 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18044 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18045 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18046 _mm_maskz_getmant_ss): New intrinsics.
18047 (__builtin_ia32_getexpss128_mask): Changed to ...
18048 __builtin_ia32_getexpss128_round ... this.
18049 (__builtin_ia32_getexpsd128_mask): Changed to ...
18050 __builtin_ia32_getexpsd128_round ... this.
18051 * config/i386/i386-builtin-types.def
18052 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18053 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18054 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18055 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18056 __builtin_ia32_getmantss_mask_round): New builtins.
18057 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18058 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18059 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18060 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18061 * config/i386/sse.md
18062 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18063 avx512f_sgetexp<mode><mask_scalar_name>
18064 <round_saeonly_scalar_name> ... this.
18065 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18066 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18067 vgetexp<ssescalarmodesuffix>
18068 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18069 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18070 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18071 avx512f_vgetmant<mode><mask_scalar_name>
18072 <round_saeonly_scalar_name> ... this.
18073 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18074 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18075 vgetmant<ssescalarmodesuffix>
18076 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18077 %0<mask_scalar_operand4>, %1, %2
18078 <round_saeonly_scalar_mask_op4>, %3} ... this.
18079 * config/i386/subst.md (mask_scalar_operand4,
18080 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18081 round_saeonly_scalar_nimm_predicate): New subst attributes.
18083 2017-07-06 Julia Koval <julia.koval@intel.com>
18085 * config/i386/i386.c (ix86_erase_embedded_rounding):
18086 Remove code for old rounding pattern.
18088 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18090 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18092 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18094 * doc/sourcebuild.texi (Test Directives, Variants of
18095 dg-require-support): Add documentation for dg-require-stack-check.
18097 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18099 * config/i386/subst.md (mask_scalar, round_scalar,
18100 round_saeonly_scalar): New meta-templates.
18101 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18102 round_scalar_mask_operand3, round_scalar_mask_op3,
18103 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18104 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18105 round_saeonly_scalar_constraint,
18106 round_saeonly_scalar_prefix): New subst attribute.
18107 * config/i386/sse.md
18108 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18109 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18110 <round_scalar_name> ... this.
18111 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18112 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18113 <round_scalar_name> ... this.
18114 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18115 <sse>_vm<code><mode>3<mask_scalar_name>
18116 <round_saeonly_scalar_name> ... this.
18117 (v<plusminus_mnemonic><ssescalarmodesuffix>
18118 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18119 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18120 v<plusminus_mnemonic><ssescalarmodesuffix>
18121 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18122 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18123 (v<multdiv_mnemonic><ssescalarmodesuffix>
18124 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18125 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18126 v<multdiv_mnemonic><ssescalarmodesuffix>
18127 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18128 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18129 (v<maxmin_float><ssescalarmodesuffix>
18130 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18131 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18132 v<maxmin_float><ssescalarmodesuffix>
18133 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18134 %0<mask_scalar_operand3>, %1, %<iptr>2
18135 <round_saeonly_scalar_mask_op3>} ... this.
18137 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18139 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18140 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18142 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18143 Alan Hayward <alan.hayward@arm.com>
18144 David Sherwood <david.sherwood@arm.com>
18146 * combine.c (simplify_if_then_else): Remove "enum" before
18148 * compare-elim.c (can_eliminate_compare): Likewise.
18149 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18151 (aarch64_lookup_simd_builtin_type): Likewise.
18152 (aarch64_simd_builtin_type): Likewise.
18153 (aarch64_init_simd_builtin_types): Likewise.
18154 (aarch64_simd_expand_args): Likewise.
18155 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18157 (aarch64_reverse_mask): Likewise.
18158 (aarch64_simd_emit_reg_reg_move): Likewise.
18159 (aarch64_gen_adjusted_ldpstp): Likewise.
18160 (aarch64_ccmp_mode_to_code): Likewise.
18161 (aarch64_operands_ok_for_ldpstp): Likewise.
18162 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18163 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18165 (aarch64_min_divisions_for_recip_mul): Likewise.
18166 (aarch64_reassociation_width): Likewise.
18167 (aarch64_get_condition_code_1): Likewise.
18168 (aarch64_simd_emit_reg_reg_move): Likewise.
18169 (aarch64_simd_attr_length_rglist): Likewise.
18170 (aarch64_reverse_mask): Likewise.
18171 (aarch64_operands_ok_for_ldpstp): Likewise.
18172 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18173 (aarch64_gen_adjusted_ldpstp): Likewise.
18174 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18176 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18177 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18178 (arm_lookup_simd_builtin_type): Likewise.
18179 (arm_simd_builtin_type): Likewise.
18180 (arm_init_simd_builtin_types): Likewise.
18181 (arm_expand_builtin_args): Likewise.
18182 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18183 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18184 (ft32_setup_incoming_varargs): Likewise.
18185 (ft32_function_arg): Likewise.
18186 (ft32_function_arg_advance): Likewise.
18187 (ft32_pass_by_reference): Likewise.
18188 (ft32_arg_partial_bytes): Likewise.
18189 (ft32_valid_pointer_mode): Likewise.
18190 (ft32_addr_space_pointer_mode): Likewise.
18191 (ft32_addr_space_legitimate_address_p): Likewise.
18192 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18194 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18195 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18196 (iamcu_alignment): Likewise.
18197 (canonicalize_vector_int_perm): Likewise.
18198 (ix86_noce_conversion_profitable_p): Likewise.
18199 (ix86_mpx_bound_mode): Likewise.
18200 (ix86_operands_ok_for_move_multiple): Likewise.
18201 * config/microblaze/microblaze-protos.h
18202 (microblaze_expand_conditional_branch_reg): Likewise.
18203 * config/microblaze/microblaze.c
18204 (microblaze_expand_conditional_branch_reg): Likewise.
18205 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18207 (rs6000_reassociation_width): Likewise.
18208 (rs6000_invalid_binary_op): Likewise.
18209 (fusion_p9_p): Likewise.
18210 (emit_fusion_p9_load): Likewise.
18211 (emit_fusion_p9_store): Likewise.
18212 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18214 (riscv_hard_regno_mode_ok_p): Likewise.
18215 (riscv_address_insns): Likewise.
18216 (riscv_split_symbol): Likewise.
18217 (riscv_legitimize_move): Likewise.
18218 (riscv_function_value): Likewise.
18219 (riscv_hard_regno_nregs): Likewise.
18220 (riscv_expand_builtin): Likewise.
18221 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18222 (riscv_build_integer): Likewise.
18223 (riscv_split_integer): Likewise.
18224 (riscv_legitimate_constant_p): Likewise.
18225 (riscv_cannot_force_const_mem): Likewise.
18226 (riscv_regno_mode_ok_for_base_p): Likewise.
18227 (riscv_valid_base_register_p): Likewise.
18228 (riscv_valid_offset_p): Likewise.
18229 (riscv_valid_lo_sum_p): Likewise.
18230 (riscv_classify_address): Likewise.
18231 (riscv_legitimate_address_p): Likewise.
18232 (riscv_address_insns): Likewise.
18233 (riscv_load_store_insns): Likewise.
18234 (riscv_force_binary): Likewise.
18235 (riscv_split_symbol): Likewise.
18236 (riscv_force_address): Likewise.
18237 (riscv_legitimize_address): Likewise.
18238 (riscv_move_integer): Likewise.
18239 (riscv_legitimize_const_move): Likewise.
18240 (riscv_legitimize_move): Likewise.
18241 (riscv_address_cost): Likewise.
18242 (riscv_subword): Likewise.
18243 (riscv_output_move): Likewise.
18244 (riscv_canonicalize_int_order_test): Likewise.
18245 (riscv_emit_int_order_test): Likewise.
18246 (riscv_function_arg_boundary): Likewise.
18247 (riscv_pass_mode_in_fpr_p): Likewise.
18248 (riscv_pass_fpr_single): Likewise.
18249 (riscv_pass_fpr_pair): Likewise.
18250 (riscv_get_arg_info): Likewise.
18251 (riscv_function_arg): Likewise.
18252 (riscv_function_arg_advance): Likewise.
18253 (riscv_arg_partial_bytes): Likewise.
18254 (riscv_function_value): Likewise.
18255 (riscv_pass_by_reference): Likewise.
18256 (riscv_setup_incoming_varargs): Likewise.
18257 (riscv_print_operand): Likewise.
18258 (riscv_elf_select_rtx_section): Likewise.
18259 (riscv_save_restore_reg): Likewise.
18260 (riscv_for_each_saved_reg): Likewise.
18261 (riscv_register_move_cost): Likewise.
18262 (riscv_hard_regno_mode_ok_p): Likewise.
18263 (riscv_hard_regno_nregs): Likewise.
18264 (riscv_class_max_nregs): Likewise.
18265 (riscv_memory_move_cost): Likewise.
18266 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18267 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18268 (rl78_addr_space_address_mode): Likewise.
18269 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18271 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18272 (rs6000_reassociation_width): Likewise.
18273 (rs6000_invalid_binary_op): Likewise.
18274 (fusion_p9_p): Likewise.
18275 (emit_fusion_p9_load): Likewise.
18276 (emit_fusion_p9_store): Likewise.
18277 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18278 (ok_for_simple_move_operands): Likewise.
18279 (ok_for_simple_move_strict_operands): Likewise.
18280 (ok_for_simple_arith_logic_operands): Likewise.
18281 (visium_legitimize_reload_address): Likewise.
18282 (visium_select_cc_mode): Likewise.
18283 (output_cbranch): Likewise.
18284 (visium_split_double_move): Likewise.
18285 (visium_expand_copysign): Likewise.
18286 (visium_expand_int_cstore): Likewise.
18287 (visium_expand_fp_cstore): Likewise.
18288 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18289 (visium_function_arg): Likewise.
18290 (visium_function_arg_advance): Likewise.
18291 (visium_libcall_value): Likewise.
18292 (visium_setup_incoming_varargs): Likewise.
18293 (visium_legitimate_constant_p): Likewise.
18294 (visium_legitimate_address_p): Likewise.
18295 (visium_legitimize_address): Likewise.
18296 (visium_secondary_reload): Likewise.
18297 (visium_register_move_cost): Likewise.
18298 (visium_memory_move_cost): Likewise.
18299 (prepare_move_operands): Likewise.
18300 (ok_for_simple_move_operands): Likewise.
18301 (ok_for_simple_move_strict_operands): Likewise.
18302 (ok_for_simple_arith_logic_operands): Likewise.
18303 (visium_function_value_1): Likewise.
18304 (rtx_ok_for_offset_p): Likewise.
18305 (visium_legitimize_reload_address): Likewise.
18306 (visium_split_double_move): Likewise.
18307 (visium_expand_copysign): Likewise.
18308 (visium_expand_int_cstore): Likewise.
18309 (visium_expand_fp_cstore): Likewise.
18310 (visium_split_cstore): Likewise.
18311 (visium_select_cc_mode): Likewise.
18312 (visium_split_cbranch): Likewise.
18313 (output_cbranch): Likewise.
18314 (visium_print_operand_address): Likewise.
18315 * expmed.c (flip_storage_order): Likewise.
18316 * expmed.h (emit_cstore): Likewise.
18317 (flip_storage_order): Likewise.
18318 * genrecog.c (validate_pattern): Likewise.
18319 * hsa-gen.c (gen_hsa_addr): Likewise.
18320 * internal-fn.c (expand_arith_overflow): Likewise.
18321 * ira-color.c (allocno_copy_cost_saving): Likewise.
18322 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18323 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18324 (process_invariant_for_inheritance): Likewise.
18325 * lra-eliminations.c (move_plus_up): Likewise.
18326 * omp-low.c (lower_oacc_reductions): Likewise.
18327 * simplify-rtx.c (simplify_subreg): Likewise.
18328 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18329 (TARGET_CHKP_BOUND_MODE): Likewise..
18330 * targhooks.c (default_chkp_bound_mode): Likewise.
18331 (default_setup_incoming_vararg_bounds): Likewise.
18332 * targhooks.h (default_chkp_bound_mode): Likewise.
18333 (default_setup_incoming_vararg_bounds): Likewise.
18334 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18335 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18336 (have_whole_vector_shift): Likewise.
18337 * tree-vect-stmts.c (vectorizable_load): Likewise.
18338 * doc/tm.texi: Regenerate.
18340 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18342 Graceful degrade if Binutils PR21472 is not available.
18345 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18346 .rodata in flash test fails.
18347 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18348 * confgure: Regenerate.
18349 * config.in: Regenerate.
18350 * config/avr/avr.c (avr_asm_named_section)
18351 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18352 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18353 (avr_asm_init_sections): Same.
18355 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18357 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18358 (fma<VH:mode>4_intrinsic): Likewise.
18359 (*fmsub<VCVTF:mode>4): Likewise.
18360 (*fmsub<VH:mode>4_intrinsic): Likewise.
18362 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18365 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18366 Don't depend on "optimize > 0".
18367 (out_movhi_r_mr, out_movqi_mr_r): Same.
18368 (out_movhi_mr_r, out_movqi_r_mr): Same.
18369 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18370 io_address_operand on "optimize > 0".
18372 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18374 * tree-loop-distribution.c: Add general explanantion on the pass.
18375 (generate_loops_for_partition): Mark distributed loop.
18376 (pg_add_dependence_edges): New parameter. Handle alias data
18377 dependence specially and record it in the parameter if asked.
18378 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18379 (init_partition_graph_vertices, add_partition_graph_edge): New.
18380 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18381 (free_partition_graph_vdata, build_partition_graph): New.
18382 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18383 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18384 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18385 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18386 (version_for_distribution_p, finalize_partitions): New.
18387 (distribute_loop): Handle alias data dependence specially. Factor
18388 out loop fusion code as functions and call these functions.
18390 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18392 * tree-loop-distribution.c (classify_partition): New parameter and
18393 better handle reduction statement.
18394 (rdg_build_partitions): Revise comment.
18395 (distribute_loop): Compute statements in all partitions and pass it
18396 to classify_partition.
18398 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18400 * tree-loop-distribution.c (enum partition_type): New.
18401 (struct partition): New field type.
18402 (partition_merge_into): Add parameter. Update partition type.
18403 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18404 (build_rdg_partition_for_vertex): Compute partition type.
18405 (rdg_build_partitions): Dump partition type.
18406 (distribute_loop): Update calls to partition_merge_into.
18408 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18410 * tree-loop-distribution.c (struct ddr_hasher): New.
18411 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18413 (classify_partition): Call get_data_dependence.
18414 (pg_add_dependence_edges): Ditto.
18415 (distribute_loop): Release data dependence hash table.
18417 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18419 * tree-loop-distribution.c (ref_base_address): Delete.
18420 (similar_memory_accesses): Rename ...
18421 (share_memory_accesses): ... to this. Check if partitions access
18422 the same memory reference.
18423 (distribute_loop): Call share_memory_accesses.
18425 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18427 * tree-loop-distribution.c (struct partition): New field recording
18428 its data reference.
18429 (partition_alloc, partition_free): Init and release data refs.
18430 (partition_merge_into): Merge data refs.
18431 (build_rdg_partition_for_vertex): Collect data refs for partition.
18432 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18434 (distribute_loop): Remve data refs from vertice data of partition
18437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18439 * tree-loop-distribution.c (params.h): Include header file.
18440 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18441 (datarefs_vec): New global var.
18442 (create_rdg_vertices): Use datarefs_vec directly.
18443 (free_rdg): Don't free data references.
18444 (build_rdg): Update use. Don't free data references.
18445 (distribute_loop): Compute global variable for data references.
18446 Bail out if there are too many data references.
18448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18450 * tree-loop-distribution.c (loop_nest): New global var.
18451 (build_rdg): Use loop directly, rather than loop nest.
18452 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18454 (distribute_loop): Compute global variable loop nest. Update use.
18456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18458 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18459 (partition_merge_into): New parameter. Dump reason for fusion.
18460 (distribute_loop): Update use of partition_merge_into.
18462 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18464 * tree-loop-distribution.c (bb_top_order_index): New.
18465 (bb_top_order_index_size, bb_top_order_cmp): New.
18466 (stmts_from_loop): Use topological order.
18467 (pass_loop_distribution::execute): Compute and release topological
18468 order for basic blocks.
18470 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18472 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18477 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18478 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18479 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18480 * internal-fn.def (LOOP_DIST_ALIAS): New.
18481 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18482 (fold_loop_internal_call): ... this.
18483 (vect_loop_dist_alias_call): New function.
18484 (set_uid_loop_bbs): Call fold_loop_internal_call.
18485 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18488 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18491 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18492 Require dead FLAGS_REG at the beginning of a peephole.
18494 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18497 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18498 arguments in the call to __builtin_ia32_sbb_u32.
18499 (_subborrow_u64): Swap _X and _Y arguments in the call to
18500 __builtin_ia32_sbb_u64.
18502 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18505 * tree-vrp.c (compare_assert_loc): Turn into a function template
18506 with stable template parameter. Only test if a->e is NULL,
18507 !a->e == !b->e has been verified already. Use e == NULL or
18508 e != NULL instead of e or ! e tests. If stable is true, don't use
18509 iterative_hash_expr, on the other side allow a or b or both NULL
18510 and sort the NULLs last.
18511 (process_assert_insertions): Sort using compare_assert_loc<false>
18512 instead of compare_assert_loc, later sort using
18513 compare_assert_loc<true> before calling process_assert_insertions_for
18514 in a loop. Use break instead of continue once seen NULL pointer.
18516 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18518 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18519 Cortex-R7 and Cortex-R8 processors.
18521 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18523 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18524 uninitialized while src is not.
18526 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18528 * common/config/arm/arm-common.c: Adjust include path for
18530 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18531 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18532 (arm-cpu-data.h): Likewise.
18533 (arm-cpu-cdata.h): Likewise.
18534 * config/arm/arm-cpu.h: Delete.
18535 * config/arm/arm-cpu-cdata.h: Delete.
18536 * config/arm/arm-cpu-data.h: Delete.
18538 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18540 * config/arm/arm-cpus.in (cortex-a55): New.
18541 (cortex-a75): Likewise.
18542 (cortex-a75.cortex-a55): Likewise.
18543 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18545 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18546 * config/arm/arm-cpu-cdata.h: Regenerate.
18547 * config/arm/arm-cpu-data.h: Regenerate.
18548 * config/arm/arm-cpu.h: Regenerate.
18549 * config/arm/arm-tables.opt: Regenerate.
18550 * config/arm/arm-tune.md: Regenerate.
18552 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18554 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18556 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18558 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18561 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18563 PR tree-optimization/81292
18564 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18565 full_string_p, also call adjust_related_strinfos if the adjustment
18566 is simple, otherwise invalidate related strinfos.
18568 2017-07-04 Martin Liska <mliska@suse.cz>
18571 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18572 newly created variable as DECL_IGNORED_P.
18574 2017-07-04 Martin Liska <mliska@suse.cz>
18577 * ipa-inline.c (inline_small_functions):
18578 Use xstrdup_for_dump.
18580 2017-07-04 Tom de Vries <tom@codesourcery.com>
18582 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18584 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18587 * config/darwin.c (darwin_function_switched_text_sections):
18590 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18592 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18594 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18596 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18598 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18600 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18601 min_profitable_iters, and th as inclusive lower bounds.
18602 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18603 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18604 for min_profitable_iters and min_profitable_estimate.
18605 (vect_transform_loop): Treat th as an inclusive lower bound.
18606 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18608 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18611 * config/darwin.c (darwin_function_switched_text_sections):
18612 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18613 in two pieces, and suppress the use of buf.
18615 2017-07-03 Nathan Sidwell <nathan@acm.org>
18617 * hash-table.h (hash_table_mod1): Fix indentation.
18619 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18621 PR middle-end/81290
18622 * predict.c (force_edge_cold): Be more careful about propagation
18624 * profile-count.h (profile_probability::guessed,
18625 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18627 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18629 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18631 * doc/invoke.texi (rcpc architecture extension): Document it.
18633 2017-07-03 Richard Biener <rguenther@suse.de>
18635 PR tree-optimization/60510
18636 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18637 the scalar reduction PHI and use it.
18638 (vectorizable_reduction): Properly guard the single_defuse_cycle
18639 path for non-SLP reduction chains where we cannot use it.
18640 Rework reduc_def/index and vector type deduction. Rework
18641 vector operand gathering during reduction op code-gen.
18642 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18643 chains dissolve the chain and leave it to non-SLP reduction
18646 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18648 * tree-data-ref.h (dr_alignment): Declare.
18649 * tree-data-ref.c (dr_alignment): New function.
18650 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18651 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18653 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18655 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18657 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18658 and base_misalignment fields.
18659 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18660 * tree-data-ref.c: Include builtins.h.
18661 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18662 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18663 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18664 * tree-vect-data-refs.c: Include tree-cfg.h.
18665 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18666 fields instead of calculating an alignment here.
18667 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18668 innermost_loop_behavior fields.
18670 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18672 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18674 (DR_STEP_ALIGNMENT): New macro.
18675 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18676 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18677 (create_data_ref): Print it.
18678 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18679 to tell whether the step preserves vector (mis)alignment.
18680 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18681 Move the check for an integer step and generalise to all INTEGER_CST.
18682 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18683 Print the outer step alignment.
18685 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18687 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18688 with offset_alignment.
18689 (DR_ALIGNED_TO): Delete.
18690 (DR_OFFSET_ALIGNMENT): New macro.
18691 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18692 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18693 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18694 (create_data_ref): Likewise.
18695 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18696 (vect_analyze_data_refs): Likewise.
18697 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18698 creating dummy innermost behavior.
18700 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18702 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18703 with a "innermost_loop_behavior *" and refeence tree.
18704 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18705 (create_data_ref): Update call accordingly.
18706 * tree-predcom.c (find_looparound_phi): Likewise.
18708 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18710 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18711 fields with dr_wrt_vec_loop.
18712 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18713 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18714 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18715 (vect_dr_behavior): New function.
18716 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18717 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18718 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18719 track whether the step preserves the misalignment.
18720 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18721 Use vect_dr_behavior.
18722 (vect_setup_realignment): Update call accordingly.
18723 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18724 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18725 call to vect_create_addr_base_for_vector_ref.
18726 (vect_create_cond_for_align_checks): Likewise.
18727 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18728 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18729 (vect_recog_mask_conversion_pattern): Likewise.
18730 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18731 (new_stmt_vec_info): Remove redundant zeroing.
18733 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18735 * common/config/arm/arm-common.c (arm_be8_option): New function.
18736 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18737 (ISA_ARMv6): Add isa_bit_be8.
18738 * config/arm/arm.h (arm_be8_option): Add prototype.
18739 (BE8_SPEC_FUNCTION): New define.
18740 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18741 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18742 (mlittle-endian): Similarly.
18743 (mbe8, mbe32): New options.
18744 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18745 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18747 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18749 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18751 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18753 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18754 (cleanup_tree_cfg_bb): Use it.
18755 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18757 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18759 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18762 * loop-doloop.c (add_test): Update profile.
18764 2017-07-03 Martin Liska <mliska@suse.cz>
18767 * sanopt.c (rewrite_usage_of_param): New function.
18768 (sanitize_rewrite_addressable_params): Likewise.
18769 (pass_sanopt::execute): Call rewrite_usage_of_param.
18771 2017-07-03 Richard Biener <rguenther@suse.de>
18773 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18774 back to using VIEW_CONVERT_EXPR.
18776 2017-07-03 Martin Liska <mliska@suse.cz>
18779 * doc/extend.texi: Document when a resolver function is
18780 generated for target_clones.
18782 2017-07-03 Martin Liska <mliska@suse.cz>
18784 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18785 and shadow memory of auto variables which are subject of
18786 use-after-scope sanitization.
18787 (asan_expand_mark_ifn): Add do set only when is_poison.
18789 2016-07-03 Richard Biener <rguenther@suse.de>
18791 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18793 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18794 (vectorizable_reduction): Perform reduction PHI creation when
18795 visiting a reduction PHI and adjust and simplify code generation
18796 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18797 (vect_transform_loop): Visit reduction PHIs.
18798 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18799 defs into the SLP tree.
18800 (vect_build_slp_tree): Reduction defs terminate the recursion.
18801 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18803 (vect_get_vec_defs_for_stmt_copy): Export.
18804 (vect_get_vec_defs): Likewise.
18805 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18807 (vect_get_vec_defs_for_stmt_copy): Declare.
18808 (vect_get_vec_defs): Likewise.
18810 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18812 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18813 parameter with a "loop" parameter and use it instead of the
18814 loop containing DR_STMT. Don't check simple_iv when doing
18815 BB analysis. Describe the two analysis modes in the comment.
18817 2017-07-03 Tom de Vries <tom@codesourcery.com>
18819 PR tree-optimization/69468
18820 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18821 (find_same_succ_bb): Handle ignore_edge_flags.
18823 2017-07-03 Tom de Vries <tom@codesourcery.com>
18825 PR tree-optimization/81192
18826 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18828 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18830 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18832 2017-07-03 Tom de Vries <tom@codesourcery.com>
18834 PR tree-optimization/81192
18835 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18836 BB_SAME_SUCC (bb) == NULL.
18838 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18840 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18843 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18845 * dumpfile.c: Include profile-count.h
18846 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18848 (insert_cond_bb): Update profile.
18849 * tree-cfg.h (insert_cond_bb): Update prototype.
18850 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18851 * tree-dump.c: Do not include tree-cfg.
18853 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18855 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18857 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18859 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18862 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18864 * tree-complex.c (expand_complex_div_wide): update profile.
18866 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18867 Alan Hayward <alan.hayward@arm.com>
18868 David Sherwood <david.sherwood@arm.com>
18870 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18871 (CORETYPES_H): New define.
18872 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18873 (insn-modes-inline.h, s-modes-inline-h): New rules.
18874 (generated_files): Add insn-modes-inline.h.
18875 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18876 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18877 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18878 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18879 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18880 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18881 (build/gencodes.o, build/genconditions.o): Likewise.
18882 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18883 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18884 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18885 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18886 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18887 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18888 * coretypes.h: Include everything up to real.h for generators.
18889 Include insn-modes.h first. Include wide-int-print.h after
18890 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18891 * machmode.h: Don't include insn-modes.h here.
18892 * function-tests.c: Remove includes of signop.h, machmode.h,
18893 double-int.h and wide-int.h.
18895 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18897 * optc-save-gen.awk: Likewise.
18898 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18899 * godump.c: Remove include of wide-int-print.h.
18900 * pretty-print.h: Likewise.
18901 * wide-int-print.cc: Likewise.
18902 * wide-int.cc: Likewise.
18903 * hash-map-tests.c: Remove include of signop.h.
18904 * hash-set-tests.c: Likewise.
18905 * rtl-tests.c: Likewise.
18906 * mkconfig.sh: Remove include of machmode.h.
18907 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18909 (emit_insn_modes_inline_h): ...this new function. Emit the code
18910 into an insn-modes-inline.h header file, adding appropriate
18911 include guards and end comments.
18912 (emit_insn_modes_c_header): Remove include of machmode.h.
18913 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18915 (main): Handle -i flag and call emit_insn_modes_inline_h when
18918 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18920 * tree-ssa-strlen.c (strinfo): Rename the length field to
18921 nonzero_chars. Add a full_string_p field.
18922 (compare_nonzero_chars, zero_length_string_p): New functions.
18923 (get_addr_stridx): Add an offset_out parameter.
18924 Use compare_nonzero_chars.
18925 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18926 (new_strinfo): Update after above changes to strinfo.
18927 (set_endptr_and_length): Set full_string_p.
18928 (get_string_length): Update after above changes to strinfo.
18929 (unshare_strinfo): Update call to new_strinfo.
18930 (maybe_invalidate): Likewise.
18931 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18932 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18933 as a uhwi instead of an shwi. Update after above changes to
18934 strinfo and new_strinfo.
18935 (zero_length_string): Assert that chainsi contains full strings.
18936 Use zero_length_string_p. Update call to new_strinfo.
18937 (adjust_related_strinfos): Update after above changes to strinfo.
18938 Copy full_string_p from origsi.
18939 (adjust_last_stmt): Use zero_length_string_p.
18940 (handle_builtin_strlen): Update after above changes to strinfo and
18941 new_strinfo. Install the lhs as the string length if the previous
18942 entry didn't describe a full string.
18943 (handle_builtin_strchr): Update after above changes to strinfo
18945 (handle_builtin_strcpy): Likewise.
18946 (handle_builtin_strcat): Likewise.
18947 (handle_builtin_malloc): Likewise.
18948 (handle_pointer_plus): Likewise.
18949 (handle_builtin_memcpy): Likewise. Track nonzero characters
18950 that aren't necessarily followed by a nul terminator.
18951 (handle_char_store): Likewise.
18953 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18955 PR tree-optimization/80769
18956 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18957 for malloc and calloc. Document the new invariant that all related
18958 strinfos have delayed lengths or none do.
18959 (verify_related_strinfos): Move earlier in file.
18960 (set_endptr_and_length): New function, split out from...
18961 (get_string_length): ...here. Also set the lengths of related
18963 (zero_length_string): Assert that chainsi has known (rather than
18965 (adjust_related_strinfos): Likewise.
18967 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18969 PR tree-optimization/81136
18970 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18971 assert that two references with the same misalignment have the same
18972 compile-time misalignment if those compile-time misalignments
18975 2017-07-01 Andi Kleen <ak@linux.intel.com>
18977 * print-tree.c (print_node): Print all attributes.
18979 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18981 * cfg.c (scale_bbs_frequencies): New function.
18982 * cfg.h (scale_bbs_frequencies): Declare it.
18983 * cfgloopanal.c (single_likely_exit): Cleanup.
18984 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18986 (scale_loop_profile): Likewise.
18987 (loop_version): Likewise.
18988 (create_empty_loop_on_edge): Update.
18989 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18990 scale_loop_frequencies, scale_loop_profile, loopify,
18991 loop_version): Update prototypes.
18992 * modulo-sched.c (sms_schedule): Update.
18993 * predict.c (unlikely_executed_edge_p): Also check probability.
18994 (probably_never_executed_edge_p): Fix typo.
18995 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18996 * tree-parloops.c (gen_parallel_loop): Update.
18997 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18998 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18999 * tree-ssa-loop-split.c (split_loop): Update.
19000 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19001 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19002 (vect_loop_versioning): Update.
19003 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19005 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19007 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19009 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19011 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19012 to keep profile consistent.
19014 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19016 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19017 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19018 * profile-count.h (max_safe_multiplier): Make unsigned.
19019 (profile_count::guessed_zero): New.
19021 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19023 * bb-reorder.c (fix_up_crossing_landing_pad,
19024 fix_crossing_conditional_branches): Use make_single_succ_edge
19025 to keep profile consistent.
19027 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19029 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19032 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19035 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19036 the right scopes, make sure cond_jump isn't preserved between multiple
19037 iterations. Search for fallthru edge whenever there are 3+ edges and
19038 use find_fallthru_edge for it.
19040 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19042 Patch by Alexander Monakov <amonakov@ispras.ru>
19043 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19044 probabilities consistently.
19046 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19048 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19049 emit_cmp_and_jump_insns.
19051 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19054 * tree-inline.c (expand_call_inline): Combine profile statuses.
19056 2017-06-30 Andrew Pinski <apinski@cavium.com>
19058 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19059 fold_stmt returned true.
19061 2017-06-30 Nathan Sidwell <nathan@acm.org>
19063 * ggc.h (empty_string): Delete.
19064 * cfgexpand.c (expand_asm_stmt): Use plain "".
19065 * optabs.c (expand_asm_memory_barrier): Likewise.
19066 * stringpool.c (empty_string): Delete.
19067 (digit_vector, digit_string): Delete.
19068 (ggc_alloc_string): Use plain "", don't optimize single digit
19069 strings. Use ggc_alloc_atomic.
19071 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19073 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19074 comparison set and one other set, use the cost of the non-comparison
19077 2017-06-30 Nathan Sidwell <nathan@acm.org>
19079 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19082 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19084 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19085 loops. Remove now unneeded calls to gimple_switch_set_label() that
19086 just set removed labels to NULL_TREE.
19088 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19090 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19091 (set_range_info): ...here. Only call set_range_info_raw if domain
19093 (set_nonzero_bits): Call set_range_info_raw.
19094 * tree-ssanames.h (set_range_info_raw): New.
19096 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19099 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19100 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19101 of nonimmediate_operand and <store_mask_constraint> instead of m
19102 for the input operand. For V8FI iterator, always split if input
19103 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19104 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19105 <store_mask_predicate> instead of register_operand and
19106 <store_mask_constraint> instead of v for the input operand. Make
19107 sure both operands aren't MEMs for if not <mask_applied>.
19109 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19111 * lto-wrapper.c (copy_file) Close both file descriptors before
19114 2017-06-30 Martin Liska <mliska@suse.cz>
19117 * multiple_target.c (create_dispatcher_calls): Make ifunc
19118 also for function that don't have calls or are not referenced.
19120 2017-06-30 Richard Biener <rguenther@suse.de>
19122 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19123 analyze the first scalar stmt. Move vector type computation
19124 for the BB case here from ...
19125 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19126 live operation processing in the SLP case properly.
19128 2017-06-30 Richard Biener <rguenther@suse.de>
19130 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19132 2017-06-30 Martin Liska <mliska@suse.cz>
19135 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19136 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19138 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19140 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19143 2017-06-30 Richard Biener <rguenther@suse.de>
19145 PR tree-optimization/81249
19146 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19147 condition reduction result to original scalar type.
19149 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19151 * profile-count.h (enum profile_quality): Fix typos and whitespace
19154 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19156 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19157 type for branch probabilities.
19159 2017-06-29 Julian Brown <julian@codesourcery.com>
19160 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19162 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19163 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19164 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19165 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19167 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19169 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19170 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19171 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19172 CC usage from generic code to here.
19173 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19174 CC usage into the target macros.
19176 2017-06-29 Maya Rashish <coypu@sdf.org>
19178 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19181 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19183 * arm/arm-builtins.c: Include profile-count.h
19184 * except.c (sjlj_emit_function_enter): Use
19185 profile_probability::unlikely.
19187 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19189 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19190 and tocrel_offset be pointer args rather than implicitly using
19192 (legitimate_constant_pool_address_p, rs6000_emit_move,
19193 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19194 tocrel_offset and use in toc_relative_expr_p call.
19195 (print_operand, print_operand_address): Use static tocrel_base_oac
19196 and tocrel_offset_oac.
19197 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19200 2017-06-29 Maya Rashish <coypu@sdf.org>
19202 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19204 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19206 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19207 objects, take into account only the alignment of 'op0' and 'mode1' if
19210 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19212 * ccmp.c (ccmp_tree_comparison_p): New function.
19213 (ccmp_candidate_p): Update to use above function.
19214 (get_compare_parts): New function.
19215 (expand_ccmp_next): Update to use new functions.
19216 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19218 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19219 take mode as argument.
19220 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19221 * expr.c (expand_expr_real_1): Pass mode as argument.
19223 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19225 * combine.c (combine_instructions): Print insns to dump_file, together
19228 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19230 * asan.c (asan_emit_stack_protection): Update.
19231 (create_cond_insert_point): Update.
19232 * auto-profile.c (afdo_propagate_circuit): Update.
19233 * basic-block.h (struct edge_def): Turn probability to
19234 profile_probability.
19235 (EDGE_FREQUENCY): Update.
19236 * bb-reorder.c (find_traces_1_round): Update.
19237 (better_edge_p): Update.
19238 (sanitize_hot_paths): Update.
19239 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19240 (make_single_succ_edge): Update.
19241 (check_bb_profile): Update.
19242 (dump_edge_info): Update.
19243 (update_bb_profile_for_threading): Update.
19244 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19245 probabilitycount to 0.
19246 * cfgbuild.c (compute_outgoing_frequencies): Update.
19247 * cfgcleanup.c (try_forward_edges): Update.
19248 (outgoing_edges_match): Update.
19249 (try_crossjump_to_edge): Update.
19250 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19251 (expand_gimple_tailcall): Update.
19252 (construct_init_block): Use make_single_succ_edge.
19253 (construct_exit_block): Use make_single_succ_edge.
19254 * cfghooks.c (verify_flow_info): Update.
19255 (redirect_edge_succ_nodup): Update.
19256 (split_edge): Update.
19257 (account_profile_record): Update.
19258 * cfgloopanal.c (single_likely_exit): Update.
19259 * cfgloopmanip.c (scale_loop_profile): Update.
19260 (set_zero_probability): Remove.
19261 (duplicate_loop_to_header_edge): Update.
19262 * cfgloopmanip.h (loop_version): Update prototype.
19263 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19264 (force_nonfallthru_and_redirect): Update.
19265 (update_br_prob_note): Update.
19266 (rtl_verify_edges): Update.
19267 (purge_dead_edges): Update.
19268 (rtl_lv_add_condition_to_bb): Update.
19269 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19270 * cgraphunit.c (init_lowered_empty_function): Update.
19271 (cgraph_node::expand_thunk): Update.
19272 * cilk-common.c: Include profile-count.h
19273 * dojump.c (inv): Remove.
19274 (jumpifnot): Update.
19275 (jumpifnot_1): Update.
19276 (do_jump_1): Update.
19278 (do_jump_by_parts_greater_rtx): Update.
19279 (do_compare_rtx_and_jump): Update.
19280 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19281 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19282 * dwarf2cfi.c: Include profile-count.h
19283 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19284 (sjlj_emit_dispatch_table): Likewise.
19285 * explow.c: Include profile-count.h
19286 * expmed.c (emit_store_flag_force): Update.
19287 (do_cmp_and_jump): Update.
19288 * expr.c (compare_by_pieces_d::generate): Update.
19289 (compare_by_pieces_d::finish_mode): Update.
19290 (emit_block_move_via_loop): Update.
19291 (store_expr_with_bounds): Update.
19292 (store_constructor): Update.
19293 (expand_expr_real_2): Update.
19294 (expand_expr_real_1): Update.
19295 * expr.h (try_casesi, try_tablejump): Update prototypes.
19296 * gimple-pretty-print.c (dump_probability): Update.
19297 (dump_profile): New.
19298 (dump_gimple_label): Update.
19299 (dump_gimple_bb_header): Update.
19300 * graph.c (draw_cfg_node_succ_edges): Update.
19301 * hsa-gen.c (convert_switch_statements): Update.
19302 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19303 (find_if_case_1): Update.
19304 (find_if_case_2): Update.
19305 * internal-fn.c (expand_arith_overflow_result_store): Update.
19306 (expand_addsub_overflow): Update.
19307 (expand_neg_overflow): Update.
19308 (expand_mul_overflow): Update.
19309 (expand_vector_ubsan_overflow): Update.
19310 * ipa-cp.c (good_cloning_opportunity_p): Update.
19311 * ipa-split.c (split_function): Use make_single_succ_edge.
19312 * ipa-utils.c (ipa_merge_profiles): Update.
19313 * loop-doloop.c (add_test): Update.
19314 (doloop_modify): Update.
19315 * loop-unroll.c (compare_and_jump_seq): Update.
19316 (unroll_loop_runtime_iterations): Update.
19317 * lra-constraints.c (lra_inheritance): Update.
19318 * lto-streamer-in.c (input_cfg): Update.
19319 * lto-streamer-out.c (output_cfg): Update.
19320 * mcf.c (adjust_cfg_counts): Update.
19321 * modulo-sched.c (sms_schedule): Update.
19322 * omp-expand.c (expand_omp_for_init_counts): Update.
19323 (extract_omp_for_update_vars): Update.
19324 (expand_omp_ordered_sink): Update.
19325 (expand_omp_for_ordered_loops): Update.
19326 (expand_omp_for_generic): Update.
19327 (expand_omp_for_static_nochunk): Update.
19328 (expand_omp_for_static_chunk): Update.
19329 (expand_cilk_for): Update.
19330 (expand_omp_simd): Update.
19331 (expand_omp_taskloop_for_outer): Update.
19332 (expand_omp_taskloop_for_inner): Update.
19333 * omp-simd-clone.c (simd_clone_adjust): Update.
19334 * optabs.c (expand_doubleword_shift): Update.
19335 (expand_abs): Update.
19336 (emit_cmp_and_jump_insn_1): Update.
19337 (expand_compare_and_swap_loop): Update.
19338 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19339 * predict.c (predictable_edge_p): Update.
19340 (edge_probability_reliable_p): Update.
19341 (set_even_probabilities): Update.
19342 (combine_predictions_for_insn): Update.
19343 (combine_predictions_for_bb): Update.
19344 (propagate_freq): Update.
19345 (estimate_bb_frequencies): Update.
19346 (force_edge_cold): Update.
19347 * profile-count.c (profile_count::dump): Add missing space into dump.
19348 (profile_count::debug): Add newline.
19349 (profile_count::differs_from_p): Explicitly convert to unsigned.
19350 (profile_count::stream_in): Update.
19351 (profile_probability::dump): New member function.
19352 (profile_probability::debug): New member function.
19353 (profile_probability::differs_from_p): New member function.
19354 (profile_probability::differs_lot_from_p): New member function.
19355 (profile_probability::stream_in): New member function.
19356 (profile_probability::stream_out): New member function.
19357 * profile-count.h (profile_count_quality): Rename to ...
19358 (profile_quality): ... this one.
19359 (profile_probability): New.
19360 (profile_count): Update.
19361 * profile.c (compute_branch_probabilities): Update.
19362 * recog.c (peep2_attempt): Update.
19363 * sched-ebb.c (schedule_ebbs): Update.
19364 * sched-rgn.c (find_single_block_region): Update.
19365 (compute_dom_prob_ps): Update.
19366 (schedule_region): Update.
19367 * sel-sched-ir.c (compute_succs_info): Update.
19368 * stmt.c (struct case_node): Update.
19369 (do_jump_if_equal): Update.
19370 (get_outgoing_edge_probs): Update.
19371 (conditional_probability): Update.
19372 (emit_case_dispatch_table): Update.
19373 (expand_case): Update.
19374 (expand_sjlj_dispatch_table): Update.
19375 (emit_case_nodes): Update.
19376 * targhooks.c: Update.
19377 * tracer.c (better_p): Update.
19378 (find_best_successor): Update.
19379 * trans-mem.c (expand_transaction): Update.
19380 * tree-call-cdce.c: Update.
19381 * tree-cfg.c (gimple_split_edge): Upate.
19382 (move_sese_region_to_fn): Upate.
19383 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19384 * tree-eh.c (lower_resx): Upate.
19385 (cleanup_empty_eh_move_lp): Upate.
19386 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19387 * tree-inline.c (copy_edges_for_bb): Update.
19388 (copy_cfg_body): Update.
19389 * tree-parloops.c (gen_parallel_loop): Update.
19390 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19391 (gimple_gen_time_profiler): Update.
19392 * tree-ssa-dce.c (remove_dead_stmt): Update.
19393 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19394 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19395 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19396 (unloop_loops): Update.
19397 (try_peel_loop): Update.
19398 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19399 * tree-ssa-loop-split.c (connect_loops): Update.
19400 (split_loop): Update.
19401 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19402 (hoist_guard): Update.
19403 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19404 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19405 (value_replacement): Update.
19406 * tree-ssa-reassoc.c (branch_fixup): Update.
19407 * tree-ssa-tail-merge.c (replace_block_by): Update.
19408 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19409 (create_edge_and_update_destination_phis): Update.
19410 (compute_path_counts): Update.
19411 (recompute_probabilities): Update.
19412 (update_joiner_offpath_counts): Update.
19413 (freqs_to_counts_path): Update.
19414 (duplicate_thread_path): Update.
19415 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19416 (struct switch_conv_info): Update.
19417 (gen_inbound_check): Update.
19418 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19419 (vect_do_peeling): Update.
19420 (vect_loop_versioning): Update.
19421 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19422 (optimize_mask_stores): Update.
19423 * ubsan.c (ubsan_expand_null_ifn): Update.
19424 * value-prof.c (gimple_divmod_fixed_value): Update.
19425 (gimple_divmod_fixed_value_transform): Update.
19426 (gimple_mod_pow2): Update.
19427 (gimple_mod_pow2_value_transform): Update.
19428 (gimple_mod_subtract): Update.
19429 (gimple_mod_subtract_transform): Update.
19430 (gimple_ic): Update.
19431 (gimple_stringop_fixed_value): Update.
19432 (gimple_stringops_transform): Update.
19433 * value-prof.h: Update.
19435 2017-06-29 Carl Love <cel@us.ibm.com>
19437 * config/rs6000/rs6000-c.c: Add support for built-in functions
19438 vector signed int vec_signed (vector float);
19439 vector signed long long vec_signed (vector double);
19440 vector signed int vec_signed2 (vector double, vector double);
19441 vector signed int vec_signede (vector double);
19442 vector signed int vec_signedo (vector double);
19443 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19444 instruction generator.
19445 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19446 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19447 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19449 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19450 vunsignede_v2df): Add define_expands.
19451 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19452 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19453 VEC_UNSIGNEDO): Add definitions.
19454 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19455 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19456 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19457 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19458 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19459 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19460 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19461 vec_unsignede, vec_unsignedo): Add builtin defines.
19462 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19464 * doc/extend.texi: Update the built-in documentation file for the
19465 new built-in functions.
19467 2017-06-29 Richard Biener <rguenther@suse.de>
19469 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19470 reduction chains to LOOP_VINFO_REDUCTIONS.
19471 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19472 SLP reductions after processing reduction chains.
19474 2017-06-29 Nathan Sidwell <nathan@acm.org>
19476 * builtins.c (fold_builtin_FUNCTION): Use
19477 lang_hooks.decl_printable_name.
19479 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19481 PR middle-end/81194
19482 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19483 with only one label.
19484 * stmt.c (expand_case): Assert NCASES is greater than one.
19486 2017-06-29 Richard Biener <rguenther@suse.de>
19488 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19490 (group_case_labels): Likewise.
19491 (find_taken_edge): Push sanity checking on val to workers...
19492 (find_taken_edge_cond_expr): ... here
19493 (find_taken_edge_switch_expr): ... and here, handle cases
19494 with just a default label.
19495 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19496 (group_case_labels): Likewise.
19497 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19498 group_case_labels does anything cleanup the CFG again.
19500 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19502 PR tree-optimization/81196
19503 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19504 exit condition comparing two IVs.
19506 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19508 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19509 profile to the dummy entry at the end of the list of architectures.
19510 * config/arm/arm-cpu-cdata.h: Regenerated.
19512 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19513 Michael Collison <michael.collison@arm.com>
19516 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19518 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19519 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19520 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19521 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19522 of shift when the shift amount is masked with constant equal to
19523 the size of the mode.
19524 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19527 2017-06-29 Martin Liska <mliska@suse.cz>
19529 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19531 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19534 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19537 (edge_growth_cache_entry::edge_growth_cache_entry): New
19539 (reset_edge_growth_cache): Update to use constructor.
19541 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19543 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19544 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19545 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19547 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19549 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19550 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19552 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19554 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19555 (*-linux-uclibc*): Add t-uclibc tmake_file.
19556 * config/t-musl: New.
19557 * config/t-uclibc: New.
19559 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19561 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19563 (gen_comm_data): Emit architectural setting of arch_prof.
19564 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19566 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19567 (armv8-m.base, armv8-m.main): Likewise.
19568 * arm-protos.h (arm_build_target): Add profile field.
19569 (arch_option): Likewise.
19570 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19572 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19573 arm_active_target.profile.
19575 2017-06-28 Richard Biener <rguenther@suse.de>
19577 PR middle-end/81227
19578 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19579 TYPE_OVERFLOW_WRAPS.
19580 * match.pd (negate_expr_p): Likewise.
19581 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19582 fold_build2, not fold_binary.
19584 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19586 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19587 Convert memory address to Pmode.
19588 (aarch64_print_operand): Assert MEM operands are always Pmode.
19590 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19593 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19594 Remove redundant if.
19595 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19596 * config/arm/aarch-common-protos.h
19597 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19598 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19600 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19603 * multiple_target.c (create_dispatcher_calls): Set the default
19604 clone to be static, not public.
19606 2017-06-28 Richard Biener <rguenther@suse.de>
19608 * tree-vect-loop.c (vectorizable_reduction): Move special
19609 cond reduction IV var creation ...
19610 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19611 parameter. Use STMT_VINFO_VECTYPE.
19612 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19615 2017-06-28 Martin Liska <mliska@suse.cz>
19618 * ipa-visibility.c (non_local_p): Handle visibility.
19620 2017-06-28 Martin Liska <mliska@suse.cz>
19623 * common.opt: Add IntegerRange to various options.
19624 * opt-functions.awk (integer_range_info): New function.
19625 * optc-gen.awk: Add integer_range_info to cl_options struct.
19626 * opts-common.c (decode_cmdline_option): Handle
19627 CL_ERR_INT_RANGE_ARG.
19628 (cmdline_handle_error): Likewise.
19629 * opts.c (print_filtered_help): Show valid interval in
19630 when --help is provided.
19631 * opts.h (struct cl_option): Add range_min and range_max fields.
19632 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19634 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19636 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19637 (x * C EQ/NE y * C): New transformation.
19639 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19641 * genmultilib (combination_space): Accept '+' in option names.
19643 2017-06-28 Martin Liska <mliska@suse.cz>
19646 * asan.c (instrument_derefs): Bail out inner references
19647 that are hard register variables.
19649 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19652 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19653 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19655 2017-06-28 Richard Biener <rguenther@suse.de>
19657 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19658 (vect_get_slp_defs): Adjust.
19659 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19661 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19663 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19664 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19665 (vectorizable_reduction): Adjust.
19666 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19668 (vect_get_slp_defs): Likewise.
19669 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19670 (vectorizable_bswap): Adjust.
19671 (vectorizable_call): Likewise.
19672 (vectorizable_conversion): Likewise.
19673 (vectorizable_assignment): Likewise.
19674 (vectorizable_shift): Likewise.
19675 (vectorizable_operation): Likewise.
19676 (vectorizable_store): Likewise.
19677 (vectorizable_condition): Likewise.
19678 (vectorizable_comparison): Likewise.
19680 2017-06-28 Michael Collison <michael.collison@arm.com>
19683 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19685 (arm_gen_movmemqi): Removed unused variable 'i'.
19686 Convert 'for' loop into 'while' loop.
19687 (arm_expand_prologue): Remove last unnecessary set of insn.
19688 (thumb_pop): Remove unused variable 'pushed_words'.
19689 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19691 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19693 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19694 * config/s390/s390-protos.h: Add prototype of
19695 s390_rel_address_ok_p.
19696 * config/s390/s390.c (s390_got_symbol): New function.
19697 (s390_rel_address_ok_p): New function.
19698 (legitimize_pic_address): Use s390_rel_address_ok_p.
19699 (s390_load_got): Use s390_got_symbol.
19700 (s390_option_override): Issue error if
19701 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19702 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19704 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19706 2017-06-27 Andrew Pinski <apinski@cavium.com>
19708 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19709 (X * copysign (1.0, X)): New pattern.
19710 (X * copysign (1.0, -X)): New pattern.
19711 (copysign (-1.0, CST)): New pattern.
19713 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19715 * genmultilib (combination_space): Remove variable.
19716 Validate reuse rules against regular expression for any sequence
19717 of multilib options in any order.
19719 2017-06-27 Michael Collison <michael.collison@arm.com>
19721 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19722 call aarch64_split_simd_combine.
19723 * (aarch64_combine_internal<mode>): Delete pattern.
19724 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19725 Allow register and subreg operands.
19727 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19729 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19730 specific need, just fallback on defaults.
19731 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19733 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19734 Olivier Hainque <hainque@adacore.com>
19736 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19738 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19739 targets. Pick a default if no particular attempt applied.
19740 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19743 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19745 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19746 (x86_64-wrs-vxworks7): Likewise.
19748 2017-06-27 Marek Polacek <polacek@redhat.com>
19751 * ubsan.c (instrument_null): Check get_base_address's result for null.
19753 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19755 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19757 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19759 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19760 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19761 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19762 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19763 New function types.
19764 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19765 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19766 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19767 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19768 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19769 BUILT_IN_FEUPDATEENV): New builtins.
19770 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19771 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19772 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19773 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19775 (builtin_structptr_types): Adjust size.
19776 * tree.c (builtin_structptr_types): Add four entries.
19778 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19779 Olivier Hainque <hainque@adacore.com>
19781 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19782 (TLS_SYM): New local macro, forcing reference to __tls__ on
19783 link command lines for VxWorks 7 RTPs, triggering initialization
19785 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19786 OS features TLS support, true for RTPs on VxWorks 7.
19787 * config/vxworks.c (vxworks_override_options): Setup emutls
19790 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19792 * predict.c (test_prediction_value_range): Use -1U instead of -1
19793 to avoid narrowing conversion warning.
19794 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19795 to avoid narrowing conversion warning.
19796 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19798 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19800 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19802 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19803 64bit configurations.
19804 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19805 (SIZE_TYPE): Likewise.
19806 * config/vxworks.c (vxworks_emutls_var_fields): Use
19807 long_unsigned_type_node instead of unsigned_type_node as the offset
19808 field type, which is "pointer" mode in emutls.c.
19810 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19813 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19815 PR middle-end/81207
19816 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19817 gimple_vuse copying separately from gimple_vdef copying.
19819 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19821 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19824 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19825 Olivier Hainque <hainque@adacore.com>
19827 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19828 for all vxworks7 targets.
19829 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19830 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19831 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19832 variations for VX6/VX7 and 32/64bits later on in ...
19833 (VXWORKS_LIB_SPEC): Leverage new macros.
19834 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19835 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19837 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19839 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19840 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19842 2017-06-26 Carl Love <cel@us.ibm.com>
19844 * config/rs6000/rs6000-c.c: Add support for built-in functions
19845 vector bool char vec_reve (vector bool char);
19846 vector signed char vec_reve (vector signed char);
19847 vector unsigned char vec_reve (vector unsigned char);
19848 vector bool int vec_reve (vector bool int);
19849 vector signed int vec_reve (vector signed int);
19850 vector unsigned int vec_reve (vector unsigned int);
19851 vector bool long long vec_reve (vector bool long long);
19852 vector signed long long vec_reve (vector signed long long);
19853 vector unsigned long long vec_reve (vector unsigned long long);
19854 vector bool short vec_reve (vector bool short);
19855 vector signed short vec_reve (vector signed short);
19856 vector double vec_reve (vector double);
19857 vector float vec_reve (vector float);
19858 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19859 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19860 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19861 (altivec_vreve): New pattern.
19862 * config/rs6000/altivec.h (vec_reve): New define.
19863 * doc/extend.texi (vec_rev): Update the built-in documentation file
19864 for the new built-in functions.
19866 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19868 PR tree-optimization/71815
19869 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19871 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19873 (slsr_process_phi): Likewise.
19874 (replace_uncond_cands_and_profitable_phis): Don't replace a
19875 multiply candidate with a stride of 1 (copy or cast).
19876 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19878 (lowest_cost_path): Likewise.
19879 (total_savings): Likewise.
19881 2017-06-26 Richard Biener <rguenther@suse.de>
19884 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19885 Use def_builtin_pure for all gather builtins.
19887 2017-06-26 Richard Biener <rguenther@suse.de>
19889 PR tree-optimization/81203
19890 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19891 non-dominating BBs.
19893 2017-06-26 Marek Polacek <polacek@redhat.com>
19896 * doc/invoke.texi: Document -Wmultistatement-macros.
19898 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19900 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19901 arm_neon_ok_no_float_abi effective target.
19903 2017-06-26 Richard Biener <rguenther@suse.de>
19905 PR tree-optimization/80928
19906 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19907 (copy_bbs): Set BB_DUPLICATED flag early.
19908 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19910 (execute_on_shrinking_pred): Likewise.
19911 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19912 BB_DUPLICATED blocks.
19913 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19914 iterate over all PHIs considering removal of *gsi.
19916 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19918 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19921 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19923 * config/rs6000/rs6000-string.c: (expand_block_clear,
19924 do_load_for_compare, select_block_compare_mode,
19925 compute_current_alignment, expand_block_compare,
19926 expand_strncmp_align_check, expand_strn_compare,
19927 expand_block_move, rs6000_output_load_multiple)
19928 Move functions related to string/block move/compare
19929 to a separate file.
19930 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19931 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19932 for this function which is now used in two files.
19933 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19934 * config.gcc: Add rs6000-string.o to extra_objs for
19935 targets powerpc*-*-* and rs6000*-*-*.
19937 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19940 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19941 32-bit, since indexed is not valid for DImode.
19942 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19943 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19944 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19945 (define_peephole2 for Altivec d-form store): Likewise.
19948 * multiple_target.c (create_dispatcher_calls): Only create the
19949 dispatcher call if the function is the default clone of a
19950 versioned function.
19952 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19954 PR middle-end/80902
19955 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19956 a call, force the call to not be a tail call.
19958 2017-06-23 Jeff Law <law@redhat.com>
19960 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19963 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19965 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19966 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19967 handling for early expansion of vector shifts (sl,sr,sra,rl).
19968 (builtin_function_type): Add vector shift right instructions
19969 to the unsigned argument list.
19971 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19973 rtl-optimizatoin/79286
19974 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19975 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19976 trap. PIC register plus a const unspec without offset can never trap.
19978 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19980 * tree.h (builtin_structptr_type): New type.
19981 (builtin_structptr_types): Declare new array.
19982 * tree.c (builtin_structptr_types): New array.
19983 (free_lang_data, build_common_tree_nodes): Use it.
19985 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19988 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19991 2017-06-22 Matt Turner <mattst88@gmail.com>
19993 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19994 Lake models to skylake case. Assume skylake for unknown
19995 models with clflushopt.
19997 2017-06-22 Jeff Law <law@redhat.com>
19999 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20000 frame sizes that do not satisfy aarch64_uimm12_shift.
20002 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20004 * profile-count.h (apply_probability,
20005 apply_scale, probability_in): Fix checks for zero.
20007 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20009 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20010 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20012 2016-06-22 Richard Biener <rguenther@suse.de>
20014 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20015 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20016 REDUC_MAX_EXPR support.
20017 (vectorizable_reduction): Likewise.
20018 (vect_create_epilog_for_reduction): Likewise.
20020 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20022 * match.pd (A / (1 << B) -> A >> B): New.
20023 * generic-match-head.c: Include optabs-tree.h.
20024 * gimple-match-head.c: Likewise.
20025 * optabs-tree.h (target_supports_op_p): New.
20026 * optabs-tree.c (target_supports_op_p): New.
20028 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20030 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20031 $gcc_cv_ld --help output.
20032 (gcc_cv_ld_demangle): Likewise.
20033 (gcc_cv_ld_eh_frame_hdr): Likewise.
20034 (gcc_cv_ld_pie): Likewise.
20035 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20036 (gcc_cv_ld_buildid): Likewise.
20037 (gcc_cv_ld_sysroot): Likewise.
20038 (ld_bndplt_support): Likewise.
20039 (ld_pushpopstate_support): Likewise.
20040 * configure: Regenerate.
20041 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20043 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20046 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20047 operands indexes to avoid gap between operands and match_dups.
20049 2017-06-21 Andrew Pinski <apinski@cavium.com>
20051 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20052 Increment Arith_shift and Arith_shift_reg by 1.
20053 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20055 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20056 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20057 (aarch64_strip_extend): Add new argument and test for it.
20058 (aarch64_cheap_mult_shift_p): New function.
20059 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20060 add a cost if it is true.
20061 Update calls to aarch64_strip_extend.
20062 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20064 2017-06-21 Andrew Pinski <apinski@cavium.com>
20066 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20068 (thunderxt88): Likewise.
20069 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20070 (thunderx_prefetch_tune): New variable.
20071 (thunderx2t99_prefetch_tune): Update for the correct values.
20072 (thunderxt88_tunings): New variable.
20073 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20074 generic_prefetch_tune.
20075 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20077 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20079 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20080 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20081 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20082 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20083 (aarch64_atomic_cas<mode>, GPI): Likewise.
20085 2017-06-21 Martin Liska <mliska@suse.cz>
20087 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20088 statements on cold and hot labels.
20089 * predict.c (tree_estimate_probability_bb): Remove the
20090 prediction from this place.
20092 2017-06-21 Martin Liska <mliska@suse.cz>
20094 PR tree-optimization/79489
20095 * gimplify.c (maybe_add_early_return_predict_stmt): New
20097 (gimplify_return_expr): Call the function.
20098 * predict.c (tree_estimate_probability_bb): Remove handling
20100 * predict.def: Update comment about early return predictor.
20101 * gimple-predict.h (is_gimple_predict): New function.
20102 * predict.def: Change default value of early return to 66.
20103 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20105 * passes.def: Put pass_strip_predict_hints to the beginning of
20108 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20110 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20111 FUNCTION_DECL declarations.
20112 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20114 (dwaf2out_decl): Likewise.
20115 * godump.c (go_early_global_decl): Skip call to the real debug hook
20116 for FUNCTION_DECL declarations.
20117 * passes.c (rest_of_decl_compilation): Skip call to the
20118 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20119 -fdump-go-spec is passed.
20121 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20123 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20124 Reorder for compactness.
20125 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20126 (def_builtin_pure, def_builtin_pure2): New functions.
20127 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20129 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20131 * match.pd (nop_convert): New predicate.
20132 ((A +- CST1) +- CST2): Allow some NOP conversions.
20134 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20137 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20138 with ctors/dtors if GOVD_SHARED is set.
20140 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20142 * config/aarch64/aarch64.md (movti_aarch64):
20143 Emit mov rather than orr.
20144 (movtf_aarch64): Likewise.
20145 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20146 Emit mov rather than orr.
20148 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20150 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20151 Swap alternatives, make integer dup more expensive.
20153 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20155 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20156 Return true for non-tls symbols.
20158 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20160 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20161 (cortex-a75): Likewise.
20162 (cortex-a75.cortex-a55): Likewise.
20163 * config/aarch64/aarch64-tune.md: Regenerate.
20164 * doc/invoke.texi (-mtune): Document new values for -mtune.
20166 2017-06-21 Tom de Vries <tom@codesourcery.com>
20168 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20169 stack_size feature.
20170 (Effective-Target Keywords, Other attributes): Suggest using
20171 dg-add-options stack_size feature to get stack limit in stack_size
20172 effective target documentation.
20174 2017-06-21 Julian Brown <julian@codesourcery.com>
20175 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20177 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20178 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20179 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20181 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20182 attribute type list for neon_multiply.
20183 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20184 attribute type list for neon_multiply.
20185 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20186 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20187 attribute type list for neon_multiply.
20188 * config/arm/types.md (crypto_pmull): Add.
20189 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20190 attribute type list.
20192 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20194 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20197 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20199 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20200 to make sure not to dereference a NULL cost_classes_ptr pointer.
20202 2017-06-20 Carl Love <cel@us.ibm.com>
20204 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20205 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20206 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20207 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20208 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20209 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20210 VMULOSW): New enum "unspec" values.
20211 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20212 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20213 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20214 altivec_vmulosw): New patterns.
20215 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20216 VMULOSW): Add definitions.
20218 2017-06-20 Julia Koval <julia.koval@intel.com>
20220 * config/i386/i386.c: Fix rounding expand for new pattern.
20221 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20223 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20225 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20226 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20228 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20230 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20233 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20235 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20236 architecture, then by implementer ID.
20237 * config/aarch64/aarch64-tune.md: Regenerate.
20239 2017-06-20 Richard Biener <rguenther@suse.de>
20241 PR middle-end/81097
20242 * fold-const.c (split_tree): Fold to type before negating.
20244 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20246 * diagnostic-show-locus.c
20247 (selftest::test_fixit_deletion_affecting_newline): New function.
20248 (selftest::diagnostic_show_locus_c_tests): Call it.
20250 2017-06-20 Andreas Schwab <schwab@suse.de>
20253 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20256 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20258 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20259 __ARM_FEATURE_COPROC according to support.
20261 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20263 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20264 Rewritten to avoid overflow for > 32-bit pointers.
20267 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20268 by removing enum keyword.
20269 (ubsan_type_descriptor): Likewise. Formatting fix.
20272 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20273 splitter): Require TARGET_SSE2 in the condition.
20275 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20278 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20279 for doing vector set of SFmode on ISA 3.0.
20280 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20281 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20283 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20284 SFmode value into a V4SF variable that was extracted from another
20285 V4SF variable without converting the element to double precision
20286 and back to single precision vector format.
20287 (vsx_insert_extract_v4sf_p9_2): Likewise.
20289 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20291 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20292 in UWHI to avoid undefined overflow.
20295 * ubsan.h (enum ubsan_encode_value_phase): New.
20296 (ubsan_encode_value): Change second argument to
20297 enum ubsan_encode_value_phase with default value of
20298 UBSAN_ENCODE_VALUE_GENERIC.
20299 * ubsan.c (ubsan_encode_value): Change second argument to
20300 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20301 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20302 create_tmp_var_raw instead of create_tmp_var and use a
20304 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20305 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20306 ubsan_encode_value callers.
20309 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20310 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20311 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20313 2017-06-19 Richard Biener <rguenther@suse.de>
20315 PR middle-end/81118
20316 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20317 estimates if we changed anything.
20319 2017-06-19 Richard Biener <rguenther@suse.de>
20321 PR tree-optimization/80887
20322 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20323 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20324 simplified lookups, then reset mprts_hook.
20325 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20327 (try_to_simplify): Likewise.
20329 2017-06-19 Martin Liska <mliska@suse.cz>
20332 * gimplify.c (gimplify_switch_expr):
20333 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20335 2017-06-19 Martin Liska <mliska@suse.cz>
20337 * doc/install.texi: Document that PGO runs in 4 stages.
20339 2017-06-19 Martin Liska <mliska@suse.cz>
20342 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20343 to dispatcher function name.
20344 * multiple_target.c (replace_function_decl): New function.
20345 (create_dispatcher_calls): Redirect both edges and references.
20347 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20349 * profile-count.c (profile_count::dump): Dump quality.
20350 (profile_count::differs_from_p): Update for unsigned val.
20351 * profile-count.h (profile_count_quality): New enum.
20352 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20354 2017-06-19 Richard Biener <rguenther@suse.de>
20356 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20357 struct function as arg.
20358 (estimate_numbers_of_iterations): Export overload with loop arg.
20359 (free_numbers_of_iterations_estimates_loop): Use an overload of
20360 free_numbers_of_iterations_estimates instead.
20361 * tree-cfg.c (remove_bb): Adjust.
20362 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20363 * tree-parloops.c (gen_parallel_loop): Likewise.
20364 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20366 (tree_unroll_loops_completely): Likewise.
20367 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20368 Use an overload instead and export.
20369 (estimated_loop_iterations): Adjust.
20370 (max_loop_iterations): Likewise.
20371 (likely_max_loop_iterations): Likewise.
20372 (estimate_numbers_of_iterations): Take struct function as arg
20374 (loop_exits_before_overflow): Adjust.
20375 (free_numbers_of_iterations_estimates_loop): Use an overload.
20376 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20377 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20379 2017-06-19 Richard Biener <rguenther@suse.de>
20382 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20383 RANGE_EXPR conservatively.
20385 2017-06-16 Carl Love <cel@us.ibm.com>
20387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20388 definitions for vec_float, vec_float2, vec_floato,
20389 vec_floate built-ins.
20390 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20391 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20393 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20394 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20395 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20396 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20397 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20398 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20399 vec_floato): Add builtin defines.
20400 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20401 Update the built-in documentation file for the new built-in
20404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20406 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20407 (mthumb): Mark as the negative of -marm.
20409 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20411 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20413 (ARM Options, -mtune): Document that this accepts the same
20414 extension options as -mcpu.
20415 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20419 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20420 permitted extensions.
20422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20424 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20425 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20426 (armv8-m.main): Add option +nodsp.
20427 * config/arm/arm-cpu-cdata.h: Regenerated.
20429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20431 * config/arm/t-fuchsia: New file.
20432 * config.gcc (arm*-*-fuchsia*): Use it.
20434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20436 * config/arm/t-symbian: Rewrite for new option infrastructure.
20438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20440 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20441 (MULTILIB_REQUIRED): Likewise.
20443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20445 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20446 (MULTILIB_RESUE): Likewise.
20447 (MULTILIB_MATCHES): Likewise.
20448 (MULTLIB_REQUIRED): Likewise.
20450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20452 * config/arm/t-rtems: Rewrite for new option framework.
20454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20456 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20457 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20458 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20459 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20460 * config/arm/t-multilib: ... here.
20461 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20462 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20463 armv7-a and armv8*-a when A-profile libraries have not been built.
20464 * config/arm/t-rmprofile: Rewrite.
20466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20468 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20469 with a backslash. Remove the backslash after substituting unescaped
20471 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20473 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20475 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20476 the driver to pass through the default setting of -mfloat-abi.
20477 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20479 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20480 (all_feat_combs): New rule.
20481 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20483 * config/arm/t-aprofile: Rewrite.
20485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20487 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20488 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20489 fpu is not specified by the user/command-line.
20490 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20491 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20492 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20493 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20494 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20495 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20499 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20500 * config/arm/t-arm-elf: Rewritten.
20502 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20504 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20505 have some floating-point instructions.
20506 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20507 (TARGET_MAYBE_HARD_FLOAT): New macro.
20508 * config/arm/arm-builtins.c (arm_init_builtins): Use
20509 TARGET_MAYBE_HARD_FLOAT.
20510 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20514 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20515 (configargs.h): Include it.
20516 (arm_print_hint_for_fpu_option): New function.
20517 (arm_parse_fpu_option): New function.
20518 (candidate_extension): New class.
20519 (arm_canon_for_multilib): New function.
20520 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20521 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20522 (ARCH_CANONICAL_SPECS): New macro.
20523 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20527 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20528 are set after handling multilib fragments. Set target_cpu_default2
20531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20533 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20535 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20536 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20538 (check_arch): Likewise.
20539 * config/arm/arm.c (arm_configure_build_target): Handle
20540 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20541 options in the default.
20543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20545 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20546 when an option is an alias of another.
20547 * config/arm/parsecpu.awk (optalias): New parser token.
20548 (gen_comm_data): Mark non-alias options as such. Emit entries
20549 for extension aliases.
20550 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20551 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20552 (armv6kz, armv6zk, armv6t2): Likewise.
20553 (armv7): Make vfpv3-d16 an alias.
20554 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20556 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20557 Sort in canonical order.
20558 (armv8-a): Sort in canonical order.
20559 (armv8.1-a, armv8.2-a): Likewise.
20560 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20562 (cortex-a9): Sort in canonical order.
20563 * config/arm/arm.c (selftests.h): Include it.
20564 (arm_test_cpu_arch_data): New function.
20565 (arm_run_self_tests): New function.
20566 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20567 (targetm): Move declaration to the end of the file.
20568 * arm-cpu-cdata.h: Regenerated.
20570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20572 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20573 call to target_mode_check describing the type of option passed.
20574 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20575 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20576 arm_parse_cpu_option_name to match parameters against list of
20578 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20579 arm_arch_core_flags data structure.
20580 * config/arm/arm-cpu_cdata.h: Regenerated.
20582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20584 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20586 (arm_print_hint_for_cpu_option): Likewise.
20587 (arm_print_hint_for_arch_option): Likewise.
20588 (arm_parse_cpu_option_name): Likewise.
20589 (arm_parse_arch_option_name): Likewise.
20590 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20591 of entries in the all_fpus list.
20592 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20593 (arm_parse_cpu_option_name): Declare.
20594 (arm_parse_arch_option_name): Declare.
20595 (arm_parse_option_features): Declare.
20596 (arm_intialize_isa): Declare.
20597 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20599 (gen_comm_data): ... here. Make definitions non-static.
20600 * config/arm/arm-cpu-data.h: Regenerated.
20601 * config/arm/arm-cpu-cdata.h: Regenerated.
20603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20605 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20606 (cpu_arch_extension): New structure.
20607 (cpu_arch_option, arch_option, cpu_option): New structures.
20608 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20609 architecture types.
20610 (gen_data): Generate new format data tables.
20611 * config/arm/arm.c (cpu_tune): New structure.
20612 (cpu_option, processors): Delete.
20613 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20614 (arm_print_hint_for_cpu_option): ... this and ...
20615 (arm_print_hint_for_arch_option): ... this.
20616 (arm_parse_arch_cpu_name): Delete. Replace with ...
20617 (arm_parse_cpu_option_name): ... this and ...
20618 (arm_parse_arch_option_name): ... this.
20619 (arm_unrecognized_feature): Change type of target parameter to
20621 (arm_parse_arch_cpu_features): Delete. Replace with ...
20622 (arm_parse_option_features): ... this.
20623 (arm_configure_build_target): Rework to use new configuration data
20625 (arm_print_tune_info): Rework for new configuration data tables.
20626 * config/arm/arm-cpu-data.h: Regenerated.
20627 * config/arm/arm-cpu.h: Regenerated.
20629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20631 * Makefile.in (OBJS): Move sbitmap.o from here ...
20632 (OBJS-libcommon): ... to here.
20634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20636 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20637 (ISA_ALL_CRYPTO): New macro.
20638 (ISA_ALL_SIMD): New macro
20639 (ISA_ALL_FP): New macro.
20640 * config/arm/arm.c (fpu_bitlist): Update initializer.
20641 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20643 (arm9e): Add fpu. Add option for nofp
20644 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20645 (arm926ej-s, arm1026ej-s): Likewise.
20646 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20647 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20648 neon-fp16, neon-vfpv4, nofp and nosimd.
20649 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20650 (cortex-a8): Add fpu. Add option for nofp.
20651 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20652 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20653 (cortex-r4f): Add fpu.
20654 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20655 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20657 (cortex-r8): Likewise.
20658 (cortex-m4): Add fpu. Add option for nofp.
20659 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20660 (cortex-a17.cortex-a7): Likewise.
20661 (cortex-a32): Add fpu. Add options for crypto and nofp.
20662 (cortex-a35, cortex-a53): Likewise.
20663 (cortex-a57): Add fpu. Add option for crypto.
20664 (cortex-a72, cortex-a73): Likewise.
20665 (exynos-m1): Likewise.
20666 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20667 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20668 (cortex-m33): Add fpu. Add option for nofp.
20669 * config/arm/arm-cpu-cdata.h: Regenerated
20670 * config/arm/arm-cpu-data.h: Regenerated.
20672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20674 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20675 (armv5te, armv5tej): Likewise.
20676 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20677 (armv7): Add options fp and vfpv3-d16.
20678 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20679 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20681 (armv7ve): Likewise.
20682 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20683 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20684 (armv8-a): Add nocrypto option.
20685 (armv8.1-a, armv8.2-a): Likewise.
20686 (armv8-m.main): add options fp, fp.dp and nofp.
20688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20690 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20692 (armv8-a+crc): Delete.
20693 (armv8.1-a): Add options simd, crypto and nofp.
20694 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20695 (armv8.2-a+fp16): Delete.
20696 (armv8-m.main): Add option dsp.
20697 (armv8-m.main+dsp): Delete.
20698 (cortex-a8): Add fpu. Add nofp option.
20699 (cortex-a9): Add fpu. Add nofp and nosimd options.
20700 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20701 link to main cpu and architecture data structures.
20702 (gen_comm_data): Only put isa attributes from the main architecture
20704 (option): New statement for architecture and CPU entries.
20705 * arm.c (struct cpu_option): New structure.
20706 (struct processors): Add entry for options.
20707 (arm_unrecognized_feature): New function.
20708 (arm_parse_arch_cpu_name): Ignore any characters after the first
20710 (arm_parse_arch_cpu_feature): New function.
20711 (arm_configure_build_target): Separate out any CPU and architecture
20712 features and parse separately. Don't error out if -mfpu=auto is
20713 used with only an architecture string.
20714 (arm_print_asm_arch_directives): New function.
20715 (arm_file_start): Call it.
20716 * config/arm/arm-cpu-cdata.h: Regenerated.
20717 * config/arm/arm-cpu-data.h: Likewise.
20718 * config/arm/arm-tables.opt: Likewise.
20720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20722 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20723 assembler when it is not -mfpu=auto.
20725 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20727 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20728 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20729 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20730 (ASM_CPU_SPEC): Rewrite.
20731 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20732 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20733 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20734 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20735 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20736 copied string is NUL-terminated. Also strip any characters prefixed
20738 (arm_rewrite_selected_arch): New function.
20739 (arm_rewrite_march): New function.
20741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20743 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20744 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20745 (march, mcpu, mtune): Convert to string-based options.
20746 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20747 (arm_parse_arch_cpu_name): New function.
20748 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20749 identify selected architecture or CPU.
20750 (arm_option_save): New function.
20751 (TARGET_OPTION_SAVE): Redefine.
20752 (arm_option_restore): Restore string options.
20753 (arm_option_print): Print string options.
20755 2017-06-16 Martin Sebor <msebor@redhat.com>
20757 PR tree-optimization/80933
20758 PR tree-optimization/80934
20759 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20760 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20761 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20762 (gimple_fold_builtin): Call them.
20764 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20766 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20767 as unlikely; update profile.
20769 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20771 * predict.c (force_edge_cold): Handle declaring edges impossible
20774 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20776 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20778 (try_unroll_loop_completely): Fix reporting.
20780 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20782 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20784 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20787 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20788 if given a non-constant argument for an intrinsic which requires a
20791 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20793 * profile.c (compare_freqs): New function.
20794 (branch_prob): Sort edge list.
20795 (find_spanning_tree): Assume that the list is priority sorted.
20797 2017-06-16 Richard Biener <rguenther@suse.de>
20799 PR tree-optimization/81090
20800 * passes.def (pass_record_bounds): Remove.
20801 * tree-pass.h (make_pass_record_bounds): Likewise.
20802 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20803 make_pass_record_bounds): Likewise.
20804 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20805 not free niter estimates at the beginning but at the end.
20806 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20808 2017-06-16 Richard Biener <rguenther@suse.de>
20810 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20811 initializer to workaround ICE in host GCC 4.8.
20813 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20815 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20817 (clone_inlined_nodes): Update.
20819 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20821 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20822 prefetch settings, and enable prefetching by default at -O3.
20824 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20826 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20827 Set flag_prefetch_loop_arrays according to tuning data.
20829 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20831 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20832 New tune structure.
20833 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20834 [Unrelated to main purpose of the patch] Place the pointer field last
20835 to enable type checking errors when tune structure are wrongly merged.
20836 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20837 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20838 (thunderx2t99_prefetch_tune): New tune constants.
20839 (tune_params *_tunings): Update all tunings (no functional change).
20840 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20841 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20842 from tunings structures.
20844 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20847 * ubsan.c (instrument_null): Add T argument, use it instead
20848 of computing it based on IS_LHS.
20849 (instrument_object_size): Likewise.
20850 (pass_ubsan::execute): Adjust instrument_null and
20851 instrument_object_size callers to pass gimple_get_lhs or
20852 gimple_assign_rhs1 result to it. Use instrument_null instead of
20853 calling get_base_address and instrument_mem_ref. Handle
20854 aggregate call arguments for object-size sanitization.
20856 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20858 PR tree-optimization/81089
20859 * tree-vrp.c (is_masked_range_test): Validate operands of
20862 2017-06-15 Martin Sebor <msebor@redhat.com>
20865 * dumpfile.c (dump_register): Avoid calling memset to initialize
20866 a class with a default ctor.
20867 * gcc.c (struct compiler): Remove const qualification.
20868 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20869 * hash-table.h: Ditto.
20870 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20872 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20873 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20875 * params.h (struct param_info): Make struct members non-const.
20876 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20877 with default initialization.
20878 * vec.h (vec_copy_construct, vec_default_construct): New helper
20880 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20881 with vec_copy_construct.
20882 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20883 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20884 * doc/invoke.texi (-Wclass-memaccess): Document.
20886 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20888 * emit-rtl.h (is_leaf): Update comment about local
20889 register allocator.
20891 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20894 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20895 for a variable to have a section before checking if the section has a
20897 Set section to.persistent if persistent attribute is set.
20898 Warn if .persistent attribute is used on an automatic variable.
20900 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20902 PR rtl-optimization/80474
20903 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20905 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20907 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20910 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20912 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20913 (casesi): Emit bounds checking as RTL.
20914 (casesi_internal_mips16_<mode>): Remove bounds checking.
20916 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20918 * config/xtensa/xtensa.c (xtensa_option_override): Append
20919 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20920 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20921 xtensa_doloop_hooks): Define unconditionally.
20922 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20924 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20925 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20926 for it in xtensa_option_override.
20927 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20928 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20930 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20932 * doc/cppopts.texi: Document '-' special value to -MF.
20934 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20936 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20937 (cortex_a53_fconst): Likewise.
20938 (cortex_a53_fpmul): Likewise.
20939 (cortex_a53_f_load_64): Likewise.
20940 (cortex_a53_f_load_many): Likewise.
20941 (cortex_a53_advsimd_alu): Likewise.
20942 (cortex_a53_advsimd_alu_q): Likewise.
20943 (cortex_a53_advsimd_mul): Likewise.
20944 (cortex_a53_advsimd_mul_q): Likewise.
20945 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20946 Add missing fmul, r2f_cvt and fconst cases.
20948 2017-06-14 Richard Biener <rguenther@suse.de>
20950 PR middle-end/81088
20951 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20953 (fold_binary_loc): When associating do not treat pre-existing
20954 TREE_OVERFLOW on literal constants as a reason to allow
20955 TREE_OVERFLOW on associated literal constants.
20957 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20959 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20960 (MASK_FEATURES): New macro.
20961 * config/sparc/sparc.c (sparc_option_override): Remove the special
20962 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20964 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20966 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20967 a division of 0 if non-call exceptions are enabled.
20969 2017-06-14 Andrew Pinski <apinski@cavium.com>
20970 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20973 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20974 Improve vector initialization code gen for only variable case.
20976 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20978 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20980 2017-06-14 Richard Biener <rguenther@suse.de>
20982 PR tree-optimization/81083
20983 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20986 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20988 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20989 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20990 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20991 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20992 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20993 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20995 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20997 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20998 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21000 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21002 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21004 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21006 * config/rs6000/t-rtems: Don't handle SPE.
21008 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21010 * config/rs6000/t-linux: Don't handle SPE.
21012 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21014 * config/rs6000/eabispe.h: Delete file.
21016 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21018 * config/rs6000/t-spe: Delete file.
21020 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21022 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21023 (rs6000_legitimate_offset_address_p): Return false for anything in
21024 V2SImode or V2SFmode.
21026 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21028 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21029 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21030 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21032 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21033 (rs6000_legitimate_offset_address_p): Ditto.
21034 (rs6000_emit_move): Ditto.
21035 (rs6000_init_builtins): Remove V4HI_type_node.
21037 2017-06-13 Martin Liska <mliska@suse.cz>
21040 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21041 (gate_asan): Likewise.
21042 * asan.h (asan_no_sanitize_address_p): Remove the function.
21043 (sanitize_flags_p): New function.
21044 * builtins.def: Fix coding style.
21045 * common.opt: Use renamed enum value.
21046 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21047 * doc/extend.texi: Document no_sanitize attribute.
21048 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21049 to SANITIZE_UNDEFINED_NONDEFAULT.
21050 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21051 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21052 Use sanitize_flags_p.
21053 * gimplify.c (gimplify_function_tree): Likewise.
21054 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21055 * opts.c (parse_no_sanitize_attribute): New function.
21056 (common_handle_option): Use renamed enum value.
21057 * opts.h (parse_no_sanitize_attribute): Declare.
21058 * tree.c (sanitize_flags_p): New function.
21059 * tree.h: Declared here.
21060 * tsan.c: Use sanitize_flags_p.
21061 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21062 (instrument_mem_ref): Likewise.
21063 (instrument_bool_enum_load): Likewise.
21064 (do_ubsan_in_current_function): Remove the function.
21065 (pass_ubsan::execute): Use sanitize_flags_p.
21066 * ubsan.h: Remove do_ubsan_in_current_function
21067 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21068 (dump_function_to_file): Use it here.
21070 2017-06-13 Martin Jambor <mjambor@suse.cz>
21072 PR tree-optimization/80803
21073 PR tree-optimization/81063
21074 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21075 (propagate_subaccesses_across_link): Enqueue subtree whenever
21076 necessary instead of relying on the caller.
21078 2017-06-13 Martin Jambor <mjambor@suse.cz>
21080 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21081 that have a first_link.
21082 (sort_and_splice_var_accesses): Do not check first_link before
21084 (subtree_mark_written_and_enqueue): Likewise.
21085 (propagate_all_subaccesses): Likewise and do not stop at first
21086 parent with a first_link.
21088 2017-06-13 Martin Jambor <mjambor@suse.cz>
21090 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21093 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21095 * match.pd: New pattern.
21097 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21099 * tree-vrp.c (is_masked_range_test): New function.
21100 (register_edge_assert_for): Determine ranges for
21103 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21105 PR tree-optimization/67328
21106 * fold-const.c (maskable_range_p): New function.
21107 (build_range_check): Generate bittests if possible.
21109 2017-06-13 Martin Liska <mliska@suse.cz>
21111 * gimple-pretty-print.c (dump_probability): Add new argument.
21112 (dump_edge_probability): Dump both probability and count.
21113 (dump_gimple_label): Likewise.
21114 (dump_gimple_bb_header): Likewise.
21116 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21119 * config/avr/avr-devices.c: Fix indentation.
21120 * config/avr/gen-avr-mmcu-specs.c: Dito.
21122 2017-06-13 Richard Biener <rguenther@suse.de>
21124 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21125 instead get vector type from stmt_info.
21126 (vectorizable_reduction): Adjust. Remove dead code.
21128 2017-06-13 Richard Biener <rguenther@suse.de>
21130 PR middle-end/81065
21131 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21132 case of C * (x * C2 + C3).
21133 (fold_addr_of_array_ref_difference): Properly fold index difference.
21135 2017-06-12 David S. Miller <davem@davemloft.net>
21138 * config/sparc/sparc.md (return expander): Emit frame blockage if
21139 function uses alloca.
21141 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21143 * combine.c (make_field_assignment): Check len rather than the mode
21144 precision when calling force_to_mode.
21146 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21148 Support multilibs and devices that see flash in RAM address range.
21151 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21152 (avr_mcu_t) <flash_pm_offset>: New field.
21153 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21154 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21155 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21156 (AVR_TINY_PM_OFFSET): Remove macro.
21157 * config/avr/avr.opt (-mshort-calls): New option.
21158 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21159 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21160 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21161 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21162 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21163 instead of avr_arch->have_jmp_call.
21164 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21165 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21166 avr_arch->flash_pm_offset to define.
21167 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21168 new field flash_pm_offset. Add entry for avrxmega3.
21169 (avr_texinfo): Add entry for avrxmega3.
21170 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21171 attiny212, attiny214,
21172 attiny412, attiny414, attiny416, attiny417,
21173 attiny814, attiny816, attiny817,
21174 attiny1614, attiny1616, attiny1617,
21175 attiny3214, attiny3216, attiny3217.
21176 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21177 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21178 (avr_print_operand_address) [AVR_TINY]: Same.
21179 (avr_asm_init_sections) <readonly_data_section>: Only patch
21180 callback if avr_arch->flash_pm_offset = 0.
21181 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21182 for rodata if avr_arch->flash_pm_offset != 0.
21183 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21184 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21185 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21186 (m_options): Append opt_rcall.
21187 (m_dirnames): Append dir_rcall.
21188 * config/avr/t-multilib: Regenerate.
21190 * configure.ac [target=avr]: Check whether avrxmega3 default
21191 linker description file works as needed.
21192 * configure: Regenerate.
21193 * doc/avr-mmcu.texi: Regenerate.
21194 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21195 <__AVR_ARCH__>: Document avrxmega3 and 103.
21196 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21197 <__AVR_SHORT_CALLS__>: Document it.
21198 <__AVR_PM_BASE_ADDRESS__>: Document it.
21199 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21200 (AVR Variable Attributes) <progmem>: Document this is
21201 not needed for avrxmega3.
21202 (AVR Named Address Spaces) <__flash>: Dito.
21204 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21206 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21208 2017-06-12 Doug Rupp <rupp@adacore.com>
21210 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21211 Append vxworks-stdint.h to the tm_file list.
21212 * config/vxworks-stdint.h: New file.
21214 2017-06-12 Martin Liska <mliska@suse.cz>
21216 PR tree-optimization/81041
21217 * tree-profile.c (gimple_gen_ic_func_profiler):
21218 Create an extra BB in profile-generate
21219 (gimple_gen_time_profiler): Likewise.
21221 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21223 PR tree-optimization/81003
21224 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21225 (update_range_test): Use it instead of force_gimple_operand_gsi.
21227 2017-06-12 Richard Biener <rguenther@suse.de>
21229 PR tree-optimization/81053
21230 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21231 with backedge value not defined in loop. Simplify def stmt
21234 2017-06-11 Tom de Vries <tom@codesourcery.com>
21237 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21239 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21240 nvptx_cannot_force_const_mem.
21242 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21244 * opts.c (finish_options): Move test for flag_split_stack after
21245 it has been initialized.
21247 2017-06-11 Jason Merrill <jason@redhat.com>
21249 * tree.h (id_equal): New.
21250 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21251 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21252 instead of strcmp of IDENTIFIER_POINTER.
21254 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21256 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21257 (mark_all_inlined_calls_cdtor): Fix formating.
21258 (inline_transform): Rescale profile before inlining.
21260 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21262 * cgraph.h (cgraph_edge::clone): Update prototype.
21263 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21264 (cgraph_node::create_clone): Update.
21265 (cgraph_node::create_version_clone): Update.
21266 * tree-inline.c (copy_bb): Update.
21267 (expand_call_inline): Update.
21269 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21271 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21272 factored out from ...
21273 (rs6000_emit_prologue): ... here.
21275 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21277 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21278 factored out from ...
21279 (rs6000_emit_prologue): ... here.
21281 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21283 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21285 (handle_missing_profiles): Fix computation of tp_first_run.
21286 (counts_to_freqs): Do not touch freqs when count is 0.
21288 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21290 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21293 2017-06-10 Tom de Vries <tom@codesourcery.com>
21295 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21296 attributes): Document signal effective target.
21298 2017-06-10 Tom de Vries <tom@codesourcery.com>
21300 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21301 Document effective target stack_size.
21303 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21305 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21306 to the edit_context if they can be auto-applied.
21308 2017-06-9 Ian Lance Taylor <iant@golang.org>
21310 * opts.c (finish_options): If -fsplit-stack, disable implicit
21311 -forder-blocks-and-partition.
21312 * doc/invoke.texi (Optimize Options): Document that when using
21313 -fsplit-stack -forder-blocks-and-partition is not implicitly
21316 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21318 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21319 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21320 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21321 * builtins.def (abort, trap, unreachable): Declare cold.
21322 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21323 * tree-core.h (ECF_COLD): New.
21324 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21325 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21327 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21329 * predict.c (unlikely_executed_stmt_p): Cleanup.
21331 2017-06-09 Richard Biener <rguenther@suse.de>
21333 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21334 model if the ref is always written to.
21336 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21338 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21340 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21342 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21345 2017-06-09 Tom de Vries <tom@codesourcery.com>
21348 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21349 "target cannot support label values" when encountering LABEL_REF.
21351 2017-06-09 Martin Liska <mliska@suse.cz>
21353 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21354 (gimple_gen_ic_func_profiler): Emit direct comparison
21355 of __gcov_indirect_call_callee with NULL.
21356 (gimple_gen_time_profiler): Change probability from
21357 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21359 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21361 * profile.c (edge_gcov_counts): Turn to pointer.
21362 (compute_branch_probabilities, compute_branch_probabilities): Update.
21363 (branch_prob): Do not clear edge_gcov_count.
21364 * profile.h (edge_gcov_counts): Turn to pointer.
21365 (edge_gcov_count): Update.
21367 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21369 * gimple.h (gimple_check_failed): Mark cold.
21371 2017-06-09 Richard Biener <rguenther@suse.de>
21373 PR tree-optimization/66623
21374 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21375 refactor check_reduction into two parts, properly computing
21376 whether we have to check reduction validity for outer loop
21379 2017-06-09 Richard Biener <rguenther@suse.de>
21381 PR tree-optimization/79483
21382 * graphite-scop-detection.c (order): New global.
21383 (get_order): Compute bb to order mapping that satisfies code
21384 generation constraints.
21385 (cmp_pbbs): New helper.
21386 (build_scops): Start domwalk at entry block, sort generated
21389 2017-06-09 Richard Biener <rguenther@suse.de>
21391 PR middle-end/81007
21392 * ipa-polymorphic-call.c
21393 (ipa_polymorphic_call_context::restrict_to_inner_class):
21394 Skip FIELD_DECLs with error_mark_node type.
21395 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21398 2017-06-09 Martin Liska <mliska@suse.cz>
21400 * predict.c (struct branch_predictor): New struct.
21401 (test_prediction_value_range): New test.
21402 (predict_c_tests): New function.
21403 * selftest-run-tests.c (selftest::run_tests): Run the function.
21404 * selftest.h: Declare new tests.
21406 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21409 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21410 gen_add3_insn did not fail.
21411 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21412 r0, construct that number in a temporary reg and add that reg to r0.
21413 If asked to put the result in r0 as well, fail.
21415 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21417 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21418 for early expansion of vec_eqv.
21420 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21422 PR middle-end/81005
21423 * ubsan.c (instrument_null): Avoid pointless code temporary.
21424 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21426 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21430 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21433 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21436 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21438 * predict.c (unlikely_executed_edge_p): Move ahead.
21439 (probably_never_executed_edge_p): Use it.
21441 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21443 PR middle-end/79988
21444 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21445 gimple_call_builtin_p call.
21447 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21449 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21450 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21451 rtl_check_failed_type2, rtl_check_failed_code1,
21452 rtl_check_failed_code2, rtl_check_failed_code_mode,
21453 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21454 rtvec_check_failed_bounds, rtl_check_failed_flag,
21455 _fatal_insn_not_found, _fatal_insn): Likewise.
21456 * tree.h (tree_contains_struct_check_failed,
21457 tree_check_failed, tree_not_check_failed,
21458 tree_class_check_failed, tree_range_check_failed,
21459 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21460 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21461 tree_operand_check_failed, omp_clause_check_failed,
21462 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21465 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21467 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21468 flag_branch_probabilities.
21469 * ipa-inline.c (edge_badness): Likewise.
21470 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21471 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21472 * predict.c (maybe_hot_frequency_p): Likewise.
21473 (probably_never_executed): Likewise.
21474 * sched-ebb.c (schedule_ebbs): Likewise.
21475 * sched-rgn.c (find_single_block_region): Likewise.
21476 * tracer.c (tail_duplicate): Likewise.
21478 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21480 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21481 longer requires x_flag_profile_use.
21483 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21485 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21486 instead of flag_reorder_blocks_and_partition.
21487 * dbxout.c (dbxout_function_end): Likewise.
21488 * dwarf2out.c (gen_subprogram_die): Likewise.
21489 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21490 * hw-doloop.c (reorg_loops): Likewise.
21491 * varasm.c (assemble_start_function,
21492 assemble_end_function): Likewise.
21493 (decide_function_section): Do not check for
21494 flag_reorder_blocks_and_partition.
21496 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21498 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21500 (chkp_get_hard_register_fake_addr_expr): Ditto.
21501 (chkp_build_addr_expr): Add check for hard reg case.
21502 (chkp_parse_array_and_component_ref): Ditto.
21503 (chkp_find_bounds_1): Ditto.
21504 (chkp_process_stmt): Don't generate bounds store for
21507 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21509 * predict.c (maybe_hot_bb_p): Do not check profile status.
21510 (maybe_hot_edge_p): Likewise.
21511 (probably_never_executed): Check for zero counts even if profile
21513 (unlikely_executed_edge_p): New function.
21514 (unlikely_executed_stmt_p): New function.
21515 (unlikely_executed_bb_p): New function.
21516 (set_even_probabilities): Use unlikely predicates.
21517 (combine_predictions_for_bb): Likewise.
21518 (predict_paths_for_bb): Likewise.
21519 (predict_paths_leading_to_edge): Likewise.
21520 (determine_unlikely_bbs): New function.
21521 (estimate_bb_frequencies): Use it.
21522 (compute_function_frequency): Use zero counts even if profile is
21524 * profile-count.h: Fix typo.
21526 2017-08-08 Julia Koval <julia.koval@intel.com>
21528 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21529 _mm512_mask_cvtsepi16_storeu_epi8,
21530 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21531 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21532 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21533 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21534 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21535 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21536 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21537 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21538 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21539 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21540 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21541 __builtin_ia32_pmovuswb256mem_mask,
21542 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21543 __builtin_ia32_pmovwb512mem_mask): New builtins.
21545 2017-08-08 Julia Koval <julia.koval@intel.com>
21547 PR target/73350,80862
21548 * config/i386/subst.md (round): Fix round pattern.
21549 * config/i386/i386.c (ix86_erase_embedded_rounding):
21550 Fix erasing rounding for the fixed pattern.
21552 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21554 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21556 2017-06-08 Martin Liska <mliska@suse.cz>
21558 PR gcov-profile/80911
21559 * gcov.c (block_info::block_info): New constructor.
21561 2017-06-07 Carl Love <cel@us.ibm.com>
21563 * config/rs6000/rs6000-c: The return type of the following
21564 built-in functions was implemented as int not long long. Fix sign
21565 of return value for the unsigned version of vec_mulo and vec_mule.
21566 vector unsigned long long vec_bperm (vector unsigned long long,
21567 vector unsigned char)
21568 vector signed long long vec_mule (vector signed int,
21570 vector unsigned long long vec_mule (vector unsigned int,
21571 vector unsigned int)
21572 vector signed long long vec_mulo (vector signed int,
21574 vector unsigned long long vec_mulo (vector unsigned int,
21575 vector unsigned int)
21576 * doc/extend.texi: Fix the documentation for the built-in
21579 2017-06-07 Carl Love <cel@us.ibm.com>
21582 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21585 2017-06-07 Carl Love <cel@us.ibm.com>
21587 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21588 support, Generate doublehv for signed int/float for BE case only.
21590 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21592 * doc/invoke.texi (mcx16): Rewrite.
21594 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21596 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21597 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21598 *mov<mode>_softfloat, and an anonymous splitter): Use
21599 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21603 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21605 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21606 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21607 (rs6000_debug_reg_global): Adjust.
21608 (rs6000_init_hard_regno_mode_ok): Adjust.
21609 (rs6000_dbx_register_number): Adjust.
21610 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21611 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21612 Remove SPE_ACC and SPEFSCR.
21613 (REG_ALLOC_ORDER): Ditto.
21614 (FRAME_POINTER_REGNUM): Change to 111.
21615 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21616 (REG_CLASS_NAMES): Ditto.
21617 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21618 (REGISTER_NAMES): Ditto.
21619 (ADDITIONAL_REG_NAMES): Ditto.
21620 (rs6000_reg_names): Ditto.
21621 * config/rs6000/rs6000.md: Renumber some register number
21624 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21626 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21628 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21629 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21631 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21632 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21633 Delete the SPE high registers.
21634 (REG_ALLOC_ORDER): Ditto.
21635 (enum reg_class): Remove SPE_HIGH_REGS.
21636 (REG_CLASS_NAMES): Ditto.
21637 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21638 (REGISTER_NAMES): Ditto.
21639 (rs6000_reg_names): Ditto.
21640 * doc/tm.texi.in: Remove SPE as example.
21641 * doc/tm.texi: Regenerate.
21643 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21645 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21646 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21647 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21648 * config/rs6000/rs6000.md (type): Remove "brinc".
21650 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21652 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21653 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21654 * config/rs6000/linuxspe.h: Delete file.
21655 * config/rs6000/rs6000.md: Don't include spe.md.
21656 * config/rs6000/spe.h: Delete file.
21657 * config/rs6000/spe.md: Delete file.
21658 * config/rs6000/t-rs6000: Remove spe.md.
21660 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21662 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21663 (reg_or_none500mem_operand): Delete.
21664 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21665 instead of reg_or_none500mem_operand.
21667 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21669 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21670 handling of SPE flags.
21671 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21673 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21677 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21678 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21679 paired_divv2sf3): Similar.
21680 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21681 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21682 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21684 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21685 Rename the paired_* instruction patterns.
21686 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21688 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21689 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21690 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21691 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21692 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21693 PAIRED_VECTOR_MODE.
21694 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21695 (spe_func_has_64bit_regs_p): Delete.
21696 (spe_expand_predicate_builtin): Delete.
21697 (spe_expand_evsel_builtin): Delete.
21698 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21699 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21700 (invalid_e500_subreg): Delete.
21701 (rs6000_legitimize_address): Always force_reg op2 as well, for
21702 paired single memory accesses.
21703 (rs6000_member_type_forces_blk): Delete.
21704 (rs6000_spe_function_arg): Delete.
21705 (rs6000_expand_unop_builtin): Delete SPE handling.
21706 (rs6000_expand_binop_builtin): Ditto.
21707 (spe_expand_stv_builtin): Delete.
21708 (bdesc_2arg_spe): Delete.
21709 (spe_expand_builtin): Delete.
21710 (spe_expand_predicate_builtin): Delete.
21711 (spe_expand_evsel_builtin): Delete.
21712 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21713 (spe_init_builtins): Delete.
21714 (spe_func_has_64bit_regs_p): Delete.
21715 (savres_routine_name): Delete "info" parameter. Adjust callers.
21716 (rs6000_emit_stack_reset): Ditto.
21717 (rs6000_dwarf_register_span): Delete.
21718 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21719 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21720 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21722 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21724 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21725 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21726 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21727 mulv2sf3, divv2sf3): Delete expanders.
21729 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21731 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21735 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21736 * config/rs6000/rs6000.c: Ditto.
21738 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21740 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21741 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21742 comparison_operator.
21744 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21746 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21747 * config/rs6000/rs6000.opt: Ditto.
21748 * config/rs6000/t-rtems: Ditto.
21750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21752 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21753 TARGET_E500_SINGLE by 0, simplify.
21754 * config/rs6000/rs6000.c: Ditto.
21755 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21756 (spe_build_register_parallel): Delete.
21757 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21758 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21759 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21760 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21761 (E500_CONVERT): Delete.
21762 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21764 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21766 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21767 * config/rs6000/dfp.md: Ditto.
21768 (negdd2, *negdd2_fpr): Merge.
21769 (absdd2, *absdd2_fpr): Merge.
21770 (negtd2, *negtd2_fpr): Merge.
21771 (abstd2, *abstd2_fpr): Merge.
21772 * config/rs6000/e500.h: Delete file.
21773 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21774 TARGET_FPRS by 1 and simplify.
21775 * config/rs6000/rs6000-c.c: Ditto.
21776 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21777 TARGET_DF_SPE by 0.
21778 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21780 * config/rs6000/rs6000.md: Ditto.
21781 (floatdidf2, *floatdidf2_fpr): Merge.
21782 (move_from_CR_gt_bit): Delete.
21783 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21784 (E500_CR_IOR_COMPARE): Delete.
21785 (All patterns that require !TARGET_FPRS): Delete.
21786 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21788 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21790 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21792 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21794 * graphds.c (add_edge): Intitialize edge's attached data.
21795 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21796 pointer parameter. Call pointed function on each edge during
21797 graph traversing. Skip traversing the edge when the function
21799 (graphds_dfs, graphds_scc): Ditto.
21800 (for_each_edge): New parameter. Pass the new parameter to callback
21802 * graphds.h (skip_edge_callback): New function pointer type.
21803 (graphds_dfs, graphds_scc): New function pointer parameter.
21804 (graphds_edge_callback, for_each_edge): New parameter.
21806 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21808 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21809 out code checking if runtime alias check is possible to below ...
21810 Call the new function.
21811 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21812 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21814 2017-06-07 Marek Polacek <polacek@redhat.com>
21817 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21818 TYPE_OVERFLOW_WRAPS check.
21820 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21822 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21823 if versioning is required.
21824 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21825 peeling with the check for versioning.
21827 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21829 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21830 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21831 Set true to new parameter if new ssa variable is defined.
21832 (vect_gen_vector_loop_niters): Refactor. Set range information
21833 for the new vector loop bound variable.
21834 (vect_do_peeling): Ditto.
21836 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21838 * tree-affine.c (ssa.h): Include header file.
21839 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21840 has wrapping overflow behavior.
21842 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21844 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21846 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21848 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21849 (tree_to_aff_combination): ... here.
21851 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21853 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21854 reg_pressure model function.
21855 (ivopts_global_cost_for_size): Delete.
21856 (determine_set_costs, iv_ca_recount_cost): Call new model function
21857 ivopts_estimate_reg_pressure.
21859 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21861 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21862 expensive than udiv. Remove floating point cases from mod.
21864 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21866 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21867 Increase idiv cost.
21869 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21871 * config/aarch64/aarch64.md
21872 (copysignsf3): Fix mask generation.
21874 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21876 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21878 (class dump_manager): Add register_dumps method.
21879 * dumpfile.c: Include langhooks.h.
21880 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21881 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21882 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21883 (dump_manager::dump_register): Start with 512 entries instead of 32.
21884 (dump_manager::register_dumps): New method.
21885 * toplev.c (general_init): Instead of invoking register_dumps
21886 langhook, invoke register_dumps method on the dump manager.
21887 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21890 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21892 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21893 Say that # requires an associated define_split to exist, and that
21894 the define_split must be suitable for use after register allocation.
21896 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21898 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21899 (compute_outgoing_frequencies): Also initialize zero counts.
21900 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21901 around loops; preserve more of profile when nothing changes.
21903 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21905 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21907 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21908 * config/arm/arm-cpu-cdata.h: Regenerate.
21909 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21910 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21911 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21912 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21913 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21915 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21917 * config/arm/t-rmprofile: Likewise.
21918 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21920 2017-06-06 David S. Miller <davem@davemloft.net>
21923 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21924 blockage if function uses alloca.
21926 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21928 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21929 New "uid" fields to hold pretty-print IDs of group and ref.
21930 Memory references are now identified as <group_id>:<ref_id>
21931 instead of using [random] addresses.
21932 (dump_mem_details): Simplify, no functional change.
21933 (dump_mem_ref): Simplify and make output more concise.
21934 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21935 (find_or_create_group): Initialize group uid.
21936 (record_ref): Initialize ref uid. Improve debug output.
21937 (prune_group_by_reuse, should_issue_prefetch_p,)
21938 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21939 (mark_nontemporal_store, determine_loop_nest_reuse):
21940 Improve debug output.
21942 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21944 * dbgcnt.def (prefetch): New debug counter.
21945 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21946 (schedule_prefetches): Stop issueing prefetches if debug counter
21949 2017-06-06 Tom de Vries <tom@codesourcery.com>
21951 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21952 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21954 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21956 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21957 Use aarch64_reg_or_zero predicate for operand 4.
21958 (aarch64_compare_and_swap<mode> define_insn_and_split):
21959 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21960 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21962 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21964 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21965 (arm_compute_save_core_reg_mask): This.
21966 (thumb1_compute_save_reg_mask): Rename into ...
21967 (thumb1_compute_save_core_reg_mask): This.
21968 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21969 (arm_compute_frame_layout): Likewise.
21971 2017-06-06 Richard Biener <rguenther@suse.de>
21973 PR tree-optimization/80974
21974 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21975 keep or clear leaders SSA info.
21977 2017-06-06 Tom de Vries <tom@codesourcery.com>
21979 * config/nvptx/nvptx.c (split_mode_p): New function.
21980 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21982 2017-06-06 Tom de Vries <tom@codesourcery.com>
21984 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21986 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21989 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21992 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21994 * shrink-wrap.c (handle_simple_exit): Update profile.
21995 (try_shrink_wrapping): Upate profile.
21997 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21999 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22000 (tree_guess_outgoing_edge_probabilities): New.
22001 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22002 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22004 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22006 * ipa-split.c (split_function): Initialize return bb profile.
22008 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22010 * profile.c (compute_branch_probabilities): Also initialize
22011 EXIT_BLOCK profile.
22013 2017-06-06 Richard Biener <rguenther@suse.de>
22015 PR tree-optimization/80928
22016 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22017 (vect_analyze_loop_operations): Properly guard analysis for
22019 (vect_transform_loop): Likewise.
22020 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22021 (vect_model_induction_cost): Do not cost for pure SLP.
22022 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22023 of induction in inner loop vectorization.
22024 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22025 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22026 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22028 (vect_analyze_slp_cost_1): Cost induction.
22029 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22030 (vect_get_slp_vect_defs): Likewise.
22031 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22032 (vect_transform_stmt): Handle SLP reductions.
22033 * tree-vectorizer.h (vectorizable_induction): Adjust.
22035 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22037 * config/rs6000/rs6000.c (make_resolver_func): Update
22038 init_lowered_empty_function call.
22040 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22042 * doc/invoke.texi: Document the -fprofile-abs-path option.
22043 * common.opt (fprofile-abs-path): New option.
22044 * gcov-io.h (gcov_write_filename): Declare.
22045 * gcov-io.c (gcov_write_filename): New function.
22046 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22047 * profile.c (output_location): Likewise.
22049 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22051 * shring-wrap.c: Revert accidental commit.
22053 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22055 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22057 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22059 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22061 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22062 profile in callgraph edge.
22063 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22064 (apply_scale): Likewise.
22065 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22066 Also scale profile when inlining function with zero profile.
22067 (initialize_cfun): Update exit block profile even when it is zero.
22068 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22069 when profile is read.
22071 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22073 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22074 (CLONE_*): New constants to define the processors we can generate
22075 code for with the target_clone attribute.
22076 (rs6000_clone_map): New array to identify which clone processors
22077 the current program is running on.
22078 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22079 target_clone attribute.
22080 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22081 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22082 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22083 (cpu_expand_builtin): Add support for target_clone attribute.
22084 (rs6000_valid_attribute_p): Allow "default" attribute.
22085 (get_decl_name): New debug function to simplify printing the
22086 current function name in debugging statements.
22087 (rs6000_clone_priority): New functions to support the target_clone
22088 attribute, and be able to generate code to switch between ISA 2.05
22089 through ISA 3.0 (power6 through power9).
22090 (rs6000_compare_version_priority): Likewise.
22091 (rs6000_get_function_versions_dispatcher): Likewise.
22092 (make_resolver_func): Likewise.
22093 (add_condition_to_bb): Likewise.
22094 (dispatch_function_versions): Likewise.
22095 (rs6000_generate_version_dispatcher_body): Likewise.
22096 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22097 (fusion_gpr_load_p): Fix a spacing issue.
22098 * doc/extend.texi (Common Function Attributes): Document that the
22099 PowerPC supports the target_clone attribute.
22101 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22103 * config/arm/arm.h: explain F symbol found in description of ARM
22104 register allocation in its legend.
22106 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22108 * config/mips/frame-header-opt.c: Include profile-count.h.
22109 * config/riscv/riscv.c: Include profile-count.h
22111 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22113 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22115 (sm_set_flag_if_changed): Add bbs field.
22116 (execute_sm_if_changed_flag_set): Pass BBS.
22117 (execute_sm): Update.
22119 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22121 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22124 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22126 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22127 (peephole2): New peephole2 to emit the above.
22128 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22130 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22132 * config/aarch64/aarch64.c (define_peephole2 above
22133 *sub_<shift>_<mode>): New peephole.
22135 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22137 * config/i386/i386.c (make_resolver_func): Update.
22138 * Makefile.in: Add profile-count.h and profile-count.o
22139 * auto-profile.c (afdo_indirect_call): Update to new API.
22140 (afdo_set_bb_count): Update.
22141 (afdo_propagate_edge): Update.
22142 (afdo_propagate_circuit): Update.
22143 (afdo_calculate_branch_prob): Update.
22144 (afdo_annotate_cfg): Update.
22145 * basic-block.h: Include profile-count.h
22146 (struct edge_def): Turn count to profile_count.
22147 (struct basic_block_def): Likewie.
22148 (REG_BR_PROB_BASE): Move to profile-count.h
22149 (RDIV): Move to profile-count.h
22150 * bb-reorder.c (max_entry_count): Turn to profile_count.
22151 (find_traces): Update.
22152 (rotate_loop):Update.
22153 (connect_traces):Update.
22154 (sanitize_hot_paths):Update.
22155 * bt-load.c (migrate_btr_defs): Update.
22156 * cfg.c (RDIV): Remove.
22157 (init_flow): Use alloc_block.
22158 (alloc_block): Uninitialize count.
22159 (unchecked_make_edge): Uninitialize count.
22160 (check_bb_profile): Update.
22161 (dump_edge_info): Update.
22162 (dump_bb_info): Update.
22163 (update_bb_profile_for_threading): Update.
22164 (scale_bbs_frequencies_int): Update.
22165 (scale_bbs_frequencies_gcov_type): Update.
22166 (scale_bbs_frequencies_profile_count): New.
22167 * cfg.h (update_bb_profile_for_threading): Update.
22168 (scale_bbs_frequencies_profile_count): Declare.
22169 * cfgbuild.c (compute_outgoing_frequencies): Update.
22170 (find_many_sub_basic_blocks): Update.
22171 * cfgcleanup.c (try_forward_edges): Update.
22172 (try_crossjump_to_edge): Update.
22173 * cfgexpand.c (expand_gimple_tailcall): Update.
22174 (construct_exit_block): Update.
22175 * cfghooks.c (verify_flow_info): Update.
22176 (dump_bb_for_graph): Update.
22177 (split_edge): Update.
22178 (make_forwarder_block): Update.
22179 (duplicate_block): Update.
22180 (account_profile_record): Update.
22181 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22182 (get_estimated_loop_iterations): Update.
22183 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22184 (single_likely_exit): Update.
22185 * cfgloopmanip.c (scale_loop_profile): Update.
22187 (set_zero_probability): Update.
22188 (lv_adjust_loop_entry_edge): Update.
22189 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22190 (purge_dead_edges): Update.
22191 (rtl_account_profile_record): Update.
22192 * cgraph.c (cgraph_node::create): Uninitialize count.
22193 (symbol_table::create_edge): Uninitialize count.
22194 (cgraph_update_edges_for_call_stmt_node): Update.
22195 (cgraph_edge::dump_edge_flags): Update.
22196 (cgraph_node::dump): Update.
22197 (cgraph_edge::maybe_hot_p): Update.
22198 * cgraph.h: Include profile-count.h
22199 (create_clone), create_edge, create_indirect_edge): Update.
22200 (cgraph_node): Turn count to profile_count.
22201 (cgraph_edge0: Likewise.
22202 (make_speculative, clone): Update.
22203 (create_edge): Update.
22204 (init_lowered_empty_function): Update.
22205 * cgraphclones.c (cgraph_edge::clone): Update.
22206 (duplicate_thunk_for_node): Update.
22207 (cgraph_node::create_clone): Update.
22208 * cgraphunit.c (cgraph_node::analyze): Update.
22209 (cgraph_node::expand_thunk): Update.
22210 * final.c (dump_basic_block_info): Update.
22211 * gimple-streamer-in.c (input_bb): Update.
22212 * gimple-streamer-out.c (output_bb): Update.
22213 * graphite.c (print_global_statistics): Update.
22214 (print_graphite_scop_statistics): Update.
22215 * hsa-brig.c: Include basic-block.h.
22216 * hsa-dump.c: Include basic-block.h.
22217 * hsa-gen.c (T sum_slice): Update.
22218 (convert_switch_statements):Update.
22219 * hsa-regalloc.c: Include basic-block.h.
22220 * ipa-chkp.c (chkp_produce_thunks): Update.
22221 * ipa-cp.c (struct caller_statistics): Update.
22222 (init_caller_stats): Update.
22223 (gather_caller_stats): Update.
22224 (ipcp_cloning_candidate_p): Update.
22225 (good_cloning_opportunity_p): Update.
22226 (get_info_about_necessary_edges): Update.
22227 (dump_profile_updates): Update.
22228 (update_profiling_info): Update.
22229 (update_specialized_profile): Update.
22230 (perhaps_add_new_callers): Update.
22231 (decide_about_value): Update.
22232 (ipa_cp_c_finalize): Update.
22233 * ipa-devirt.c (struct odr_type_warn_count): Update.
22234 (struct decl_warn_count): Update.
22235 (struct final_warning_record): Update.
22236 (possible_polymorphic_call_targets): Update.
22237 (ipa_devirt): Update.
22238 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22239 * ipa-icf.c (sem_function::merge): Update.
22240 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22241 * ipa-inline.c (compute_uninlined_call_time): Update.
22242 (compute_inlined_call_time): Update.
22243 (want_inline_small_function_p): Update.
22244 (want_inline_self_recursive_call_p): Update.
22245 (edge_badness): Update.
22246 (lookup_recursive_calls): Update.
22247 (recursive_inlining): Update.
22248 (inline_small_functions): Update.
22249 (dump_overall_stats): Update.
22250 (dump_inline_stats): Update.
22251 * ipa-profile.c (ipa_profile_generate_summary): Update.
22252 (ipa_propagate_frequency): Update.
22253 (ipa_profile): Update.
22254 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22255 * ipa-utils.c (ipa_merge_profiles): Update.
22256 * loop-doloop.c (doloop_modify): Update.
22257 * loop-unroll.c (report_unroll): Update.
22258 (unroll_loop_runtime_iterations): Update.
22259 * lto-cgraph.c (lto_output_edge): Update.
22260 (lto_output_node): Update.
22261 (input_node): Update.
22262 (input_edge): Update.
22263 (merge_profile_summaries): Update.
22264 * lto-streamer-in.c (input_cfg): Update.
22265 * lto-streamer-out.c (output_cfg): Update.
22266 * mcf.c (create_fixup_graph): Update.
22267 (adjust_cfg_counts): Update.
22268 (sum_edge_counts): Update.
22269 * modulo-sched.c (sms_schedule): Update.
22270 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22271 * predict.c (maybe_hot_count_p): Update.
22272 (probably_never_executed): Update.
22273 (dump_prediction): Update.
22274 (combine_predictions_for_bb): Update.
22275 (propagate_freq): Update.
22276 (handle_missing_profiles): Update.
22277 (counts_to_freqs): Update.
22278 (rebuild_frequencies): Update.
22279 (force_edge_cold): Update.
22280 * predict.h: Include profile-count.h
22281 (maybe_hot_count_p, counts_to_freqs): UPdate.
22282 * print-rtl-function.c: Do not include cfg.h
22283 * print-rtl.c: Include basic-block.h
22284 * profile-count.c: New file.
22285 * profile-count.h: New file.
22286 * profile.c (is_edge_inconsistent): Update.
22287 (correct_negative_edge_counts): Update.
22288 (is_inconsistent): Update.
22289 (set_bb_counts): Update.
22290 (read_profile_edge_counts): Update.
22291 (compute_frequency_overlap): Update.
22292 (compute_branch_probabilities): Update; Initialize and deinitialize
22294 (branch_prob): Update.
22295 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22296 (edge_gcov_count): New.
22297 (bb_gcov_count): New.
22298 * shrink-wrap.c (try_shrink_wrapping): Update.
22299 * tracer.c (better_p): Update.
22300 * trans-mem.c (expand_transaction): Update.
22301 (ipa_tm_insert_irr_call): Update.
22302 (ipa_tm_insert_gettmclone_call): Update.
22303 * tree-call-cdce.c: Update.
22304 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22305 (gimple_duplicate_sese_tail): Update.
22306 (gimple_account_profile_record): Update.
22307 (execute_fixup_cfg): Update.
22308 * tree-inline.c (copy_bb): Update.
22309 (copy_edges_for_bb): Update.
22310 (initialize_cfun): Update.
22311 (freqs_to_counts): Update.
22312 (copy_cfg_body): Update.
22313 (expand_call_inline): Update.
22314 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22315 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22316 (try_unroll_loop_completely): Update.
22317 (try_peel_loop): Update.
22318 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22319 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22320 * tree-ssa-loop-split.c (connect_loops): Update.
22321 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22322 * tree-ssa-reassoc.c (branch_fixup): Update.
22323 * tree-ssa-tail-merge.c (replace_block_by): Update.
22324 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22325 (compute_path_counts): Update.
22326 (update_profile): Update.
22327 (recompute_probabilities): Update.
22328 (update_joiner_offpath_counts): Update.
22329 (estimated_freqs_path): Update.
22330 (freqs_to_counts_path): Update.
22331 (clear_counts_path): Update.
22332 (ssa_fix_duplicate_block_edges): Update.
22333 (duplicate_thread_path): Update.
22334 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22335 (struct switch_conv_info): Update.
22336 * tree-tailcall.c (decrease_profile): Update.
22337 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22338 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22339 * value-prof.c (check_counter): Update.
22340 (gimple_divmod_fixed_value): Update.
22341 (gimple_mod_pow2): Update.
22342 (gimple_mod_subtract): Update.
22343 (gimple_ic_transform): Update.
22344 (gimple_stringop_fixed_value): Update.
22345 * value-prof.h (gimple_ic): Update.
22347 2017-06-02 Carl Love <cel@us.ibm.com>
22349 * config/rs6000/rs6000-c: Add support for built-in functions
22350 vector double vec_doublee (vector signed int);
22351 vector double vec_doublee (vector unsigned int);
22352 vector double vec_doublee (vector float);
22353 vector double vec_doubleh (vector signed int);
22354 vector double vec_doubleh (vector unsigned int);
22355 vector double vec_doubleh (vector float);
22356 vector double vec_doublel (vector signed int);
22357 vector double vec_doublel (vector unsigned int);
22358 vector double vec_doublel (vector float);
22359 vector double vec_doubleo (vector signed int);
22360 vector double vec_doubleo (vector unsigned int);
22361 vector double vec_doubleo (vector float);.
22362 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22363 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22365 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22366 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22367 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22369 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22370 vec_doublel, vec_doubleh.
22371 * doc/extend.texi: Update the built-in documentation file for the
22372 new built-in functions.
22374 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22377 * ipa-inline-analysis.c (free_growth_caches): Set
22378 edge_removal_hook_holder to NULL after removing it.
22380 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22382 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22383 comparision with zero.
22385 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22387 for early expansion of vec_min and vec_max builtins.
22388 (builtin_function_type): Add min/max unsigned variants to those
22389 identified as having unsigned arguments.
22391 2017-06-02 Olivier Hainque <hainque@adacore.com>
22393 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22395 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22397 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22398 Use VALL_F16 iterator rather than VALL.
22400 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22402 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22403 Emit CBNZ inside loop when doing a strong exchange and comparing
22404 against zero. Generate the CC flags after the loop.
22406 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22408 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22409 (dl_section_ref): New.
22410 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22411 On AIX, append an expression to subtract the size of the
22412 section length to dl_section_ref.
22414 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22416 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22417 for early expansion of vector absolute builtins.
22419 2017-06-02 Richard Biener <rguenther@suse.de>
22421 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22422 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22424 2017-06-02 Richard Biener <rguenther@suse.de>
22426 PR tree-optimization/80948
22427 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22428 stmt order as well.
22430 2017-06-02 Richard Biener <rguenther@suse.de>
22432 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22434 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22435 for inductions as relevant.
22437 2017-06-02 Richard Biener <rguenther@suse.de>
22439 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22440 (vectorizable_induction): ... this. Remove dead code.
22442 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22444 * builtins. (expand_builtin_alloca): Remove second parameter and
22445 infer its value from the first parameter instead.
22446 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22448 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22450 PR rtl-optimization/80903
22451 * loop-doloop.c (add_test): Unshare sequence.
22453 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22455 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22457 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22459 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22461 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22462 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22463 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22464 (xlogue_layout::get_stub_rtx): Make static.
22465 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22466 (xlogue_layout::compute_stub_managed_regs): Rename to...
22467 (xlogue_layout::count_stub_managed_regs): ...this.
22468 (xlogue_layout::is_stub_managed_reg): New function.
22469 (xlogue_layout::m_stub_names): Rename to...
22470 (xlogue_layout::s_stub_names): ...this, make static.
22471 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22472 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22473 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22474 xlogue_layout::s_stub_names): Instantiate statics.
22475 (stub_managed_regs): Remove.
22476 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22477 (disable_call_ms2sysv_xlogues): Rename to...
22478 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22479 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22481 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22482 change after reload_completed.
22483 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22485 (ix86_expand_prologue): Likewise.
22486 (ix86_expand_epilogue): Likewise.
22487 (ix86_expand_split_stack_prologue): Likewise.
22488 (ix86_compute_frame_layout): Remove frame parameter ...
22489 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22490 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22492 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22493 (ix86_frame): Move from here ...
22494 * config/i386/i386.h (ix86_frame): ... to here.
22495 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22496 complete ix86_frame data structure instead. Remove some_ld_name.
22498 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22500 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22501 symbols that hold a DECL_VALUE_EXPR.
22503 2017-06-01 Martin Jambor <mjambor@suse.cz>
22505 PR tree-optimization/80898
22506 * tree-sra.c (process_subtree_disqualification): Removed.
22507 (disqualify_candidate): Do not acll
22508 process_subtree_disqualification.
22509 (subtree_mark_written_and_enqueue): New function.
22510 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22511 RHS has been disqualified and re-queue LHS if necessary. Apart
22512 from that, ignore disqualified RHS.
22514 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22516 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22517 address fetch for z10 or later.
22519 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22521 * config/arc/arc.md (tst_movb): Add guard when splitting.
22523 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22525 * config/arc/arc.c (arc_can_eliminate): Test against
22526 arc_frame_pointer_needed.
22528 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22530 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22531 to prevent store reordering.
22532 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22533 (type): Add block type.
22534 (stack_tie): Define special instruction to be used in
22537 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22539 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22540 constraint. It is not valid for the pattern.
22541 (noncommutative_binary_comparison): Likewise.
22543 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22545 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22548 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22550 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22551 be used by the reg-alloc.
22553 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22555 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22556 reg-alloc when having mul64 or mul32x16 instructions.
22557 (mulsidi3): Likewise.
22558 (umulsidi3): Likewise.
22559 (mulsi32x16): New pattern.
22560 (mulsi64): Likewise.
22561 (mulsidi64): Likewise.
22562 (umulsidi64): Likewise.
22563 (MUL32x16_REG): Define.
22564 (mul64_600): Use MUL32x16_REG.
22565 (mac64_600): Likewise.
22566 (umul64_600): Likewise.
22567 (umac64_600): Likewise.
22569 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22571 * config/arc/arc.md (mulsi3_700): Make it commutative.
22573 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22575 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22577 (*sign_extendsidi2_insn): Likewise for movstosw.
22579 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22581 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22582 the type of the input discriminant value. Convert the
22583 discriminant value of signedness vary.
22585 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22587 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22588 Add to -Wall section.
22590 2017-06-01 Richard Biener <rguenther@suse.de>
22592 PR middle-end/66313
22593 * fold-const.c (fold_plusminus_mult_expr): If the factored
22594 factor may be zero use a wrapping type for the inner operation.
22595 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22596 and handle moved defs.
22597 (process_assignment): Properly guard the unary op case. Return a
22598 tri-state indicating that moving the stmt before the call may allow
22599 to continue. Pass through to_move.
22600 (find_tail_calls): Handle moving unrelated defs before
22603 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22606 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22607 splitter result in the canonical way.
22609 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22611 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22612 also for 32bit target. Update insn attributes.
22613 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22615 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22617 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22618 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22619 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22620 (_mm_maskz_min_ss): New intrinsics.
22622 2017-05-31 Martin Liska <mliska@suse.cz>
22624 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22625 Change comment style to one we normally use.
22626 (vectorizable_reduction): Likewise.
22627 (vectorizable_induction): Likewise.
22628 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22629 (vectorizable_call): Likewise.
22630 (vectorizable_simd_clone_call): Likewise.
22631 (vectorizable_conversion): Likewise.
22632 (vectorizable_assignment): Likewise.
22633 (vectorizable_shift): Likewise.
22634 (vectorizable_operation): Likewise.
22635 (vectorizable_store): Likewise.
22636 (vectorizable_load): Likewise.
22637 * tree-vectorizer.h: Likewise.
22639 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22641 * passes.c (emergency_dump_function): New.
22642 * tree-pass.h (emergency_dump_function): Declare.
22643 * plugin.c (plugins_internal_error_function): Remove.
22644 * plugin.h (plugins_internal_error_function): Remove declaration.
22645 * toplev.c (internal_error_function): New static function. Use it...
22646 (general_init): ...here.
22648 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22650 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22651 (arc_rtx_costs): Add costs for new patterns.
22652 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22653 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22655 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22657 * tree-ssa-strlen.c (get_next_strinfo): New function.
22658 (get_stridx_plus_constant): Use it.
22659 (zero_length_string): Likewise.
22660 (adjust_related_strinfos): Likewise.
22661 (adjust_last_stmt): Likewise.
22663 2017-05-31 Richard Biener <rguenther@suse.de>
22666 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22667 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22669 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22671 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22672 loop_vinfo argument and use of dependence distance vectors.
22673 Check instead whether the two references differ only in their
22674 initial value and assume that they have the same alignment if the
22675 difference is a multiple of the vector alignment.
22676 (vect_analyze_data_refs_alignment): Update call accordingly.
22678 2017-05-31 Martin Liska <mliska@suse.cz>
22681 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22683 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22685 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22686 (create_intersect_range_checks): Move from ...
22687 * tree-data-ref.c (create_intersect_range_checks_index)
22688 (create_intersect_range_checks): ... to here.
22689 (create_runtime_alias_checks): New function factored from ...
22690 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22691 here. Call above function.
22692 * tree-data-ref.h (create_runtime_alias_checks): New function.
22694 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22696 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22697 segment length for dr_b and compute it in wide_int.
22699 2017-05-31 Richard Biener <rguenther@suse.de>
22701 PR tree-optimization/80906
22702 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22703 and pass through iv_map.
22704 (copy_bb_and_scalar_dependences): Adjust.
22705 (translate_pending_phi_nodes): Likewise.
22706 (copy_loop_close_phi_args): Handle code-generating IVs instead
22709 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22711 * diagnostic-color.c (color_dict): Add "type-diff".
22712 (parse_gcc_colors): Update comment.
22713 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22714 -fdiagnostics-show-template-tree and -fno-elide-type.
22715 (GCC_COLORS): Add type-diff to example.
22717 (-fdiagnostics-show-template-tree): New.
22718 (-fno-elide-type): New.
22719 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22720 the pp_format_decoder callback. Call any m_format_postprocessor's
22722 (pretty_printer::pretty_printer): Initialize
22723 m_format_postprocessor.
22724 (pretty_printer::~pretty_printer): Delete any
22725 m_format_postprocessor.
22726 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22727 (class format_postprocessor): New class.
22728 (struct pretty_printer::format_decoder): Document the new parameters.
22729 (struct pretty_printer::m_format_postprocessor): New field.
22730 * tree-diagnostic.c (default_tree_printer): Update for new
22731 bool and const char ** params.
22732 * tree-diagnostic.h (default_tree_printer): Likewise.
22734 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22736 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22737 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22739 rs6000_gen_cell_microcode code.
22740 (rs6000_final_prescan_insn): Delete.
22741 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22742 "warn-cell-microcode" entries.
22743 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22744 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22745 throughout. Change cc_reg_not_micro_cr0_operand to
22746 cc_reg_not_cr0_operand throughout.
22747 (*extendhi<mode>2_noload): Delete.
22748 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22749 (mwarn-cell-microcode): Delete.
22750 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22751 -mgen-cell-microcode and -mwarn-cell-microcode.
22753 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22756 * config/i386/constraints.md (Yd): New constraint.
22758 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22759 and (?Yd, r) alternatives. Update insn attributes.
22760 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22761 and (?*Yd, r) alternatives. Update insn attributes.
22762 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22764 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22766 * gimplify.c (gimplify_modify_expr): Don't create a
22767 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22770 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22772 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22774 2017-05-30 Richard Biener <rguenther@suse.de>
22776 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22777 and reduc_def fields.
22778 (STMT_VINFO_REDUC_TYPE): New define.
22779 (STMT_VINFO_REDUC_DEF): Likewise.
22780 (vect_force_simple_reduction): Adjust prototype.
22781 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22782 (vect_is_simple_reduction): Remove check_reduction argument.
22783 (vect_force_simple_reduction): Adjust and set
22784 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22785 (vectorizable_reduction): Do not re-do reduction analysis
22786 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22787 * tree-parloops.c (gather_scalar_reductions): Adjust.
22789 2017-05-30 Richard Biener <rguenther@suse.de>
22791 PR middle-end/80901
22792 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22795 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22797 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22798 Introduce unknown_misalignment parameter and remove vf.
22799 (vect_peeling_hash_get_lowest_cost):
22800 Pass unknown_misalignment parameter.
22801 (vect_enhance_data_refs_alignment):
22802 Fix unsupportable data ref treatment.
22804 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22806 * tree-vect-data-refs.c (vect_get_data_access_cost):
22807 Workaround for SLP handling.
22808 (vect_enhance_data_refs_alignment):
22809 Compute costs for doing no peeling at all, compare to the best
22810 peeling costs so far and avoid peeling if cheaper.
22812 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22814 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22815 Return peeling info and set costs to zero for unlimited cost
22817 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22818 with unknown misalignment. Compute and costs for unknown
22819 misalignment, compare them to the costs for known misalignment
22820 and choose the cheapest for peeling.
22822 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22824 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22825 (vect_get_peeling_costs_all_drs): Create function.
22826 (vect_peeling_hash_get_lowest_cost):
22827 Use vect_get_peeling_costs_all_drs.
22828 (vect_peeling_supportable): Create function.
22829 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22831 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22833 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22834 DR_HAS_NEGATIVE_STEP.
22835 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22836 (vect_enhance_data_refs_alignment): Use.
22837 (vect_duplicate_ssa_name_ptr_info): Use.
22838 * tree-vectorizer.h (dr_misalignment): Use.
22839 (known_alignment_for_access_p): Use.
22841 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22844 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22846 (has_section_name): New function.
22848 2017-05-30 Martin Liska <mliska@suse.cz>
22851 * auto-profile.c (get_function_decl_from_block): Fix
22854 2017-05-30 Richard Biener <rguenther@suse.de>
22856 PR middle-end/80876
22857 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22859 2017-05-30 Martin Liska <mliska@suse.cz>
22861 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22862 * dumpfile.h (struct dump_file_info): Remove ctors.
22864 2017-05-30 Martin Liska <mliska@suse.cz>
22866 * predict.def: Fix GNU coding style.
22868 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22870 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22871 Mark 'to' argument with ATTRIBUTE_UNUSED.
22873 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22875 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22876 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22877 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22880 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22882 * doc/install.texi (Options specification): Restore entry of
22883 --enable-sjlj-exceptions.
22885 2017-05-27 Michael Eager <eager@eagercon.com>
22888 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22890 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22892 * config/microblaze/microblaze.h
22893 (FIXED_REGISTERS): Update in macro.
22894 (CALL_USED_REGISTERS): Update in macro.
22896 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22898 * doc/install.texi: Add links to macOS binary distributions.
22900 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22904 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22906 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22908 2017-05-26 Martin Liska <mliska@suse.cz>
22910 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22912 2017-05-26 Martin Liska <mliska@suse.cz>
22914 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22915 always leading ';; '.
22916 (dump_bb_info): Likewise.
22917 (brief_dump_cfg): Likewise.
22918 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22919 * dumpfile.c: Remove usage of TDF_VERBOSE.
22920 * dumpfile.h (enum dump_kind): Likewise.
22921 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22922 * print-tree.c (debug_verbose): Remove.
22923 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22924 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22925 * tree-diagnostic.c (default_tree_printer): Replace
22926 TDF_DIAGNOSTIC with TDF_SLIM.
22928 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22930 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22931 in parameter loop, rather than loop_vinfo.
22932 (create_intersect_range_checks): Ditto.
22933 (vect_create_cond_for_alias_checks): Update call to above functions.
22935 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22937 PR tree-optimization/80815
22938 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22939 for merging runtime alias checks. Handle negative DR_STEPs.
22941 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22943 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22945 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22946 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22947 out code pruning runtime alias checks.
22948 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22949 factored out from above.
22950 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22952 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22954 (prune_runtime_alias_test_list): New decalaration.
22956 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22958 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22959 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22960 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22961 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22962 (operator==, comp_dr_with_seg_len_pair): Ditto.
22963 (vect_prune_runtime_alias_test_list): Ditto.
22965 2017-05-26 Martin Liska <mliska@suse.cz>
22968 * params.def: Bound partial-inlining-entry-probability param.
22970 2017-05-26 Marek Polacek <polacek@redhat.com>
22973 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22976 2017-05-26 Richard Biener <rguenther@suse.de>
22978 PR tree-optimization/80842
22979 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22982 2017-05-26 Richard Biener <rguenther@suse.de>
22984 PR tree-optimization/80844
22985 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22987 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22989 * doc/md.texi (Machine Constraints): Update x86 family
22990 machine constraints section to match 'config/i386/constraints.md'.
22992 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22994 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22996 2017-05-25 Nathan Sidwell <nathan@acm.org>
22998 * doc/invoke.texi (--enable-languages): Update documentation.
23000 2017-05-25 Martin Liska <mliska@suse.cz>
23002 * dumpfile.c: Add TDF_FOLDING.
23003 * dumpfile.h (enum dump_kind): Likewise.
23004 * genmatch.c (dt_simplify::gen_1): Use it.
23006 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23008 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23010 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23012 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23013 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23015 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23017 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23018 * match.pd (X == C): Rewrite it here.
23019 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23020 with_certain_nonzero_bits2): New predicates.
23021 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23023 2017-05-24 Nathan Sidwell <nathan@acm.org>
23025 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23028 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23031 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23033 * config/powerpcspe: New port. Files are copied from the rs6000
23034 port, with "rs6000" in filenames replaced by "powerpcspe".
23036 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23038 PR rtl-optimization/80754
23039 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23041 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23043 * config/sparc/sparc.md (length): Return the correct value for -mflat
23044 sibcalls to match output_sibcall.
23046 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23050 * config/rs6000/rs6000.c (struct machine_function): Add new field
23052 (rs6000_get_separate_components): Init that field, use it.
23053 (rs6000_components_for_bb): Use the field.
23055 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23057 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23059 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23061 PR middle-end/80823
23062 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23064 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23067 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23068 address against address_operand predicate.
23069 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23071 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23073 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23074 parameters passed indirectly.
23076 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23078 * config/i386/i386.md (*movdi_internal): Remove SSE4
23079 alternative 18 (?r, *v). Update insn attributes.
23080 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23081 Update insn attributes.
23082 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23083 Update insn attributes.
23084 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23085 alternative 1 (r, v). Remove isa attribute.
23086 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23087 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23088 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23090 2017-05-23 Tom de Vries <tom@codesourcery.com>
23092 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23095 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23097 * cgraphunit.c (symbol_table::process_new_functions): Update.
23098 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23099 (inline_generate_summary): Rename to ...
23100 (ipa_fn_summary_generate): ... this one.
23101 (inline_read_summary): Rename to ...
23102 (ipa_fn_summary_read): ... this one.
23103 (inline_write_summary): Rename to ...
23104 (ipa_fn_summary_write): ... this one.
23105 (inline_free_summary): Rename to ...
23106 (ipa_free_fn_summary): ... this one.
23107 (pass_data_local_fn_summary, pass_local_fn_summary,
23108 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23109 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23110 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23111 make_pass_ipa_fn_summary): New.
23112 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23113 inline_write_summary, inline_free_summary): Remove.
23114 (ipa_free_fn_summary) : New.
23115 * ipa-inline.c (ipa_inline): Update.
23116 (pass_ipa_inline): Do not generate summaries.
23117 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23119 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23120 and add pass_ipa_fn_summary.
23121 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23123 (make_pass_inline_parameters): Remove.
23125 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23127 * omp-low.c (struct omp_context): Remove "default_kind" member.
23130 * omp-offload.c (execute_oacc_device_lower): Remove the
23131 parallelism dimensions function attributes for unparallelized
23132 OpenACC kernels constructs.
23134 2017-05-23 Martin Liska <mliska@suse.cz>
23136 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23138 (cgraph_edge::make_speculative): Likewise.
23139 (cgraph_edge::resolve_speculation): Likewise.
23140 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23141 (cgraph_node::dump): Likewise.
23142 * cgraph.h: Likewise.
23143 * cgraphunit.c (analyze_functions): Likewise.
23144 (symbol_table::compile): Likewise.
23145 * ipa-cp.c (print_all_lattices): Likewise.
23146 (determine_versionability): Likewise.
23147 (initialize_node_lattices): Likewise.
23148 (ipcp_verify_propagated_values): Likewise.
23149 (estimate_local_effects): Likewise.
23150 (update_profiling_info): Likewise.
23151 (create_specialized_node): Likewise.
23152 (perhaps_add_new_callers): Likewise.
23153 (decide_about_value): Likewise.
23154 (decide_whether_version_node): Likewise.
23155 (identify_dead_nodes): Likewise.
23156 (ipcp_store_bits_results): Likewise.
23157 * ipa-devirt.c (dump_targets): Likewise.
23158 (ipa_devirt): Likewise.
23159 * ipa-icf.c (sem_item::dump): Likewise.
23160 (sem_function::equals): Likewise.
23161 (sem_variable::equals): Likewise.
23162 (sem_item_optimizer::read_section): Likewise.
23163 (sem_item_optimizer::execute): Likewise.
23164 (congruence_class::dump): Likewise.
23165 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23166 (dump_inline_summary): Likewise.
23167 (estimate_node_size_and_time): Likewise.
23168 (inline_analyze_function): Likewise.
23169 * ipa-inline-transform.c (inline_call): Likewise.
23170 * ipa-inline.c (report_inline_failed_reason): Likewise.
23171 (want_early_inline_function_p): Likewise.
23172 (edge_badness): Likewise.
23173 (update_edge_key): Likewise.
23174 (inline_small_functions): Likewise.
23175 * ipa-profile.c (ipa_profile): Likewise.
23176 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23177 (ipa_make_edge_direct_to_target): Likewise.
23178 (remove_described_reference): Likewise.
23179 (ipa_impossible_devirt_target): Likewise.
23180 (propagate_controlled_uses): Likewise.
23181 (ipa_print_node_params): Likewise.
23182 (ipcp_transform_function): Likewise.
23183 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23184 (propagate_pure_const): Likewise.
23185 * ipa-reference.c (generate_summary): Likewise.
23186 (read_write_all_from_decl): Likewise.
23187 (propagate): Likewise.
23188 (ipa_reference_read_optimization_summary): Likewise.
23189 * ipa-utils.c (ipa_merge_profiles): Likewise.
23190 * ipa.c (walk_polymorphic_call_targets): Likewise.
23191 (symbol_table::remove_unreachable_nodes): Likewise.
23192 (ipa_single_use): Likewise.
23193 * passes.c (execute_todo): Likewise.
23194 * predict.c (drop_profile): Likewise.
23195 * symtab.c (symtab_node::get_dump_name): New function.
23196 (symtab_node::dump_name): Likewise.
23197 (symtab_node::dump_asm_name): Likewise.
23198 (symtab_node::dump_references): Likewise.
23199 (symtab_node::dump_referring): Likewise.
23200 (symtab_node::dump_base): Likewise.
23201 (symtab_node::debug_symtab): Likewise.
23202 * tree-sra.c (convert_callers_for_node): Likewise.
23203 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23204 * value-prof.c (init_node_map): Likewise.
23206 2017-05-23 Martin Liska <mliska@suse.cz>
23208 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23209 and symtab_node::debug_symtab to symbol_table::debug.
23210 * cgraphunit.c (analyze_functions): Use the renamed function.
23211 (symbol_table::compile): Likewise.
23212 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23213 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23214 * passes.c (execute_todo): Likewise.
23215 * symtab.c (symbol_table::dump): New function.
23216 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23218 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23220 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23221 that nonconst implies exec.
23223 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23225 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23226 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23227 (inline_edge_summary_vec): Turn into ...
23228 (ipa_call_summaries): ... this one.
23229 (redirect_to_unreachable, edge_set_predicate,
23230 evaluate_properties_for_edge, inline_summary_alloc,
23231 reset_ipa_call_summary, reset_inline_summary,
23232 inline_summary_t::duplicate): Update.
23233 (inline_edge_duplication_hook): Turn to ...
23234 (ipa_call_summary_t::duplicate): ... this one.
23235 (inline_edge_removal_hook): Turn to ...
23236 (ipa_call_summary_t::remove): ... this one.
23237 (dump_inline_edge_summary): Turn to ...
23238 (dump_ipa_call_summary): ... this one.
23239 (estimate_function_body_sizes): Update.
23240 (inline_update_callee_summaries): Update.
23241 (remap_edge_change_prob): Update.
23242 (remap_edge_summaries): Update.
23243 (inline_merge_summary): Update.
23244 (do_estimate_edge_time): Update.
23245 (inline_generate_summary): Update.
23246 (inline_read_section): Update.
23247 (inline_read_summary): Update.
23248 (inline_free_summary): Update.
23249 * ipa-inline.c (can_inline_edge_p): Update.
23250 (compute_inlined_call_time): Update.
23251 (want_inline_small_function_p): Update.
23252 (edge_badness): Update.
23253 (early_inliner): Update.
23254 * ipa-inline.h (inline_edge_summary): Turn to ...
23255 (ipa_call_summary): ... this one.
23256 (ipa_call_summary_t): New class.
23257 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23258 (ipa_call_summaries): New.
23259 (inline_edge_summary): Remove.
23260 (estimate_edge_growth): Update.
23261 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23262 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23263 * ipa-split.c (execute_split_functions): Update.
23264 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23266 2017-05-23 Tom de Vries <tom@codesourcery.com>
23268 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23269 attributes): Document rdrand effective target.
23271 2017-05-23 Tom de Vries <tom@codesourcery.com>
23273 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23274 attributes): Sort alphabetically.
23276 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23278 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23280 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23283 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23284 V2DF/V2DI splat into two separate patterns, one that handles
23285 registers, and the other that only handles memory. Drop support
23286 for splatting from a GPR on ISA 2.07 and then splitting the
23287 splat into direct move and splat.
23288 (vsx_splat_<mode>_reg): Likewise.
23289 (vsx_splat_<mode>_mem): Likewise.
23291 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23293 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23295 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23297 PR middle-end/80809
23298 * omp-low.c (finish_taskreg_remap): New function.
23299 (finish_taskreg_scan): If unit size of ctx->record_type
23300 is non-constant, unshare the size expression and replace
23301 decls in it with possible outer var refs.
23303 PR middle-end/80809
23304 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23305 GOVD_SHARED rather than GOVD_PRIVATE with it.
23306 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23307 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23309 PR middle-end/80853
23310 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23311 as last argument to build_outer_var_ref for pointer bases of array
23312 section reductions.
23314 2017-05-19 Martin Sebor <msebor@redhat.com>
23316 * print-tree.c (print_node): Print DECL_READ_P flag.
23318 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23320 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23321 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23322 * cgraph.c: Likewise.
23323 * cgraphunit.c: Likewise.
23324 * gengtype.c: Likewise.
23325 * ipa-cp.c: Likewise.
23326 * ipa-devirt.c: Likewise.
23327 * ipa-icf.c: Likewise.
23328 * ipa-predicate.c: Likewise.
23329 * ipa-profile.c: Likewise.
23330 * ipa-prop.c: Likewise.
23331 * ipa-split.c: Likewise.
23333 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23334 edge_predicate_pool, dump_inline_hints,
23335 inline_summary::account_size_time, redirect_to_unreachable,
23336 edge_set_predicate, set_hint_predicate,
23337 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23338 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23339 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23340 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23341 ipa_call_summary_t::remove, initialize_growth_caches,
23342 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23343 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23344 mark_modified, unmodified_parm_1, unmodified_parm,
23345 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23346 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23347 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23348 will_be_nonconstant_predicate, record_modified_bb_info,
23349 get_minimal_bb, record_modified, param_change_prob,
23350 phi_result_unknown_predicate, predicate_for_phi_result,
23351 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23352 estimate_function_body_sizes, compute_inline_parameters,
23353 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23354 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23355 inline_update_callee_summaries, remap_edge_change_prob,
23356 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23357 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23358 inline_analyze_function, inline_summary_t::insert,
23359 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23360 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23361 inline_free_summary): Move to ipa-fnsummary.h
23362 (predicate_t): Remove.
23363 * ipa-fnsummary.c: New file.
23364 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23365 (enum inline_hints_vals, inline_hints, agg_position_info,
23366 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23367 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23368 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23369 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23370 inline_read_summary, inline_write_summary, inline_free_summary,
23371 inline_analyze_function, initialize_inline_failed,
23372 inline_merge_summary, inline_update_overall_summary,
23373 compute_inline_parameters): Move to ipa-fnsummary.h
23374 * ipa-fnsummary.h: New file.
23375 * ipa-inline-transform.h: Include ipa-inline.h.
23376 * ipa-inline.c: LIkewise.
23378 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23380 * ipa-inline.c (edge_badness): Use inlined_time instead of
23381 inline_summaries->get.
23383 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23385 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23387 2017-05-22 Nathan Sidwell <nathan@acm.org>
23389 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23390 (fdump-lang): Document 'raw' option.
23391 * dumpfile.h (TDI_tu): Delete.
23392 * dumpfile.c (dump_files): Remove translation-unit.
23393 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23395 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23397 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23398 command option from $(AWK) call.
23399 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23400 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23401 [FORMAT]: Remove handling of variable.
23402 * config/avr/t-multilib: Regenerate.
23404 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23406 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23408 (dump_inline_summary): Do not print self_time.
23409 (estimate_function_body_sizes): Do not set self_time.
23410 (compute_inline_parameters): Likewise.
23411 (inline_read_section, inline_write_summary): Do not stream self_time.
23412 * ipa-inline.h (inline_summary): Drop self_time.
23414 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23416 * ipa-inline-analysis.c (account_size_time): Rename to ...
23417 (inline_summary::account_size_time): ... this one.
23418 (reset_ipa_call_summary): Turn to ...
23419 (ipa_call_summary::reset): ... this one.
23420 (reset_inline_summary): Turn to ...
23421 (inline_summary::reset): ... this one.
23422 (inline_summary_t::remove): Update.
23423 (inline_summary_t::duplicate): Update.
23424 (ipa_call_summary_t::remove): Update.
23425 (dump_inline_summary): Update.
23426 (estimate_function_body_sizes): Update.
23427 (compute_inline_parameters): Update.
23428 (estimate_node_size_and_time): Update.
23429 (inline_merge_summary): Update.
23430 (inline_update_overall_summary): Update.
23431 (inline_read_section): Update.
23432 (inline_write_summary): Update.
23433 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23434 add account_size_time and reset member functions.
23435 (ipa_call_summary): Add reset function.
23436 * ipa-predicate.h (predicate::operator &): Constify.
23438 2017-05-22 Richard Biener <rguenther@suse.de>
23440 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23442 2017-05-19 Jason Merrill <jason@redhat.com>
23444 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23446 2017-05-19 Marek Polacek <polacek@redhat.com>
23449 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23450 TYPE_OVERFLOW_WRAPS checks.
23452 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23454 * tree-core.h (enum omp_clause_default_kind): Add
23455 "OMP_CLAUSE_DEFAULT_PRESENT".
23456 * tree-pretty-print.c (dump_omp_clause): Handle it.
23457 * gimplify.c (enum gimplify_omp_var_data): Add
23458 "GOVD_MAP_FORCE_PRESENT".
23459 (gimplify_adjust_omp_clauses_1): Map it to
23460 "GOMP_MAP_FORCE_PRESENT".
23461 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23463 * gimplify.c (oacc_default_clause): Clarify.
23465 2017-05-19 Nathan Sidwell <nathan@acm.org>
23467 LANG_HOOK_REGISTER_DUMPS
23468 * toplev.c (general_init): Call register dump lang hook.
23469 * doc/invoke.texi: Document -fdump-lang option family.
23470 * dumpfile.c (dump_files): Remove class dump here.
23471 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23472 * dumpfile.h (tree_dump_index): Remove TDI_class.
23473 * langhooks-def.h (lhd_register_dumps): Declare.
23474 (LANG_HOOKS_REGISTER_DUMPS): Define.
23475 (LANG_HOOKS_INITIALIZER): Add it.
23476 * langhooks.c (lhd_register_dumps): Define.
23477 * langhooks.h (struct lang_hooks): Add register_dumps.
23479 2017-05-19 Nathan Sidwell <nathan@acm.org>
23481 * context.h (context::set_passes): New.
23482 * context.c (context::context): Do not create pass manager.
23483 * toplev.c (general_init): Create pass manager here.
23485 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23487 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23488 use this splitter if two add or or instructions would also work for
23489 the constant we want to generate.
23491 2017-05-19 Richard Biener <rguenther@suse.de>
23494 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23495 predicate evaluation.
23497 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23499 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23501 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23504 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23506 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23507 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23508 (inline_edge_summary_vec): Turn into ...
23509 (ipa_call_summaries): ... this one.
23510 (redirect_to_unreachable, edge_set_predicate,
23511 evaluate_properties_for_edge, inline_summary_alloc,
23512 reset_ipa_call_summary, reset_inline_summary,
23513 inline_summary_t::duplicate): Update.
23514 (inline_edge_duplication_hook): Turn to ...
23515 (ipa_call_summary_t::duplicate): ... this one.
23516 (inline_edge_removal_hook): Turn to ...
23517 (ipa_call_summary_t::remove): ... this one.
23518 (dump_inline_edge_summary): Turn to ...
23519 (dump_ipa_call_summary): ... this one.
23520 (estimate_function_body_sizes): Update.
23521 (inline_update_callee_summaries): Update.
23522 (remap_edge_change_prob): Update.
23523 (remap_edge_summaries): Update.
23524 (inline_merge_summary): Update.
23525 (do_estimate_edge_time): Update.
23526 (inline_generate_summary): Update.
23527 (inline_read_section): Update.
23528 (inline_read_summary): Update.
23529 (inline_free_summary): Update.
23530 * ipa-inline.c (can_inline_edge_p): Update.
23531 (compute_inlined_call_time): Update.
23532 (want_inline_small_function_p): Update.
23533 (edge_badness): Update.
23534 (early_inliner): Update.
23535 * ipa-inline.h (inline_edge_summary): Turn to ...
23536 (ipa_call_summary): ... this one.
23537 (ipa_call_summary_t): New class.
23538 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23539 (ipa_call_summaries): New.
23540 (inline_edge_summary): Remove.
23541 (estimate_edge_growth): Update.
23542 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23543 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23544 * ipa-split.c (execute_split_functions): Update.
23545 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23547 2017-05-19 Richard Biener <rguenther@suse.de>
23549 PR middle-end/80764
23550 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23552 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23554 * config/rs6000/rs6000.c (struct machine_function): Add field
23555 fpr_is_wrapped_separately.
23556 (rs6000_get_separate_components): Use 64 components. Handle the
23557 new FPR components.
23558 (rs6000_components_for_bb): Handle the FPR components.
23559 (rs6000_emit_prologue_components): Handle the FPR components.
23560 (rs6000_emit_epilogue_components): Handle the FPR components.
23561 (rs6000_set_handled_components): Handle the FPR components.
23562 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23563 that are already separately shrink-wrapped.
23564 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23565 that are already separately shrink-wrapped.
23567 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23570 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23573 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23574 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23575 cases where the register allocator uses a move and an offsettable
23576 memory operation to/from a FPR register on ISA 2.06/2.07.
23577 (define_peephole2 for Altivec d-form store): Likewise.
23579 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23582 * config/i386/mmx.md (*mov<mode>_internal): Enable
23583 alternatives 11, 12, 13 and 14 also for 32bit targets.
23584 Remove alternatives 15, 16, 17 and 18.
23585 * config/i386/sse.md (vec_concatv2di): Change
23586 alternative (!x, *y) to (x, ?!*Yn).
23588 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23590 * dumpfile.h (enum dump_kind): Remove stray comma.
23592 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23594 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23595 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23596 predicate::num_conditions
23597 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23598 (CHANGED): turn into predicate::changed.
23599 (agg_position_info): Move to ipa-predicate.h
23600 (add_condition, predicate::add_clause, predicate::operator &=,
23601 predicate::or_with, predicate::evaluate, predicate::probability,
23602 dump_condition, dump_clause, predicate::dump,
23603 predicate::remap_after_duplication, predicate::remap_after_inlining,
23604 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23605 (evaluate_conditions_for_known_args): Update.
23606 (set_cond_stmt_execution_predicate): Update.
23607 * ipa-inline.h: Include ipa-predicate.h
23608 (condition, inline_param_summary, conditions, agg_position_info,
23609 predicate): Move to ipa-predicate.h
23610 * ipa-predicate.c: New file.
23611 * ipa-predicate.h: New file.
23613 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23615 * final.c (leaf_function_p): Check we are not in a sequence.
23617 2017-05-18 Martin Liska <mliska@suse.cz>
23619 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23620 * dumpfile.c (dump_register): Use new enum dump_kind.
23621 (get_dump_file_name): Likewise.
23622 (dump_enable_all): Likewise.
23623 (dump_switch_p_1): Likewise.
23624 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23625 * dumpfile.h (enum dump_kind): New enum type.
23626 (struct dump_file_info): Create constructor and
23627 format fields and comments.
23628 * passes.c (pass_manager::register_one_dump_file):
23630 * statistics.c (statistics_early_init): Likewise.
23631 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23632 TDF_TREE with TDF_SLIM.
23633 (gather_memory_references_ref): Likewise.
23635 2017-05-18 Martin Liska <mliska@suse.cz>
23637 * vec.h (struct vnull): Use it.
23639 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23641 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23642 (true_predicate, false_predicate, true_predicate_p,
23643 false_predicate_p): Remove.
23644 (single_cond_predicate, not_inlined_predicate): Turn to member function
23646 (add_condition): Update.
23647 (add_clause): Turn to...
23648 (predicate::add_clause): ... this one; update; allow passing NULL
23650 (and_predicates): Turn to ...
23651 (predicate::operator &=): ... this one.
23652 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23653 (or_predicates): Turn to ...
23654 (predicate::or_with): ... this one.
23655 (evaluate_predicate): Turn to ...
23656 (predicate::evaluate): ... this one.
23657 (predicate_probability): Turn to ...
23658 (predicate::probability): ... this one.
23659 (dump_condition): Update.
23660 (dump_predicate): Turn to ...
23661 (predicate::dump): ... this one.
23662 (account_size_time): Update.
23663 (edge_set_predicate): Update.
23664 (set_hint_predicate): UPdate.
23665 (evaluate_conditions_for_known_args): Update.
23666 (evaluate_properties_for_edge): Update.
23667 (remap_predicate_after_duplication): Turn to...
23668 (predicate::remap_after_duplication): ... this one.
23669 (remap_hint_predicate_after_duplication): Update.
23670 (inline_summary_t::duplicate): UPdate.
23671 (dump_inline_edge_summary): Update.
23672 (dump_inline_summary): Update.
23673 (set_cond_stmt_execution_predicate): Update.
23674 (set_switch_stmt_execution_predicate): Update.
23675 (compute_bb_predicates): Update.
23676 (will_be_nonconstant_expr_predicate): Update.
23677 (will_be_nonconstant_predicate): Update.
23678 (phi_result_unknown_predicate): Update.
23679 (predicate_for_phi_result): Update.
23680 (array_index_predicate): Update.
23681 (estimate_function_body_sizes): Update.
23682 (estimate_node_size_and_time): Update.
23683 (estimate_ipcp_clone_size_and_time): Update.
23684 (remap_predicate): Rename to ...
23685 (predicate::remap_after_inlining): ... this one.
23686 (remap_hint_predicate): Update.
23687 (inline_merge_summary): Update.
23688 (inline_update_overall_summary): Update.
23689 (estimate_size_after_inlining): Update.
23690 (read_predicate): Rename to ...
23691 (predicate::stream_in): ... this one.
23692 (read_inline_edge_summary): Update.
23693 (write_predicate): Rename to ...
23694 (predicate::stream_out): ... this one.
23695 (write_inline_edge_summary): Update.
23696 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23697 (clause_t): Turn to uint32_t
23698 (predicate): Turn to class; implement constructor and operators
23700 (size_time_entry): Update.
23701 (inline_summary): Update.
23702 (inline_edge_summary): Update.
23704 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23706 * fold-const.c (fold_binary_loc): Move transformation...
23707 * match.pd (C - X CMP X): ... here.
23709 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23711 * config/sparc/sparc.c (sparc_option_override): Set function
23712 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23713 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23715 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23717 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23719 2017-05-18 Marek Polacek <polacek@redhat.com>
23722 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23723 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23724 gimple_assign_load_p.
23726 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23728 PR middle-end/80692
23729 * real.c (do_compare): Give decimal_do_compare preference over
23730 comparing just the signs.
23732 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23734 * doc/md.texi (Canonicalization of Instructions): Describe the
23735 canonical form of instructions that inherently set a condition
23738 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23740 PR middle-end/80775
23741 * tree-cfg.c: Move deletion of unreachable case statements to after
23742 the merging of consecutive case labels.
23744 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23746 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23747 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23748 restoring of callee-saved registers.
23750 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23752 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23753 * config/visium/visium.c (single_set_and_flags): Likewise.
23754 * config/visium/visium.md (Substitutions): Likewise.
23756 2017-05-17 Martin Liska <mliska@suse.cz>
23758 * cfg.c: Introduce dump_flags_t type and
23759 use it instead of int type.
23761 * cfghooks.c: Likewise.
23762 * cfghooks.h (struct cfg_hooks): Likewise.
23763 * cfgrtl.c: Likewise.
23764 * cfgrtl.h: Likewise.
23765 * cgraph.c (cgraph_node::get_body): Likewise.
23766 * coretypes.h: Likewise.
23767 * domwalk.c: Likewise.
23768 * domwalk.h: Likewise.
23769 * dumpfile.c (struct dump_option_value_info): Likewise.
23770 (dump_enable_all): Likewise.
23771 (dump_switch_p_1): Likewise.
23772 (opt_info_switch_p): Likewise.
23773 * dumpfile.h (enum tree_dump_index): Likewise.
23774 (struct dump_file_info): Likewise.
23775 * genemit.c: Likewise.
23776 * generic-match-head.c: Likewise.
23777 * gengtype.c (open_base_files): Likewise.
23778 * gimple-pretty-print.c: Likewise.
23779 * gimple-pretty-print.h: Likewise.
23780 * graph.c (print_graph_cfg): Likewise.
23781 * graphite-scop-detection.c (dot_all_sese): Likewise.
23782 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23783 * loop-unroll.c (report_unroll): Likewise.
23784 * passes.c (pass_manager::register_one_dump_file): Likewise.
23785 * print-tree.c: Likewise.
23786 * statistics.c: Likewise.
23787 * tree-cfg.c: Likewise.
23788 * tree-cfg.h: Likewise.
23789 * tree-dfa.c: Likewise.
23790 * tree-dfa.h: Likewise.
23791 * tree-dump.c (dump_function): Likewise.
23792 * tree-dump.h (struct dump_info): Likewise.
23793 * tree-pretty-print.c: Likewise.
23794 * tree-pretty-print.h: Likewise.
23795 * tree-ssa-live.c: Likewise.
23796 * tree-ssa-live.h: Likewise.
23797 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23798 * tree-vect-loop.c: Likewise.
23799 * tree-vect-slp.c: Likewise.
23801 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23802 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23804 PR tree-optimization/80457
23805 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23806 of all arguments to a statement as scalar_to_vec operations.
23807 (vectorizable_call): Adjust call to vect_model_simple_cost for
23809 (vectorizable_conversion): Likewise.
23810 (vectorizable_assignment): Likewise.
23811 (vectorizable_shift): Likewise.
23812 (vectorizable_operation): Likewise.
23813 (vectorizable_comparison): Likewise.
23814 (vect_is_simple_cond): Record the def types for operands.
23815 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23816 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23817 for statement argument count.
23819 2017-05-16 Carl Love <cel@us.ibm.com>
23821 * config/rs6000/rs6000-c: Add support for built-in functions
23822 vector unsigned long long vec_bperm (vector unsigned long long,
23823 vector unsigned char)
23824 vector signed long long vec_mule (vector signed int,
23826 vector unsigned long long vec_mule (vector unsigned int,
23827 vector unsigned int)
23828 vector signed long long vec_mulo (vector signed int,
23830 vector unsigned long long vec_mulo (vector unsigned int,
23831 vector unsigned int)
23832 vector signed char vec_sldw (vector signed char,
23833 vector signed char,
23835 vector unsigned char vec_sldw (vector unsigned char,
23836 vector unsigned char,
23838 vector signed short vec_sldw (vector signed short,
23839 vector signed short,
23841 vector unsigned short vec_sldw (vector unsigned short,
23842 vector unsigned short,
23844 vector signed int vec_sldw (vector signed int,
23847 vector unsigned int vec_sldw (vector unsigned int,
23848 vector unsigned int,
23850 vector signed long long vec_sldw (vector signed long long,
23851 vector signed long long,
23853 vector unsigned long long vec_sldw (vector unsigned long long,
23854 vector unsigned long long,
23856 * config/rs6000/rs6000-c: Add support for built-in functions
23857 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23858 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23859 * doc/extend.texi: Update the built-in documentation for the
23860 new built-in functions.
23862 2017-05-16 Marek Polacek <polacek@redhat.com>
23866 * tree.c (save_expr): Don't fold the expression.
23868 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23870 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23871 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23872 and (?*y,m). Update insn attributes.
23874 2017-05-16 Martin Liska <mliska@suse.cz>
23876 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23877 flags argument of print_gimple_stmt, print_gimple_expr,
23878 print_generic_stmt and print_generic_expr.
23879 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23880 * coretypes.h: Likewise.
23881 * except.c (dump_eh_tree): Likewise.
23882 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23883 * gimple-pretty-print.h: Likewise.
23884 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23885 (backprop::push_to_worklist): Likewise.
23886 (backprop::pop_from_worklist): Likewise.
23887 (backprop::process_use): Likewise.
23888 (backprop::intersect_uses): Likewise.
23889 (note_replacement): Likewise.
23890 * gimple-ssa-store-merging.c
23891 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23892 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23893 (pass_store_merging::execute): Likewise.
23894 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23895 (ssa_base_cand_dump_callback): Likewise.
23896 (dump_incr_vec): Likewise.
23897 (replace_refs): Likewise.
23898 (replace_mult_candidate): Likewise.
23899 (create_add_on_incoming_edge): Likewise.
23900 (create_phi_basis): Likewise.
23901 (insert_initializers): Likewise.
23902 (all_phi_incrs_profitable): Likewise.
23903 (introduce_cast_before_cand): Likewise.
23904 (replace_one_candidate): Likewise.
23905 * gimplify.c (gimplify_expr): Likewise.
23906 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23907 (set_rename): Likewise.
23908 (rename_uses): Likewise.
23909 (copy_loop_phi_nodes): Likewise.
23910 (add_close_phis_to_merge_points): Likewise.
23911 (copy_loop_close_phi_args): Likewise.
23912 (copy_cond_phi_args): Likewise.
23913 (graphite_copy_stmts_from_block): Likewise.
23914 (translate_pending_phi_nodes): Likewise.
23915 * graphite-poly.c (print_pdr): Likewise.
23916 (dump_gbb_cases): Likewise.
23917 (dump_gbb_conditions): Likewise.
23918 (print_scop_params): Likewise.
23919 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23920 (build_cross_bb_scalars_use): Likewise.
23921 (gather_bbs::before_dom_children): Likewise.
23922 * hsa-dump.c (dump_hsa_immed): Likewise.
23923 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23924 (get_replacement_map): Likewise.
23925 * ipa-inline-analysis.c (dump_condition): Likewise.
23926 (estimate_function_body_sizes): Likewise.
23927 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23928 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23929 * ipa-prop.c (ipa_dump_param): Likewise.
23930 (ipa_print_node_jump_functions_for_edge): Likewise.
23931 (ipa_modify_call_arguments): Likewise.
23932 (ipa_modify_expr): Likewise.
23933 (ipa_dump_param_adjustments): Likewise.
23934 (ipa_dump_agg_replacement_values): Likewise.
23935 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23936 * ipa-pure-const.c (check_stmt): Likewise.
23937 (pass_nothrow::execute): Likewise.
23938 * ipa-split.c (execute_split_functions): Likewise.
23939 * omp-offload.c (dump_oacc_loop_part): Likewise.
23940 (dump_oacc_loop): Likewise.
23941 * trans-mem.c (tm_log_emit): Likewise.
23942 (tm_memopt_accumulate_memops): Likewise.
23943 (dump_tm_memopt_set): Likewise.
23944 (dump_tm_memopt_transform): Likewise.
23945 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23946 (print_loop): Likewise.
23947 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23948 (chkp_gather_checks_info): Likewise.
23949 (chkp_get_check_result): Likewise.
23950 (chkp_remove_check_if_pass): Likewise.
23951 (chkp_use_outer_bounds_if_possible): Likewise.
23952 (chkp_reduce_bounds_lifetime): Likewise.
23953 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23954 (chkp_mark_completed_bounds): Likewise.
23955 (chkp_register_incomplete_bounds): Likewise.
23956 (chkp_mark_invalid_bounds): Likewise.
23957 (chkp_maybe_copy_and_register_bounds): Likewise.
23958 (chkp_build_returned_bound): Likewise.
23959 (chkp_get_bound_for_parm): Likewise.
23960 (chkp_build_bndldx): Likewise.
23961 (chkp_get_bounds_by_definition): Likewise.
23962 (chkp_generate_extern_var_bounds): Likewise.
23963 (chkp_get_bounds_for_decl_addr): Likewise.
23964 * tree-chrec.c (chrec_apply): Likewise.
23965 * tree-data-ref.c (dump_data_reference): Likewise.
23966 (dump_subscript): Likewise.
23967 (dump_data_dependence_relation): Likewise.
23968 (analyze_overlapping_iterations): Likewise.
23969 * tree-inline.c (expand_call_inline): Likewise.
23970 (tree_function_versioning): Likewise.
23971 * tree-into-ssa.c (dump_defs_stack): Likewise.
23972 (dump_currdefs): Likewise.
23973 (dump_names_replaced_by): Likewise.
23974 (dump_update_ssa): Likewise.
23975 (update_ssa): Likewise.
23976 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23977 * tree-parloops.c (build_new_reduction): Likewise.
23978 (try_create_reduction_list): Likewise.
23979 (ref_conflicts_with_region): Likewise.
23980 (oacc_entry_exit_ok_1): Likewise.
23981 (oacc_entry_exit_single_gang): Likewise.
23982 * tree-pretty-print.h: Likewise.
23983 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23984 (get_scalar_evolution): Likewise.
23985 (add_to_evolution): Likewise.
23986 (get_loop_exit_condition): Likewise.
23987 (analyze_evolution_in_loop): Likewise.
23988 (analyze_initial_condition): Likewise.
23989 (analyze_scalar_evolution): Likewise.
23990 (instantiate_scev): Likewise.
23991 (number_of_latch_executions): Likewise.
23992 (gather_chrec_stats): Likewise.
23993 (final_value_replacement_loop): Likewise.
23994 (scev_const_prop): Likewise.
23995 * tree-sra.c (dump_access): Likewise.
23996 (disqualify_candidate): Likewise.
23997 (create_access): Likewise.
23998 (reject): Likewise.
23999 (maybe_add_sra_candidate): Likewise.
24000 (create_access_replacement): Likewise.
24001 (analyze_access_subtree): Likewise.
24002 (analyze_all_variable_accesses): Likewise.
24003 (sra_modify_assign): Likewise.
24004 (initialize_constant_pool_replacements): Likewise.
24005 (find_param_candidates): Likewise.
24006 (decide_one_param_reduction): Likewise.
24007 (replace_removed_params_ssa_names): Likewise.
24008 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24009 * tree-ssa-copy.c (dump_copy_of): Likewise.
24010 (copy_prop_visit_cond_stmt): Likewise.
24011 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24012 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24013 (record_equivalences_from_stmt): Likewise.
24014 * tree-ssa-dse.c (compute_trims): Likewise.
24015 (delete_dead_call): Likewise.
24016 (delete_dead_assignment): Likewise.
24017 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24018 (forward_propagate_into_cond): Likewise.
24019 (pass_forwprop::execute): Likewise.
24020 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24021 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24023 (move_computations_worker): Likewise.
24024 (execute_sm): Likewise.
24025 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24026 (remove_exits_and_undefined_stmts): Likewise.
24027 (remove_redundant_iv_tests): Likewise.
24028 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24029 (adjust_iv_update_pos): Likewise.
24030 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24031 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24032 (value_replacement): Likewise.
24033 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24034 * tree-ssa-pre.c (print_pre_expr): Likewise.
24035 (get_representative_for): Likewise.
24036 (create_expression_by_pieces): Likewise.
24037 (insert_into_preds_of_block): Likewise.
24038 (eliminate_insert): Likewise.
24039 (eliminate_dom_walker::before_dom_children): Likewise.
24040 (eliminate): Likewise.
24041 (remove_dead_inserted_code): Likewise.
24042 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24043 * tree-ssa-reassoc.c (get_rank): Likewise.
24044 (eliminate_duplicate_pair): Likewise.
24045 (eliminate_plus_minus_pair): Likewise.
24046 (eliminate_not_pairs): Likewise.
24047 (undistribute_ops_list): Likewise.
24048 (eliminate_redundant_comparison): Likewise.
24049 (update_range_test): Likewise.
24050 (optimize_range_tests_var_bound): Likewise.
24051 (optimize_vec_cond_expr): Likewise.
24052 (rewrite_expr_tree): Likewise.
24053 (rewrite_expr_tree_parallel): Likewise.
24054 (linearize_expr): Likewise.
24055 (break_up_subtract): Likewise.
24056 (linearize_expr_tree): Likewise.
24057 (attempt_builtin_powi): Likewise.
24058 (attempt_builtin_copysign): Likewise.
24059 (transform_stmt_to_copy): Likewise.
24060 (transform_stmt_to_multiply): Likewise.
24061 (dump_ops_vector): Likewise.
24062 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24063 (print_scc): Likewise.
24064 (set_ssa_val_to): Likewise.
24065 (visit_reference_op_store): Likewise.
24066 (visit_use): Likewise.
24067 (sccvn_dom_walker::before_dom_children): Likewise.
24068 (run_scc_vn): Likewise.
24069 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24071 (expr_hash_elt::print): Likewise.
24072 (const_and_copies::pop_to_marker): Likewise.
24073 (const_and_copies::record_const_or_copy_raw): Likewise.
24074 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24075 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24076 (dump_predicates): Likewise.
24077 (find_uninit_use): Likewise.
24078 (warn_uninitialized_phi): Likewise.
24079 (pass_late_warn_uninitialized::execute): Likewise.
24080 * tree-ssa.c (verify_vssa): Likewise.
24081 (verify_ssa): Likewise.
24082 (maybe_optimize_var): Likewise.
24083 * tree-vrp.c (dump_value_range): Likewise.
24084 (dump_all_value_ranges): Likewise.
24085 (dump_asserts_for): Likewise.
24086 (register_edge_assert_for_2): Likewise.
24087 (vrp_visit_cond_stmt): Likewise.
24088 (vrp_visit_switch_stmt): Likewise.
24089 (vrp_visit_stmt): Likewise.
24090 (vrp_visit_phi_node): Likewise.
24091 (simplify_cond_using_ranges_1): Likewise.
24092 (fold_predicate_in): Likewise.
24093 (evrp_dom_walker::before_dom_children): Likewise.
24094 (evrp_dom_walker::push_value_range): Likewise.
24095 (evrp_dom_walker::pop_value_range): Likewise.
24096 (execute_early_vrp): Likewise.
24098 2017-05-16 Richard Biener <rguenther@suse.de>
24100 * dwarf2out.c (loc_list_from_tree_1): Do not create
24101 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24103 2017-05-16 Richard Biener <rguenther@suse.de>
24105 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24107 (note_variable_value_in_expr): If we resolved the decl ref
24108 do not push to the stack.
24110 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24112 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24113 operations in fast-math mode.
24114 (vaddq_f16): Likewise.
24115 (vmul_f16): Likewise.
24116 (vmulq_f16): Likewise.
24117 (vsub_f16): Likewise.
24118 (vsubq_f16): Likewise.
24119 * config/arm/neon.md (add<mode>3): New.
24121 (fma:<VH:mode>3): New. Also remove outdated comment.
24124 2017-05-16 Martin Liska <mliska@suse.cz>
24128 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24129 (odr_types_equivalent_p): Likewise.
24131 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24133 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24135 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24138 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24139 non-interunit SSE move alternatives with '?'.
24140 (zero-extendsidi peephole2): New peephole to skip intermediate
24141 general register in SSE zero-extend sequence.
24143 2017-05-15 Jeff Law <law@redhat.com>
24145 * reorg.c (relax_delay_slots): Create a new variable to hold
24146 the temporary target rather than clobbering TARGET_LABEL.
24148 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24149 missing argument to extract_bit_field call.
24150 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24152 2017-05-15 Martin Liska <mliska@suse.cz>
24155 * gcc.c (process_command): Do not allow empty argument of -o option.
24157 2017-05-15 Renlin Li <renlin.li@arm.com>
24159 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24160 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24161 * config/aarch64/constraints.md (Usf): Add long call check.
24162 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24163 (call_value): Likewise.
24164 (sibcall): Likewise.
24165 (sibcall_value): Likewise.
24167 (call_value_insn): New.
24168 (sibcall_insn): Update rtx pattern.
24169 (sibcall_value_insn): Likewise.
24170 (call_internal): Remove.
24171 (call_value_internal): Likewise.
24172 (sibcall_internal): Likewise.
24173 (sibcall_value_internal): Likewise.
24174 (call_reg): Likewise.
24175 (call_symbol): Likewise.
24176 (call_value_reg): Likewise.
24177 (call_value_symbol): Likewise.
24179 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24182 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24184 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24186 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24187 compatible with CCGOCmode and with CCZmode.
24189 2017-05-14 Martin Sebor <msebor@redhat.com>
24191 PR middle-end/77671
24192 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24193 (gimple_fold_builtin_snprintf): Same.
24194 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24195 (gimple_fold_builtin_snprintf): Same.
24196 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24197 of character types.
24198 (is_call_safe): New function.
24199 (try_substitute_return_value): Call it.
24200 (try_simplify_call): New function.
24201 (pass_sprintf_length::handle_gimple_call): Call it.
24203 2017-05-14 Martin Sebor <msebor@redhat.com>
24205 PR middle-end/80669
24206 * builtins.c (expand_builtin_stpncpy): Simplify.
24208 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24210 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24211 * config/i386/i386.h
24212 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24213 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24214 (struct machine_function): Add new members call_ms2sysv,
24215 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24216 (struct machine_frame_state): New fields sp_realigned and
24217 sp_realigned_offset.
24218 * config/i386/i386.c
24219 (enum xlogue_stub): New enum.
24220 (enum xlogue_stub_sets): New enum.
24221 (class xlogue_layout): New class.
24222 (struct ix86_frame): New fields stack_realign_allocate_offset,
24223 stack_realign_offset and outlined_save_offset. Modify comments to
24224 detail stack layout when using out-of-line stubs.
24225 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24226 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24227 -mcall-ms2sysv-xlogues.
24228 (stub_managed_regs): New static variable.
24229 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24230 registers managed by out-of-line stub.
24231 (disable_call_ms2sysv_xlogues): New function.
24232 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24233 m->call_ms2sysv when appropriate and compute frame layout for
24235 (sp_valid_at, fp_valid_at): New inline functions.
24236 (choose_basereg): New function.
24237 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24239 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24240 Use align parameter of choose_baseaddr to generated aligned SSE movs
24242 (pro_epilogue_adjust_stack): Modify to track
24243 machine_frame_state::sp_realigned.
24244 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24245 (ix86_nsaved_sseregs): Likewise.
24246 (ix86_emit_save_regs): Likewise.
24247 (ix86_emit_save_regs_using_mov): Likewise.
24248 (ix86_emit_save_sse_regs_using_mov): Likewise.
24249 (get_scratch_register_on_entry): Likewise.
24250 (gen_frame_set): New function.
24251 (gen_frame_load): Likewise.
24252 (gen_frame_store): Likewise.
24253 (emit_outlined_ms2sysv_save): Likewise.
24254 (emit_outlined_ms2sysv_restore): Likewise.
24255 (ix86_expand_prologue): Modify stack re-alignment code and call
24256 emit_outlined_ms2sysv_save when appropriate.
24257 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24258 parameter rtx_insn *insn, which allows the function to be used to only
24259 generate the notes.
24260 (ix86_expand_epilogue): Modify validity checks of frame and stack
24261 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24262 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24263 * config/i386/predicates.md
24264 (save_multiple): New predicate.
24265 (restore_multiple): Likewise.
24266 * config/i386/sse.md
24267 (save_multiple<mode>): New pattern.
24268 (save_multiple_realign<mode>): Likewise.
24269 (restore_multiple<mode>): Likewise.
24270 (restore_multiple_and_return<mode>): Likewise.
24271 (restore_multiple_leave_return<mode>): Likewise.
24272 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24274 2017-05-14 Julia Koval <julia.koval@intel.com>
24276 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24277 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24278 (__builtin_ia32_xsetbv): New builtins.
24279 * config/i386/i386.c (ix86_expand_special_args_builtin):
24281 (ix86_expand_builtin): Special expand for new intrinsics.
24282 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24283 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24284 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24286 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24288 * cfganal.c (inverted_post_order_compute): Change argument type
24290 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24291 * df-core.c (rest_of_handle_df_initialize): Adjust.
24292 (rest_of_handle_df_finish): Likewise.
24293 (df_analyze_1): Likewise.
24294 (df_analyze): Likewise.
24295 (loop_inverted_post_order_compute): Change argument to be a vec *.
24296 (df_analyze_loop): Adjust.
24297 (df_get_n_blocks): Likewise.
24298 (df_get_postorder): Likewise.
24299 * df.h (struct df_d): Change field to be a vec.
24300 * lcm.c (compute_laterin): Adjust.
24301 (compute_available): Likewise.
24302 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24303 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24304 * tree-ssa-pre.c (compute_antic): Likewise.
24306 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24308 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24309 (depth_first_search::depth_first_search): Change structure init
24310 function to this constructor.
24311 (depth_first_search::add_bb): Rename function to this member.
24312 (depth_first_search::execute): Likewise.
24313 (flow_dfs_compute_reverse_finish): Adjust.
24315 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24317 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24318 (longest_simple_path): Likewise.
24319 * shrink-wrap.c (spread_components): Likewise.
24320 (disqualify_problematic_components): Likewise.
24321 (emit_common_heads_for_components): Likewise.
24322 (emit_common_tails_for_components): Likewise.
24323 (insert_prologue_epilogue_for_components): Likewise.
24325 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24327 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24330 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24332 * df-core.c (df_set_blocks): Start using auto_bitmap.
24333 (df_compact_blocks): Likewise.
24334 * df-problems.c (df_rd_confluence_n): Likewise.
24335 * df-scan.c (df_insn_rescan_all): Likewise.
24336 (df_process_deferred_rescans): Likewise.
24337 (df_update_entry_block_defs): Likewise.
24338 (df_update_exit_block_uses): Likewise.
24339 (df_entry_block_bitmap_verify): Likewise.
24340 (df_exit_block_bitmap_verify): Likewise.
24341 (df_scan_verify): Likewise.
24342 * lra-constraints.c (lra_constraints): Likewise.
24343 (undo_optional_reloads): Likewise.
24344 (lra_undo_inheritance): Likewise.
24345 * lra-remat.c (calculate_gen_cands): Likewise.
24346 (do_remat): Likewise.
24347 * lra-spills.c (assign_spill_hard_regs): Likewise.
24348 (spill_pseudos): Likewise.
24349 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24350 (bitmap_set_subtract_values): Likewise.
24352 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24354 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24355 management with auto_bitmap.
24356 (fix_inter_tick): Likewise.
24357 (fix_recovery_deps): Likewise.
24358 * ira.c (add_store_equivs): Likewise.
24359 (find_moveable_pseudos): Likewise.
24360 (split_live_ranges_for_shrink_wrap): Likewise.
24361 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24362 (rtx_reuse_manager::seen_def_p): Likewise.
24363 (rtx_reuse_manager::set_seen_def): Likewise.
24364 * print-rtl.h (class rtx_reuse_manager): Likewise.
24366 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24368 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24370 (migrate_btr_def): Likewise.
24371 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24372 * df-core.c (loop_post_order_compute): Likewise.
24373 (loop_inverted_post_order_compute): Likewise.
24374 * hsa-common.h: Likewise.
24375 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24376 * init-regs.c (initialize_uninitialized_regs): Likewise.
24377 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24378 (inline_small_functions): Likewise.
24379 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24380 * ira.c (combine_and_move_insns): Likewise.
24381 (build_insn_chain): Likewise.
24382 * loop-invariant.c (find_invariants): Likewise.
24383 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24384 * predict.c (tree_predict_by_opcode): Likewise.
24385 (predict_paths_leading_to): Likewise.
24386 (predict_paths_leading_to_edge): Likewise.
24387 (estimate_loops_at_level): Likewise.
24388 (estimate_loops): Likewise.
24389 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24390 (spread_components): Likewise.
24391 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24392 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24393 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24394 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24395 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24396 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24397 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24398 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24399 (mark_threaded_blocks): Likewise.
24400 (thread_through_all_blocks): Likewise.
24401 * tree-ssa.c (verify_ssa): Likewise.
24402 (execute_update_addresses_taken): Likewise.
24403 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24405 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24407 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24409 (post_order_compute): Likewise.
24410 (inverted_post_order_compute): Likewise.
24411 (pre_and_rev_post_order_compute_fn): Likewise.
24413 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24415 * genrecog.c (int_set::int_set): Explicitly construct our
24416 auto_vec base class.
24417 * vec.h (auto_vec::auto_vec): New constructor.
24419 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24421 * bitmap.h (class auto_bitmap): New constructor taking
24422 bitmap_obstack * argument.
24424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24426 * bitmap.h (class auto_bitmap): Change type of m_bits to
24427 bitmap_head, and adjust ctor / dtor and member operators.
24429 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24431 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24432 when returned register mode doesn't match original mode.
24434 2017-05-12 Jeff Law <law@redhat.com>
24435 Jakub Jelinek <jakub@redhat.com>
24437 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24438 we look for cc setter after the compare-elim changes.
24439 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24440 within the vector to match what compare-elim now expects.
24441 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24442 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24444 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24445 after the compare-elim changes.
24446 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24447 the vector to match what compare-elim now expects.
24448 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24449 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24450 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24451 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24452 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24454 * config/visium/visium.c (single_set_and_flags): Fix where
24455 we look for cc setter after the compare-elim changes.
24456 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24457 with the vector to match what compare-elim now expects.
24458 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24459 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24460 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24461 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24462 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24463 (neg<mode>2_insn_set_overflow): Likewise.
24465 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24467 PR middle-end/79794
24468 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24469 maybe_expand_insn call, set ops[0].target. If still set after call,
24470 set alt_rtl. Add extra arg to recursive calls.
24471 (extract_bit_field): Add alt_rtl argument. Pass to
24473 * expmed.h (extract_bit_field): Fix prototype.
24474 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24475 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24476 to extract_bit_field_calls.
24477 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24478 Pass alt_rtl to extract_bit_field calls.
24479 * calls.c (store_unaligned_arguments_into_psuedos)
24480 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24481 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24483 * optabs.h (struct expand_operand): Add target bitfield.
24485 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24487 * compare-elim.c (try_eliminate_compare): Canonicalize
24488 operation with embedded compare to
24489 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24490 (set (reg) (operation)].
24492 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24494 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24497 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24498 cost of adding a carry flag for ADC instruction.
24499 [case MINUS]: Ignore the cost of subtracting a carry flag
24500 for SBB instruction.
24502 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24504 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24506 * config/rs6000/bmiintrin.h: New file.
24507 * config/rs6000/bmi2intrin.h: New file.
24508 * config/rs6000/x86intrin.h: New file.
24510 2017-05-12 Jeff Law <law@redhat.com>
24512 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24515 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24517 PR middle-end/80707
24518 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24520 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24523 early expansion of vector divide builtins.
24524 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24525 builtins identified as having unsigned arguments.
24527 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24529 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24530 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24531 expansion of vector logical operations (and, andc, or, xor,
24534 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24536 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24537 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24539 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24541 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24542 early GIMPLE expansion of vector multiplies.
24544 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24546 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24547 TARGET_HAVE_MOVT conditional.
24548 (movt splitter): Likewise.
24550 2017-05-12 Richard Biener <rguenther@suse.de>
24552 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24553 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24554 Fold all stmts not inplace.
24556 2017-05-12 Richard Biener <rguenther@suse.de>
24558 PR tree-optimization/80713
24559 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24560 inserted_exprs bit for not removed stmts.
24562 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24564 PR middle-end/69921
24565 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24566 parallelized" attribute for parallelized OpenACC kernels.
24567 * omp-offload.c (execute_oacc_device_lower): Use it.
24569 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24570 Set "oacc kernels" attribute.
24571 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24572 parameter. Adjust all users.
24573 (oacc_fn_attrib_kernels_p): Remove function.
24574 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24575 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24576 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24577 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24578 assert "oacc kernels" attribute is set.
24580 2017-05-11 Carl Love <cel@us.ibm.com>
24582 * config/rs6000/rs6000-c: Add support for built-in functions
24583 vector unsigned char vec_popcnt (vector signed char)
24584 vector unsigned char vec_popcnt (vector unsigned char)
24585 vector unsigned short vec_popcnt (vector signed short)
24586 vector unsigned short vec_popcnt (vector unsigned short)
24587 vector unsigned int vec_popcnt (vector signed int)
24588 vector unsigned int vec_popcnt (vector unsigned int)
24589 vector unsigned long long vec_popcnt (vector signed long long)
24590 vector unsigned long long vec_popcnt (vector unsigned long long)
24591 vector signed long long vec_slo (vector signed long long,
24592 vector signed char)
24593 vector signed long long vec_slo (vector signed long long,
24594 vector unsigned char)
24595 vector unsigned long long vec_slo (vector unsigned long long,
24596 vector signed char)
24597 vector unsigned long long vec_slo (vector unsigned long long,
24598 vector unsigned char)
24599 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24600 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24601 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24602 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24603 * doc/extend.texi: Update the built-in documentation file for the
24604 new built-in functions.
24606 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24608 * attribs.h (sorted_attr_string): Move machine independent
24609 functions for target clone support from the i386 port to common
24610 code. Rename ix86_function_versions to common_function_versions.
24611 Rename make_name to make_unique_name.
24612 (common_function_versions): Likewise.
24613 (make_unique_name): Likewise.
24614 (make_dispatcher_decl): Likewise.
24615 (is_function_default_version): Likewise.
24616 * attribs.c (attr_strcmp): Likewise.
24617 (sorted_attr_string): Likewise.
24618 (common_function_versions): Likewise.
24619 (make_unique_name): Likewise.
24620 (make_dispatcher_decl): Likewise.
24621 (is_function_default_version): Likewise.
24622 * config/i386/i386.c (attr_strcmp): Likewise.
24623 (sorted_attr_string): Likewise.
24624 (ix86_function_versions): Likewise.
24625 (make_name): Likewise.
24626 (make_dispatcher_decl): Likewise.
24627 (is_function_default_version): Likewise.
24628 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24630 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24633 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24634 Account for direct move costs for vec_construct of integer
24637 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24640 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24641 (UNSPEC_STX_ATOMIC): Ditto.
24642 (loaddi_via_sse): New insn.
24643 (storedi_via_sse): Ditto.
24644 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24645 Update corresponding peephole2 patterns.
24646 (atomic_storedi_fpu): Ditto.
24648 2017-05-11 Julia Koval <julia.koval@intel.com>
24650 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24651 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24653 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24654 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24655 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24657 2017-05-11 Nathan Sidwell <nathan@acm.org>
24659 * graphite-poly.c: Include dumpfile.h.
24661 * dumpfle.h (dump_function): Declare here ...
24662 * tree-dump.h (dump_function): ... not here.
24663 * dumpfile.c: #include tree-cfg.h.
24664 (dump_function): Move here from ...
24665 * tree-dump.c (dump_function): ... here.
24666 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24667 * graphite-poly.c: Don't include tree-dump.h.
24668 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24669 * print-tree.c: Likewise.
24670 * stor-layout.c: Likewise.
24671 * tree-nested.c: Likewise.
24673 * dumpfile.c (dump_start): Use TDF_FLAGS.
24674 (dump_enable_all): Fix TDF_KIND check thinko.
24676 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24678 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24679 array entries to represent two legal parameterizations of the
24680 overloaded __builtin_cmpb function, as represented by the
24681 P6_OV_BUILTIN_CMPB constant.
24682 (altivec_resolve_overloaded_builtin): Add special case handling
24683 for the __builtin_cmpb function, as represented by the
24684 P6_OV_BUILTIN_CMPB constant.
24685 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24686 (BU_P6_64BIT_2): New macro.
24687 (BU_P6_OVERLOAD_2): New macro
24688 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24689 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24690 (CMPB): Add overload support to represent both 32-bit and 64-bit
24691 compare-bytes function.
24692 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24693 support for TARGET_CMPB.
24694 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24695 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24696 documentation of the __builtin_cmpb overloaded built-in function.
24698 2017-05-11 Richard Biener <rguenther@suse.de>
24700 PR tree-optimization/80705
24701 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24702 bases are not vectorizable.
24704 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24706 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24707 when counting register pressure.
24709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24711 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24712 (iv_ca_more_deps): Renamed to ...
24713 (iv_ca_compare_deps): ... this.
24714 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24718 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24720 (determine_group_iv_costs): ... here.
24721 (find_inv_vars_cb): Record inv var if it's not recorded before.
24723 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24725 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24726 (get_shiftadd_cost): Ditto.
24728 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24730 * tree-ssa-address.c: Include header file.
24731 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24733 (add_to_parts): Refactor.
24734 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24735 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24740 PR tree-optimization/53090
24741 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24743 (extract_cond_operands): Detect condition with IV on both sides
24744 and return COMP_IV_EXPR_2.
24745 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24746 (rewrite_use_compare): Simplify by removing call to function
24747 extract_cond_operands.
24749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24751 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24752 (extract_cond_operands): Detect condition comparing against non-
24753 invariant bound and return appropriate enum value.
24754 (find_interesting_uses_cond): Update use of extract_cond_operands.
24755 Handle its return value accordingly.
24756 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24758 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24760 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24761 nonlinear iv_use computation in loop invariant sensitive way.
24763 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24765 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24766 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24768 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24770 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24771 (dump_cand): Support iv_cand.inv_exprs.
24772 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24774 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24777 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24779 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24781 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24782 as local function. Include necessary header files.
24783 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24787 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24791 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24792 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24793 RSHIFT_EXPR and BIT_NOT_EXPR.
24795 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24797 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24798 (adjust_setup_cost): New parameter supporting round up adjustment.
24799 (struct address_cost_data): Delete.
24800 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24801 (split_address_cost, ptr_difference_cost): Delete.
24802 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24803 (struct ainc_cost_data): New struct.
24804 (get_address_cost_ainc): New function.
24805 (get_address_cost, get_computation_cost): Reimplement.
24806 (determine_group_iv_cost_address): Record inv_expr for all uses of
24808 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24809 (iv_ca_has_deps): Reimplemented to ...
24810 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24812 (iv_ca_extend): Call iv_ca_more_deps.
24814 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24816 * tree-ssa-address.c (struct mem_address): Move to header file.
24817 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24818 * tree-ssa-address.h (struct mem_address): Move from C file.
24819 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24823 * tree-affine.h (aff_combination_type): New interface.
24824 (aff_combination_zero_p): Remove static.
24825 (aff_combination_const_p): New interface.
24826 (aff_combination_singleton_var_p): New interfaces.
24828 2017-05-11 Richard Biener <rguenther@suse.de>
24830 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24831 Skip unreachable blocks and destinations.
24832 (eliminate): Move stmt removal and fixup ...
24833 (fini_eliminate): ... here. Skip inserted exprs.
24834 (pass_pre::execute): Move fini_pre after fini_eliminate.
24835 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24836 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24837 PRE to get rid of dead code that has invalid SSA form and
24838 split critical edges again.
24840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24842 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24844 2017-05-11 Richard Biener <rguenther@suse.de>
24846 * passes.c (execute_function_todo): Verify loops if they are
24847 said to be up-to-date.
24848 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24849 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24851 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24854 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24855 handle calling assemble_external ourself.
24858 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24859 modes with zero size. Enhance comment.
24861 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24863 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24864 built-ins for vec_xl and vec_xst with short and char pointer
24867 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24869 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24870 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24871 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24872 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24873 (_mm_maskz_min_round_ss): New intrinsics.
24874 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24875 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24876 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24877 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24878 (__builtin_ia32_minss_mask_round): New builtins.
24879 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24880 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24881 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24883 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24884 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24886 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24889 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24891 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24892 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24893 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24894 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24895 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24896 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24897 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24898 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24899 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24900 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24901 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24902 (__builtin_ia32_mulss_mask_round): New builtins.
24903 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24904 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24905 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24907 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24908 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24910 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24913 2017-05-10 Julia Koval <julia.koval@intel.com>
24915 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24916 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24917 (_mm256_setr_m128i): New intrinsics.
24919 2017-05-10 Julia Koval <julia.koval@intel.com>
24921 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24922 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24923 (_mm_maskz_rcp14_ss): New intrinsics.
24924 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24925 (__builtin_ia32_rcp14ss_mask): New builtins.
24926 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24928 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24930 PR tree-optimization/51513
24931 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24932 (assert_unreachable_fallthru_edge_p): Use it.
24933 (group_case_labels_stmt): Likewise.
24934 * tree-cfg.h: Prototype it.
24935 * stmt.c: Include cfghooks.h and tree-cfg.h.
24936 (emit_case_dispatch_table) <gap_label>: New local variable.
24937 Use it to fill dispatch table gaps.
24938 Test for default_label before updating probabilities.
24939 (expand_case) <default_label>: Remove unneeded initialization.
24940 Test for unreachable default case statement and remove its edge.
24941 Set default_label accordingly.
24942 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24944 2017-05-10 Carl Love <cel@us.ibm.com>
24946 * config/rs6000/rs6000-c: Add support for built-in functions
24947 vector signed char vec_neg (vector signed char)
24948 vector signed short int vec_neg (vector short int)
24949 vector signed int vec_neg (vector signed int)
24950 vector signed long long vec_neg (vector signed long long)
24951 vector float vec_neg (vector float)
24952 vector double vec_neg (vector double)
24953 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24955 * config/rs6000/altivec.h: Add define for vec_neg
24956 * doc/extend.texi: Update the built-in documentation for the
24957 new built-in functions.
24959 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24961 PR tree-optimization/77644
24962 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24964 2017-05-10 Nathan Sidwell <nathan@acm.org>
24966 * dumpfile.h (TDI_lang_all): New.
24967 (TDF_KIND): New. Renumber others
24968 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24970 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24972 (get_dump_file_name): Adjust suffix generation.
24973 (dump_enable_all): Use TDF_KIND.
24974 * doc/invoke.texi (-fdump-lang-all): Document.
24976 * dumpfile.h: Tabify.
24978 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24981 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24982 Move member access before delete.
24984 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24986 * tree-inline.c (expand_call_inline): Split block at stmt
24989 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24992 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24993 are now unused after splitting mov{sf,sd}_hardfloat.
24994 (f32_lr2): Likewise.
24995 (f32_lm): Likewise.
24996 (f32_lm2): Likewise.
24997 (f32_li): Likewise.
24998 (f32_li2): Likewise.
24999 (f32_lv): Likewise.
25000 (f32_sr): Likewise.
25001 (f32_sr2): Likewise.
25002 (f32_sm): Likewise.
25003 (f32_sm2): Likewise.
25004 (f32_si): Likewise.
25005 (f32_si2): Likewise.
25006 (f32_sv): Likewise.
25007 (f32_dm): Likewise.
25008 (f32_vsx): Likewise.
25009 (f32_av): Likewise.
25010 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25011 For movsf, order stores so the VSX stores occur before the GPR
25012 store which encourages the register allocator to use a traditional
25013 FPR instead of a GPR. For movsd, order the stores so that the GPR
25014 store comes before the VSX stores to allow the power6 to work.
25015 This is due to the power6 not having a 32-bit integer store
25016 instruction from a FPR.
25017 (movsf_hardfloat): Likewise.
25018 (movsd_hardfloat): Likewise.
25020 2017-05-09 Martin Sebor <msebor@redhat.com>
25022 PR translation/80280
25023 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25026 PR translation/80280
25027 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25028 data member added in r247778.
25029 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25031 2017-05-09 Nathan Sidwell <nathan@acm.org>
25033 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25035 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25038 2017-05-09 Marek Polacek <polacek@redhat.com>
25040 * doc/invoke.texi: Fix typo.
25042 2017-05-09 Richard Biener <rguenther@suse.de>
25044 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25045 (vrp_val_is_min): Likewise.
25046 (set_value_range_to_value): Likewise.
25047 (set_value_range_to_nonnegative): Likewise.
25048 (gimple_assign_nonzero_p): Likewise.
25049 (gimple_stmt_nonzero_p): Likewise.
25050 (vrp_int_const_binop): Likewise. Remove unreachable case.
25051 (adjust_range_with_scev): Adjust comments.
25052 (compare_range_with_value): Likewise.
25053 (extract_range_from_phi_node): Likewise.
25054 (test_for_singularity): Likewise.
25056 2017-05-09 Richard Biener <rguenther@suse.de>
25058 * tree-vrp.c (get_single_symbol): Add assert that we don't
25059 get overflowed constants as invariant part.
25060 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25061 checks. Use wi::cmp instead of recursing for integer constants.
25062 (compare_values): Just ignore whether we assumed undefined
25063 overflow instead of failing the compare.
25064 (extract_range_for_var_from_comparison_expr): Add comment before the
25065 TREE_NO_WARNING sets.
25066 (test_for_singularity): Likewise.
25067 (extract_range_from_comparison): Do not disable optimization
25068 when we assumed undefined overflow.
25069 (extract_range_basic): Remove init of unused var.
25071 2017-05-09 Richard Biener <rguenther@suse.de>
25073 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25074 (extract_range_from_multiplicative_op_1): Adjust.
25075 (extract_range_from_binary_expr_1): Use int_const_binop.
25077 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25080 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25081 rs6000_store_data_bypass_p in seven define_bypass directives and
25082 in several comments.
25083 * config/rs6000/rs6000-protos.h: Add prototype for
25084 rs6000_store_data_bypass_p function.
25085 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25086 function implements slightly different (rs6000-specific) semantics
25087 than store_data_bypass_p, returning false rather than aborting
25088 with assertion error when arguments do not satisfy the
25089 requirements of store data bypass.
25090 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25091 rs6000_store_data_bypass_p.
25093 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25095 * config/xtensa/xtensa-protos.h
25096 (xtensa_initial_elimination_offset): New declaration.
25097 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25098 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25099 macro definition, add case for FRAME_POINTER_REGNUM when
25100 FRAME_GROWS_DOWNWARD.
25101 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25102 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25103 xtensa_initial_elimination_offset.
25105 2017-05-08 Nathan Sidwell <nathan@acm.org>
25107 * doc/invoke.texi: Alphabetize -fdump options.
25109 2017-05-08 Martin Sebor <msebor@redhat.com>
25111 PR translation/80280
25112 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25114 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25116 * target.def (compute_frame_layout): New optional target hook.
25117 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25118 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25119 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25121 * reload1.c (verify_initial_elim_offsets): Likewise.
25122 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25123 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25124 (arm_get_frame_offsets): Split up into this ...
25125 (arm_compute_frame_layout): ... and this function.
25127 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25129 * config/aarch64/constraints.md (Usa): New constraint.
25130 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25132 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25134 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25135 with_multilib_list after it has been checked.
25137 2017-05-08 Richard Biener <rguenther@suse.de>
25139 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25140 (bitmap_set_subtract_values): Likewise.
25142 2017-05-08 Richard Biener <rguenther@suse.de>
25144 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25145 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25147 (gimple_stmt_nonzero_warnv_p): Rename to ...
25148 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25150 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25151 (extract_range_basic): Adjust, do not disable propagation on
25152 strict overflow sensitive simplification.
25153 (vrp_visit_cond_stmt): Likewise.
25155 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25157 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25158 body size unconditionally.
25160 2017-05-07 Jeff Law <law@redhat.com>
25163 2017-05-06 Jeff Law <law@redhat.com>
25164 PR tree-optimization/78496
25165 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25168 PR tree-optimization/78496
25169 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25170 (simplify_stmt_using_ranges): Call it.
25171 (vrp_dom_walker::before_dom_children): Extract equivalences
25172 from an ASSERT_EXPR with an equality comparison against a
25175 2017-05-06 Jeff Law <law@redhat.com>
25177 PR tree-optimization/78496
25178 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25181 PR tree-optimization/78496
25182 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25183 (simplify_stmt_using_ranges): Call it.
25184 (vrp_dom_walker::before_dom_children): Extract equivalences
25185 from an ASSERT_EXPR with an equality comparison against a
25188 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25190 * lra-constraints.c (lra_copy_reg_equiv): New function.
25191 (split_reg): Use it to copy equivalence information from the
25192 original register to the spill register.
25194 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25196 PR rtl-optimization/75964
25197 * simplify-rtx.c (simplify_const_relational_operation): Remove
25198 invalid handling of comparisons of integer ABS.
25200 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25202 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25203 initialize to zero.
25204 (init_regs): Remove declaration.
25205 (function_arg_advance_32): Initialize error_p as boolean variable.
25207 2017-05-05 Nathan Sidwell <nathan@acm.org>
25209 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25210 lines. Use for (;;).
25212 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25214 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25215 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25216 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25217 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25218 VF=2 that require versioning.
25220 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25222 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25225 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25227 * diagnostic.h (diagnostic_override_option_index): Convert from
25228 macro to inline function.
25230 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25232 * diagnostic.c (last_module_changed_p): New function.
25233 (set_last_module): New function.
25234 (diagnostic_report_current_module): Convert macro usage to
25235 the above functions.
25236 * diagnostic.h (diagnostic_context::last_module): Strengthen
25237 from const line_map * to const line_map_ordinary *.
25238 (diagnostic_last_module_changed): Delete macro.
25239 (diagnostic_set_last_module): Delete macro.
25241 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25243 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25244 with diagnostic_report_diagnostic.
25245 (diagnostic_n_impl_richloc): Likewise.
25246 * diagnostic.h (report_diagnostic): Delete macro.
25247 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25248 with diagnostic_report_diagnostic.
25249 * substring-locations.c (format_warning_va): Likewise.
25251 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25253 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25254 save/restor of format_spec. Move option-printing code to...
25255 (print_option_information): ...this new function, and
25256 reimplement by simply printing to the pretty_printer,
25257 rather than appending to the format string.
25259 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25261 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25262 handling logic into...
25263 (update_effective_level_from_pragmas): ...this new function.
25265 2017-05-04 Andrew Waterman <andrew@sifive.com>
25267 * config/riscv/riscv.opt (mstrict-align): New option.
25268 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25269 (SLOW_UNALIGNED_ACCESS): Define.
25270 (riscv_slow_unaligned_access): Declare.
25271 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25273 (riscv_slow_unaligned_access): New variable.
25274 (rocket_tune_info): Set slow_unaligned_access to true.
25275 (optimize_size_tune_info): Set slow_unaligned_access to false.
25276 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25277 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25278 (riscv_option_override): Set riscv_slow_unaligned_access.
25279 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25281 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25283 * config/riscv/riscv.md: Unify indentation.
25285 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25290 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25292 (extendsi<mode>2): Add support for doing sign extension via
25293 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25294 don't have ISA 3.0 instructions.
25295 (extendsi<mode>2 splitter): Likewise.
25296 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25297 generate the normal insns since SImode can now go in vector
25298 registers. Disallow the special UNSPECs needed for previous
25299 machines to hide SImode being used. Add new insns
25300 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25301 (fix_trunc<mode>si2_stfiwx): Likewise.
25302 (fix_trunc<mode>si2_internal): Likewise.
25303 (fixuns_trunc<mode>si2): Likewise.
25304 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25305 (fctiw<u>z_<mode>_smallint): Likewise.
25306 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25307 of floating point to 32-bit integer from doing a direct move to
25308 the GPR registers to do a store.
25309 (fctiwz_<mode>): Break long line.
25311 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25313 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25314 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25315 (addr_list, addr_offset_valid_p): New.
25316 (split_address_groups): Check offset validity with above function.
25317 (gt-tree-ssa-loop-ivopts.h): Include header file.
25319 2017-05-05 Nathan Sidwell <nathan@acm.org>
25321 * config.gcc (arm*-*-*): Add missing 'fi'.
25323 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25325 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25326 included in -fopt-info does not matter.
25327 * doc/optinfo.texi (-fopt-info): Fix description of default
25328 behavour. Explicitly say order of options included in -fopt-info
25331 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25333 * config.gcc: Allow combinations of aprofile and rmprofile values for
25334 --with-multilib-list.
25335 * config/arm/t-multilib: New file.
25336 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25337 variables. Remove setting of ISA and floating-point ABI in
25338 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25339 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25340 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25341 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25343 * config/arm/t-rmprofile: Likewise except for the matches changes.
25344 * doc/install.texi (--with-multilib-list): Document the combination of
25345 aprofile and rmprofile values and warn about pitfalls in doing that.
25347 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25349 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25350 (movdi_aarch64): Likewise.
25352 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25354 PR tree-optimization/80632
25355 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25357 (build_arrays): Initialize it for virtual phis.
25358 (fix_phi_nodes): Use it for virtual phis.
25360 PR tree-optimization/80558
25361 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25362 [x, y] op z into [x op, y op z] for op & or | if conditions
25365 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25366 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25369 * config/arm/arm.md (use_literal_pool): Remove.
25370 (64-bit immediate split): No longer takes cost into consideration
25371 if arm_disable_literal_pool is enabled.
25372 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25373 used when arm_disable_literal_pool is enabled.
25374 (arm_max_const_double_inline_cost): Remove use of
25375 arm_disable_literal_pool.
25376 (push_minipool_fix): Add assert.
25377 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25378 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25379 (no_literal_pool_sf_immediate): New.
25381 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25383 PR tree-optimization/80613
25384 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25385 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25387 2017-05-05 Richard Biener <rguenther@suse.de>
25389 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25391 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25393 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25394 of this flag from insn conditions due to removal from r247495.
25396 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25398 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25400 (arm_early_store_addr_dep_ptr): Likewise.
25401 * config/arm/aarch-common-protos.h
25402 (arm_early_load_addr_dep_ptr): Add prototype.
25403 (arm_early_store_addr_dep_ptr): Likewise.
25404 * config/arm/cortex-a53.md: Add new bypasses.
25406 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25408 * tree.c (next_type_uid): Change type to unsigned.
25409 (type_hash_canon): Decrement back next_type_uid if
25410 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25411 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25414 2017-05-04 Martin Sebor <msebor@redhat.com>
25416 * builtins.c: Fix a trivial typo in a comment.
25418 PR middle-end/79234
25419 * builtins.c (check_sizes): Adjust to handle reading past the end.
25420 Avoid printing excessive upper bound of ranges. Use %E to print
25421 tree nodes instead of converting them to %wu.
25422 (expand_builtin_memchr): New function.
25423 (compute_dest_size): Rename...
25424 (compute_objsize): ...to this.
25425 (expand_builtin_memcpy): Adjust.
25426 (expand_builtin_mempcpy): Adjust.
25427 (expand_builtin_strcat): Adjust.
25428 (expand_builtin_strcpy): Adjust.
25429 (check_strncat_sizes): Adjust.
25430 (expand_builtin_strncat): Adjust.
25431 (expand_builtin_strncpy): Adjust and simplify.
25432 (expand_builtin_memset): Adjust.
25433 (expand_builtin_bzero): Adjust.
25434 (expand_builtin_memcmp): Adjust.
25435 (expand_builtin): Handle memcmp.
25436 (maybe_emit_chk_warning): Check strncat just once.
25438 2017-05-04 Martin Sebor <msebor@redhat.com>
25440 PR preprocessor/79214
25441 PR middle-end/79222
25442 PR middle-end/79223
25443 * builtins.c (check_sizes): Add inlining context and issue
25444 warnings even when -Wno-system-headers is set.
25445 (check_strncat_sizes): Same.
25446 (expand_builtin_strncat): Same.
25447 (expand_builtin_memmove): New function.
25448 (expand_builtin_stpncpy): Same.
25449 (expand_builtin): Handle memmove and stpncpy.
25451 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25453 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25454 which is not used any more.
25456 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25458 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25460 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25462 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25463 (cortexa53_tunings): Likewise.
25464 (cortexa57_tunings): Likewise.
25465 (cortexa72_tunings): Likewise.
25466 (cortexa73_tunings): Likewise.
25468 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25470 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25471 Set loop alignment to 8.
25473 2017-05-04 Martin Sebor <msebor@redhat.com>
25475 PR translation/80280
25476 * builtins.c (expand_builtin_object_size): Add missing quoting to
25477 %D and like directives.
25478 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25479 (hsa_type_for_tree_type): Same.
25480 (verify_function_arguments): Same.
25481 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25482 * varasm.c (get_section): Same.
25485 2017-05-04 Martin Sebor <msebor@redhat.com>
25487 PR translation/80280
25488 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25490 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25492 * config/aarch64/aarch64.c (generic_addrcost_table):
25493 Change HI/TI mode setting.
25495 2017-05-04 Martin Jambor <mjambor@suse.cz>
25497 PR tree-optimization/80622
25498 * tree-sra.c (comes_initialized_p): New function.
25499 (build_accesses_from_assign): Only set write lazily when
25500 comes_initialized_p is false.
25501 (analyze_access_subtree): Use comes_initialized_p.
25502 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25503 instead of testing for PARM_DECL.
25505 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25507 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25508 constraint on operand 0 to allow more general addressing modes.
25509 Adjust output template.
25510 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25512 * config/aarch64/aarch64-protos.h
25513 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25514 * config/aarch64/constraints.md (Dp): New address constraint.
25515 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25518 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25520 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25521 update use of estimate_ipcp_clone_size_and_time.
25522 (estimate_local_effects): Update use of
25523 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25524 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25525 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25526 Return nonspecialized time.
25528 2017-05-04 Richard Biener <rguenther@suse.de>
25530 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25531 for the last VUSE which def dominates the PHI. Directly call
25533 (get_continuation_for_phi_1): Remove.
25535 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25537 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25538 to explain the use of truncating division. Cap the number of
25539 iterations to the maximum given by nb_iterations_upper_bound,
25542 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25544 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25545 * configure: Regenerate.
25546 * config.in: Regenerate.
25547 * config/i386/driver-mingw32.c: new file.
25548 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25549 * config.host: Link driver-mingw32.o on MinGW host.
25550 * doc/install.texi: Document new --enable-mingw-wildcard configure
25553 2017-05-04 Marek Polacek <polacek@redhat.com>
25555 PR tree-optimization/80612
25556 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25558 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25559 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25561 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25562 (movt splitter): Likewise.
25563 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25564 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25565 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25566 block for Thumb-1 with MOVT.
25567 (thumb2_legitimate_address_p): Move code block ...
25568 (can_avoid_literal_pool_for_label_p): ... into this new function.
25569 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25571 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25572 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25573 "M-profile targets with the MOVT instruction".
25575 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25577 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25578 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25579 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25581 2017-05-04 Martin Liska <mliska@suse.cz>
25583 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25584 variable cond_code.
25586 2017-05-04 Richard Biener <rguenther@suse.de>
25588 * tree.c (array_at_struct_end_p): Handle arrays at struct
25589 end with flexarrays more conservatively. Refactor and treat
25590 arrays of arrays or aggregates more strict. Fix
25591 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25592 * tree.c (array_at_struct_end_p): Adjust prototype.
25593 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25594 * gimple-fold.c (get_range_strlen): Likewise.
25595 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25597 2017-05-04 Richard Biener <rguenther@suse.de>
25599 PR tree-optimization/31130
25600 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25602 (supports_overflow_infinity): Likewise.
25603 (is_negative_overflow_infinity): Likewise.
25604 (is_positive_overflow_infinity): Likewise.
25605 (is_overflow_infinity): Likewise.
25606 (stmt_overflow_infinity): Likewise.
25607 (overflow_infinity_range_p): Likewise.
25608 (usable_range_p): Remove as always returning true.
25609 (make_overflow_infinity): Remove.
25610 (negative_overflow_infinity): Likewise.
25611 (positive_overflow_infinity): Likewise.
25612 (avoid_overflow_infinity): Likewise.
25613 (set_value_range): Adjust accordingly.
25614 (set_value_range_to_nonnegative): Likewise, remove now unused
25615 overflow_infinity arg.
25616 (vrp_operand_equal_p): Adjust.
25617 (update_value_range): Likewise.
25618 (range_int_cst_singleton_p): Likewise.
25619 (operand_less_p): Likewise.
25620 (compare_values_warnv): Likewise.
25621 (extract_range_for_var_from_comparison_expr): Likewise.
25622 (vrp_int_const_binop): Likewise.
25623 (zero_nonzero_bits_from_vr): Likewise.
25624 (extract_range_from_multiplicative_op_1): Likewise.
25625 (extract_range_from_binary_expr_1): Likewise.
25626 (extract_range_from_unary_expr): Likewise.
25627 (extract_range_from_comparison): Likewise.
25628 (extract_range_basic): Likewise.
25629 (adjust_range_with_scev): Likewise.
25630 (compare_ranges): Likewise.
25631 (compare_range_with_value): Likewise.
25632 (dump_value_range): Likewise.
25633 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25635 (simplify_cond_using_ranges): Adjust.
25637 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25639 * brig-builtins.def: Added a builtin for class_f64.
25640 * builtin-types.def: Added a builtin type needed by class_f64.
25642 2017-05-03 Jason Merrill <jason@redhat.com>
25644 * timevar.def: Add TV_CONSTEXPR.
25646 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25648 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25650 2017-05-03 Martin Jambor <mjambor@suse.cz>
25652 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25653 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25654 * ipa-cp.c (ipcp_propagate_stage): Do not call
25655 ipa_update_after_lto_read.
25656 * ipa-inline.c (ipa_inline): Likewise.
25658 2017-05-03 Martin Jambor <mjambor@suse.cz>
25660 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25661 tag. Added a default constructor and a destructor.
25662 (ipa_edge_args_sum_t): New class;
25663 (ipa_edge_args_sum): Declare.
25664 (ipa_edge_args_vector): Remove declaration.
25665 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25666 (ipa_free_edge_args_substructures): Remove declaration.
25667 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25668 (ipa_edge_args_info_available_for_edge_p): Likewise.
25669 * ipa-prop.c (ipa_edge_args_vector): Removed.
25670 (edge_removal_hook_holder): Likewise.
25671 (edge_duplication_hook_holder): Likewise.
25672 (ipa_edge_args_sum): New variable.
25673 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25674 ipa_edge_args_vector.
25675 (ipa_free_edge_args_substructures): Likewise.
25676 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25677 ipa_edge_args_vector.
25678 (ipa_edge_removal_hook): Turned into method
25679 ipa_edge_args_sum_t::remove.
25680 (ipa_edge_duplication_hook): Turned into method
25681 ipa_edge_args_sum_t::duplicate.
25682 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25683 registering edge hooks.
25684 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25685 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25686 ipa_edge_args_sum instead of ipa_edge_args_vector.
25687 * ipa-profile.c (ipa_profile): Likewise.
25689 2017-05-03 Martin Jambor <mjambor@suse.cz>
25691 * symbol-summary.h (function_summary): New method exists.
25692 (function_summary::symtab_removal): Deallocate through release.
25693 (call_summary): New class.
25694 (gt_ggc_mx): New overload.
25695 (gt_pch_nx): Likewise.
25696 (gt_pch_nx): Likewise.
25698 2017-05-03 Jeff Law <law@redhat.com>
25700 PR tree-optimization/78496
25701 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25702 from simplify_cond_using_ranges. Split off code to walk
25703 backwards through casts into ...
25704 (simplify_cond_using_ranges_2): New function.
25705 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25706 (execute_vrp): After identifying jump threads, call
25707 simplify_cond_using_ranges_2.
25709 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25712 * ipa-inline.h (inline_summary): Add ctor.
25713 (create_ggc): Do not use ggc_cleared_alloc.
25715 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25716 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25718 * gcc.c (handle_braces): Support escaping in switch matching
25720 * doc/invoke.texi (Spec Files): Document it.
25721 Remove superfluous @code markup in items.
25723 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25725 * diagnostic-show-locus.c (struct column_range): New struct.
25726 (get_affected_columns): New function.
25727 (get_printed_columns): New function.
25728 (struct correction): New struct.
25729 (correction::ensure_capacity): New function.
25730 (correction::ensure_terminated): New function.
25731 (struct line_corrections): New struct.
25732 (line_corrections::~line_corrections): New dtor.
25733 (line_corrections::add_hint): New function.
25734 (layout::print_trailing_fixits): Reimplement in terms of the new
25736 (selftest::test_overlapped_fixit_printing): New function.
25737 (selftest::diagnostic_show_locus_c_tests): Call it.
25739 2017-05-03 Nathan Sidwell <nathan@acm.org>
25741 Canonicalize canonical type hashing
25742 * tree.h (type_hash_canon_hash): Declare.
25743 * tree.c (type_hash_list, attribute_hash_list): Move into
25744 type_hash_canon_hash.
25745 (build_type_attribute_qual_variant): Break out hash code calc into
25746 type_hash_canon_hash.
25747 (type_hash_canon_hash): New. Generic type hash computation.
25748 (build_range_type_1, build_array_type_1, build_function_type,
25749 build_method_type_directly, build_offset_type, build_complex_type,
25750 make_vector_type): Call it.
25752 2017-05-03 Richard Biener <rguenther@suse.de>
25754 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25755 When all DRs have unknown misaligned do not always peel
25756 when there is a store but apply the same costing model as if
25757 there were only loads.
25759 2017-05-03 Richard Biener <rguenther@suse.de>
25762 PR tree-optimization/80492
25763 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25764 compare_base_decls returning dont-know properly.
25766 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25768 * config/arm/iterators.md (CCSI): New mode iterator.
25769 (arch): New mode attribute.
25770 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25771 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25772 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25773 code iterator for success result mode.
25774 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25775 the corresponding new insn generators.
25777 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25780 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25781 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25783 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25785 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25786 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25787 (DDR_A): Wrap DDR argument in brackets.
25788 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25789 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25790 (DDR_REVERSED_P): Likewise.
25792 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25794 PR tree-optimization/79472
25795 * tree-switch-conversion.c (struct switch_conv_info): Add
25796 contiguous_range and default_case_nonstandard fields.
25797 (collect_switch_conv_info): Compute contiguous_range and
25798 default_case_nonstandard fields, don't clear final_bb if
25799 contiguous_range and only the default case doesn't have the required
25801 (check_all_empty_except_final): Set default_case_nonstandard instead
25802 of failing if contiguous_range and the default case doesn't have empty
25804 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25805 and only the default case doesn't have the required constants. Skip
25807 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25808 if default_case_nonstandard.
25809 (build_constructors): Build constant 1 just once. Assert that default
25810 values aren't inserted in between cases if contiguous_range. Skip
25812 (build_arrays): Skip virtual phis.
25813 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25814 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25815 Handle virtual phis.
25816 (gen_inbound_check): Handle default_case_nonstandard case.
25817 (process_switch): Adjust check_final_bb caller. Call
25818 gather_default_values with the first non-default case instead of
25819 default case if default_case_nonstandard.
25821 2017-05-02 Nathan Sidwell <nathan@acm.org>
25823 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25824 check. Fix formatting.
25826 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25828 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25829 errors when comparing specialized and unspecialized times.
25831 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25833 * diagnostic-show-locus.c
25834 (layout::should_print_annotation_line_p): Make private.
25835 (layout::print_annotation_line): Make private.
25836 (layout::annotation_line_showed_range_p): Make private.
25837 (layout::show_ruler): Make private.
25838 (layout::print_source_line): Make private. Pass in line and
25839 line_width, rather than calling location_get_source_line. Drop
25841 (layout::print_leading_fixits): New method.
25842 (layout::print_any_fixits): Rename to...
25843 (layout::print_trailing_fixits): ...this, and make private.
25844 Don't print newline fixits.
25845 (diagnostic_show_locus): Move logic for printing one row into...
25846 (layout::print_line): ...this new function. Move the
25847 location_get_source_line call and error-handling from
25848 print_source_line to here. Call print_leading_fixits, and rename
25849 print_any_fixits to print_trailing_fixits.
25850 (selftest::test_fixit_insert_containing_newline): Update now that
25851 newlines are partially supported.
25852 (selftest::test_fixit_insert_containing_newline_2): New test.
25853 (selftest::test_fixit_replace_containing_newline): Update comments.
25854 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25855 * edit-context.c (class added_line): New class.
25856 (class edited_line): Describe newline handling in comment.
25857 (edited_line::actually_edited_p): New method.
25858 (edited_line::print_content): Delete redundant decl.
25859 (edited_line::m_predecessors): New field.
25860 (edited_file::print_content): Call edited_line::print_content.
25861 (edited_file::print_diff): Update to support newlines.
25862 (edited_file::print_diff_hunk): Likewise.
25863 (edited_file::print_run_of_changed_lines): New function.
25864 (edited_file::print_diff_line): Convert to...
25865 (print_diff_line): ...this.
25866 (edited_file::get_effective_line_count): New function.
25867 (edited_line::edited_line): Initialize new field m_predecessors.
25868 (edited_line::~edited_line): Clean up m_predecessors.
25869 (edited_line::apply_fixit): Handle newlines.
25870 (edited_line::get_effective_line_count): New function.
25871 (edited_line::print_content): New function.
25872 (edited_line::print_diff_lines): New function.
25873 (selftest::test_applying_fixits_insert_containing_newline): New
25875 (selftest::test_applying_fixits_replace_containing_newline): New
25877 (selftest::insert_line): New function.
25878 (selftest::test_applying_fixits_multiple_lines): Add example of
25880 (selftest::edit_context_c_tests): Call the new tests.
25882 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25884 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25885 parameter cand. Update dump information.
25886 (get_computation_cost): Update uses.
25888 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25890 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25891 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25892 (get_computation_at, rewrite_use_address): Update use of
25893 get_computation_aff.
25895 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25897 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25898 (get_computation): Delete.
25899 (get_computation_cost): Implement like get_computation_cost_at.
25900 Use get_computation_at.
25901 (get_computation_cost_at): Delete.
25902 (rewrite_use_nonlinear_expr): Use get_computation_at.
25903 (rewrite_use_compare, remove_unused_ivs): Ditto.
25905 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25907 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25911 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25912 (ivopts_global_cost_for_size): Rename parameter and update uses.
25913 (iv_ca_recount_cost): Update uses.
25914 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25915 candidates seperately in n_invs and n_cands.
25916 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25918 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25920 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25921 (find_inv_vars_cb): New.
25922 (find_depends): Renamed to ...
25923 (find_inv_vars): ... this.
25924 (add_candidate_1, force_var_cost): Call find_inv_vars.
25925 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25927 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25929 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25930 inv_vars. Add inv_exprs.
25931 (struct iv_cand): Rename depends_on to inv_vars.
25932 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25933 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25934 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25935 (dump_cand): Dump inv_vars.
25936 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25937 (record_invariant, find_depends, add_candidate_1): Ditto.
25938 (set_group_iv_cost, force_var_cost): Ditto.
25939 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25940 (get_computation_cost_at, get_computation_cost): Ditto.
25941 (determine_group_iv_cost_generic): Ditto.
25942 (determine_group_iv_cost_address): Ditto.
25943 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25944 (determine_group_iv_costs): Ditto.
25945 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25946 (iv_ca_set_remove_invariants): Renamed to ...
25947 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25948 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25949 (iv_ca_set_add_invariants): Renamed to ...
25950 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25951 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25952 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25953 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25954 (create_new_ivs): Remove useless dump.
25956 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25958 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25960 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25961 (iv_ca_set_no_cp, create_new_iv): Ditto.
25963 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25965 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25967 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25969 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25970 function tree_check2.
25972 2017-05-02 Martin Liska <mliska@suse.cz>
25974 * doc/gcov.texi: Add missing preposition.
25975 * gcov.c (function_info::function_info): Properly fill up
25976 all member variables.
25978 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25980 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25982 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25984 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25986 2017-05-02 Martin Liska <mliska@suse.cz>
25989 * lto-streamer-in.c (lto_read_tree_1): Remove
25990 LTO_STREAMER_DEBUG.
25991 * lto-streamer.c (struct tree_hash_entry): Likewise.
25992 (struct tree_entry_hasher): Likewise.
25993 (tree_entry_hasher::hash): Likewise.
25994 (tree_entry_hasher::equal): Likewise.
25995 (lto_streamer_init): Likewise.
25996 (lto_orig_address_map): Likewise.
25997 (lto_orig_address_get): Likewise.
25998 (lto_orig_address_remove): Likewise.
25999 * lto-streamer.h: Likewise.
26000 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26001 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26003 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26005 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26006 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26007 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26008 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26009 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26010 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26011 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26012 (mm_maskz_sub_ss): New intrinsics.
26013 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26014 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26015 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26016 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26017 (__builtin_ia32_subss_mask_round): New builtins.
26018 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26019 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26020 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26022 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26023 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26025 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26028 2017-05-02 Martin Jambor <mjambor@suse.cz>
26030 PR tree-optimization/78687
26031 * tree-sra.c (access): New field parent.
26032 (process_subtree_disqualification): New function.
26033 (disqualify_candidate): Call it.
26034 (build_accesses_from_assign): Reset write flag if creating an
26036 (build_access_subtree): Fill in parent field and also prpagate
26037 down grp_write flag.
26038 (create_artificial_child_access): New parameter set_grp_write, set
26039 grp_write to its value.
26040 (propagate_subaccesses_across_link): Also propagate grp_write flag
26042 (propagate_all_subaccesses): Push the closest parent back to work
26043 queue if add_access_to_work_queue returned true.
26045 2017-05-02 Richard Biener <rguenther@suse.de>
26047 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26048 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26049 -fstrict-overflow documentation.
26050 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26051 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26052 flag_strict_overflow.
26053 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26054 * lto-opts.c (lto_write_options): Do not stream it.
26055 * lto-wrapper.c (merge_and_complain): Do not handle it.
26056 * opts.c (default_options_table): Do not set -fstrict-overflow.
26057 (finish_options): Likewise do not clear it when sanitizing.
26058 * simplify-rtx.c (simplify_const_relational_operation): Do not
26059 test flag_strict_overflow.
26061 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26063 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26064 using enabled attribute.
26065 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26066 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26067 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26068 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26069 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26070 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26071 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26072 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26073 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26074 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26076 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26078 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26080 2017-05-02 Richard Biener <rguenther@suse.de>
26082 PR tree-optimization/80591
26084 2017-04-10 Richard Biener <rguenther@suse.de>
26086 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26089 2017-05-02 Richard Biener <rguenther@suse.de>
26091 PR tree-optimization/80549
26092 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26093 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26094 headers if they do not have a preheader.
26096 2017-05-02 Martin Liska <mliska@suse.cz>
26099 * common.opt: Fix typo.
26100 * doc/invoke.texi: Likewise.
26102 2017-05-01 Jan Beulich <jbeulich@suse.com>
26104 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26105 swapping, add (x,x,m,x,n) alternative.
26107 2017-05-01 Nathan Sidwell <nathan@acm.org>
26109 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26110 unnecessary unadjusted_alignment check.
26112 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26115 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26117 * gimplify.c (gimplify_cilk_detach): New function.
26118 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26119 * tree-core.h: Document EXPR_CILK_SPAWN.
26120 * tree.h (EXPR_CILK_SPAWN): Define.
26122 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26124 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26125 to use new fixit_hint representation, using the "replace" logic.
26126 (get_line_span_for_fixit_hint): Likewise.
26127 (layout::print_any_fixits): Likewise.
26128 (selftest::test_one_liner_many_fixits): Rename to...
26129 (selftest::test_one_liner_many_fixits_1): ...this, and update
26130 comment and expected output to reflect that the multiple fix-it
26131 hints are now consolidated into one insertion.
26132 (selftest::test_one_liner_many_fixits_2): New test.
26133 (selftest::test_diagnostic_show_locus_one_liner): Update for
26135 (selftest::test_fixit_consolidation): Update for fix-it API
26137 * diagnostic.c (print_parseable_fixits): Likewise.
26138 * edit-context.c (edited_line::m_line_events): Convert from
26139 auto_vec <line_event *> to auto_vec <line_event>.
26140 (class line_event): Convert from abstract base class to a concrete
26141 class, taking over the role of replace_event.
26142 (class insert_event): Delete.
26143 (class replace_event): Rename to class line_event. Convert to
26145 (edit_context::add_fixits): Reimplement.
26146 (edit_context::apply_insert): Delete.
26147 (edit_context::apply_replace): Rename to...
26148 (edit_context::apply_fixit): ...this. Convert to half-open range.
26149 (edited_file::apply_insert): Delete.
26150 (edited_file::apply_replace): Rename to...
26151 (edited_file::apply_fixit): ...this.
26152 (edited_line::~edited_line): Drop deletion of events.
26153 (edited_line::apply_insert): Delete.
26154 (edited_line::apply_replace): Rename to...
26155 (edited_line::apply_fixit): ...this. Convert to half-open range.
26156 Update for change to type of m_line_events.
26157 * edit-context.h (edit_context::apply_insert): Delete.
26158 (edit_context::apply_replace): Rename to...
26159 (edit_context::apply_fixit): ...this.
26161 2017-05-01 Martin Sebor <msebor@redhat.com>
26163 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26166 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26169 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26170 __get_cpuid_max returns 0.
26171 (__get_cpuid_count): Ditto.
26173 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26175 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26176 replacement expression is another instance of one of its arguments.
26178 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26181 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26182 check for stack push/pop autoinc.
26183 * config/i386/i386.c (ix86_agi_dependent): Return false
26184 if the only reason why modified_in_p returned true is that
26185 addr is SP based and set_insn is a push or pop.
26187 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26189 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26192 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26195 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26196 (account_size_time): Use two predicates - exec_pred and
26198 (evaluate_conditions_for_known_args): Compute both clause and
26200 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26201 (inline_summary_t::duplicate): Update.
26202 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26204 (compute_inline_parameters): Likewise.
26205 (estimate_edge_size_and_time): Update caluclation of time.
26206 (estimate_node_size_and_time): Compute both time and nonspecialized
26208 (estimate_ipcp_clone_size_and_time): Update.
26209 (inline_merge_summary): Update.
26210 (do_estimate_edge_time): Update.
26211 (do_estimate_edge_size): Update.
26212 (do_estimate_edge_hints): Update.
26213 (inline_read_section, inline_write_summary): Stream both new predicates.
26214 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26216 (compute_inlined_call_time): Cleanup.
26217 (big_speedup_p): Update.
26218 (edge_badness): Update.
26219 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26220 (size_time_entry): Replace predicate by exec_predicate and
26221 nonconst_predicate.
26222 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26223 (estimate_edge_time): Return also nonspec_time.
26224 (reset_edge_growth_cache): Update.
26226 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26228 PR rtl-optimization/80491
26229 * ifcvt.c (noce_process_if_block): When looking for x setter
26230 with missing else_bb, don't check only the insn right before
26231 cond_earliest, but look for the last insn that x is modified in
26232 within the same bb.
26234 PR rtl-optimization/80491
26235 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26237 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26239 PR tree-optimization/80487
26240 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26242 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26244 PR tree-optimization/79697
26245 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26246 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26247 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26249 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26250 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26252 2017-04-28 Martin Sebor <msebor@redhat.com>
26254 PR tree-optimization/80523
26255 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26256 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26258 (maybe_warn, format_directive, parse_directive): Use new functions.
26259 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26261 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26263 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26265 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26267 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26268 target_header_dir): Set correctly.
26269 * configure: Regenerated.
26270 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26271 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26272 instead of SYSTEM_HEADER_DIR.
26274 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26276 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26277 (estimate_local_effects): Likewise.
26278 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26279 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26280 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26281 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26282 do_estimate_edge_time, estimate_edge_time): Likewise.
26283 * ipa-inline-analysis.c (estimate_node_size_and_time,
26284 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26285 (estimate_time_after_inlining): Remove.
26287 2017-04-28 Martin Liska <mliska@suse.cz>
26289 * doc/gcov.texi: Enhance documentation of gcov.
26291 2017-04-28 Martin Liska <mliska@suse.cz>
26293 * doc/gcov.texi: Sort options in alphabetic order.
26294 * doc/gcov-dump.texi: Likewise.
26295 * doc/gcov-tool.texi: Likewise.
26296 * gcov.c (print_usage): Likewise.
26297 * gcov-dump.c (print_usage): Likewise.
26298 * gcov-tool.c (print_merge_usage_message): Likewise.
26299 (print_rewrite_usage_message): Likewise.
26300 (print_overlap_usage_message): Likewise.
26302 2017-04-28 Martin Liska <mliska@suse.cz>
26304 PR gcov-profile/53915
26305 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26307 2017-04-28 Martin Liska <mliska@suse.cz>
26309 PR gcov-profile/79891
26310 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26311 is marked by compiler as living on a line.
26312 (get_cycles_count): Remove usage of the union.
26313 (output_intermediate_file): Likewise.
26314 (find_source): Fix GNU coding style.
26315 (accumulate_line_counts): Remove old non-all block mode.
26316 (output_lines): Remove usage of the union.
26317 * profile.c (output_location): Include all BBs, even if
26318 belonging to a same line (and file) as a previous BB.
26320 2017-04-28 Martin Liska <mliska@suse.cz>
26322 * gcov.c (process_args): Handle new argument 'w'.
26323 (read_graph_file): Assign ID to BBs.
26324 (output_branch_count): Display BB # if verbose flag is set.
26325 (output_lines): Likewise for arcs.
26326 (print_usage): Add '--verbose' option help.
26327 * doc/gcov.texi: Document --verbose (-w) option.
26329 2017-04-28 Martin Liska <mliska@suse.cz>
26331 * gcov.c (struct block_location_info): New struct.
26332 (process_file): Fill up the new structure.
26333 (read_graph_file): Replace usage of encoding by the newly added
26335 (add_line_counts): Likewise.
26336 (accumulate_line_counts): Remove usage of the union.
26337 (function_info::function_info): New function.
26338 (function_info::~function_info): Likewise.
26339 (process_file): Call delete instead of release_function.
26340 (release_function): Release the function.
26341 (release_structures): Call delete instead of release_function.
26342 (solve_flow_graph): Replace usage of num_blocks.
26343 (find_exception_blocks): Likewise.
26344 (output_lines): Fix GNU coding style.
26346 2017-04-28 Martin Liska <mliska@suse.cz>
26349 * coverage.c (coverage_remove_note_file): New function.
26350 * coverage.h: Declare the function.
26351 * toplev.c (finalize): Clean if an error has been seen.
26353 2017-04-28 Martin Liska <mliska@suse.cz>
26355 PR gcov-profile/80031
26356 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26357 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26358 * gcov.c (read_graph_file): Read just number of blocks.
26359 * profile.c (branch_prob): Do not stream 0 flags per a basic
26362 2017-04-28 Martin Liska <mliska@suse.cz>
26364 * gcov-dump.c (tag_*): Add new argument to declarations.
26365 (dump_gcov_file): Likewise.
26366 (tag_blocks): Add and use new argument depth.
26367 (tag_arcs): Likewise.
26368 (tag_lines): Likewise.
26369 (tag_counters): Likewise.
26370 (tag_summary): Likewise.
26371 (dump_working_sets): Use depth to do a proper indentation.
26373 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26376 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26377 * symtab.c (symtab_node::debug_symtab): Move definition here.
26379 2017-04-28 Richard Biener <rguenther@suse.de>
26381 * lto-streamer.h (LTO_major_version): Bump to 7.
26383 2017-04-28 Richard Biener <rguenther@suse.de>
26385 * tree-vrp.c (assert_info): New struct.
26386 (add_assert_info): New helper.
26387 (register_edge_assert_for_2): Refactor to add asserts to a vector
26389 (register_edge_assert_for_1): Likewise.
26390 (register_edge_assert_for): Likewise.
26391 (finish_register_edge_assert_for): New helper actually registering
26392 asserts where live on edge.
26393 (find_conditional_asserts): Adjust.
26394 (find_switch_asserts): Likewise.
26395 (evrp_dom_walker::try_find_new_range): Generalize.
26396 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26398 2017-04-27 Marek Polacek <polacek@redhat.com>
26401 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26402 arg10 and arg11 to itype.
26404 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26406 * doc/extend.texi (Object Size Checking): Improve grammar.
26408 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26411 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26412 that the logic for permitting reciprocal estimates matches that
26415 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26418 * tree.c (type_cache_hasher::equal): Only compare
26419 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26420 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26421 non-aggregate element types.
26422 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26423 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26425 2017-04-27 Richard Biener <rguenther@suse.de>
26427 PR middle-end/80533
26428 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26429 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26430 keeping a reference to a trailing array.
26432 2017-04-27 Richard Biener <rguenther@suse.de>
26434 PR middle-end/80539
26435 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26436 being in loop-closed SSA form conservatively.
26437 (chrec_fold_multiply_poly_poly): Likewise.
26439 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26441 PR middle-end/79665
26442 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26443 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26445 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26448 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26449 (aarch64_function_arg_alignment): Return unsigned int again, but still
26450 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26451 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26452 Don't emit -Wpsabi note.
26453 (aarch64_function_arg_boundary): Likewise.
26454 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26457 2017-04-26 Nathan Sidwell <nathan@acm.org>
26459 * tree.h (crc32_unsigned_n): Declare.
26460 (crc32_unsigned, crc32_unsigned): Make inline.
26461 * tree.c (crc32_unsigned_bits): Replace with ...
26462 (crc32_unsigned_n): ... this.
26463 (crc32_unsigned, crc32_byte): Remove.
26464 (crc32_string): Remove unnecessary braces.
26466 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26468 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26469 * ipa-inline-analysis.c (MAX_TIME): Remove.
26470 (account_size_time): Use sreal for time.
26471 (dump_inline_summary): Update.
26472 (estimate_function_body_sizes): Update.
26473 (estimate_edge_size_and_time): Update.
26474 (estimate_calls_size_and_time): Update.
26475 (estimate_node_size_and_time): Update.
26476 (inline_merge_summary): Update.
26477 (inline_update_overall_summary): Update.
26478 (estimate_time_after_inlining): Update.
26479 (inline_read_section): Update.
26480 (inline_write_summary): Update.
26481 * ipa-inline.c (compute_uninlined_call_time): Update.
26482 (compute_inlined_call_time): Update.
26483 (recursive_inlining): Update.
26484 (inline_small_functions): Update.
26485 (dump_overall_stats): Update.
26486 * ipa-inline.h: Include sreal.h.
26487 (size_time_entry): Turn time to sreal.
26488 (inline_summary): Turn self_time nad time to sreal.
26490 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26492 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26494 (sreal::stream_out, sreal::stream_in): New.
26495 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26497 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26499 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26502 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26505 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26507 (dimode_scalar_chain::compute_convert_gain): Ditto.
26508 (dimode_scalar_chain::make_vector_copies): Ditto.
26509 (dimode_scalar_chain::convert_reg): Ditto.
26510 (dimode_scalar_chain::convert_insn): Ditto.
26511 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26512 (VI248_AVX512BW_1): New mode iterator.
26513 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26514 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26517 2017-04-25 Martin Sebor <msebor@redhat.com>
26519 PR tree-optimization/80497
26520 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26521 constants are representable in HOST_WIDE_INT.
26522 (parse_directive): Ditto.
26524 2017-04-25 Martin Sebor <msebor@redhat.com>
26527 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26528 (new_zero_array): Adjust signature.
26529 (dom_info::dom_init): Used unsigned rather that size_t.
26530 (dom_info::dom_info): Same.
26532 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26533 Jakub Jelinek <jakub@redhat.com>
26536 * config/arm/arm.c: Include gimple.h.
26537 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26538 returns negative, increment ncrn only if it returned positive.
26539 (arm_needs_doubleword_align): Return int instead of bool,
26540 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26541 members, but if there is any such non-FIELD_DECL
26542 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26543 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26544 returns negative, increment nregs only if it returned positive.
26545 (arm_setup_incoming_varargs): Likewise.
26546 (arm_function_arg_boundary): Emit -Wpsabi note if
26547 arm_needs_doubleword_align returns negative, return
26548 DOUBLEWORD_ALIGNMENT only if it returned positive.
26550 2017-04-25 Marek Polacek <polacek@redhat.com>
26553 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26554 first argument to type.
26556 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26560 type checks to test for compatibility instead of equality.
26562 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26563 Jakub Jelinek <jakub@redhat.com>
26566 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26568 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26569 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26570 the alignment computation, but return their maximum in warn_alignment.
26571 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26572 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26574 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26575 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26578 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26580 * config/arc/simdext.md (dmpyh): Fix typo.
26582 2017-04-25 Richard Biener <rguenther@suse.de>
26584 PR tree-optimization/80492
26585 * alias.c (compare_base_decls): Handle registers with asm
26586 specification conservatively.
26587 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26588 compare_base_decls returning dont-know properly.
26590 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26592 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26593 (legitimate_offset_address_p): New function.
26594 (arc_legitimate_address_p): Use above function.
26596 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26598 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26600 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26602 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26603 ACCH registers whenever they are available.
26605 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26607 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26608 double regs fix when not used.
26610 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26612 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26614 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26615 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26617 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26619 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26620 when emitting short ADD instructions.
26622 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26624 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26626 (cmpsi_cc_c_insn): Likewise.
26627 (cbranchsi4_scratch): Compute proper instruction length using
26628 compact_hreg_operand.
26629 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26631 2017-04-25 Richard Biener <rguenther@suse.de>
26633 PR middle-end/80509
26634 * passes.c (pass_manager::pass_manager): Initialize
26635 m_name_to_pass_map.
26637 2017-04-25 Richard Biener <rguenther@suse.de>
26639 PR tree-optimization/79201
26640 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26642 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26645 * config/s390/vector.md: Split MEM->GPR vector moves for
26646 non-s_operand addresses.
26648 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26651 * config/s390/predicates.md (reload_const_wide_int_operand): New
26653 * config/s390/s390.md ("movti"): Remove d/P alternative.
26654 ("movti_bigconst"): New pattern definition.
26656 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26659 * s390-protos.h (s390_expand_cs_hqi): Removed.
26660 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26661 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26662 modes as well as CCZ1mode and CCZmode.
26663 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26664 signature of s390_emit_compare_and_swap.
26665 (s390_expand_cs_hqi): Likewise, make static.
26666 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26667 compare-and-swap, in some cases.
26668 (s390_expand_cs): Wrapper function.
26669 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26671 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26672 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26673 patterns for small and large integers. Forbid symref memory operands.
26674 Move expander to s390.c. Require cc register.
26675 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26676 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26677 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26678 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26679 symref memory operands. Remove CC mode and call s390_match_ccmode
26681 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26683 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26685 * config/s390/s390.md (define_peephole2): New peephole to help
26686 combining the load-and-test pattern with volatile memory.
26688 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26690 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26691 with CCZmode for TARGET_Z196.
26693 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26695 PR rtl-optimization/80501
26696 * combine.c (make_compound_operation_int): Set subreg_code to SET
26697 even for AND with mask of the sign bit of mode.
26699 PR rtl-optimization/80500
26700 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26701 sum's initial value.
26703 2017-04-25 Julian Brown <julian@codesourcery.com>
26704 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26706 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26708 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26710 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26712 2017-04-25 Julian Brown <julian@codesourcery.com>
26713 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26715 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26716 (thunderx2t99_sha): New Reservation.
26718 2017-04-25 Julian Brown <julian@codesourcery.com>
26719 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26721 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26722 type for 1-element load.
26724 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26726 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26728 2017-04-24 Martin Jambor <mjambor@suse.cz>
26730 PR tree-optimization/80293
26731 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26732 char arrays not totally scalarizable if it is false.
26733 (analyze_all_variable_accesses): Pass correct value in the new
26734 parameter. Add a statistics counter.
26736 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26738 PR middle-end/79931
26739 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26741 2017-04-24 Richard Biener <rguenther@suse.de>
26743 PR tree-optimization/80494
26744 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26745 out for complex types.
26747 2017-04-24 Richard Biener <rguenther@suse.de>
26749 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26750 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26751 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26752 (DFS): Adjust and never fail.
26753 (sccvn_dom_walker::fail): Remove.
26754 (sccvn_dom_walker::before_dom_children): Adjust.
26755 (run_scc_vn): Likewise and never fail.
26756 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26757 (pass_fre::execute): Likewise.
26759 2017-04-24 Richard Biener <rguenther@suse.de>
26761 PR tree-optimization/79725
26762 * tree-ssa-sink.c (statement_sink_location): Return whether
26763 failure reason was zero uses. Move that check later.
26764 (sink_code_in_bb): Deal with zero uses by removing the stmt
26767 2017-04-24 Richard Biener <rguenther@suse.de>
26770 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26771 pointer-based references.
26773 2017-04-24 Richard Biener <rguenther@suse.de>
26776 * pass_manager.h (pass_manager::operator new): Remove.
26777 (pass_manager::operator delete): Likewise.
26778 * passes.c (pass_manager::operator new): Remove.
26779 (pass_manager::operator delete): Likewise.
26780 (pass_manager::pass_manager): Zero individual pass members.
26782 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26785 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26786 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26787 Check "XEXP (src, 1)" operand here.
26788 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26789 Check "XEXP (src, 1)" operand here.
26790 (dimode_scalar_chain::make_vector_copies): Detect count register
26791 of a shift instruction. Zero extend count register from QImode
26792 to DImode to satisfy vector shift pattern count operand predicate.
26793 Substitute vector shift count operand with a DImode copy.
26794 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26797 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26799 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26800 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26801 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26802 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26803 (UNSPEC_NOREX_MEM): Remove definition.
26805 2017-04-21 Richard Biener <rguenther@suse.de>
26807 PR tree-optimization/79547
26808 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26809 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26810 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26811 without any constraints.
26813 2017-04-21 Richard Biener <rguenther@suse.de>
26815 PR tree-optimization/78847
26816 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26818 2017-04-21 Richard Biener <rguenther@suse.de>
26820 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26821 (build_distinct_type_copy): Likewise.
26822 (build_variant_type_copy): Likewise.
26823 * tree.c (build_qualified_type): Pass down mem-stat info.
26824 (build_distinct_type_copy): Likewise.
26825 (build_variant_type_copy): Likewise.
26827 2017-04-21 Richard Biener <rguenther@suse.de>
26829 PR tree-optimization/80237
26830 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26832 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26833 for a simplified result.
26835 2016-04-21 Richard Biener <rguenther@suse.de>
26837 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26838 sth as strict as a simple_iv but a chrec without symbols and an
26839 operand defined in the loop we are peeling (and not some subloop).
26840 (propagate_constants_for_unrolling): Propagate all constants.
26842 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26845 * config/i386/i386.c (print_reg): Remove assert for disalowed
26846 regno values, call output_operand_lossage instead.
26848 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26851 * config/i386/constraints.md (Yc): New register constraint.
26852 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26853 Use Yc constraint for alternative 2 of operand 0. Remove
26854 preferred_for_speed attribute.
26856 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26858 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26859 lastprivate clauses in SIMT case.
26861 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26863 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26865 2017-04-20 Richard Biener <rguenther@suse.de>
26867 PR tree-optimization/57796
26868 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26869 as N scalar stores.
26870 (vect_model_load_cost): Cost gathers as N scalar loads.
26872 2017-04-20 Richard Biener <rguenther@suse.de>
26874 * ggc-page.c (ggc_allocated_p): Rename to ...
26875 (safe_lookup_page_table_entry): ... this and return the lookup
26877 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26879 2017-04-20 Richard Biener <rguenther@suse.de>
26881 PR tree-optimization/80453
26882 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26883 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26884 from the conditions.
26885 (vn_phi_eq): Pass them down.
26886 (vn_phi_lookup): Record them.
26887 (vn_phi_insert): Likewise.
26889 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26891 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26892 uninitialized variable warning to avoid buffer overrun.
26894 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26897 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26898 is suppressed for '{ 0 }' in C.
26900 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26902 * BASE-VER: Set to 8.0.0.
26904 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26906 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26907 priority .init_array and .fini_array section with SECTION_NOTYPE
26910 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26912 PR middle-end/80423
26913 * tree.h (build_array_type): Add typeless_storage default argument.
26914 * tree.c (type_cache_hasher::equal): Also compare
26915 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26916 (build_array_type): Add typeless_storage argument, set
26917 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26919 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26920 (build_array_type): Likewise. Add typeless_storage argument.
26922 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26923 Jakub Jelinek <jakub@redhat.com>
26925 PR tree-optimization/80426
26926 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26927 operation on symbolic operands, also compute the overflow for the
26928 invariant part when the operation degenerates into a negation.
26930 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26933 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26934 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26937 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26939 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26942 * config/avr/avr.c (tree.h): Include it.
26943 (cgraph.h): Include it.
26944 (avr_encode_section_info): Don't warn for uninitialized progmem
26945 variable if it's just an alias.
26947 2017-04-19 Richard Biener <rguenther@suse.de>
26950 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26951 when needed by AutoPGO.
26953 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26956 * doc/lto.texi: Remove an extra 'that'.
26958 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26960 PR rtl-optimization/80429
26961 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26962 are only used in debug insns.
26964 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26965 Vladimir Makarov <vmakarov@redhat.com>
26967 * config/sparc/predicates.md (input_operand): Add comment. Return
26968 true for any memory operand when LRA is in progress.
26969 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26971 2017-04-18 Jeff Law <law@redhat.com>
26974 * mips.md ({return,simple_return}_internal): Do not overwrite
26977 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26979 PR tree-optimization/80443
26980 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26981 instead of adding 1, subtract -1 and similarly instead of subtracting
26984 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26986 PR rtl-optimization/80357
26987 * haifa-sched.c (tmp_bitmap): New variable.
26988 (model_recompute): Handle duplicate use records.
26989 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26990 (free_global_sched_pressure_data): Free it.
26992 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26995 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26996 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26997 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26998 instead of SYSTEM_HEADER_DIR.
27000 2017-04-18 Jeff Law <law@redhat.com>
27002 PR middle-end/80422
27003 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27004 predecessors after walking up the insn chain.
27006 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27009 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27010 sizetype type into debug info.
27012 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27015 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27016 unneeded test for TARGET_UPPER_REGS_SF.
27017 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27019 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27022 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27023 instead of gsi_after_labels.
27025 2017-04-18 Jeff Law <law@redhat.com>
27027 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27031 2017-04-13 Jeff Law <law@redhat.com>
27032 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27033 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27035 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27038 * config/avr/avr.c (intl.h): Include it.
27039 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27041 2017-04-18 Martin Liska <mliska@suse.cz>
27043 PR gcov-profile/78783
27044 * gcov-tool.c (gcov_output_files): Validate that destination
27045 file is either removed by the tool or by a user.
27047 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27048 Guy Benyei <guybe@mellanox.com>
27050 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27051 block, and do not negate it, the stored id is already negative.
27053 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27055 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27057 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27060 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27061 masks of options that should be turned off if the VSX vector
27062 options are turned off.
27063 (OTHER_P8_VECTOR_MASKS): Likewise.
27064 (OTHER_VSX_VECTOR_MASKS): Likewise.
27065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27066 rs6000_disable_incompatible_switches to validate no type switches
27068 (rs6000_incompatible_switch): New function to disallow turning on
27069 other vector options if -mno-vsx, -mno-power8-vector, or
27070 -mno-power9-vector are specified.
27072 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27074 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27076 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27078 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27079 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27080 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27081 (ARG_POINTER_CFA_OFFSET): Likewise.
27083 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27085 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27086 conditions to take advantage of various optimizations.
27088 2017-04-13 Jeff Law <law@redhat.com>
27090 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27091 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27092 (zero_extendsidi2_dext): Likewise.
27094 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27097 * fold-const.c (fold_ternary_loc): Revert
27098 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27101 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27103 PR rtl-optimization/80343
27104 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27105 new scratch pseudo.
27107 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27110 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27111 to ubsan_encode_value.
27113 2017-04-13 Jeff Law <law@redhat.com>
27115 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27116 appearing in DEBUG_INSNs.
27118 2017-04-13 Martin Liska <mliska@suse.cz>
27120 PR gcov-profile/80413
27121 * gcov-io.c (gcov_write_string): Copy to buffer just when
27122 allocated size is greater than zero.
27124 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27127 * dwarf2out.c (decls_for_scope): Ignore declarations of
27128 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27130 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27133 * ipa-visibility.c (non_local_p): Fix typos.
27134 (localize_node): When localizing symbol in same comdat group,
27135 dissolve the group only when we know external symbols are going
27137 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27139 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27141 PR tree-optimization/79390
27142 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27143 order does not result in usable sequence, retry with reversed operand
27149 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27150 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27151 op0 instead of fold_convert_loc (loc, type, arg0).
27153 2017-04-12 Jeff Law <law@redhat.com>
27155 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27156 has a delay slot in the generated code.
27158 * config/cris/cris.md (cris_preferred_reload_class): Return
27159 GENNONACR_REGS rather than GENERAL_REGS.
27161 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27164 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27165 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27166 signedness of the result type.
27168 2017-04-12 Richard Biener <rguenther@suse.de>
27169 Jeff Law <law@redhat.com>
27171 PR tree-optimization/80359
27172 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27173 trim stores to TARGET_MEM_REFs.
27175 2017-04-12 Richard Biener <rguenther@suse.de>
27177 PR tree-optimization/79390
27178 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27179 threading case even more.
27181 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27184 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27185 for quad_address_p for TImode, instead of just not indexed_address.
27187 2017-04-12 Richard Biener <rguenther@suse.de>
27188 Bernd Edlinger <bernd.edlinger@hotmail.de>
27190 PR middle-end/79671
27191 * alias.c (component_uses_parent_alias_set_from): Handle
27192 TYPE_TYPELESS_STORAGE.
27193 (get_alias_set): Likewise.
27194 * tree-core.h (tree_type_common): Add typeless_storage flag.
27195 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27196 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27197 for types containing members with TYPE_TYPELESS_STORAGE.
27198 (place_field): Likewise.
27199 (layout_type): Likewise for ARRAY_TYPE.
27200 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27201 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27202 TYPE_TYPELESS_STORAGE.
27203 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27205 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27208 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27209 first argument to type.
27211 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27215 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27216 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27217 (rs6000_expand_binop_builtin): Likewise.
27218 (rs6000_expand_ternop_builtin): Likewise; also add missing
27219 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27220 vshasigma built-ins.
27221 * doc/extend.texi: Document that vec_xxpermdi's third argument
27222 must be a constant.
27224 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27226 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27227 Use shift_const cost parameter when calculating gain of STV shifts.
27229 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27231 PR rtl-optimization/70478
27232 * lra-constraints.c (process_alt_operands): Check memory for
27233 disfavoring memory insn operand.
27235 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27237 PR middle-end/80100
27238 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27239 left shift in unsigned HOST_WIDE_INT type.
27241 PR rtl-optimization/80385
27242 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27243 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27246 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27247 if they have any depend clauses.
27249 2017-04-11 Martin Liska <mliska@suse.cz>
27252 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27253 * ipa-split.c (split_function): Create a local comdat symbol
27254 if caller is in a comdat group.
27256 2017-04-11 Martin Liska <mliska@suse.cz>
27259 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27262 2017-04-11 Martin Sebor <msebor@redhat.com>
27264 PR middle-end/80364
27265 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27266 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27268 (directive::set_width, directive::set_precision, format_character):
27270 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27273 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27276 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27277 conflict, set target->arch_name instead of target->cpu_name.
27279 2017-04-11 Richard Biener <rguenther@suse.de>
27281 PR tree-optimization/80374
27282 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27283 build_zero_cst, remove fold_convertible_p check again.
27285 2017-04-11 Martin Liska <mliska@suse.cz>
27288 * ubsan.c (instrument_object_size): Do not instrument register
27291 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27294 * config/i386/i386-builtin-types.def
27295 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27296 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27297 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27298 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27299 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27300 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27301 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27302 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27303 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27304 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27305 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27306 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27307 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27308 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27309 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27310 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27311 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27312 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27313 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27314 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27315 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27316 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27317 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27318 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27319 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27320 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27321 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27322 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27323 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27324 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27325 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27326 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27327 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27328 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27329 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27330 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27331 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27332 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27333 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27334 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27335 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27336 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27337 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27338 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27339 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27340 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27342 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27343 flag to second_arg_count, handle 4 argument function type _COUNT
27344 aliases, handle second_arg_count on second argument rather than last.
27346 2017-04-10 Jeff Law <law@redhat.com>
27348 PR tree-optimization/80374
27349 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27350 record anything if we can not convert integer_zero_node to the
27353 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27356 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27357 Enhance special handling given to the TARGET_P9_MINMAX option in
27358 relation to certain other options.
27360 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27362 PR tree-optimization/80153
27363 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27364 remove POINTER_PLUS_EXPR's base part directly, rather than through
27367 2017-04-10 Richard Biener <rguenther@suse.de>
27368 Bin Cheng <bin.cheng@arm.com>
27370 PR tree-optimization/80153
27371 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27372 the first element of pointer type aff_tree. Build result expr in
27374 (add_elt_to_tree): Convert to type unconditionally. Remove other
27375 fold_convert calls.
27376 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27377 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27379 2017-04-10 Richard Biener <rguenther@suse.de>
27381 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27384 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27386 PR rtl-optimization/70478
27387 * lra-constraints.c (curr_small_class_check): New.
27388 (update_and_check_small_class_inputs): New.
27389 (process_alt_operands): Update curr_small_class_check. Disfavor
27390 alternative insn memory operands. Check available regs for small
27393 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27396 * config/mips/mips.opt (-mvirt): Update description.
27397 * doc/invoke.texi (-mvirt): Likewise.
27399 2017-04-10 Richard Biener <rguenther@suse.de>
27401 PR middle-end/80362
27402 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27403 looking for NEGATE_EXPR in -A / -B to A / B folding.
27405 2017-04-10 Martin Liska <mliska@suse.cz>
27407 PR gcov-profile/80224
27408 * gcov.c (print_usage): Fix usage string.
27409 (get_gcov_intermediate_filename): Remove.
27410 (output_gcov_file): Use both for normal and intermediate format.
27411 (generate_results): Do not initialize special file for
27412 intermediate format.
27414 2017-04-10 Richard Biener <rguenther@suse.de>
27416 PR tree-optimization/80304
27417 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27420 2017-04-10 Nathan Sidwell <nathan@acm.org>
27423 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27424 (rs6000_init_builtins): Use it.
27426 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27428 * config/arm/arm.md (<mrc>): Add mode to SET source.
27429 (<mrrc>): Likewise.
27431 2017-04-10 Richard Biener <rguenther@suse.de>
27433 PR middle-end/80344
27434 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27436 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27439 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27440 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27441 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27442 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27443 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27444 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27445 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27446 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27447 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27448 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27449 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27450 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27451 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27452 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27453 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27454 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27455 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27456 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27457 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27458 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27459 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27460 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27461 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27463 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27465 PR rtl-optimization/70478
27466 * lra-constraints.c: Reverse the last patch.
27468 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27470 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27471 Add comment for WCHAR_T.
27473 2017-04-08 Martin Liska <mliska@suse.cz>
27476 2017-04-07 Martin Liska <mliska@suse.cz>
27479 * ipa-split.c (split_function): Add function part to a same comdat
27482 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27485 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27487 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27489 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27490 (vec_pairing): ...to this.
27491 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27492 (rs6000_sched_init): Adjust for name change.
27493 (struct rs6000_sched_context): Likewise.
27494 (rs6000_init_sched_context): Likewise.
27495 (rs6000_set_sched_context): Likewise.
27497 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27503 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27505 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27506 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27507 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27509 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27511 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27513 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27515 PR rtl-optimization/70703
27516 * ira-color.c (update_conflict_hard_regno_costs): Use
27517 int64_t instead of HOST_WIDE_INT.
27519 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27521 PR rtl-optimization/70478
27522 * lra-constraints.c (process_alt_operands): Disfavor alternative
27523 insn memory operands.
27525 2017-04-07 Jeff Law <law@redhat.com>
27527 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27528 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27530 2017-04-07 Martin Liska <mliska@suse.cz>
27533 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27534 Show error message instead of an ICE.
27536 2017-04-07 Martin Liska <mliska@suse.cz>
27539 * ipa-split.c (split_function): Add function part to a same comdat
27542 2017-04-07 Richard Biener <rguenther@suse.de>
27544 PR middle-end/80341
27545 * tree.c (get_unwidened): Also handle ! for_type case for
27547 * convert.c (do_narrow): Split out from ...
27548 (convert_to_integer_1): ... here. Do not pass final truncation
27549 type to get_unwidened for TRUNC_DIV_EXPR.
27551 2017-04-07 Richard Biener <rguenther@suse.de>
27553 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27555 (aff_combination_const): Adjust.
27556 (aff_combination_scale): Likewise.
27557 (aff_combination_add_elt): Likewise.
27558 (aff_combination_add_cst): Likewise.
27559 (aff_combination_convert): Likewise.
27560 (add_elt_to_tree): Likewise. Remove unused argument.
27561 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27563 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27565 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27567 * config/arm/arm.c (arm_default_short_enums): Use
27568 ARM_DEFAULT_SHORT_ENUMS.
27569 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27571 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27574 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27575 members with redundant out-of-class redeclaration.
27577 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27580 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27581 * config/i386/i386.md (*zero_extendsidi2):
27582 Add (?*x,*x) and (?*v,*v) alternatives.
27584 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27587 * config/i386/i386.c (ix86_expand_builtin)
27588 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27589 mode from insn data. Convert operands to insn operand mode.
27590 Copy operands that don't satisfy insn predicate to a register.
27592 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27594 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27597 2017-04-06 Richard Biener <rguenther@suse.de>
27599 PR tree-optimization/80334
27600 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27601 preserve alignment of accesses.
27603 2017-04-06 Richard Biener <rguenther@suse.de>
27605 PR tree-optimization/80262
27606 * tree-sra.c (build_ref_for_offset): Preserve address-space
27608 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27609 Drop useless address-space information on MEM_REF offsets.
27611 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27613 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27615 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27617 PR rtl-optimization/70703
27618 * ira-color.c (update_conflict_hard_regno_costs): Use
27619 HOST_WIDE_INT instead of long.
27621 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27624 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27625 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27627 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27628 for x86_64 target. Handle -m3dnowa option.
27630 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27632 PR rtl-optimization/70703
27633 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27634 (update_conflict_hard_regno_costs): Use long instead of unsigned
27635 arithmetic for cost calculation.
27637 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27638 Bernd Edlinger <bernd.edlinger@hotmail.de>
27641 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27644 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27647 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27648 ptr_mode with Pmode throughout.
27649 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27650 into probe_stack_range and use DImode.
27652 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27655 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27656 call_eh_return is true.
27658 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27660 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27661 Initialize last_match_fntype_index.
27663 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27666 * tree-nvr.c: Include internal-fn.h.
27667 (pass_return_slot::execute): Ignore internal calls without
27670 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27671 Richard Biener <rguenther@suse.de>
27674 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27675 captures used multiple times, except for the last use.
27676 * generic-match-head.c: Include gimplify.h.
27678 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27680 PR tree-optimization/79390
27681 * target.h (struct noce_if_info): Declare.
27682 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27683 * target.def (noce_conversion_profitable_p): New target hook.
27684 * ifcvt.h (struct noce_if_info): New type, moved from ...
27685 * ifcvt.c (struct noce_if_info): ... here.
27686 (noce_conversion_profitable_p): Renamed to ...
27687 (default_noce_conversion_profitable_p): ... this. No longer
27689 (noce_try_store_flag_constants, noce_try_addcc,
27690 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27691 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27692 instead of noce_conversion_profitable_p.
27693 * config/i386/i386.c: Include ifcvt.h.
27694 (ix86_option_override_internal): Don't override
27695 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27696 (ix86_noce_conversion_profitable_p): New function.
27697 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27698 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27699 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27700 * doc/tm.texi: Regenerated.
27702 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27704 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27707 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27710 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27711 instructions for small multiply cores.
27713 2017-04-04 Jeff Law <law@redhat.com>
27715 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27717 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27718 that are not set by the loop over the elements.
27720 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27723 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27724 int mode, convert_modes it to mode as unsigned, otherwise use
27725 lowpart_subreg to mode rather than SImode.
27726 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27727 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27728 Use DImode instead of SImode for the shift count operand.
27729 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27732 2017-04-04 Richard Biener <rguenther@suse.de>
27734 PR middle-end/80281
27735 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27736 arithmetic done for the negate or the plus. Simplify.
27737 (A - (-B) -> A + B): Likewise.
27738 * fold-const.c (split_tree): Make sure to not negate pointers.
27740 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27742 PR rtl-optimization/60818
27743 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27744 a compare of comparisons with the thing compared if this results
27745 in a different machine mode.
27747 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27749 * alias.c (base_alias_check): Fix typo in comment.
27750 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27751 * cgraphunit.c (symbol_table::compile): Likewise.
27752 * collect2.c (maybe_run_lto_and_relink): Likewise.
27753 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27754 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27755 * config/avr/avr.c (avr_map_op_t): Likewise.
27756 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27757 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27758 * config/epiphany/epiphany.md (movcc): Likewise.
27759 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27760 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27762 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27763 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27764 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27765 * config/sh/sh.c (sh_rtx_costs): Likewise.
27766 * fold-const.c (fold_truth_andor): Likewise.
27767 * genautomata.c (collapse_flag): Likewise.
27768 * gengtype.h (struct type::u::s): Likewise.
27769 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27770 * input.c (FORMAT_AMOUNT): Likewise.
27771 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27772 (known_aggs_to_agg_replacement_list): Likewise.
27773 * ipa-inline-analysis.c: Likewise.
27774 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27775 * ipa-polymorphic-call.c
27776 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27777 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27778 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27780 * modulo-sched.c (apply_reg_moves): Likewise.
27781 * omp-expand.c (build_omp_regions_1): Likewise.
27782 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27783 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27784 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27785 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27786 * value-prof.c: Likewise.
27787 * var-tracking.c (val_reset): Likewise.
27789 2017-04-03 Richard Biener <rguenther@suse.de>
27791 PR tree-optimization/80275
27792 * fold-const.c (split_address_to_core_and_offset): Handle
27795 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27797 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27798 descriptors is at least equal to that of functions.
27800 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27802 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27804 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27807 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27808 (mov<IMOD4:mode>): New expander.
27809 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27811 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27813 PR rtl-optimization/79405
27814 * fwprop.c (propagations_left): New variable.
27815 (forward_propagate_into): Decrement it.
27816 (fwprop_init): Initialize it.
27817 (fw_prop): If the variable has reached zero, stop propagating.
27818 (fwprop_addr): Ditto.
27820 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27823 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27824 a FUNCTION_DECL, pass it as decl instead of origin to
27827 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27829 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27832 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27835 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27836 TARGET_VSX_SMALL_INTEGER.
27838 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27840 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27841 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27843 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27845 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27846 extraction from odd-numbered MSA register.
27848 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27850 PR middle-end/80173
27851 * expmed.c (store_bit_field_1): Don't attempt to create
27852 a word subreg out of hard registers wider than word if they
27853 have HARD_REGNO_NREGS of 1 for their mode.
27855 PR middle-end/80163
27856 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27857 conversions to integer types wider than word and pointer.
27860 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27861 (rtx_equal_for_cselib_p): Pass 0 to it.
27862 * cselib.c (cselib_hasher::equal): Likewise.
27863 (rtx_equal_for_cselib_1): Add depth argument. If depth
27864 is 128, don't look up VALUE locs and punt. Increment
27865 depth in recursive calls when walking VALUE locs.
27867 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27869 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27870 (make_gcov_file_name): Use the canonical path name for generating
27872 (read_line): Fix handling of files with ascii null bytes.
27874 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27876 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27877 to initialise a vector register instead
27878 of using a const_int.
27880 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27882 PR translation/80189
27883 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27884 diagnostic messages.
27886 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27889 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27890 (dfp_diex_<mode>): Update mode of operand 1.
27891 * doc/extend.texi (dxex, dxexq): Document change to return type.
27892 (diex, diexq): Document change to argument type.
27894 2017-03-30 Martin Jambor <mjambor@suse.cz>
27897 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27898 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27899 it reflects the signature changes performed at the callee side.
27900 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27901 to cgraph_build_function_type_skip_args.
27902 (build_function_decl_skip_args): Adjust call to the above function.
27904 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27907 * config/i386/sse.md
27908 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27909 register as dest whenever it is a MEM not rtx_equal_p to the
27910 corresponding dup operand, and when forcing into reg move the
27911 reg into the memory afterwards.
27912 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27913 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27914 for the force_reg mode.
27915 (avx512vl_vextractf128<mode>): Use register as dest either
27916 always when a MEM, or when it is a MEM not rtx_equal_p to the
27917 corresponding dup operand, or even not when it is a CONST_VECTOR
27918 depending on the mode and lo vs. hi.
27919 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27921 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27922 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27923 Likewise. Require that operands[2] is even.
27924 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27925 Remove extraneous parens. Require that operands[2] is a multiple
27927 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27928 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27929 disallow memory then.
27931 2017-03-30 Richard Biener <rguenther@suse.de>
27933 PR tree-optimization/77498
27934 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27935 to non-constants over backedges.
27937 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27939 PR rtl-optimization/80233
27940 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27941 as last_combined_insn. Do not test for BARRIER_P separately.
27943 2017-03-29 Andreas Schwab <schwab@suse.de>
27946 * calls.c (prepare_call_address): Convert funexp to Pmode before
27947 copying to temp reg.
27949 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27951 PR tree-optimization/80158
27952 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27953 Handle possible future case of more than one alternate
27955 (replace_rhs_if_not_dup): Likewise.
27956 (replace_one_candidate): Likewise.
27958 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27960 PR rtl-optimization/80193
27961 * ira.c (ira): Do not check allocation for LRA.
27963 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27965 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27966 (nvptx_output_simt_exit): Declare.
27967 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27968 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27969 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27970 (nvptx_declare_function_name): ...here. Emit declaration of local
27971 memory space buffer for omp_simt_enter insn.
27972 (nvptx_output_unisimt_switch): New.
27973 (nvptx_output_softstack_switch): New.
27974 (nvptx_output_simt_enter): New.
27975 (nvptx_output_simt_exit): New.
27976 * config/nvptx/nvptx.h (struct machine_function): New fields
27977 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27978 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27979 (UNSPECV_SIMT_EXIT): Ditto.
27980 (omp_simt_enter_insn): New insn.
27981 (omp_simt_enter): New expansion.
27982 (omp_simt_exit): New insn.
27983 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27985 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27986 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27987 (expand_GOMP_SIMT_EXIT): New.
27988 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27989 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27990 (GOMP_SIMT_EXIT): Ditto.
27991 * target-insns.def (omp_simt_enter): New insn.
27992 (omp_simt_exit): Ditto.
27993 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27995 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27996 (lower_rec_input_clauses): Likewise.
27997 (lower_lastprivate_clauses): Handle SIMT privatization.
27999 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28000 (ompdevlow_adjust_simt_enter): New.
28001 (find_simtpriv_var_op): New.
28002 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28003 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28005 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28006 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28007 (copy_decl_for_dup_finish): Ditto.
28009 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28011 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28014 * config/i386/i386.c (ix86_option_override_internal): Always
28015 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28017 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28019 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28021 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28023 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28024 mark new edge's irreducible flag accordign to it.
28025 (vect_do_peeling): Check loop preheader edge's irreducible flag
28026 and pass it to function slpeel_add_loop_guard.
28028 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28030 PR tree-optimization/80218
28031 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28032 Update block frequencies and counts.
28034 2017-03-28 Richard Biener <rguenther@suse.de>
28036 PR tree-optimization/78644
28037 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28038 of a simplification result we may not use it at all.
28040 2017-03-28 Richard Biener <rguenther@suse.de>
28043 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28044 without arguments, generate default definition of a SSA name.
28046 2017-03-28 Richard Biener <rguenther@suse.de>
28048 PR middle-end/80222
28049 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28050 TYPE_REF_CAN_ALIAS_ALL references.
28051 * fold-const.c (fold_indirect_ref_1): Likewise.
28053 2017-03-28 Martin Liska <mliska@suse.cz>
28056 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28057 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28059 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28060 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28062 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28063 (EXTRA_SPECS): Define.
28064 (SUBTARGET_EXTRA_SPECS): Likewise.
28065 (SUBTARGET_CPP_SPEC): Likewise.
28066 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28067 SUBTARGET_EXTRA_SPECS.
28068 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28070 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28072 * config/arc/simdext.md (vst64_insn): Update pattern.
28073 (vld32wh_insn): Likewise.
28074 (vld32wl_insn): Likewise.
28075 (vld64_insn): Likewise.
28076 (vld32_insn): Likewise.
28078 2017-03-28 Marek Polacek <polacek@redhat.com>
28081 * fold-const.c (fold_comparison): Use protected_set_expr_location
28082 instead of SET_EXPR_LOCATION.
28084 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28086 * tree.c (add_expr): Avoid name lookup warning.
28088 2017-03-27 Jeff Law <law@redhat.com>
28090 PR tree-optimization/80216
28091 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28092 function name. Limit recursion depth.
28093 (record_temporary_equivalences): Corresponding changes.
28095 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28097 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28100 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28103 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28105 * cfgcleanup.c (reg_note_cfa_p): New array.
28106 (insns_have_identical_cfa_notes): New function.
28107 (old_insns_match_p): Don't cross-jump in between /f
28108 and non-/f instructions. If both i1 and i2 are frame related,
28109 verify all CFA notes, their order and content.
28111 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28114 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28115 HImode and SImode with zero extend to DImode to one insn.
28116 (bswap<mode>2_extenddi): Likewise.
28117 (bswapsi2_extenddi): Likewise.
28118 (bswaphi2_extendsi): Likewise.
28119 (bswaphi2): Combine bswap HImode and SImode into one insn.
28120 Separate memory insns from swapping register.
28121 (bswapsi2): Likewise.
28122 (bswap<mode>2): Likewise.
28123 (bswaphi2_internal): Delete, no longer used.
28124 (bswapsi2_internal): Likewise.
28125 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28126 store, and gpr<-gpr swap insns.
28127 (bswap<mode>2_store): Likewise.
28128 (bswaphi2_reg): Register only splitter, combine with the splitter.
28129 (bswaphi2 splitter): Likewise.
28130 (bswapsi2_reg): Likewise.
28131 (bswapsi2 splitter): Likewise.
28132 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28133 the insns into load, store, and register/register insns.
28134 (bswapdi2_ldbrx): Likewise.
28135 (bswapdi2_load): Likewise.
28136 (bswapdi2_store): Likewise.
28137 (bswapdi2_reg): Likewise.
28139 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28141 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28142 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28144 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28147 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28149 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28150 special handling for target option conflicts between dform
28151 options (-mpower9-dform, -mpower9-dform-vector,
28152 -mpower9-dform-scalar) and -mno-direct-move.
28154 2017-03-27 Richard Biener <rguenther@suse.de>
28156 PR tree-optimization/80181
28157 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28159 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28161 * config/arc/predicates.md (move_double_src_operand): Replace the
28162 call to move_double_src_operand with a call to address_operand.
28164 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28166 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28167 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28168 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28170 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28172 * config/arc/predicates.md (long_immediate_loadstore_operand):
28173 Consider scaled addresses cases.
28175 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28177 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28178 restored when in interrupt.
28179 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28180 doesn't have delay slot.
28182 2017-03-27 Richard Biener <rguenther@suse.de>
28185 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28186 inlined thunk clones.
28188 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28191 * asan.c (instrument_derefs): Copy over last operand from
28192 original COMPONENT_REF to the new COMPONENT_REF with
28193 DECL_BIT_FIELD_REPRESENTATIVE.
28194 * ubsan.c (instrument_object_size): Likewise.
28196 2017-03-27 Richard Biener <rguenther@suse.de>
28198 PR tree-optimization/80170
28199 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28200 sure DR/SCEV didnt fold in constants we do not see when looking
28201 at the reference base alignment.
28203 2017-03-27 Richard Biener <rguenther@suse.de>
28205 PR middle-end/80171
28206 * gimple-fold.c (fold_ctor_reference): Properly guard against
28207 NULL return value from canonicalize_constructor_val.
28209 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28212 * config/i386/i386.c (ix86_expand_builtin)
28213 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28214 flags reg setting and flags reg using instructions.
28215 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28216 clobbering instructions to zero extend op2.
28218 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28220 * doc/install.texi (Configuration) <--with-aix-soname>:
28221 Update link to AIX ld.
28223 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28225 PR rtl-optimization/80160
28226 PR rtl-optimization/80159
28227 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28228 reg_alternate_class into account.
28230 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28233 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28234 to consider in curr_insn_transform.
28236 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28238 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28239 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28240 and emit_mode_inner.
28242 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28244 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28245 argument to the overloaded builtin variants. Use the new flag to
28246 deprecate certain builtin variants.
28247 * config/s390/s390-builtin-types.def: Add new builtin types.
28248 * config/s390/s390-builtins.h: Support new flags field for
28249 overloaded builtins.
28250 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28251 (s390_macro_to_expand): Enable vector float data type.
28252 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28253 builtins by incrementing the __VEC__ version number.
28254 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28256 (s390_resolve_overloaded_builtin): Emit error messages depending
28257 on the builtin flags.
28258 * config/s390/s390.c (s390_expand_builtin): Support additional
28259 flags argument. Change error message to match the messages
28260 emitted in s390-c.c.
28261 * config/s390/s390.md: New UNSPEC_* constants.
28262 (op_type): Add new instruction types.
28263 * config/s390/vecintrin.h: Add new builtins and test data class
28265 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28266 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28267 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28268 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28269 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28270 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28272 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28273 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28274 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28275 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28277 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28278 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28279 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28280 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28282 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28283 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28284 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28285 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28286 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28287 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28288 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28290 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28291 ("vec_scatter_element<V_HW_4:mode>_DI")
28292 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28293 ("vec_fpint<mode>", "vflls")
28294 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28295 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28296 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28297 ("*vec_cmphe<mode>_cc"): ... these.
28299 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28300 mode constant instead of magic value.
28302 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28304 * config/s390/s390.c (s390_expand_vec_compare): Support other
28305 vector floating point modes than just V2DF.
28306 (s390_expand_vcond): Likewise.
28307 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28308 (s390_cannot_change_mode_class): Prevent mode changes between TF
28309 and V1TF in vector registers.
28310 * config/s390/s390.md (DF, SF): New mode attributes.
28311 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28312 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28313 SFmode support for VRs.
28314 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28316 (VFT, VF_HW): New mode iterators.
28317 (vw, sdx): New mode attributes.
28318 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28319 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28320 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28321 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28322 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28323 also the new vector floating point modes. Renaming to ...
28325 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28326 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28327 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28328 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28329 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28330 ("vec_unordered<mode>"): ... these.
28332 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28333 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28334 ("*vec_extendv2df"): New insn definitions.
28336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28338 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28339 ("mulditi3_2", "*muldi3_sign"): New patterns.
28340 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28341 rename the pattern definition.
28343 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28345 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28347 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28351 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28352 instruction if possible.
28353 * config/s390/vector.md (vec_halfnumelts): New mode
28355 ("*vec_vllezlf<mode>"): New pattern.
28357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28359 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28360 ("popcountv4si2", "popcountv2di2"): Rename to ...
28361 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28362 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28364 ("popcount<mode>2_vxe"): New pattern.
28366 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28368 * common/config/s390/s390-common.c (processor_flags_table): Add
28370 * config.gcc: Add arch12.
28371 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28372 Default to arch12 for unknown CPU model numbers.
28373 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28374 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28375 PROCESSOR_max sanity check.
28376 * config/s390/s390-opts.h (enum processor_type): Add
28378 * config/s390/s390.c (processor_table): Add arch12.
28379 (s390_expand_builtin): Add check for B_VXE flag.
28380 (s390_issue_rate): Add PROCESSOR_ARCH12.
28381 (s390_get_sched_attrmask): Likewise.
28382 (s390_get_unit_mask): Likewise.
28383 (s390_sched_score): Enable z13 scheduling for arch12.
28384 (s390_sched_reorder): Likewise.
28385 (s390_sched_variable_issue): Likewise.
28386 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28388 (s390_tune_attr): Use z13 scheduling also for arch12.
28389 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28390 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28391 (TARGET_VXE_P): New macros.
28392 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28393 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28394 * config/s390/s390.opt: Add arch12 as processor_type.
28396 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28398 * config/s390/s390.md
28399 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28400 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28401 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28403 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28404 Rename expanders to ...
28406 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28407 ("fixuns_truncdddi2_emu"): ... these.
28409 ("fixuns_trunc<mode>si2_emu"): New expander.
28411 ("*fixuns_truncdfdi2_z13"): Rename to ...
28412 ("*fixuns_truncdfdi2_vx"): ... this.
28414 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28416 * config/s390/2964.md: Remove the single element vector compare
28417 instructions which are no longer used.
28418 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28420 (s390_canonicalize_comparison): Remove handling of DFmode
28422 (s390_expand_vec_compare_scalar): Remove function.
28423 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28424 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28426 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28428 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28430 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28432 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28433 will anyway by matched by mov<mode>_64dfp.
28435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28437 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28438 vlef/vstef. Add missing operand to vleif.
28440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28442 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28443 pair for all vector types with 64 bit elements.
28444 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28445 * config/s390/vector.md (V_HW_64): ... here.
28446 (V_128_NOSINGLE): New mode iterator.
28447 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28448 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28449 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28450 ("*vec_load_pairv2di"): Change to ...
28451 ("*vec_load_pair<mode>"): ... this one.
28453 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28455 * config/s390/constraints.md: Add comments.
28456 (jKK): Reject element sizes > 8 bytes.
28457 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28459 * config/s390/s390.md: Add the s_operand checks formerly in
28460 s390_split_ok_p to various splitters where they are still
28462 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28463 for 128 bit vectors. Plus two splitters.
28465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28467 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28470 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28473 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28474 error if the boundary argument is not constant.
28476 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28478 PR rtl-optimization/80112
28479 * loop-doloop.c (doloop_condition_get): Don't check condition
28480 if cmp isn't SET with IF_THEN_ELSE src.
28482 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28484 PR tree-optimization/80158
28485 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28486 replacing a candidate statement, also replace it for the
28487 candidate's alternate interpretation.
28488 (replace_rhs_if_not_dup): Likewise.
28489 (replace_one_candidate): Likewise.
28491 2017-03-24 Richard Biener <rguenther@suse.de>
28493 PR tree-optimization/80167
28494 * graphite-isl-ast-to-gimple.c
28495 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28497 (translate_isl_ast_to_gimple::get_rename): Likewise.
28499 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28502 handling of certain combinations of target options, including the
28503 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28504 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28506 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28509 * config/arm/arm.md (*load_multiple): Add reload_completed to
28510 matching condition.
28512 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28513 Richard Biener <rguenth@suse.de>
28515 PR tree-optimization/79908
28516 PR tree-optimization/80136
28517 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28518 been cast away, gimplify_and_add suffices.
28520 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28522 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28524 2017-03-23 Richard Biener <rguenther@suse.de>
28526 PR tree-optimization/80032
28527 * gimplify.c (gimple_push_cleanup): Forced unconditional
28528 cleanups still have to go to the conditional_cleanups
28531 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28533 PR tree-optimization/80072
28534 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28536 (next_operand_entry_id): Change type to unsigned int.
28537 (sort_by_operand_rank): Make sure to return the right return value
28538 even if unsigned fields are bigger than INT_MAX.
28539 (struct oecount): Change cnt and id type to unsigned int.
28540 (oecount_hasher::equal): Formatting fix.
28541 (oecount_cmp): Make sure to return the right return value
28542 even if unsigned fields are bigger than INT_MAX.
28543 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28546 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28547 TREE_READONLY on result if writing it more than once.
28550 * doc/invoke.texi (-fsanitize=thread): Document that with
28551 -fnon-call-exceptions atomics are not able to throw
28555 * tsan.c: Include tree-eh.h.
28556 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28557 maybe_clean_or_replace_eh_stmt where needed.
28558 (instrument_memory_accesses): Add cfg_changed argument.
28559 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28560 if it returned true.
28561 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28563 PR rtl-optimization/63191
28564 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28565 wrapper function, moved the whole old content into ...
28566 (ix86_delegitimize_address_1): ... this. New inline function.
28567 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28568 true as last argument instead of ix86_delegitimize_address.
28570 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28572 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28573 cortexa57_branch_cost.
28575 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28577 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28579 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28582 * doc/md.texi (Constraints): Document wA constraint.
28583 * config/rs6000/constraints.md (wA): New.
28584 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28585 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28586 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28587 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28589 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28592 * gimplify.c (is_oacc_declared): New function.
28593 (oacc_default_clause): Use it to set default flags for acc declared
28594 variables inside parallel regions.
28595 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28596 declared variables.
28597 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28598 declare attribute to any decl as necessary.
28600 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28603 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28604 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28605 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28606 (arm_arch_lpae): This.
28607 * config/arm/arm.c (arm_arch7ve): Rename into ...
28608 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28609 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28612 2017-03-22 Martin Liska <mliska@suse.cz>
28615 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28616 error message instead of an ICE.
28618 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28620 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28622 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28624 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28626 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28629 2017-03-21 Martin Sebor <msebor@redhat.com>
28631 * doc/extend.texi: Use "cannot" instead of "can't."
28632 * doc/hostconfig.texi: Same.
28633 * doc/install.texi: Same.
28634 * doc/invoke.texi: Same.
28635 * doc/loop.texi: Same.
28636 * doc/md.texi: Same.
28637 * doc/objc.texi: Same.
28638 * doc/rtl.texi: Same.
28639 * doc/tm.texi: Same.
28640 * doc/tm.texi.in: Same.
28641 * doc/trouble.texi: Same.
28643 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28646 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28647 (collect_checksum_attributes): Set it.
28648 (die_checksum_ordered): Use it.
28650 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28652 PR tree-optimization/79908
28653 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28654 change: For a VA_ARG whose LHS has been cast away, use
28655 force_gimple_operand to construct the side effects.
28657 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28659 PR translation/80001
28660 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28661 more amenable to translation.
28662 (oacc_loop_auto_partitions): Likewise.
28664 2017-03-21 Marek Polacek <polacek@redhat.com>
28665 Martin Sebor <msebor@redhat.com>
28667 PR tree-optimization/80109
28668 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28669 on INTEGRAL_TYPE_P.
28671 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28672 Segher Boessenkool <segher@kernel.crashing.org>
28675 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28676 check reg_used_between_p between insn and one of succ or succ2
28677 depending on if succ is artificial insn not inserted into insn
28680 2017-03-21 Martin Liska <mliska@suse.cz>
28682 PR gcov-profile/80081
28683 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28684 * doc/gcc.texi: Include gcov-dump stuff.
28685 * doc/gcov-dump.texi: New file.
28687 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28689 PR rtl-optimization/79150
28690 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28691 conditional jump, if the jump is the last insn of the loop.
28693 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28694 Richard Biener <rguenth@suse.de>
28696 PR tree-optimization/79908
28697 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28698 been cast away, use force_gimple_operand to construct the side
28701 2017-03-21 Martin Liska <mliska@suse.cz>
28703 PR libfortran/79956
28704 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28707 2017-03-21 Brad Spengler <spender@grsecurity.net>
28710 * plugin.c (htab_hash_plugin): New function.
28711 (add_new_plugin): Use it and adjust.
28712 (parse_plugin_arg_opt): Adjust.
28713 (init_one_plugin): Likewise.
28715 2017-03-21 Richard Biener <rguenther@suse.de>
28717 PR tree-optimization/80032
28718 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28719 if set force the cleanup to happen unconditionally.
28720 (gimplify_target_expr): Push inserted clobbers with force_uncond
28721 to avoid them being removed by control-dependent DCE.
28723 2017-03-21 Richard Biener <rguenther@suse.de>
28725 PR tree-optimization/80122
28726 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28727 va_arg_pack_len when the inlined call stmt requires pack
28729 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28731 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28734 * tsan.c (instrument_builtin_call): If the memory model argument
28735 is not a constant, assume it is valid.
28738 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28741 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28743 PR rtl-optimization/79910
28744 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28745 if its dest is used by an insn before I2 (other than the combined
28746 insns themselves, which are properly handled already).
28748 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28751 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28753 * combine.c (record_used_regs): New static function.
28754 (try_combine): Handle situations where there is an additional
28755 instruction between I2 and I3 which needs to have a LOG_LINK
28759 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28761 * combine.c (try_combine): Delete redundant i1 test. Call
28762 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28764 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28767 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28768 alternatives 13/14.
28770 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28772 PR tree-optimization/80054
28773 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28774 the optimization if a PHI or any of its arguments is not dominated
28775 by the candidate's basis. Use gphi* rather than gimple* as
28777 (replace_profitable_candidates): Clean up a gimple* variable that
28778 should be a gphi* variable.
28780 2017-03-20 Martin Sebor <msebor@redhat.com>
28783 * doc/extend.texi (attribute constructor): Document present limitation.
28785 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28788 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28789 __POWER9_VECTOR__ #ifdef control, change template definition to
28790 use Power9-specific built-in function.
28791 (vec_any_eq): Likewise.
28792 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28793 to control outcomes from this test.
28794 (vector_ae_<mode>p): For VEC_F modes, likewise.
28796 2017-03-20 Ian Lance Taylor <iant@google.com>
28798 * config/i386/i386.c (ix86_function_regparm): Save an extra
28799 register for -fsplit-stack with DECL_STATIC_CHAIN.
28801 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28804 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28805 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28807 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28809 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28811 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28814 2017-03-20 Marek Polacek <polacek@redhat.com>
28817 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28819 2017-03-20 Richard Biener <rguenther@suse.de>
28821 PR tree-optimization/80113
28822 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28823 allocate extra SSA name for PHI def.
28824 (add_close_phis_to_outer_loops): Likewise.
28825 (add_close_phis_to_merge_points): Likewise.
28826 (copy_loop_close_phi_args): Likewise.
28827 (copy_cond_phi_nodes): Likewise.
28829 2017-03-20 Martin Liska <mliska@suse.cz>
28831 PR middle-end/79753
28832 * tree-chkp.c (chkp_build_returned_bound): Do not build
28833 returned bounds for a LHS that's not a BOUNDED_P type.
28835 2017-03-20 Martin Liska <mliska@suse.cz>
28839 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28840 COMPLEX_CST and VECTOR_CST.
28842 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28845 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28846 target operand. A new splitter adds the clobber statement in case
28847 the target operand is dead anyway.
28849 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28851 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28852 to age-old versions of binutils and glibc.
28854 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28856 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28858 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28860 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28862 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28864 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28865 requirement for binutils 2.13.
28867 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28869 * combine.c (try_combine): Delete redundant i1 test. Call
28870 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28872 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28874 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28875 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28877 <riscv64-*-elf>: Re-arrange section
28878 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28879 <riscv32-*-linux>: Likewise.
28880 <riscv64-*-elf>: Likewise
28881 <riscv64-*-linux>: Likewise.
28883 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28886 * aarch64.opt(verbose-cost-dump): Fix typo.
28888 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28891 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28892 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28894 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28896 * reload.c (find_reloads): When reloading a nonoffsettable address,
28897 use RELOAD_OTHER for it and its address reloads.
28899 PR rtl-optimization/79910
28900 * combine.c (record_used_regs): New static function.
28901 (try_combine): Handle situations where there is an additional
28902 instruction between I2 and I3 which needs to have a LOG_LINK
28905 2017-03-17 Jeff Law <law@redhat.com>
28907 PR tree-optimization/71437
28908 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28909 conditional in the hash table first.
28910 (vrp_dom_walker::before_dom_children): Extract condition from
28911 ASSERT_EXPR. Record condition, its inverion and any implied
28912 conditions as well.
28914 2017-03-17 Marek Polacek <polacek@redhat.com>
28915 Markus Trippelsdorf <markus@trippelsdorf.de>
28917 PR tree-optimization/80079
28918 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28921 2017-03-17 Richard Biener <rguenther@suse.de>
28923 PR middle-end/80075
28924 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28925 Properly verify the LHS before the RHS possibly claims to be
28927 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28930 2017-03-17 Martin Jambor <mjambor@suse.cz>
28932 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28933 (List of -O2 options): Likewise.
28934 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28937 2017-03-17 Tom de Vries <tom@codesourcery.com>
28939 * gcov-dump.c (print_usage): Print bug_report_url.
28941 2017-03-17 Richard Biener <rguenther@suse.de>
28943 PR middle-end/80050
28944 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28945 (parser::peek): Likewise.
28947 2017-03-17 Richard Biener <rguenther@suse.de>
28949 PR tree-optimization/80048
28950 * sese.c (free_sese_info): Properly release rename_map and
28951 copied_bb_map elements.
28953 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28955 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28956 Add linked-list forward and backlinks. Insert on
28957 construction, remove on destruction.
28958 (class pass_store_merging): Add m_stores_head field.
28959 (pass_store_merging::terminate_and_process_all_chains):
28960 Iterate over m_stores_head list.
28961 (pass_store_merging::terminate_all_aliasing_chains):
28963 (pass_store_merging::execute): Check for debug stmts first.
28964 Push new chains onto the m_stores_head stack.
28966 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28969 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28970 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28971 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28973 2017-03-16 Jeff Law <law@redhat.com>
28975 PR tree-optimization/71437
28976 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28977 member function. Implementation moved into after_dom_children
28978 member function and into the threader's thread_outgoing_edges
28980 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28981 some code into new thread_outgoing_edges.
28982 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28983 definition. Simplify marker handling (do it here). Assume we always
28984 have the available expression and the const/copies tables.
28985 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28987 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28988 * tree-vrp.c (equiv_stack): No longer file scoped.
28989 (vrp_dom_walker): New class.
28990 (vrp_dom_walker::before_dom_children): New member function.
28991 (vrp_dom_walker::after_dom_children): Likewise.
28992 (identify_jump_threads): Setup domwalker. Use it rather than
28993 walking edges in a random order by hand. Simplify setup/finalization.
28994 (finalize_jump_threads): Remove.
28995 (vrp_finalize): Do not call identify_jump_threads here.
28996 (execute_vrp): Do it here instead and call thread_through_all_blocks
28999 PR tree-optimization/71437
29000 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29002 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29004 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29005 (dom_opt_dom_walker::thread_across_edge): Remove
29006 handle_dominating_asserts argument. All callers changed.
29007 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29008 changes. Remove calls to lhs_of_dominating_assert. Other
29009 uses of handle_dominating_asserts turn into unconditional code
29010 (simplify_control_stmt_condition_1): Likewise.
29011 (simplify_control_stmt_condition): Likewise.
29012 (thread_through_normal_block, thread_across_edge): Likewise.
29013 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29014 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29015 object if it is not an SSA_NAME.
29016 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29017 before calling into the VRP specific simplifiers.
29018 (identify_jump_threads): Remove handle_dominating_asserts
29021 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29024 * tree-diagnostic.c (default_tree_printer): No longer static.
29025 * tree-diagnostic.h (default_tree_printer): New prototype.
29027 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29029 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29030 Change ins into fmov.
29032 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29034 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29035 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29036 Use h_con constraint for operand 1.
29037 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29038 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29040 2017-03-15 Jeff Law <law@redhat.com>
29042 PR tree-optimization/71437
29043 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29044 (record_temporary_equivalences): Use it.
29046 PR tree-optimization/71437
29047 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29048 tree-ssa-scopedtables.
29049 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29050 (record_conditions, record_cond, vuse_eq): Likewise.
29051 (record_edge_info): Adjust to API tweak of record_conditions.
29052 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29053 (record_temporary_equivalences, optimize_stmt): Likewise.
29054 (eliminate_redundant_computations): Likewise.
29055 (record_equivalences_from_stmt): Likewise.
29056 * tree-ssa-scopedtables.c: Include options.h and params.h.
29057 (vuse_eq): New function, moved from tree-ssa-dom.c
29058 (build_and_record_new_cond): Likewise.
29059 (record_conditions): Likewise. Accept vector of conditions rather
29060 than edge_equivalence structure for first argument.
29061 for the first argument.
29062 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29063 from tree-ssa-dom.c.
29064 (avail_exprs_stack::record_cond): Likewise.
29065 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29066 from tree-ssa-dom.c.
29067 (avail_exprs_stack): Add new member functions lookup_avail_expr
29069 (record_conditions): Declare.
29071 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29074 * lra-constraints.c (process_alt_operands): Increase reject for
29075 reloading an input/output operand.
29077 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29080 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29081 insns to convert from signed/unsigned char/short to IEEE 128-bit
29083 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29085 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29088 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29089 subreg of inner mode for values already in registers.
29091 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29093 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29094 iteration reg is used after the loop.
29096 2017-03-14 Martin Sebor <msebor@redhat.com>
29098 PR tree-optimization/79800
29099 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29100 precision in negative-positive range.
29101 (format_floating): Call non-const overload with adjusted precision.
29103 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29106 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29109 2017-03-14 Martin Sebor <msebor@redhat.com>
29111 PR middle-end/80020
29112 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29113 * builtins.def (aligned_alloc): Use it.
29116 * Makefile.in (GTFILES): Add calls.c.
29117 * calls.c: Include "gt-calls.h".
29119 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29121 PR rtl-optimization/79728
29122 * regs.h (struct target_regs): New field
29123 x_contains_allocatable_regs_of_mode.
29124 (contains_allocatable_regs_of_mode): New macro.
29125 * reginfo.c (init_reg_sets_1): Initialize it, and change
29126 contains_reg_of_mode so it includes global regs as well.
29127 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29128 rather than contains_regs_of_mode.
29130 2017-03-14 Martin Liska <mliska@suse.cz>
29132 * doc/invoke.texi: Document options that can't be combined with
29133 -fcheck-pointer-bounds.
29135 2017-03-14 Martin Liska <mliska@suse.cz>
29137 PR middle-end/79831
29138 * doc/invoke.texi (-Wchkp): Document the option.
29140 2017-03-14 Martin Liska <mliska@suse.cz>
29142 * Makefile.in: Install gcov-dump.
29144 2017-03-14 Martin Liska <mliska@suse.cz>
29146 * multiple_target.c (expand_target_clones): Bail out for
29147 an invalid attribute.
29149 2017-03-14 Richard Biener <rguenther@suse.de>
29151 * alias.c (struct alias_set_entry): Pack properly.
29152 * cfgloop.h (struct loop): Likewise.
29153 * cse.c (struct set): Likewise.
29154 * ipa-utils.c (struct searchc_env): Likewise.
29155 * loop-invariant.c (struct invariant): Likewise.
29156 * lra-remat.c (struct cand): Likewise.
29157 * recog.c (struct change_t): Likewise.
29158 * rtl.h (struct address_info): Likewise.
29159 * symbol-summary.h (function_summary): Likewise.
29160 * tree-loop-distribution.c (struct partition): Likewise.
29161 * tree-object-size.c (struct object_size_info): Likewise.
29162 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29163 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29164 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29165 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29166 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29167 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29168 (struct _stmt_vec_info): Likewise.
29170 2017-03-14 Martin Liska <mliska@suse.cz>
29173 * multiple_target.c (create_dispatcher_calls): Check that
29174 a target can create a function dispatcher.
29176 2017-03-14 Martin Liska <mliska@suse.cz>
29179 * multiple_target.c (expand_target_clones): Drop local.local
29180 flag for default implementation.
29182 2017-03-14 Richard Biener <rguenther@suse.de>
29184 PR tree-optimization/80030
29185 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29187 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29189 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29190 gcc_fallthrough() instead of __attribute__((fallthrough));
29192 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29194 * doc/gcc.texi: Remove "up" link to (DIR).
29195 * doc/gccint.texi: Ditto.
29197 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29199 * doc/install.texi (Specific) <avr>: Remove reference to
29202 2017-03-13 Jeff Law <law@redhat.com>
29204 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29205 attribute rather than comments.
29207 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29208 match_scratch operand is highest.
29210 2017-03-13 Martin Liska <mliska@suse.cz>
29212 PR middle-end/78339
29213 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29214 is a CHKP clone, use original declaration.
29216 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29218 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29219 (arc_conditional_register_usage): Use a different allocation order
29220 when optimizing for size.
29221 * common/config/arc/arc-common.c (arc_option_optimization_table):
29222 Section anchors default on when optimizing for size.
29224 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29226 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29228 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29230 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29231 * config/arc/arc.md (cpu_facility): Add cd variant.
29232 (*movqi_insn): Add code density variant.
29233 (*movhi_insn): Likewise.
29234 (*movqi_insn): Likewise.
29235 (*addsi3_mixed): Likewise.
29236 (subsi3_insn): Likewise.
29238 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29240 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29242 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29244 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29245 expressions with MINUS and UNARY ops.
29247 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29250 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29252 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29253 between vec_select and vector argument.
29254 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29255 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29256 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29257 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29258 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29259 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29261 2017-03-13 Richard Biener <rguenther@suse.de>
29264 * params.def (vect-max-peeling-for-alignment): Fix typo.
29266 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29268 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29269 issue that only occurred with binutils below 2.18.
29271 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29273 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29274 refer to binutils 2.11/2.12 minimum.
29276 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29278 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29279 ftp.kernel.org and simplify binutils requirement.
29281 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29283 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29285 (Optimize Options): Ditto. Also remove redundancy.
29287 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29289 PR translation/79848
29290 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29292 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29293 to G_ to avoid double translation.
29295 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29297 PR translation/79923
29298 * auto-profile.c (get_combined_location): Convert leading
29299 character of diagnostics to lower case and remove trailing period.
29300 (read_profile): Likewise for various diagnostics.
29301 * config/arm/arm.c (arm_option_override): Remove trailing period
29302 from various diagnostics.
29303 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29304 (msp430_expand_delay_cycles): Likewise.
29306 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29309 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29310 full command-line argument, rather than just "str".
29311 (aarch64_validate_march): Likewise.
29312 (aarch64_validate_mtune): Likewise.
29314 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29316 PR rtl-optimization/78911
29317 * lra-assigns.c (must_not_spill_p): New function.
29318 (spill_for): Use it.
29320 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29322 PR tree-optimization/79981
29323 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29324 ATOMIC_COMPARE_EXCHANGE ifn result.
29325 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29326 IFN_ATOMIC_COMPARE_EXCHANGE.
29328 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29331 * opts.c (parse_sanitizer_options): Add missing question mark to
29332 "did you mean" message.
29334 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29336 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29338 (VMULEUH_UNS): Likewise.
29339 (VMULOUB_UNS): Likewise.
29340 (VMULOUH_UNS): Likewise.
29341 * config/rs6000/rs6000.c (builtin_function_type): Remove
29342 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29344 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29347 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29348 x with result of extra_parsing_for_operand_code_0.
29349 (function_reader::extra_parsing_for_operand_code_0): Convert
29350 return type from void to rtx, returning x. When reading
29351 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29352 larger size containing struct block_symbol.
29354 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29357 -mfloat128-hardware without -m64.
29359 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29362 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29363 entries to the case statement that marks unsigned arguments to
29364 overloaded functions.
29366 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29369 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29371 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29374 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29375 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29377 2017-03-10 Martin Liska <mliska@suse.cz>
29381 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29382 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29385 2017-03-10 Olivier Hainque <hainque@adacore.com>
29387 * tree-switch-conversion (array_value_type): Start by resetting
29388 candidate type to it's main variant.
29390 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29392 PR rtl-optimization/79909
29393 * combine.c (try_combine): Use simplify_replace_rtx on individual
29394 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29395 of the whole CALL_INSN_FUNCTION_USAGE.
29397 PR tree-optimization/79972
29398 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29399 get_range_info on SSA_NAMEs. Formatting fixes.
29401 2017-03-10 Richard Biener <rguenther@suse.de>
29402 Jakub Jelinek <jakub@redhat.com>
29404 PR tree-optimization/77975
29405 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29406 edge to be constant.
29407 (get_val_for): For constant x return it. Formatting fix.
29408 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29409 would use the same bases as the current one.
29411 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29413 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29414 instead of vec_select for V1TImode.
29415 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29417 (VSX_LE_128): Add V1TI to this mode iterator.
29418 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29419 (*vsx_le_perm_store_<mode>): Likewise.
29420 (pre-reload splitter for VSX stores): Likewise.
29421 (post-reload splitter for VSX stores): Likewise.
29422 (*vsx_xxpermdi2_le_<mode>): Likewise.
29423 (*vsx_lxvd2x2_le_<mode>): Likewise.
29424 (*vsx_stxvd2x2_le_<mode>): Likewise.
29426 2017-03-09 Michael Eager <eager@eagercon.com>
29428 Correct failures with --enable-checking=yes,rtl.
29430 * config/microblaze/microblaze.c (microblaze_expand_shift):
29431 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29432 test for const0_rtx.
29433 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29434 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29436 2017-03-09 Richard Biener <rguenther@suse.de>
29438 PR tree-optimization/79977
29439 * graphite-scop-detection.c (scop_detection::merge_sese):
29440 Handle the case of extra exits to blocks dominating the entry.
29442 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29444 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29447 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29449 PR rtl-optimization/79949
29450 * lra-constraints.c (process_alt_operands): Check memory when
29451 trying to predict a cycle. Print about the overall increase.
29453 2017-03-09 Richard Biener <rguenther@suse.de>
29455 PR middle-end/79971
29456 * gimple-expr.c (useless_type_conversion_p): Preserve
29457 TYPE_SATURATING for fixed-point types.
29459 2017-03-09 Richard Biener <rguenther@suse.de>
29462 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29463 alignment of BLKmode params.
29465 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29468 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29469 (VALL_NO_V2Q): Likewise.
29471 * config/aarch64/aarch64-simd.md
29472 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29474 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29475 VALL_NO_V2Q mode iterator.
29476 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29478 2017-03-09 Martin Liska <mliska@suse.cz>
29480 PR tree-optimization/79631
29481 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29482 tree_int_cst_sign_bit just for INTEGER constants.
29484 2017-03-09 Martin Liska <mliska@suse.cz>
29488 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29491 2017-03-09 Marek Polacek <polacek@redhat.com>
29494 * tree.c (inchash::add_expr): Handle TREE_VEC.
29496 2017-03-09 Martin Liska <mliska@suse.cz>
29499 (chkp_narrow_size_and_offset): New function.
29500 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29501 (void chkp_parse_bit_field_ref): New function.
29502 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29503 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29505 2017-03-09 Martin Liska <mliska@suse.cz>
29508 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29509 (chkp_find_bounds_1): Remove gcc_unreachable.
29511 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29514 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29515 BUILT_IN_SYNC*, determine the access type from the size suffix and
29516 always build a MEM_REF with that type. Handle forgotten
29517 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29520 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29521 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29522 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29523 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29524 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29525 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29526 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29527 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29528 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29529 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29530 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29531 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29532 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29533 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29534 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29535 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29536 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29537 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29538 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29539 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29540 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29541 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29542 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29543 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29544 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29545 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29546 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29547 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29548 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29549 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29550 definitions outside of __OPTIMIZE__ guarded section.
29553 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29554 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29555 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29556 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29559 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29561 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29562 ("vfenez<mode>"): Add missing constraints.
29564 2017-03-08 Martin Sebor <msebor@redhat.com>
29567 * config/nds32/nds32.c (nds32_option_override):
29568 Fix misspelled diagnostic.
29570 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29573 * gimplify.c (gimplify_omp_for): Replace index var in outer
29574 taskloop statement with an artificial variable and add
29575 OMP_CLAUSE_PRIVATE clause for it.
29577 2017-03-08 Richard Biener <rguenther@suse.de>
29579 PR tree-optimization/79955
29580 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29581 for accesses that are completely outside of the variable.
29583 2017-03-08 Andrew Haley <aph@redhat.com>
29585 PR tree-optimization/79943
29586 * tree-ssa-loop-split.c (compute_new_first_bound): When
29587 calculating the new upper bound, (END-BEG) should be added, not
29590 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29592 * config/avr/avr.md (setmemhi): Make sure match_dup
29593 operand number comes before match_scratch.
29595 2017-03-08 Richard Biener <rguenther@suse.de>
29597 PR tree-optimization/79920
29598 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29599 with ncopies == 1 to ...
29600 (vect_transform_slp_perm_load): ... here. Properly compute
29601 all element loads by iterating VF times over the group. Do
29602 not handle ncopies (computed in a broken way) in
29603 vect_create_mask_and_perm.
29605 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29608 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29609 is a uniform vector, use uniform_vector_p return value instead of
29610 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29612 2017-03-07 Marek Polacek <polacek@redhat.com>
29614 PR middle-end/79809
29615 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29616 (alloca_call_type): Likewise.
29618 2017-03-07 Martin Liska <mliska@suse.cz>
29620 * gcov.c (process_args): Put comment to correct location.
29622 2017-03-07 Martin Liska <mliska@suse.cz>
29624 PR middle-end/68270
29625 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29626 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29627 (chkp_narrow_bounds_for_field): Likewise.
29628 (chkp_parse_array_and_component_ref): Pass one more argument to
29631 2017-03-07 Richard Biener <rguenther@suse.de>
29633 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29636 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29638 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29639 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29641 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29644 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29645 to end of description.
29646 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29648 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29650 PR rtl-optimization/79901
29651 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29653 (*avx512f_<code><mode>3<mask_name>): ... this.
29654 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29655 iterator instead of VI8_AVX2_AVX512BW.
29657 PR rtl-optimization/79901
29658 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29659 min/max expander, expand it using expand_vec_cond_expr.
29662 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29665 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29668 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29669 to void * for PCH reasons.
29670 * dwarf2out.c (output_loc_operands, output_die): Cast
29671 v.val_vec.array to unsigned char *.
29673 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29676 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29679 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29681 PR rtl-optimization/79571
29682 * lra-constraints.c (process_alt_operands): Calculate static
29683 reject and subtract it from overall when only addresses will be
29686 2017-03-06 Julia Koval <julia.koval@intel.com>
29689 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29690 incoming stack boundary to 128 for 64-bit targets.
29692 2017-03-06 Richard Biener <rguenther@suse.de>
29694 PR tree-optimization/79894
29695 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29696 to NULL after folding it.
29698 2017-03-06 Richard Biener <rguenther@suse.de>
29700 PR tree-optimization/79824
29701 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29702 check disabling peeling for gaps.
29704 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29706 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29707 attributes): Document gettimeofday.
29709 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29711 * config/s390/s390.c (s390_option_override_internal): Set
29712 PARAM_MIN_VECT_LOOP_BOUND
29714 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29716 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29717 * config/s390/s390.md: Likewise.
29719 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29722 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29723 (<avx2_avx512>_perm<mode>): Rename to ...
29724 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29726 (<avx512>_perm<mode>_mask): Rename to ...
29727 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29729 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29730 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29731 instead of VI8F_256_512.
29732 (avx512f_perm<mode>): New define_expand.
29733 (avx512f_perm<mode>_mask): Likewise.
29734 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29735 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29737 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29739 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29740 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29742 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29744 2017-03-06 Martin Liska <mliska@suse.cz>
29747 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29748 when having a SSA NAME w/o VAR_DECL assigned to it.
29750 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29752 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29753 msa_dpsub_<su>_d): Fix MODE for vec_select.
29755 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29757 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29759 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29761 2017-03-06 Richard Biener <rguenther@suse.de>
29763 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29764 * plugin.c (register_plugin_info): Likewise.
29765 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29767 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29769 * config/i386/sse.md (sse_storehps, sse_storelps,
29770 avx_<castmode><avxsizesuffix>_<castmode>,
29771 avx512f_<castmode><avxsizesuffix>_<castmode>,
29772 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29773 in condition that at least one operand is not a MEM.
29775 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29777 PR middle-end/79805
29778 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29779 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29781 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29782 gimple_call_nothrow_p flag based on whether original builtin can throw.
29783 If it can, emit following stmts on the fallthrough edge.
29784 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29785 don't create new bb if inserting just debug stmts on the edge, try to
29786 insert them on the fallthru bb or just reset debug stmts.
29788 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29791 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29792 restore recog_data (including the operand rtxes inside it) around
29793 the call to get_insn_template.
29795 2017-03-03 Martin Sebor <msebor@redhat.com>
29797 PR tree-optimization/79699
29798 * context.c (context::~context): Free MPFR caches to avoid
29799 a memory leak on program exit.
29801 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29803 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29804 Use wide_int::ulow () instead of .elt (0).
29806 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29808 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29809 (*pushxf): Limit oF constraint to 32bit targets and add oC
29810 constraint for 64bit targets.
29811 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29812 (*pushdf): Change rmF constraint to rmC.
29814 2017-03-03 Martin Liska <mliska@suse.cz>
29816 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29817 Remove unused variable.
29819 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29822 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29823 is a memory operand, increase num_memory.
29824 (ix86_expand_args_builtin): Likewise.
29826 2017-03-03 Jan Hubicka <jh@suse.cz>
29829 * ipa-devirt.c (maybe_record_node): Properly handle
29830 __cxa_pure_virtual visibility.
29832 2017-03-03 Martin Liska <mliska@suse.cz>
29834 PR tree-optimization/79803
29835 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29837 (pass_loop_prefetch::execute): Disabled optimization if an
29838 assumption about L1 cache size is not met.
29840 2017-03-03 Martin Liska <mliska@suse.cz>
29842 PR rtl-optimization/79574
29843 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29844 (hash_scan_set): Likewise.
29845 (dump_hash_table): Likewise.
29846 (hoist_code): Likewise.
29848 2017-03-03 Richard Biener <rguenther@suse.de>
29850 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29851 in place of uhigh ().
29852 (fixed_convert_from_real): Likewise.
29854 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29857 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29859 2017-03-03 Richard Biener <rguenther@suse.de>
29861 PR middle-end/79818
29862 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29863 TYPE_OVERFLOW_UNDEFINED check.
29865 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29867 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29869 (vector_ae_<mode>_p): Likewise.
29870 (vector_nez_<mode>_p): Likewise.
29871 (vector_ne_v2di_p): Likewise.
29872 (vector_ae_v2di_p): Likewise.
29873 (vector_ne_<mode>_p): Likewise.
29874 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29876 (vsx_tsqrt<mode>2_fe): Likewise.
29878 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29881 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29883 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29885 PR rtl-optimization/79780
29886 * cprop.c (one_cprop_pass): When second and further conditional trap
29887 in a single basic block is turned into an unconditional trap, turn it
29888 into a deleted note to avoid RTL verification failures.
29890 2017-03-02 Richard Biener <rguenther@suse.de>
29892 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29894 2017-03-02 Richard Biener <rguenther@suse.de>
29896 PR tree-optimization/79345
29898 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29899 param and abort the walk, returning -1 if it is hit.
29900 (walk_aliased_vdefs): Take a limit param and pass it on.
29901 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29902 defaulting to 0 and return a signed int.
29903 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29904 (check_defs): New helper.
29905 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29906 about uninitialized memory.
29907 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29908 bogus uninitialized warning.
29909 (fixed_convert_from_real): Likewise.
29911 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29913 PR tree-optimization/66768
29914 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29915 iv_use if base object can't be determined.
29917 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29919 PR tree-optimization/79345
29920 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29921 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29922 (get_pattern_stats): Initialize it.
29923 * genemit.c (gen_expand): Verify match_scratch numbers come after
29924 match_operand/match_dup numbers.
29925 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29926 match_scratch numbers.
29927 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29929 * config/s390/s390.md (trunctdsd2): Likewise.
29931 2017-03-02 Richard Biener <rguenther@suse.de>
29933 * wide-int.h (wide_int_storage::operator=): Implement in terms
29936 2017-03-02 Richard Biener <rguenther@suse.de>
29938 PR tree-optimization/79777
29939 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29940 the to insert expression to sth existing.
29942 2017-03-01 Martin Sebor <msebor@redhat.com>
29944 PR middle-end/79692
29945 * gimple-ssa-sprintf.c
29946 (directive::known_width_and_precision): New function.
29947 (format_integer): Use it.
29948 (get_mpfr_format_length): Consider the full range of precision
29949 when computing %g output with the # flag. Set the likely byte
29950 count to 3 rather than 1 when precision is indeterminate.
29951 (format_floating): Correct the lower bound of precision.
29953 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29955 * doc/invoke.texi: Document default code model for 64-bit Linux.
29957 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29960 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29961 udiv rather than div since input pattern is unsigned.
29963 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29965 * config/i386/i386.c (print_reg): Warn for values of
29966 unsupported size in integer register.
29968 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29971 * config/rs6000/predicates.md (current_file_function_operand): Do
29972 not allow self calls to be local if the function is replaceable.
29974 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29977 * config/rs6000/altivec.h (vec_ctz and others): Change the
29978 preprocessor macro that controls conditional compilation from
29979 _ARCH_PWR9 to __POWER9_VECTOR__.
29980 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29981 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29982 control (instead of _ARCH_PWR9 control) so that template
29983 definition uses power9-specific function.
29984 (vec_any_eq): Likewise.
29985 (vec_all_ne): Change macro definition to use a power9-specific
29986 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29987 _ARCH_PWR9 control).
29988 (vec_any_eq) Likewise.
29989 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29990 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29991 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29992 support for xvcmpnedp instruction.
29993 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29994 macro expansion so that Power9 implementation of vec_all_ne does
29995 not use the AltiVec predicate framework.
29996 (VCMPNEH_P): Likewise.
29997 (VCMPNEW_P): Likewise.
29998 (VCMPNED_P): Likewise.
29999 (VCMPNEFP_P): Likewise.
30000 (VCMPNEDP_P): Likewise.
30001 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30002 implementation of vec_any_eq to not use AltiVec predicate
30004 (VCMPAEH_P): Likewise.
30005 (VCMPAEW_P): Likewise.
30006 (VCMPAED_P): Likewise.
30007 (VCMPAEFP_P): Likewise.
30008 (VCMPAEDP_P): Likewise.
30009 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30010 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30011 not use the AltiVec predicate framework.
30012 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30013 of vec_any_eq to not use AltiVec predicate framework.
30014 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30015 support for predefined __POWER9_VECTOR__ macro to indicate that
30016 Power9 instruction selection is enabled.
30017 (altivec_overloaded_builtins): Remove extraneous
30018 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30019 function argument types RS6000_BTI_bool_V16QI and
30020 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30021 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30022 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30023 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30024 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30025 Power9 for implementations of vec_cmpne. Change the signature for
30026 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30027 (representing vec_all_ne) to remove the previously described first
30028 argument of type RS6000_BTI_INTSI, as this was an artifact of
30029 reliance on the AltiVec predicate framework, which is no longer
30030 used in the implementation of these functions. Add
30031 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30032 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30033 since, unlike the AltiVec predicate framework implementation, we
30034 do not share function descriptors between vec_alle and vec_anyeq.
30035 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30036 set of modes that receive special treatment even when
30037 TARGET_P9_VECTOR is true. The special treatment emits code that
30038 does not depend on Power9 instructions.
30039 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30040 define_expand to not rely on AltiVec predicate framework.
30041 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30043 (vector_ne_v2di_p): Change this define_expand to not rely on
30044 AltiVec predicate framework.
30045 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30047 (vector_ne_<mode>_p): Change this define_expand to not rely on
30048 AltiVec predicate framework.
30049 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30051 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30052 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30053 define_insn pattern.
30054 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30055 define_insn pattern because the xvcmpne<VSs>. instruction is not
30057 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30058 instruction is not supported.
30060 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30062 * config/nvptx/nvptx.c: Include intl.h.
30064 2017-03-01 Martin Jambor <mjambor@suse.cz>
30067 * ipa-prop.h (ipa_bits): Removed field known.
30068 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30069 to pointers. Adjusted their comments to warn about their sharing.
30070 (ipcp_transformation_summary): Change bits to a vector of pointers.
30071 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30072 (ipa_get_ipa_bits_for_value): Declare.
30073 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30074 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30075 (ipa_bits_hash_table): Likewise.
30076 (ipa_vr_ggc_hash_traits): Likewise.
30077 (ipa_vr_hash_table): Likewise.
30078 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30079 being pointers and vr_known being removed.
30080 (ipa_set_jf_unknown): Likewise.
30081 (ipa_get_ipa_bits_for_value): New function.
30082 (ipa_set_jfunc_bits): Likewise.
30083 (ipa_get_value_range): New overloaded functions.
30084 (ipa_set_jfunc_vr): Likewise.
30085 (ipa_compute_jump_functions_for_edge): Use the above functions to
30086 construct bits and vr parts of jump functions.
30087 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30088 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30090 (ipcp_grow_transformations_if_necessary): Also allocate
30091 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30093 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30094 them. Fix too long lines.
30095 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30096 vr_known being removed.
30097 (ipa_read_jump_function): Use new setter functions to construct bits
30098 and vr parts of jump functions or set them to NULL.
30099 (write_ipcp_transformation_info): Adjust for bits being pointers.
30100 (read_ipcp_transformation_info): Likewise.
30101 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30103 Include gt-ipa-prop.h.
30104 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30106 (ipcp_store_bits_results): Likewise.
30107 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30108 Do not write to existing jump functions but use a temporary instead.
30110 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30113 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30114 attempt to use its first operand as BIT_FIELD_REF base.
30116 2017-03-01 Richard Biener <rguenther@suse.de>
30118 PR middle-end/79721
30119 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30120 interpolating formula in wrapping arithmetic.
30121 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30123 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30125 PR tree-optimization/79734
30126 * tree-vect-generic.c (expand_vector_condition): Optimize
30127 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30128 Handle VEC_COND_EXPR where comparison has different inner width from
30129 type's inner width.
30131 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30133 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30134 markup, and similar issues. Remove @opindex entries for things
30135 that aren't options. Add missing -mmpy-option entries.
30137 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30139 PR tree-optimization/79737
30140 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30141 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30142 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30143 instead of byte_size. Formatting fix.
30144 (shift_bytes_in_array_right): Formatting fix.
30146 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30149 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30150 condition on optimize for the leaf function test.
30152 2017-02-28 Martin Liska <mliska@suse.cz>
30155 * read-rtl-function.c (function_reader::handle_unknown_directive):
30156 Bail out when one uses -flto.
30158 2017-02-28 Martin Liska <mliska@suse.cz>
30160 * common.opt: Replace space with tabular for options of <number>
30162 * config/i386/i386.opt: Show <number> value for
30163 -mlarge-data-threshold.
30164 * opts.c (print_filtered_help): Do not display number in hexadecimal
30167 2017-02-28 Martin Liska <mliska@suse.cz>
30169 * common.opt: Fix --help=option -Q for options which are of
30172 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30174 * config/i386/i386.c (print_reg): Error out for values
30175 of 8-bit size in invalid integer register.
30177 2017-02-28 Martin Sebor <msebor@redhat.com>
30179 PR tree-optimization/79691
30180 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30182 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30185 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30186 gcc_unreachable with output_operand_lossage.
30188 2017-02-28 Richard Biener <rguenther@suse.de>
30190 PR tree-optimization/79740
30191 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30193 (visit_nary_op): Insert the nary into the hashtable if we
30194 pattern-matched sth.
30195 * tree-ssa-pre.c (eliminate_insert): Robustify.
30197 2017-02-28 Richard Biener <rguenther@suse.de>
30199 PR middle-end/79731
30200 * fold-const.c (decode_field_reference): Reject out-of-bound
30203 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30205 * config/i386/i386.c: Include intl.h.
30206 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30207 instead of just cond ? "..." : "...".
30208 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30209 * coverage.c (read_counts_file): Likewise.
30210 * omp-offload.c: Include intl.h.
30211 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30212 of just cond ? "..." : "...".
30213 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30214 of just cond ? "..." : "...".
30216 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30219 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30221 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30222 'tune for' CPU name.
30223 * config/arm/arm-cpu-data.h: Regenerated.
30225 2017-02-28 Richard Biener <rguenther@suse.de>
30227 PR tree-optimization/79732
30228 * tree-inline.c (expand_call_inline): Do not shadow var.
30230 2017-02-28 Richard Biener <rguenther@suse.de>
30232 PR tree-optimization/79723
30233 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30234 address-space properly.
30236 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30238 * doc/optinfo.texi (Optimization groups): Fix option used for
30240 * doc/invoke.texi (-fopt-info): Document "omp".
30241 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30242 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30243 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30244 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30245 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30247 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30249 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30250 OPTGROUP_OMP with "omp".
30252 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30255 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30256 for arithmetic shift of unsigned V2DI.
30258 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30260 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30261 arc/linux.h headers.
30262 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30263 (LINK_SPEC): Likewise.
30264 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30265 (EXTRA_SPECS): Likewise.
30266 (STARTFILE_SPEC): Likewise.
30267 (ENDFILE_SPEC): Likewise.
30268 (LIB_SPEC): Likewise.
30269 (TARGET_SDATA_DEFAULT): Likewise.
30270 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30271 (MULTILIB_DEFAULTS): Likewise.
30272 (DWARF2_UNWIND_INFO): Likewise.
30273 * config/arc/big.h: New file.
30274 * config/arc/elf.h: Likewise.
30275 * config/arc/linux.h: Likewise.
30276 * config/arc/t-uClibc: Remove.
30278 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30280 PR tree-optimization/77536
30281 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30282 (tree_transform_and_unroll_loop): Use above function to compute the
30283 estimated niter of unrolled loop and use it when scaling profile.
30284 Also use count info rather than frequency if it's non-zero.
30285 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30286 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30287 (vect_transform_loop): Call above function.
30289 2017-02-27 Richard Biener <rguenther@suse.de>
30291 PR tree-optimization/45397
30292 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30293 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30294 (visit_nary_op): Add pattern matching for CSEing sign-changed
30295 or truncated operations with wider ones.
30297 2017-02-27 Richard Biener <rguenther@suse.de>
30299 PR tree-optimization/79690
30300 * tree-vect-stmts.c (vectorizable_store): Use vector type
30301 built from the DR with address-space.
30303 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30305 * doc/invoke.texi (Optimize Options): Refine the description
30306 of asan-use-after-return.
30308 2017-02-25 Alan Modra <amodra@gmail.com>
30310 PR rtl-optimization/79584
30311 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30312 base, not ad->base_term, the reg within base. Remove assertion
30313 that ad->base == ad->base_term. Replace gen_int_mode using
30314 bogus mode with const0_rtx.
30316 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30318 PR middle-end/79396
30319 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30320 FMA_EXPR like tcc_binary or tcc_unary.
30322 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30325 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30327 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30328 (output_loc_operands): Handle DW_OP_call_ref and
30329 DW_OP_GNU_variable_value.
30330 (struct variable_value_struct): New type.
30331 (struct variable_value_hasher): Likewise.
30332 (variable_value_hash): New variable.
30333 (string_types): Remove.
30334 (copy_loc_descr): New function.
30335 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30336 (prepend_loc_descr_to_each): New function.
30337 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30338 instead of add_loc_descr_to_each if the first argument is single
30339 location list and the second has multiple.
30340 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30341 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30342 when looking for variable value which doesn't have other location info.
30343 (loc_list_from_tree): Formatting fix.
30344 (gen_array_type_die): Simplify DW_AT_string_length handling.
30345 (adjust_string_types): Remove.
30346 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30347 string_types. Call resolve_variable_values.
30348 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30349 (resolve_addr_in_expr): Likewise. Add A argument.
30350 (copy_deref_exprloc): Remove deref argument. Adjust for the
30351 original expression being DW_OP_GNU_variable_value with optionally
30352 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30353 optionally after it.
30354 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30355 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30356 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30357 (variable_value_hasher::hash, variable_value_hasher::equal): New
30359 (resolve_variable_value_in_expr, resolve_variable_value,
30360 resolve_variable_values, note_variable_value_in_expr,
30361 note_variable_value): New functions.
30362 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30365 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30368 * opts.h (handle_generated_option): Add GENERATED_P argument.
30369 * opts-common.c (handle_option): Adjust function comment.
30370 (handle_generated_option): Add GENERATED_P argument, pass it to
30372 (control_warning_option): Pass false to handle_generated_option
30374 * opts.c (maybe_default_option): Pass true to handle_generated_option
30376 * optc-gen.awk: Likewise.
30378 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30380 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30381 a REG, look at the REG it is a SUBREG of.
30382 (splitter for cmpeqsi_t): Ditto.
30384 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30386 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30387 the special USEs with the pattern of the insn, not the insn itself.
30389 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30392 * doc/invoke.texi: Document -mload-store-pairs.
30394 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30395 Sandra Loosemore <sandra@codesourcery.com>
30397 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30398 argument isn't a CONST_INT.
30399 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30400 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30401 (nios2_validate_compare): Bypass alternate compare logic if *op2
30402 is not a CONST_INT.
30403 (ldstwm_operation_p): Return false if first_base is not a REG or
30404 if first_offset is not a CONST_INT.
30406 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30408 * config/cris/cris.md: Use correct operand in a define_peephole2.
30410 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30412 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30414 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30416 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30417 this_insn if it is an INSN or JUMP_INSN.
30418 (force_offsettable): Look at base, not at addr.
30419 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30420 on things that aren't necessarily CONST_INTs.
30422 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30424 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30425 -mfpmath=sse is the default also for x86-32 targets with SSE2
30426 instruction set when @option{-ffast-math} is enabled
30428 2017-02-24 Jeff Law <law@redhat.com>
30430 PR rtl-optimizatoin/79286
30431 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30434 2017-02-24 Richard Biener <rguenther@suse.de>
30436 PR tree-optimization/79389
30437 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30440 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30442 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30443 function comment to reflect reality.
30444 (loop_exits_before_overflow): Fix typo in function description.
30446 2017-02-24 Richard Biener <rguenther@suse.de>
30448 PR tree-optimization/79389
30449 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30450 properly that a threading opportunity exists. Detect conditional
30451 copy/constant propagation opportunities.
30453 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30455 * config/visium/visium.md (type): Add trap.
30456 (b): New mode attribute.
30457 (*btst): Rename into...
30458 (*btst<mode>): ...this and adjust.
30459 (*cbranchsi4_btst_insn): Rename into...
30460 (*cbranch<mode>4_btst_insn): ...this and adjust.
30461 (trap): New define_insn.
30463 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30465 PR tree-optimization/79389
30466 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30467 (noce_reversed_cond_code): New function.
30468 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30469 reversed_comparison_code. Formatting fix.
30470 (noce_try_store_flag): Test rev_cond != NULL in addition to
30471 reversed_comparison_code.
30472 (noce_try_store_flag_constants): Likewise.
30473 (noce_try_store_flag_mask): Likewise.
30474 (noce_try_addcc): Use rev_cond if non-NULL instead of
30475 reversed_comparison_code.
30476 (noce_try_cmove_arith): Likewise. Formatting fixes.
30477 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30478 (noce_find_if_block): Initialize rev_cond.
30479 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30480 instead of false as last argument never attempt to reverse it
30483 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30485 PR tree-optimization/79663
30486 * tree-predcom.c (combine_chains): Process refs in reverse order
30487 only for ZERO length chains, and add explaining comment.
30489 2017-02-23 Jeff Law <law@redhat.com>
30491 PR tree-optimization/79578
30492 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30493 in call to operand_equal_p.
30495 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30498 * config/i386/cpuid.h: Fix another undefined behavior.
30500 2017-02-23 Richard Biener <rguenther@suse.de>
30502 PR tree-optimization/79683
30503 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30504 vector types for data-refs.
30506 2017-02-23 Martin Liska <mliska@suse.cz>
30508 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30510 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30512 PR middle-end/79665
30513 * internal-fn.c (get_range_pos_neg): Moved to ...
30514 * tree.c (get_range_pos_neg): ... here. No longer static.
30515 * tree.h (get_range_pos_neg): New prototype.
30516 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30517 are known to be in between 0 and signed maximum inclusive, try to
30518 expand both unsigned and signed divmod and use the cheaper one from
30521 2017-02-22 Jeff Law <law@redhat.com>
30523 PR tree-optimization/79578
30524 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30525 to compare base operands.
30527 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30530 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30531 gpc_reg_operand instead of fpr_reg_operand.
30533 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30535 * config/mips/mips.c (mips_return_in_memory): Force FP
30536 vector types to be returned in memory for o32 ABI.
30538 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30540 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30541 instead of DW_TAG_member for static data member declarations and don't
30542 set no_linkage_name for static inline data members.
30543 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30546 2017-02-22 Martin Liska <mliska@suse.cz>
30548 * doc/invoke.texi: Replace inequality signs with square brackets
30551 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30554 * lra-constraints.c (simplify_operand_subreg): Handle
30555 WORD_REGISTER_OPERATIONS targets.
30557 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30560 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30561 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30562 elimination by swapping fld*.
30564 2017-02-22 Richard Biener <rguenther@suse.de>
30566 PR tree-optimization/79673
30567 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30568 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30569 irrelevant address-space qualifiers and avoiding a
30570 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30572 2017-02-22 Richard Biener <rguenther@suse.de>
30574 PR tree-optimization/79666
30575 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30576 to not symbolically negate if that may introduce undefined
30579 2017-02-22 Martin Liska <mliska@suse.cz>
30582 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30583 * data-streamer-out.c (streamer_write_gcov_count_stream):
30585 * value-prof.c (stream_out_histogram_value): Make assert more
30586 precise based on type of counter.
30588 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30591 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30592 Use nonimmediate_operand instead of memory_operand for operand 1.
30593 (float-extend standard_x87sse_constant_load splitter): Ditto.
30595 2017-02-21 Jeff Law <law@redhat.com>
30597 PR tree-optimization/79621
30598 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30599 blocks with edges to themselves.
30601 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30604 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30605 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30606 Use gimple_call_builtin_p.
30609 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30610 on temporarily removed DEBUG_INSNs.
30612 PR tree-optimization/79649
30613 * tree-loop-distribution.c (classify_partition): Give up on
30614 non-generic address space loads/stores.
30616 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30618 * doc/loop.texi (Loop manipulation): Remove nonexistent
30619 tree_ssa_loop_version from the documentation.
30620 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30622 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30625 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30626 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30627 * config/rs6000/rs6000.c: Include except.h.
30628 (rs6000_expand_split_stack_prologue): Call
30629 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30631 2017-02-21 Martin Jambor <mjambor@suse.cz>
30634 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30635 have been analyzed.
30637 2017-02-21 Martin Jambor <mjambor@suse.cz>
30639 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30640 for backward compatibility only.
30641 * doc/invoke.texi (Option Summary): Remove all references to
30642 -fipa-cp-alignment.
30644 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30648 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30650 * lra-constraints.c (curr_insn_transform): Handle
30651 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30653 2017-02-21 Martin Liska <mliska@suse.cz>
30655 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30657 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30659 PR translation/79638
30660 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30662 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30665 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30666 (arm_function_ok_for_sibcall): Return false for an indirect call by
30667 descriptor if all the argument registers are used.
30668 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30669 alignment of the function.
30671 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30673 PR tree-optimization/61441
30674 * simplify-rtx.c (simplify_const_unary_operation): For
30675 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30676 the sNaN unmodified.
30678 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30680 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30681 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30682 instead of SYSTEM_HEADER_DIR.
30684 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30685 Martin Liška <mliska@suse.cz>
30687 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30688 Fix typos and grammar, use active voice, and clarify.
30690 2017-02-20 Marek Polacek <polacek@redhat.com>
30692 PR middle-end/79537
30693 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30696 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30698 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30701 * config/i386/i386.c (ix86_expand_builtin): Handle
30702 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30703 ix86_builtins_isa[fcode].isa as a requirement of those
30704 flags and any other flag in the bitmask.
30705 (ix86_init_mmx_sse_builtins): Use 0 instead of
30706 ~OPTION_MASK_ISA_64BIT as mask.
30707 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30708 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30709 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30710 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30712 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30715 * lra-constraints.c (split_reg): Check requested split mode
30716 is supported by the register.
30718 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30720 * lra-constraints.c (simplify_operand_subreg): Remove early
30723 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30726 * lra-constraints.c (curr_insn_transform): Tighten condition
30727 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30729 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30732 * lra-constraints.c (curr_insn_transform): Handle
30733 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30735 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30738 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30740 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30742 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30745 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30748 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30750 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30753 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30755 * final.c (last_columnnum, override_columnnum): New variables.
30756 (final_start_function): Set last_columnnum, pass it to begin_prologue
30757 hook and pass 0 to dwarf2out_begin_prologue.
30758 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30759 to source_line debug hook.
30760 (notice_source_line): Compute last_columnnum and for debug_column_info
30761 return true on column changes.
30762 * debug.h (struct gcc_debug_hooks): Add column argument to
30763 source_line and begin_prologue hooks.
30764 (debug_nothing_int_charstar_int_bool): Remove prototype.
30765 (debug_nothing_int_int_charstar,
30766 debug_nothing_int_int_charstar_int_bool): New prototypes.
30767 (dwarf2out_begin_prologue): Add column argument.
30768 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30769 begin_prologue hooks.
30770 (debug_nothing_int_charstar_int_bool): Remove.
30771 (debug_nothing_int_int_charstar,
30772 debug_nothing_int_int_charstar_int_bool): New functions.
30773 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30774 through to dwarf2out_source_line.
30775 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30776 (dwarf2out_source_line): Add column argument, emit it if requested.
30777 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30779 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30780 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30781 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30782 through to dwarf2out_begin_prologue.
30783 (vmsdbgout_source_line): Add column argument, pass it through to
30784 dwarf2out_source_line.
30785 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30786 dbxout_source_line caller.
30787 (dbxout_source_line): Add column argument.
30789 * common.opt (gno-column-info, gcolumn-info): New options.
30790 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30791 (check_die): Also test for multiple DW_AT_decl_column attributes.
30792 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30793 DW_AT_decl_column if requested.
30794 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30796 (gen_variable_die): Likewise.
30797 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30798 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30801 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30802 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30803 (ix86_handle_option): Handle OPT_m3dnowa.
30804 * doc/invoke.texi (-m3dnowa): Document.
30805 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30806 -m3dnowa instead of -m3dnow -march=athlon.
30809 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30810 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30812 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30815 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30816 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30817 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30818 generator for vsx_xxpermdi_<mode>_be.
30819 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30820 force big-endian semantics.
30821 (vsx_xxpermdi_<mode>_be): New define_expand with same
30822 implementation as previous version of vsx_xxpermdi_<mode>.
30824 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30826 PR tree-optimization/79327
30827 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30828 variable, its initialization and use.
30830 2017-02-17 Julia Koval <julia.koval@intel.com>
30832 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30833 (OPTION_MASK_ISA_PKU_UNSET): New.
30834 (ix86_handle_option): Handle -mrdpid.
30835 * config/i386/cpuid.h (bit_RDPID): New.
30836 * config/i386/driver-i386.c (host_detect_local_cpu):
30837 Detect RDPID feature.
30838 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30839 * config/i386/i386-c.c (ix86_target_macros_internal):
30841 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30842 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30843 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30844 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30845 * config/i386/i386.md (define_insn "rdpid"): New.
30846 * config/i386/i386.opt Add -mrdpid.
30847 * config/i386/immintrin.h (_rdpid_u32): New.
30849 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30851 PR rtl-optimization/79541
30852 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30853 instead of transforming it into USE.
30855 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30857 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30858 If HONOR_SNANS (SFmode) force the input to a register.
30859 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30860 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30861 an frsp or similar insn.
30863 2017-02-17 Martin Liska <mliska@suse.cz>
30865 PR rtl-optimization/79577
30866 * params.def (selsched-max-sched-times): Increase minimum to 1.
30868 2017-02-17 Martin Liska <mliska@suse.cz>
30870 PR rtl-optimization/79574
30871 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30873 2017-02-17 Martin Liska <mliska@suse.cz>
30875 PR tree-optimization/79529
30876 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30877 ssa_defined_default_def_p to handle cases which are implicitly
30879 * tree-ssa.c (ssa_defined_default_def_p): New function.
30880 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30881 which are implicitly defined.
30882 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30884 2017-02-17 Richard Biener <rguenther@suse.de>
30886 PR middle-end/79576
30887 * params.def (max-ssa-name-query-depth): Limit to 10.
30889 2017-02-17 Richard Biener <rguenther@suse.de>
30891 PR tree-optimization/79552
30892 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30895 2017-02-17 Richard Biener <rguenther@suse.de>
30898 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30900 2017-02-17 Marek Polacek <polacek@redhat.com>
30902 PR middle-end/79536
30903 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30904 (fold_negate_expr): New wrapper.
30906 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30908 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30909 Correct terminology and de-emphasize pre-standard behavior.
30911 2017-02-16 Alan Modra <amodra@gmail.com>
30913 PR rtl-optimization/79286
30914 * ira.c (def_dominates_uses): New function.
30915 (update_equiv_regs): Don't create an equivalence for insns that
30916 may trap where the register def does not dominate the use.
30918 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30920 PR rtl-optimization/78127
30921 * lra.c (lra): Call lra_eliminate before finish the loop after
30924 2017-02-16 Richard Biener <rguenther@suse.de>
30926 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30928 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30929 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30930 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30931 (isl_val_int_from_wi): New function.
30932 (extract_affine_gmp): Rename to ...
30933 (extract_affine_wi): ... this, take a widest_int.
30934 (extract_affine_int): Just wrap extract_affine_wi.
30935 (add_param_constraints): Use isl_val_int_from_wi.
30936 (add_loop_constraints): Likewise, and extract_affine_wi.
30938 2017-02-15 Jeff Law <law@redhat.com>
30940 PR middle-end/79521
30941 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30942 ira_init_register_move_cost_if_necessary.
30944 2017-02-15 Martin Sebor <msebor@redhat.com>
30946 PR middle-end/32003
30947 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30948 removed in a prior commit.
30950 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30952 PR tree-optimization/79347
30953 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30954 counters during peeling.
30956 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30958 * Makefile.in (site.exp): Remove "set ISLVER".
30960 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30963 * real.c (real_from_integer): Call real_convert even for decimal.
30965 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30968 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30970 2017-02-14 Andrew Pinski <apinski@cavium.com>
30972 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30973 cores and change the partno/implementer to be correct.
30974 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30975 the 'B" as the implementer.
30976 * config/aarch64/aarch64-tune.md: Regenerate.
30978 2017-02-14 Carl Love <cel@us.ibm.com>
30980 * config/rs6000/rs6000.c: Add case statement entry to make the
30981 xvcvuxdsp built-in argument unsigned.
30982 * config/rs6000/vsx.md: Fix the source and return operand types so they
30983 match the instruction definitions from the ISA document. Fix typo
30984 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30987 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30990 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30991 member early_clobber_alts.
30992 * lra-lives.c (reg_early_clobber_p): New.
30993 (process_bb_lives): Use it.
30994 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30995 (debug_operand_data): Initialize early_clobber_alts.
30996 (setup_operand_alternative): Set up early_clobber_alts.
30997 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30998 alternatives to new_insn_reg.
30999 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31001 (lra_update_insn_regno_info): Pass the new arg.
31003 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31005 PR middle-end/79505
31006 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31007 (new_oacc_loop_raw): Don't clear already cleared fields.
31010 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31011 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31012 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31014 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31017 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31019 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31022 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31023 the extra instruction to the right place to store 128-bit constant
31026 2017-02-14 Martin Sebor <msebor@redhat.com>
31028 PR middle-end/79448
31029 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31030 warning for strings of unknown length.
31032 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31034 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31036 2017-02-14 Jeff Law <law@redhat.com>
31039 * ira-costs.c (scan_one_insn): Initialize register move costs
31040 for pseudos seen in USE/CLOBBER insns.
31042 PR tree-optimization/79095
31043 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31044 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31045 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31046 if the operands are known to be not equal, then the resulting range
31048 (intersect_ranges): If the new range is ~[0,0] and the old range is
31049 wide, then prefer ~[0,0].
31050 * tree-vrp.c (overflow_comparison_p_1): New function.
31051 (overflow_comparison_p): New function.
31052 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31053 if NAME is used in an overflow test.
31054 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31055 overflow check that can be expressed as an equality test, then adjust
31056 ops to be that equality test.
31058 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31060 * config/s390/s390-builtin-types.def: Remove flags argument.
31061 * config/s390/s390.c (s390_init_builtins): Likewise.
31063 2017-02-14 Martin Liska <mliska@suse.cz>
31065 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31066 vector. Fix trailing white spaces.
31068 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31070 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31073 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31075 PR rtl-optimization/68664
31076 * config/arm/arm.c (arm_sched_can_speculate_insn):
31077 New function. Declare prototype.
31078 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31080 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31082 PR rtl-optimization/68664
31083 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31085 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31087 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31089 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31090 max skip bytes for function, loop and jump.
31092 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31094 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31095 ABS_EXPR for gimple dump.
31097 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31100 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31102 PR tree-optimization/79408
31103 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31104 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31105 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31106 also if rhs1 is INTEGER_CST.
31108 2017-02-14 Richard Biener <rguenther@suse.de>
31110 PR middle-end/79432
31111 * tree-into-ssa.c (insert_phi_nodes): When the function can
31112 have abnormal edges rewrite SSA names with broken use-def
31113 dominance out of SSA and register them for PHI insertion.
31115 2017-02-13 Martin Sebor <msebor@redhat.com>
31117 PR middle-end/79496
31118 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31119 clearing info.nowrite flag when snprintf size argument is a range.
31121 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31123 * cprop.c (cprop_jump): Add missing space in string literal.
31124 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31125 (get_constraint_for_component_ref): Likewise.
31126 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31127 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31128 * lra-constraints.c (process_alt_operands): Likewise.
31129 * ipa-inline.c (inline_small_functions): Likewise.
31130 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31131 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31132 * trans-mem.c (diagnose_tm_1_op): Likewise.
31133 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31134 (grid_parallel_clauses_gridifiable): Likewise.
31136 * config/nvptx/mkoffload.c (process): Add space in between
31139 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31140 "MOD4_SSE_REGS" and "ALL_REGS".
31142 * spellcheck.c (test_data): Add , in between "foo" and "food".
31144 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31147 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31148 boundary crossing check and subsequent code generation agree.
31150 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31152 * config/aarch64/aarch64.c (has_memory_op): Delete.
31153 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31156 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31158 PR rtl-optimization/79388
31159 PR rtl-optimization/79450
31160 * combine.c (distribute_notes): When removing TEM_INSN for which
31161 corresponding dest has last value recorded, invalidate that last
31164 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31166 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31167 of explicit '@'. Add missing assembly comment marker on branch costs
31170 2017-02-13 Nathan Sidwell <nathan@acm.org>
31172 * gengtype-lex.l (<in_struct>): Add '/'.
31174 2017-02-13 Martin Liska <mliska@suse.cz>
31177 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31179 2017-02-13 Richard Biener <rguenther@suse.de>
31181 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31183 * configure: Re-generate.
31184 * config.in: Likewise.
31185 * graphite-dependences.c: Simplify as if
31186 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31187 * graphite-isl-ast-to-gimple.c: Likewise.
31188 * graphite-optimize-isl.c: Likewise.
31189 * graphite-poly.c: Likewise.
31190 * graphite-sese-to-poly.c: Likewise.
31191 * graphite.h: Likewise.
31192 * toplev.c: Include isl/version.h and use isl_version () for
31193 printing the ISL version.
31194 * doc/install.texi: Update ISL requirement.
31196 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31198 * doc/standards.texi (Standards): Update reference to
31201 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31203 * doc/extend.texi (Named Address Spaces): sourceware.org now
31205 * doc/install.texi (Binaries): Ditto.
31208 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31210 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31211 standard terminology "stringize"/"stringizing" throughout.
31212 * doc/cppinternals.texi: Likewise.
31214 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31216 * doc/extend.texi: Fix some spelling mistakes and typos.
31217 * doc/invoke.texi: Likewise.
31219 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31222 * params.def (inline-min-speedup) Change from 10 to 8.
31224 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31226 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31229 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31232 * ipa-inline-analysis.c (get_minimal_bb): New function.
31233 (record_modified): Use it.
31234 (remap_edge_change_prob): Handle also ancestor functions.
31236 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31238 * doc/contrib.texi (Contributors): Remove broken link into
31239 the Mauve CVS repository.
31241 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31243 PR middle-end/79454
31244 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31245 result computation whenever lhs doesn't have vector mode, not
31246 just when it has BLKmode.
31248 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31250 * doc/makefile.texi (profiledbootstrap): Refer to the
31251 installation instructions only in textual form.
31253 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31256 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31258 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31260 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31261 (Specific): Update mingw-w64 reference.
31263 (Specific): Remove broken link to Renesas RX processor.
31265 2017-02-10 Richard Biener <rguenther@suse.de>
31267 * toplev.c (process_options): Do not mention obsolete graphite
31268 options when printing sorry message about missing graphite support.
31269 Mention -floop-nest-optimize.
31271 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31273 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31274 (vtst_p16): Likewise.
31275 (vtstq_p8): Likewise.
31276 (vtstq_p16): Likewise.
31278 (vtstq_p64): Likewise.
31279 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31280 (vset_lane_p64): New.
31281 (vsetq_lane_p64): New.
31283 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31285 PR tree-optimization/79411
31286 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31287 stmt operands are SSA_NAMEs used in abnormal phis.
31288 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31291 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31294 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31297 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31299 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31301 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31303 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31307 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31308 not arbitrary TREE_CONSTANT.
31311 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31312 "omp declare target link" attribute unless is_global_var.
31313 * omp-offload.c (find_link_var_op): Likewise.
31315 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31316 Chung-Lin Tang <cltang@codesourcery.com>
31318 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31320 (gimplify_adjust_omp_clauses): Don't delete TILE.
31321 (gimplify_omp_for): Deal with TILE.
31322 * internal-fn.c (expand_GOACC_TILE): New function.
31323 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31325 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31326 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31328 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31329 avoid DIV for outermost collapse var.
31330 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31331 Remove out of date comments, fix whitespace.
31332 * omp-general.c (omp_extract_for_data): Deal with tiling.
31333 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31334 adjust OLF_DIM_BASE value.
31335 (struct omp_for_data): Add tiling field.
31336 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31337 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31338 for auto loops. Remove default auto determining, moved to
31339 oacc_loop_fixed_partitions.
31340 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31341 stmts, add e_mask field.
31342 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31343 (oacc_thread_numbers): Use oacc_dim_call.
31344 (oacc_xform_tile): New.
31345 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31346 (finish_oacc_loop): Adjust for ifns vector.
31347 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31348 add case for GOACC_TILE fns.
31349 (oacc_loop_xform_loop): Delete.
31350 (oacc_loop_process): Iterate over call list directly, and add
31351 handling for GOACC_TILE fns.
31352 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31354 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31355 vector partitioning to outer loops. Assign 2 partitions to loops
31356 when available. Add TILE handling.
31357 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31358 (execite_oacc_device_lower): Process GOACC_TILE fns,
31359 ignore unknown specs.
31360 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31361 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31362 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31364 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31366 * configure.ac (ACX_BUGURL): Update.
31367 * configure: Regenerate.
31369 2017-02-09 Richard Biener <rguenther@suse.de>
31371 PR tree-optimization/69823
31372 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31373 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31375 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31377 * config/arc/arc-c.def: Add __NPS400__ definition.
31378 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31379 (TARGET_NPS400): Define.
31381 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31383 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31385 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31386 pointer, arch_info.
31387 (arc_cpu_types): Fill the arch_info field with a pointer into the
31388 arc_arch_types table.
31389 (arc_selected_cpu): Declare.
31390 * config/arc/arc.c (arc_selected_cpu): Make global.
31391 (arc_selected_arch): Delete.
31392 (arc_base_cpu): Delete.
31393 (arc_override_options): Remove references to deleted variables,
31394 update access to arch information.
31395 (ARC_OPT): Update access to arch information.
31396 (ARC_OPTX): Likewise.
31397 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31398 (TARGET_ARC600): Update access to arch information.
31399 (TARGET_ARC601): Likewise.
31400 (TARGET_ARC700): Likewise.
31401 (TARGET_EM): Likewise.
31402 (TARGET_HS): Likewise.
31403 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31406 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31409 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31410 condition/operands for integer GE/LE/GEU/LEU operations.
31412 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31414 PR translation/79397
31415 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31418 2017-02-08 Martin Jambor <mjambor@suse.cz>
31421 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31422 whether allocation happened.
31423 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31424 nothing was allocated.
31426 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31428 PR tree-optimization/79408
31429 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31430 constant, but SSA_NAME with a known integer range, use the minimum
31431 of that range instead of op1 to determine if modulo can be replaced
31432 with its first operand.
31434 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31436 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31438 2017-02-08 Richard Biener <rguenther@suse.de>
31440 PR tree-optimization/71824
31441 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31442 Check all loops contained in the merged region.
31444 2017-02-07 Andrew Pinski <apinski@cavium.com>
31446 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31448 2017-02-07 Andrew Pinski <apinski@cavium.com>
31450 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31451 (thunderxt88): Likewise.
31452 (thunderxt81): Disable LSE and change v8.1 to v8.
31453 (thunderxt83): Likewise.
31455 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31456 Richard Biener <rguenther@suse.de>
31458 PR middle-end/79399
31459 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31460 type from int to size_t.
31461 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31463 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31465 PR rtl-optimization/79386
31466 * cprop.c (bypass_conditional_jumps): Initialize
31467 bypass_last_basic_block already before splitting bbs after
31468 unconditional traps...
31469 (bypass_conditional_jumps): ... rather than here.
31472 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31473 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31474 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31475 fix -masm=intel patterns.
31477 2017-02-07 Richard Biener <rguenther@suse.de>
31479 PR tree-optimization/79256
31480 PR middle-end/79278
31481 * builtins.c (get_object_alignment_2): Use min_align_of_type
31482 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31483 and ADJUST_FIELD_ALIGN.
31485 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31487 * doc/tm.texi: Regenerate.
31488 * stor-layout.c (layout_decl): Adjust.
31489 (update_alignment_for_field): Likewise.
31490 (place_field): Likewise.
31491 (min_align_of_type): Likewise.
31492 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31493 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31494 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31495 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31496 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31497 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31498 * config/i386/i386.c (x86_field_alignment): Likewise.
31499 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31500 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31501 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31502 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31503 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31504 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31508 2017-01-30 Richard Biener <rguenther@suse.de>
31510 PR tree-optimization/79256
31511 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31512 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31515 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31517 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31518 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31519 builtins to SImode and emit a zero-extend, if necessary.
31521 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31523 * docs/invoke.texi (RISC-V Options): Alphabetize.
31525 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31527 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31530 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31532 * config/riscv/riscv.c: New file.
31533 * common/config/riscv/riscv-common.c: Likewise.
31534 * config.gcc: Likewise.
31535 * config/riscv/constraints.md: Likewise.
31536 * config/riscv/elf.h: Likewise.
31537 * config/riscv/generic.md: Likewise.
31538 * config/riscv/linux.h: Likewise.
31539 * config/riscv/multilib-generator: Likewise.
31540 * config/riscv/peephole.md: Likewise.
31541 * config/riscv/pic.md: Likewise.
31542 * config/riscv/predicates.md: Likewise.
31543 * config/riscv/riscv-builtins.c: Likewise.
31544 * config/riscv/riscv-c.c: Likewise.
31545 * config/riscv/riscv-ftypes.def: Likewise.
31546 * config/riscv/riscv-modes.def: Likewise.
31547 * config/riscv/riscv-opts.h: Likewise.
31548 * config/riscv/riscv-protos.h: Likewise.
31549 * config/riscv/riscv.h: Likewise.
31550 * config/riscv/riscv.md: Likewise.
31551 * config/riscv/riscv.opt: Likewise.
31552 * config/riscv/sync.md: Likewise.
31553 * config/riscv/t-elf-multilib: Likewise.
31554 * config/riscv/t-linux: Likewise.
31555 * config/riscv/t-linux-multilib: Likewise.
31556 * config/riscv/t-riscv: Likewise.
31557 * configure.ac: Likewise.
31558 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31559 Waterman as RISC-V maintainers.
31560 * doc/install.texi: Add RISC-V entries.
31561 * doc/invoke.texi: Add RISC-V options section.
31562 * doc/md.texi: Add RISC-V constraints section.
31563 * configure: Regenerated.
31565 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31568 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31569 false values to be constant vectors with all 0 or all 1 bits set.
31570 (vcondu<mode><mode>): Likewise.
31571 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31573 (fpmask_comparison_operator): Update comment.
31574 (vecint_comparison_operator): New predicate.
31575 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31576 vector conditionals when the true and false values are constant
31577 vectors with all 0 bits or all 1 bits set.
31579 2017-02-06 Martin Sebor <msebor@redhat.com>
31581 PR tree-optimization/79376
31582 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31584 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31586 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31587 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31588 to simplify split condition.
31590 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31592 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31593 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31596 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31598 PR rtl-optimization/68664
31599 * target.def (can_speculate_insn): New hook.
31600 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31601 * doc/tm.texi: Regenerate.
31602 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31603 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31604 (rs6000_sched_can_speculate_insn): New function.
31606 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31608 PR tree-optimization/79284
31609 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31610 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31611 vectorizable_mask_load_store, vectorizable_operation,
31612 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31613 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31614 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31615 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31616 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31617 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31618 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31619 is_gimple_assign (stmt). Replace another such test with
31620 is_gimple_assign (stmt).
31622 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31625 * config/avr/avr.c (rtl-iter.h): Include it.
31626 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31627 (avr_legitimate_combined_insn): ...and implementation.
31629 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31631 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31632 * config/s390/s390.c (s390_const_operand_ok)
31633 (s390_canonicalize_comparison, s390_extract_part)
31634 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31635 (s390_contiguous_bitmask_p, s390_rtx_costs)
31636 (legitimize_pic_address): Likewise.
31637 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31638 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31639 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31640 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31641 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31643 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31645 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31646 REGNO($0) == REGNO($1).
31648 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31650 * config/s390/linux.h(SIZE_TYPE): Add comment.
31652 2017-02-06 Julian Brown <julian@codesourcery.com>
31653 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31654 Virendra Pathak <virendra.pathak@broadcom.com>
31656 * config/aarch64/aarch64-cores.def: Change the scheduler
31658 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31659 * config/aarch64/thunderx2t99.md: New file.
31661 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31663 * doc/standards.texi (Go Language): Update link to language
31666 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31668 * tree-eh.c (lower_resx): Sanitize profile.
31669 (cleanup_empty_eh_move_lp): Likewise.
31671 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31674 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31676 * cfgloopmanip.h (loop_version): Update prototype.
31677 * modulo-sched.c (sms_schedule): Update call of loop_version.
31678 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31679 * tree-parloops.c (gen_parallel_loop): Likewise.
31680 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31681 * tree-ssa-loop-split.c (split_loop): Likewise.
31682 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31683 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31685 2017-02-05 Martin Liska <mliska@suse.cz>
31688 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31690 (print_operand_address): Initialize a struct to zero.
31692 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31694 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31695 garbage collector only in textual form.
31697 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31699 * doc/extend.texi (x86 specific memory model extensions for
31700 transactional memory): Simplify a phrase.
31702 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31705 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31706 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31707 (atomic_storedi_1): Likewise.
31709 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31711 PR tree-optimization/79338
31712 * tree-parloops.c (gather_scalar_reductions): Don't call
31713 vect_analyze_loop_form for loop->inner before destroying loop's
31716 2017-02-03 Martin Sebor <msebor@redhat.com>
31718 PR tree-optimization/79327
31719 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31720 when precision has resulted in leading zeros.
31721 (format_integer): Adjust the likely counter to assume an unknown
31722 argument that may be zero is non-zero.
31724 2017-02-03 Jason Merrill <jason@redhat.com>
31727 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31728 avoid copying non-taken branch.
31730 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31732 PR tree-optimization/79340
31733 * tree-vect-loop.c (vectorizable_reduction): Release
31734 vec_defs elements after safe_splicing them into other vectors.
31737 PR tree-optimization/79327
31738 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31739 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31741 (format_integer): Use wide_int_to_tree instead of build_int_cst
31742 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31743 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31744 of shortest and longest sequence.
31746 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31748 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31749 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31751 2017-02-03 Walter Lee <walt@tilera.com>
31754 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31755 after initial stackframe link reg save.
31756 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31758 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31761 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31762 wu for stxssp alternative.
31764 2017-02-03 Martin Sebor <msebor@redhat.com>
31766 PR tree-optimization/79352
31767 * gimple-fold.c (get_range_strlen): Add argument.
31768 (get_range_strlen): Change return type to bool.
31769 (get_maxval_strlen): Pass in a dummy argument.
31770 * gimple-fold.h (get_range_strlen): Change return type to bool.
31771 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31772 * tree.h (array_at_struct_end_p): Add argument.
31773 * tree.c (array_at_struct_end_p): Handle it.
31775 2017-02-03 Martin Liska <mliska@suse.cz>
31778 * multiple_target.c (create_dispatcher_calls): Redirect edge
31779 from a caller of a dispatcher.
31780 (expand_target_clones): Make the clones local.
31781 (ipa_target_clone): Do both target clones and resolvers.
31782 (ipa_dispatcher_calls): Remove the pass.
31783 (pass_dispatcher_calls::gate): Likewise.
31784 (make_pass_dispatcher_calls): Likewise.
31785 * passes.def (pass_target_clone): Put as very first IPA early
31788 2017-02-03 Martin Liska <mliska@suse.cz>
31790 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31791 in case of a function with ifunc attribute.
31793 2017-02-03 Martin Liska <mliska@suse.cz>
31795 * cgraph.c (cgraph_node::dump): Dump function version info.
31796 * symtab.c (symtab_node::dump_base): Add missing new line.
31798 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31800 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31801 (ifcombine_ifandif): Use it.
31803 2017-02-03 Martin Liska <mliska@suse.cz>
31805 * doc/invoke.texi: Document default value for
31806 use-after-scope-direct-emission-threshold.
31808 2017-02-03 Martin Liska <mliska@suse.cz>
31810 PR tree-optimization/79339
31811 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31812 (format_floating): Likewise.
31814 2017-02-03 Martin Liska <mliska@suse.cz>
31817 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31819 (ipa_node_params_t::remove): Likewise.
31820 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31821 initialization from removed ipa_node_params_t::insert.
31822 (ipa_node_params::~ipa_node_params): Move from removed
31823 ipa_node_params_t::release.
31824 * symbol-summary.h (symbol_summary::m_released): New member.
31825 Do not release a summary twice. Do not allow to call finalizer
31826 for types of a summary that live in GGC memory.
31828 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31830 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31833 2017-02-02 Martin Sebor <msebor@redhat.com>
31835 PR middle-end/79275
31836 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31837 (format_string): Tighten up the range of output for non-constant
31838 strings and correct the expected range for wide non-constant strings.
31840 2017-02-02 Martin Sebor <msebor@redhat.com>
31842 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31844 PR middle-end/32003
31845 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31847 (-fdump-tree-@var): Add to index and document how to come up
31848 with pass-specific option and dump file names.
31849 (-fdump-passes): Clarify where to look for output.
31851 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31853 PR middle-end/77445
31854 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31855 statistics of the analyzed path; allow threading for speed when
31856 any of BBs along the path are optimized for speed.
31858 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31860 PR middle-end/78468
31861 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31862 settings of the virtual registers.
31865 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31867 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31868 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31871 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31873 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31874 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31876 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31878 * config/s390/s390.md: Add missing comments with the expanded
31880 * config/s390/vector.md: Likewise.
31881 * config/s390/vx-builtins.md: Likewise.
31883 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31886 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31887 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31888 conditions on a single line.
31890 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31892 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31893 __S390_VX__ to __VX__.
31895 2017-02-01 Andrew Pinski <apinski@cavium.com>
31897 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31898 stmt_info to record_stmt_cost.
31899 (vect_get_known_peeling_cost): Pass stmt_info if known to
31901 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31902 cpu_vector_cost field into
31903 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31904 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31905 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31906 splitting of scalar_stmt_cost and vec_stmt_cost.
31907 (thunderx_vector_cost): Likewise.
31908 (cortexa57_vector_cost): LIkewise.
31909 (exynosm1_vector_cost): Likewise.
31910 (xgene1_vector_cost): Likewise.
31911 (thunderx2t99_vector_cost): Improve after the splitting of the two
31913 (aarch64_builtin_vectorization_cost): Update for the splitting of
31914 scalar_stmt_cost and vec_stmt_cost.
31916 2017-02-01 Torvald Riegel <triegel@redhat.com>
31917 Richard Henderson <rth@redhat.com>
31919 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31920 conditional on existance of a fast atomic load.
31921 * optabs-query.c (can_atomic_load_p): New function.
31922 * optabs-query.h (can_atomic_load_p): Declare it.
31923 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31924 no fast atomic load is available for the particular size of access.
31925 (expand_atomic_compare_and_swap): Likewise.
31926 (expand_atomic_load): Likewise.
31927 (expand_atomic_store): Likewise.
31928 (expand_atomic_fetch_op): Likewise.
31929 * testsuite/lib/target-supports.exp
31930 (check_effective_target_sync_int_128): Remove x86 because it provides
31931 no fast atomic load.
31932 (check_effective_target_sync_int_128_runtime): Likewise.
31934 2017-02-01 Richard Biener <rguenther@suse.de>
31936 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31937 (graphite_transform_loops): Provide opt-info for optimized nests.
31938 * tree-parloop.c (parallelize_loops): Provide opt-info for
31939 parallelized loops.
31941 2017-02-01 Richard Biener <rguenther@suse.de>
31943 PR middle-end/79315
31944 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31945 was not set before.
31947 2017-02-01 Richard Biener <rguenther@suse.de>
31949 PR tree-optimization/71824
31950 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31951 Verify the loops are valid in the merged SESE region.
31952 (scop_detection::can_represent_loop_1): Check analyzing the
31953 evolution of the number of iterations in the region succeeds.
31955 2017-01-31 Ian Lance Taylor <iant@golang.org>
31957 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31958 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31960 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31962 PR preprocessor/79210
31963 * input.c (get_substring_ranges_for_loc): Replace line_width
31964 assertion with error-handling.
31966 2017-01-31 Richard Biener <rguenther@suse.de>
31968 PR tree-optimization/77318
31969 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31970 (create_pw_aff_from_tree): Take loop parameter.
31971 (add_condition_to_pbb): Pass loop of the condition to
31972 create_pw_aff_from_tree.
31974 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31976 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31977 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31979 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31983 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31985 (convert_int_to_float128): Likewise.
31986 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31987 (convert_int_to_float128): Likewise.
31988 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31989 (UNSPEC_IEEE128_CONVERT): Likewise.
31990 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31991 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31992 Use local variables for IBM extended format.
31993 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31994 (fix_trunc<mode>si2_fprs): Likewise.
31995 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31996 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31997 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31998 to know that we can now have integers of all sizes in vector
32000 (fix<uns>_<mode>di2_hw): Likewise.
32001 (float<uns>_<mode>si2_hw): Likewise.
32002 (fix_<mode>si2_hw): Likewise.
32003 (fixuns_<mode>si2_hw): Likewise.
32004 (float<uns>_<mode>di2_hw): Likewise.
32005 (float_<mode>di2_hw): Likewise.
32006 (float_<mode>si2_hw): Likewise.
32007 (floatuns_<mode>di2_hw): Likewise.
32008 (floatuns_<mode>si2_hw): Likewise.
32009 (xscvqp<su>wz_<mode>): Delete, no longer used.
32010 (xscvqp<su>dz_<mode>): Likewise.
32011 (xscv<su>dqp_<mode>): Likewise.
32012 (ieee128_mfvsrd_64bit): Likewise.
32013 (ieee128_mfvsrd_32bit): Likewise.
32014 (ieee128_mfvsrwz): Likewise.
32015 (ieee128_mtvsrw): Likewise.
32016 (ieee128_mtvsrd_64bit): Likewise.
32017 (ieee128_mtvsrd_32bit): Likewise.
32019 2017-01-31 Martin Liska <mliska@suse.cz>
32022 * ipa-prop.c (ipa_free_all_node_params): Call release method
32023 instead of ~sumbol_summary to not to trigger double times
32026 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32028 PR tree-optimization/71691
32029 * bitmap.h (class auto_bitmap): New.
32030 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32031 is_maybe_undefined instead of ssa_undefined_value_p.
32033 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32035 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32036 __S390_ARCH_LEVEL__ to __ARCH__.
32038 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32040 PR tree-optimization/79267
32041 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32042 if should_remove_lhs_p is true.
32044 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32047 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32048 (add_alignment_attribute): New.
32049 (base_type_die): Add alignment attribute.
32050 (subrange_type_die): Likewise.
32051 (modified_type_die): Likewise.
32052 (gen_array_type_die): Likewise.
32053 (gen_descr_array_type_die: Likewise.
32054 (gen_enumeration_type_die): Likewise.
32055 (gen_subprogram_die): Likewise.
32056 (gen_variable_die): Likewise.
32057 (gen_field_die): Likewise.
32058 (gen_ptr_to_mbr_type_die): Likewise.
32059 (gen_struct_or_union_type_die): Likewise.
32060 (gen_subroutine_type_die): Likewise.
32061 (gen_typedef_die): Likewise.
32062 (base_type_cmp): Compare alignment attribute.
32064 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32067 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32068 (setb_unsigned) New pattern for setb with CCUNS.
32069 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32070 subfc./subfe sequence to avoid overflow problems. Generate a
32071 shorter sequence with cmpld/setb for power9.
32072 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32073 for generating subfc. instruction.
32074 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32075 now uses this instruction.
32077 2017-01-30 Ian Lance Taylor <iant@google.com>
32080 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32081 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32083 2017-01-30 Martin Sebor <msebor@redhat.com>
32085 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32086 Move constant to the right of a relational operator.
32087 (get_mpfr_format_length, format_character, format_string): Ditto.
32088 (should_warn_p, maybe_warn): Same.
32090 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32092 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32095 * asan.c (get_translation_unit_decl): Remove function.
32096 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32098 2017-01-30 Martin Liska <mliska@suse.cz>
32100 PR gcov-profile/79259
32101 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32102 -fprofile-generate.
32104 2017-01-30 Martin Liska <mliska@suse.cz>
32107 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32108 Initialize variables with NULL value.
32110 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32113 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32115 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32117 2017-01-30 Richard Biener <rguenther@suse.de>
32119 PR tree-optimization/79276
32120 * tree-vrp.c (process_assert_insertions): Properly adjust common
32121 when removing a duplicate.
32123 2017-01-30 Richard Biener <rguenther@suse.de>
32125 PR tree-optimization/79256
32126 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32127 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32129 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32131 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32134 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32135 ("*r<noxa>sbg_<mode>_sll_bitmask")
32136 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32137 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32138 Use contiguous_bitmask_nowrap_operand.
32140 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32143 * config/rs6000/altivec.h (vec_xl): Revise #define.
32144 (vec_xst): Likewise.
32146 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32148 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32150 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32152 PR rtl-optimization/79194
32153 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32154 traps before call to bypass_conditional_jumps.
32156 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32158 PR tree-optimization/71374
32159 * lra-constraints.c (check_conflict_input_operands): New.
32160 (match_reload): Use it.
32162 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32165 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32166 account to calculate conflict_set.
32168 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32170 PR rtl-optimization/78559
32171 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32172 other_insn in combine.
32174 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32176 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32177 uint16_type_node for BT_UINT16.
32179 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32181 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32182 "RTL Tests" to menu.
32183 (GIMPLE Tests): New node.
32184 (RTL Tests): New node.
32186 2017-01-27 Richard Biener <rguenther@suse.de>
32188 PR tree-optimization/79245
32189 * tree-loop-distribution.c (distribute_loop): Apply cost
32190 modeling also to detected patterns.
32192 2017-01-27 Richard Biener <rguenther@suse.de>
32194 PR tree-optimization/71433
32195 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32196 (compare_assert_loc): New function.
32197 (process_assert_insertions): Sort and optimize assert locations
32198 to remove duplicates and push down identical assertions on
32199 edges to their destination block.
32201 2017-01-27 Richard Biener <rguenther@suse.de>
32203 PR tree-optimization/79244
32204 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32205 out SSA names even if abnormal.
32207 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32209 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32210 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32211 instead of MPFR_RNDN.
32213 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32216 * arm.c (arm_option_override): Don't call build_target_option_node
32217 until after doing all option overrides.
32218 (arm_valid_target_attribute_tree): Likewise.
32220 2017-01-27 Martin Liska <mliska@suse.cz>
32222 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32223 for {cd}tors and C++ {cd}tors.
32225 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32227 * config/s390/s390.md ("*setmem_long_and")
32228 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32230 2017-01-26 Martin Sebor <msebor@redhat.com>
32232 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32235 2017-01-26 Martin Sebor <msebor@redhat.com>
32237 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32238 HAVE_DFmode before using XFmode or DFmode.
32239 (parse_directive): Avoid using the z length modifier to avoid
32240 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32242 PR middle-end/78703
32243 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32244 to accept adjustment as an array.
32245 (get_int_range): New function.
32246 (struct directive): Make width and prec arrays.
32247 (directive::set_width, directive::set_precision): Call get_int_range.
32248 (format_integer, format_floating): Handle width and precision ranges.
32249 (format_string, parse_directive): Same.
32251 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32254 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32255 comdat_type_p set, just clone them, but keep the children in the
32259 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32260 which have direct callers with -fvar-tracking-assignments enabled
32262 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32263 inside of type units.
32265 2017-01-26 Martin Sebor <msebor@redhat.com>
32267 PR middle-end/78703
32268 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32270 (struct format_result): Replace number_chars, number_chars_min,
32271 and number_chars_max with a single member of struct result_range.
32273 (format_result::operator+=): Adjust.
32274 (struct fmtresult): Remove bounded. Handle likely and unlikely
32276 (fmtresult::adjust_for_width_or_precision): New function.
32277 (fmtresult:type_max_digits): New function.
32278 (bytes_remaining): Handle likely and unlikely counters.
32279 (min_bytes_remaining): Remove.
32280 (format_percent): Simplify.
32281 (format_integer, format_floating): Set likely and unlikely counters.
32282 (get_string_length, format_character, format_string): Same.
32283 (format_plain, should_warn_p): New function.
32284 (maybe_warn): Call should_warn_p. Update diagnostic messages
32285 and handle those for all directives, including plain strings.
32286 (format_directive): Handle likely and unlikely counters.
32287 Remove unnecessary quoting from diagnostics. Add an informational
32289 (add_bytes): Remove.
32290 (pass_sprintf_length::compute_format_length): Simplify.
32291 (try_substitute_return_value): Handle likely and unlikely counters.
32293 2017-01-26 Carl Love <cel@us.ibm.com>
32295 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32296 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32298 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32301 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32302 endianess for subregs into account.
32303 * lra-constraints.c (lra_constraints): Do risky transformations
32304 always on the first iteration.
32305 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32306 last_call_used_reg_set.
32307 (process_bb_lives): Define and use last_call_used_reg_set.
32308 * lra.c (lra): Always continue after lra_constraints on the first
32311 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32313 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32315 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32317 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32319 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32320 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32321 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32322 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32323 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32324 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32325 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32326 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32327 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32329 2017-01-26 Marek Polacek <polacek@redhat.com>
32332 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32333 for the third operand.
32335 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32337 PR middle-end/79236
32338 * omp-low.c (struct omp_context): Add simt_stmt field.
32339 (scan_omp_for): Return omp_context *.
32340 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32341 context to the _simt_ SIMD stmt.
32342 (lower_omp_for): For combined SIMD with sibling _simt_
32343 SIMD, make sure to use the same decls in _looptemp_
32344 clauses as in the sibling.
32346 2017-01-26 David Sherwood <david.sherwood@arm.com>
32348 PR middle-end/79212
32349 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32352 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32355 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32356 emit fld b; fld a; if possible.
32358 * brig-builtins.def: Update copyright years.
32359 * config/arm/arm_acle_builtins.def: Update copyright years.
32361 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32364 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32365 constraint instead of o for the stxsd instruction.
32367 2017-01-25 Carl Love <cel@us.ibm.com>
32369 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32370 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32372 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32374 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32376 2017-01-25 Richard Biener <rguenther@suse.de>
32378 PR tree-optimization/69264
32379 * target.def (vector_alignment_reachable): Improve documentation.
32380 * doc/tm.texi: Regenerate.
32381 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32383 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32384 earlier changes with respect to TYPE_USER_ALIGN.
32385 (vector_alignment_reachable_p): Likewise. Improve dumping.
32387 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32390 * config/arm/arm.md (xordi3): Force constant operand into a register
32393 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32395 * doc/invoke.texi (-fstore-merging): Correct default optimization
32396 levels at which it is enabled.
32397 (-O): Move -fstore-merging from list to...
32400 2017-01-25 Richard Biener <rguenther@suse.de>
32403 * omp-expand.c: Include debug.h.
32404 (expand_omp_taskreg): Make sure to generate early debug before
32405 outlining anything from a function.
32406 (expand_omp_target): Likewise.
32407 (grid_expand_target_grid_body): Likewise.
32409 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32412 * asan.c (get_translation_unit_decl): New function.
32413 (asan_add_global): Extract modules file name from globals
32414 TRANSLATION_UNIT_DECL name.
32416 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32419 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32420 for long calls with APCS frame and VFP.
32422 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32424 * cfg.c (original_copy_tables_initialized_p): New function.
32425 * cfg.h (original_copy_tables_initialized_p): New decl.
32426 * cfgrtl.c (relink_block_chain): Guard the call to
32427 free_original_copy_tables with a call to
32428 original_copy_tables_initialized_p.
32429 * cgraph.h (symtab_node::native_rtl_p): New decl.
32430 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32431 (symtab_node::needed_p): Don't assert for early assembly output
32432 for __RTL functions.
32433 (cgraph_node::finalize_function): Set "force_output" for __RTL
32435 (cgraph_node::analyze): Bail out early for __RTL functions.
32436 (analyze_functions): Update assertion to support __RTL functions.
32437 (cgraph_node::expand): Bail out early for __RTL functions.
32438 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32440 * function.h (struct function): Update comment for field
32442 * gimple-expr.c: Include "tree-pass.h".
32443 (gimple_has_body_p): Return false for __RTL functions.
32444 * Makefile.in (OBJS): Add run-rtl-passes.o.
32445 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32447 (gcc::pass_manager::get_clean_slate): New accessor.
32448 * passes.c: Include "insn-addr.h".
32449 (should_skip_pass_p): Add logging. Update logic for running
32450 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32451 property-provider override so it is only done for gimple passes.
32453 (skip_pass): New function.
32454 (execute_one_pass): Call skip_pass when skipping passes.
32455 * read-md.c (md_reader::read_char): Support filtering
32456 the input to a subset of line numbers.
32457 (md_reader::md_reader): Initialize fields
32458 m_first_line and m_last_line.
32459 (md_reader::read_file_fragment): New function.
32460 * read-md.h (md_reader::read_file_fragment): New decl.
32461 (md_reader::m_first_line): New field.
32462 (md_reader::m_last_line): New field.
32463 * read-rtl-function.c (function_reader::create_function): Only
32464 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32465 curr_properties. Set DECL_INITIAL to a dummy block.
32466 (read_rtl_function_body_from_file_range): New function.
32467 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32469 * run-rtl-passes.c: New file.
32470 * run-rtl-passes.h: New file.
32472 2017-01-24 Jeff Law <law@redhat.com>
32474 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32477 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32479 PR tree-optimization/79159
32480 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32481 (record_nonwrapping_iv): Improve boundary using above function if no
32482 value range information.
32484 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32485 Martin Jambor <mjambor@suse.cz>
32487 * brig-builtins.def: New file.
32488 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32489 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32490 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32491 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32492 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32493 * builtin-types.def (BT_INT8): New.
32494 (BT_INT16): Likewise.
32495 (BT_UINT8): Likewise.
32496 (BT_UINT16): Likewise.
32497 (BT_FN_ULONG): Likewise.
32498 (BT_FN_UINT_INT): Likewise.
32499 (BT_FN_UINT_ULONG): Likewise.
32500 (BT_FN_UINT_LONG): Likewise.
32501 (BT_FN_UINT_PTR): Likewise.
32502 (BT_FN_ULONG_PTR): Likewise.
32503 (BT_FN_INT8_FLOAT): Likewise.
32504 (BT_FN_INT16_FLOAT): Likewise.
32505 (BT_FN_UINT32_FLOAT): Likewise.
32506 (BT_FN_UINT16_FLOAT): Likewise.
32507 (BT_FN_UINT8_FLOAT): Likewise.
32508 (BT_FN_UINT64_FLOAT): Likewise.
32509 (BT_FN_UINT16_UINT32): Likewise.
32510 (BT_FN_UINT32_UINT16): Likewise.
32511 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32512 (BT_FN_INT_PTR_INT): Likewise.
32513 (BT_FN_UINT_PTR_UINT): Likewise.
32514 (BT_FN_LONG_PTR_LONG): Likewise.
32515 (BT_FN_ULONG_PTR_ULONG): Likewise.
32516 (BT_FN_VOID_UINT64_UINT64): Likewise.
32517 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32518 (BT_FN_INT8_INT8_INT8): Likewise.
32519 (BT_FN_INT16_INT16_INT16): Likewise.
32520 (BT_FN_INT_INT_INT): Likewise.
32521 (BT_FN_UINT_FLOAT_UINT): Likewise.
32522 (BT_FN_FLOAT_UINT_UINT): Likewise.
32523 (BT_FN_ULONG_UINT_UINT): Likewise.
32524 (BT_FN_ULONG_UINT_PTR): Likewise.
32525 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32526 (BT_FN_UINT_UINT_UINT): Likewise.
32527 (BT_FN_VOID_UINT_PTR): Likewise.
32528 (BT_FN_UINT_UINT_PTR: Likewise.
32529 (BT_FN_UINT32_UINT64_PTR): Likewise.
32530 (BT_FN_INT_INT_UINT_UINT): Likewise.
32531 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32532 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32533 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32534 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32535 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32536 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32537 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32538 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32539 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32540 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32541 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32542 * doc/frontends.texi: List BRIG FE.
32543 * doc/install.texi (Testing): Add BRIG tesring requirements.
32544 * doc/invoke.texi (Overall Options): Mention BRIG.
32545 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32547 2017-01-24 Richard Biener <rguenther@suse.de>
32549 PR translation/79208
32550 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32552 2017-01-24 Martin Jambor <mjambor@suse.cz>
32555 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32556 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32557 and known_contexts.
32559 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32561 PR middle-end/79123
32562 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32563 casts from signed to unsigned really don't have a range.
32565 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32567 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32568 GMP_RNDx for compatiblity.
32570 2017-01-24 Martin Liska <mliska@suse.cz>
32573 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32574 that would prevent us to call alloca with -1 as argument.
32576 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32578 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32579 Avoid -Wformat-security warning.
32581 2017-01-23 Andrew Pinski <apinski@cavium.com>
32583 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32586 2017-01-23 Martin Sebor <msebor@redhat.com>
32588 PR middle-end/78703
32589 * gimple-ssa-sprintf.c (warn_level): New global.
32590 (format_integer): Use it here and throughout the rest of the file.
32591 Use the same switch to compute sign as base.
32592 (maybe_warn): New function.
32593 (format_directive): Factor out warnings into maybe_warn.
32594 Add debugging output. Use warn_level.
32595 (add_bytes): Use warn_level.
32596 (pass_sprintf_length::compute_format_length): Add debugging output.
32597 (try_substitute_return_value): Same.
32598 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32600 PR middle-end/78703
32601 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32602 (struct fmtresult, format_integer, format_floating): Adjust.
32603 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32604 (get_string_length, format_string,format_directive): Same.
32605 (pass_sprintf_length::compute_format_length): Same.
32606 (try_substitute_return_value): Simplify slightly.
32608 PR middle-end/78703
32609 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32610 (fmtresult::operator+=): Outlined.
32611 (struct fmtresult): Add ctors.
32612 (struct conversion_spec): Rename...
32613 (struct directive): ...to this. Add and remove data members.
32614 (directive::set_width, directive::set_precision): New functions.
32615 (format_percent): Use fmtresult ctor.
32616 (get_width_and_precision): Remove.
32617 (format_integer): Make naming changes. Avoid computing width and
32619 (format_floating): Same. Adjust indentation.
32620 (format_character, format_none): New functions.
32621 (format_string): Moved character handling to format_character.
32622 (format_directive): Remove arguments, change return type.
32623 (parse_directive): New function.
32624 (pass_sprintf_length::compute_format_length): Move directive
32625 parsing to parse_directive.
32627 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32629 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32630 (assign_assembler_name_if_needed): ... this.
32631 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32632 (assign_assembler_name_if_needed): ... this.
32633 (free_lang_data_in_cgraph): Adjust callers.
32634 * cgraphunit.c (cgraph_node::analyze): Likewise.
32635 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32638 2017-01-23 Richard Biener <rguenther@suse.de>
32640 PR tree-optimization/79088
32641 PR tree-optimization/79188
32642 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32643 resetting loop bounds after last path deletion. Reset loop
32644 bounds of the target loop, make code match the comments.
32645 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32646 Make sure loops need no fixups.
32648 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32650 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32651 exponent support with double type for first argument.
32652 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32653 type returned by __builtin_vec_extract_sig,
32654 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32655 functions from "vector int" to "vector unsigned int" or from
32656 "vector long long int" to "vector unsigned long long int".
32657 Changed type returned by __builtin_vec_extract_exp,
32658 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32659 functions from "vector int" to "vector unsigned int" or from
32660 "vector long long int" to "vector unsigned long long int".
32661 Changed return type of __builtin_vec_test_data_class,
32662 __builtin_vec_test_data_class_sp, and
32663 __builtin_vec_test_data_class_dp from "vector int" to
32664 "vector bool int" or from "vector long long int" to "vector bool
32665 long long int" and changed second argument type from "unsigned
32666 int" to "int". Added new overloaded function forms "vector float
32667 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32668 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32669 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32670 double, vector unsigned long long int)" and "vector double
32671 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32672 long int)". Changed return type of
32673 __builtin_scalar_test_data_class and
32674 __builtin_scalar_test_data_class_sp and
32675 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32676 int" and changed second argument from "unsigned int" to "int".
32677 Changed type returned by __builtin_scalar_test_neg,
32678 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32679 from "int" to "bool int". Added new overloaded function form
32680 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32681 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32682 exponent double-precision with floating point first argument.
32683 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32684 documentation of scalar_test_data_class, scalar_test_neg,
32685 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32686 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32687 vec_test_data_class built-in functions to reflect refinements in
32688 their type signatures.
32690 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32692 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32694 (aarch64_elf_asm_destructor): Likewise.
32696 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32698 PR rtl-optimization/78634
32699 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32700 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32701 * ifcvt.c (noce_try_cmove): Add missing cost check.
32703 PR rtl-optimization/71724
32704 * combine.c (if_then_else_cond): Look for situations where it is
32705 beneficial to undo the work of one of the recursive calls.
32707 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32709 PR tree-optimization/70754
32710 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32711 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32712 combined stmt before it if not NULL.
32713 (combine_chains): Process refs reversely and compute dominance point
32716 2017-01-23 Martin Liska <mliska@suse.cz>
32718 PR tree-optimization/79196
32719 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32720 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32722 (strlen_optimize_stmt): Call the renamed function.
32724 2017-01-23 Michael Matz <matz@suse.de>
32726 PR tree-optimization/78384
32727 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32729 2017-01-23 Richard Biener <rguenther@suse.de>
32731 PR tree-optimization/79186
32732 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32733 both incoming edges before moving an assert.
32735 2017-01-23 Martin Jambor <mjambor@suse.cz>
32737 * ipa-prop.c (load_from_param_1): Removed.
32738 (load_from_unmodified_param): Bits from load_from_param_1 put back
32740 (load_from_param): Removed.
32741 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32742 with stmt. Reverted back to use of load_from_unmodified_param.
32744 2017-01-23 Martin Jambor <mjambor@suse.cz>
32747 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32748 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32749 field a pointer to garbage collected vector, mark lattices and
32750 ipcp_orig_node with GTY((skip)).
32751 (ipa_get_param_count): Adjust to descriptors being a pointer.
32752 (ipa_get_param): Likewise.
32753 (ipa_get_type): Likewise.
32754 (ipa_get_param_move_cost): Likewise.
32755 (ipa_set_param_used): Likewise.
32756 (ipa_get_controlled_uses): Likewise.
32757 (ipa_set_controlled_uses): Likewise.
32758 (ipa_is_param_used): Likewise.
32759 (ipa_node_params_t): Move into garbage collector. New methods insert
32761 (ipa_node_params_sum): Annotate wth GTY(()).
32762 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32764 (ipa_load_from_parm_agg): Adjust declaration.
32765 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32766 * ipa-profile.c (ipa_profile): Likewise.
32767 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32768 (ipa_populate_param_decls): Make descriptors parameter garbage
32770 (ipa_dump_param): Adjust to descriptors being a pointer.
32771 (ipa_alloc_node_params): Likewise.
32772 (ipa_initialize_node_params): Likewise.
32773 (load_from_param_1): Make descriptors parameter garbage collected.
32774 (load_from_unmodified_param): Likewise.
32775 (load_from_param): Likewise.
32776 (ipa_load_from_parm_agg): Likewise.
32777 (ipa_node_params::~ipa_node_params): Removed.
32778 (ipa_free_all_node_params): Remove call to delete operator.
32779 (ipa_node_params_t::insert): New.
32780 (ipa_node_params_t::remove): Likewise.
32781 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32782 copy known_csts and known_contexts vectors.
32783 (ipa_read_node_info): Adjust to descriptors being a pointer.
32784 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32786 (ipcp_transform_function): Make descriptors variable garbage
32789 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32791 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32792 * config/i386/avx512dqintrin.h: Ditto.
32793 * config/i386/avx512fintrin.h: Ditto.
32794 * config/i386/i386.c: Handle new builtins.
32795 * config/i386/i386-builtin.def: Add new builtins.
32796 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32797 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32799 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32800 Martin Liska <mliska@suse.cz>
32802 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32803 * asan.c (asan_expand_poison_ifn): Support stores and use
32804 appropriate ASAN report function.
32805 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32806 * internal-fn.def (ASAN_POISON_USE): Declare.
32807 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32808 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32809 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32810 ASAN_POISON calls w/o LHS.
32811 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32812 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32813 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32814 * gimplify.c (asan_poison_variables): Add attribute
32815 use_after_scope_memory to variables that really needs to live
32817 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32818 having the attribute.
32820 2017-01-23 Martin Liska <mliska@suse.cz>
32822 * asan.c (create_asan_shadow_var): New function.
32823 (asan_expand_poison_ifn): Likewise.
32824 * asan.h (asan_expand_poison_ifn): New declaration.
32825 * internal-fn.c (expand_ASAN_POISON): Likewise.
32826 * internal-fn.def (ASAN_POISON): New builtin.
32827 * sanopt.c (pass_sanopt::execute): Expand
32828 asan_expand_poison_ifn.
32829 * tree-inline.c (copy_decl_for_dup_finish): Make function
32831 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32832 * tree-ssa.c (is_asan_mark_p): New function.
32833 (execute_update_addresses_taken): Rewrite local variables
32834 (identified just by use-after-scope as addressable) into SSA.
32836 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32838 * doc/install.texi (Specific): opensource.apple.com uses https
32839 now. Remove trailing slash.
32841 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32843 * README.Portability: Remove note on an Irix compatibility issue.
32845 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32847 * gcov.c (INCLUDE_ALGORITHM): Define.
32848 (INCLUDE_VECTOR): Define.
32849 No longer include <vector> and <algorithm> directly.
32851 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32853 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32855 * doc/invoke.texi (Code Gen Options): Ditto.
32857 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32860 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32862 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32864 rtl-optimization/79125
32865 * cprop.c (local_cprop_pass): Handle cases where we make an
32866 unconditional trap.
32868 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32872 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32873 read from, for big endian.
32875 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32877 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32878 register pauth builtins for LP64 only.
32880 2017-01-20 Marek Polacek <polacek@redhat.com>
32883 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32886 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32888 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32890 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32891 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32892 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32894 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32897 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32898 in big-endian mode when they are not a single duplicated value.
32900 2017-01-20 Richard Biener <rguenther@suse.de>
32902 * BASE-VER: Bump to 7.0.1.
32904 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32906 * omp-low.c (omplow_simd_context): New struct. Use it...
32907 (lower_rec_simd_input_clauses): ...here and...
32908 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32909 references to idx, lane, max_vf, is_simt.
32911 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32913 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32916 2017-01-20 Martin Jambor <mjambor@suse.cz>
32918 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32919 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32921 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32922 (GTFILES): Rename hsa.c to hsa-common.c.
32923 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32924 * hsa-dump.c: Likewise.
32925 * hsa-gen.c: Likewise.
32926 * hsa-regalloc.c: Likewise.
32927 * ipa-hsa.c: Likewise.
32928 * omp-expand.c: Likewise.
32929 * omp-low.c: Likewise.
32930 * toplev.c: Likewise.
32932 2017-01-20 Marek Polacek <polacek@redhat.com>
32935 * doc/invoke.texi: Document -Wduplicated-branches.
32936 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32937 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32938 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32939 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32940 return 0 only when not OEP_LEXICOGRAPHIC.
32941 (fold_build_cleanup_point_expr): Use the expression
32942 location when building CLEANUP_POINT_EXPR.
32943 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32944 * tree.c (add_expr): Handle error_mark_node.
32946 2017-01-20 Martin Liska <mliska@suse.cz>
32949 * tree-profile.c (init_ic_make_global_vars): Do not call
32951 (gimple_init_gcov_profiler): Likewise.
32953 2017-01-20 Martin Liska <mliska@suse.cz>
32956 * cgraph.h (maybe_create_reference): Remove argument and
32958 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32960 * ipa-cp.c (create_specialized_node): Likewise.
32961 * symtab.c (symtab_node::maybe_create_reference): Handle
32962 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32964 2017-01-20 Martin Liska <mliska@suse.cz>
32966 * read-rtl-function.c (function_reader::create_function): Use
32967 build_decl instread of build_decl_stat.
32969 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32971 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32972 * config/i386/avx512dqintrin.h: Ditto.
32973 * config/i386/avx512fintrin.h: Ditto.
32974 * config/i386/i386-builtin-types.def: Add new types.
32975 * config/i386/i386.c: Handle new types.
32976 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32977 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32978 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32979 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32980 (__builtin_ia32_kshiftridi): New.
32981 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32983 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32987 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32988 define to rs6000_init_stack_protect_guard.
32989 (rs6000_init_stack_protect_guard): New function.
32991 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32992 Yunqiang Su <yunqiang.su@imgtec.com>
32994 * config.gcc (supported_defaults): Add madd4.
32995 (with_madd4): Add validation.
32996 (all_defaults): Add madd4.
32997 * config/mips/mips.opt (mmadd4): New option.
32998 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33000 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33002 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33003 (ISA_HAS_FUSED_MADD4): Likewise.
33004 * doc/invoke.texi (-mmadd4): Document the new option.
33005 * doc/install.texi (--with-madd4): Document the new option.
33007 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33009 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33010 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33011 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33012 (aarch64_init_pauth_hint_builtins): New.
33013 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33014 (aarch64_expand_builtin): Expand new builtins.
33016 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33018 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33019 * combine-stack-adj.c (no_unhandled_cfa): Handle
33020 REG_CFA_TOGGLE_RA_MANGLE.
33021 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33022 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33023 info for return address signing.
33024 (aarch64_expand_epilogue): Likewise.
33026 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33028 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33029 * config/aarch64/aarch64-protos.h
33030 (aarch64_return_address_signing_enabled): New declaration.
33031 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33033 (aarch64_expand_prologue): Sign return address before it's pushed onto
33035 (aarch64_expand_epilogue): Authenticate return address fetched from
33037 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33038 (aarch64_attributes): New function attributes for "sign-return-address".
33039 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33040 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33041 ("*do_return"): Generate combined instructions according to key index.
33042 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33043 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33045 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33046 * config/aarch64/aarch64.opt (msign-return-address=): New.
33047 * doc/extend.texi (AArch64 Function Attributes): Documents
33048 "sign-return-address=".
33049 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33051 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33053 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33054 overall option summary.
33056 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33058 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33059 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33060 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33061 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33063 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33065 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33066 -mpower9-minmax by default for -mcpu=power9.
33067 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33068 128-bit floating point.
33070 2017-01-20 Alan Modra <amodra@gmail.com>
33072 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33073 optimizing for size.
33075 2017-01-20 Alan Modra <amodra@gmail.com>
33078 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33079 for strcmp and strncmp from corresponding builtin decl.
33081 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33083 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33084 instead of i386/rtems-64.h.
33085 * config/i386/rtems-64.h: Remove.
33087 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33091 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33093 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33095 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33097 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33098 Change int to HOST_WIDE_INT.
33099 * config/aarch64/aarch64-protos.h
33100 (aarch64_simd_gen_const_vector_dup): Likewise.
33101 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33103 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33105 * langhooks-def.h (lhd_type_for_size): New decl.
33106 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33107 * langhooks.c (lhd_type_for_size): New function, taken from
33110 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33112 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33113 define_bypass for CR latency.
33114 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33115 (power9-alu2): Add define_bypass for CR latency.
33117 (power9-mul): Update insn latency.
33118 (power9-mul-compare): Update insn latency, bypass latency and remove
33121 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33123 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33125 * config/aarch64/aarch64.md
33126 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33127 aarch64_nopcrelative_literal_loads.
33128 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33130 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33132 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33133 TARGET_LOONGSON_3A.
33134 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33136 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33139 * config.gcc (supported_defaults): Add lxc1-sxc1.
33140 (with_lxc1_sxc1): Add validation.
33141 (all_defaults): Add lxc1-sxc1.
33142 * config/mips/mips.opt (mlxc1-sxc1): New option.
33143 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33145 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33146 __mips_no_lxc1_sxc1.
33147 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33148 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33149 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33151 2017-01-19 Richard Biener <rguenther@suse.de>
33153 PR tree-optimization/72488
33154 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33155 sure to restore SSA info.
33156 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33158 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33160 PR rtl-optimization/79121
33161 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33162 of the inner type when shifting an extended value.
33164 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33167 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33168 interposable aliases.
33170 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33173 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33174 Use the evmergelohi instruction.
33175 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33176 (mov_sitf_e500_subreg8_2_be): Likewise.
33177 (mov_sitf_e500_subreg12_2_le): Likewise.
33178 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33179 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33180 (mov_sitf_e500_subreg8_2_le): Likewise.
33181 (mov_sitf_e500_subreg12_2_be): Likewise.
33183 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33185 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33186 attribute from vecsimple to vecperm.
33187 (altivec_vbpermq2): Likewise.
33189 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33192 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33194 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33195 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33196 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33197 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33198 case where N arg is SIZE_MAX.
33199 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33200 (cmpstrsi): Add pattern.
33202 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33204 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33205 __builtin_vec_revb builtins.
33206 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33207 built-in functions to support generation of the ISA 3.0 XXBR<x>
33208 vector byte reverse instructions.
33209 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33210 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33211 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33212 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33213 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33214 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33215 (P9V_BUILTIN_VEC_REVB): Likewise.
33216 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33217 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33218 (p9_xxbrq_v16qi): Likewise.
33219 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33220 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33221 (p9_xxbrh_v8hi): Likewise.
33222 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33223 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33224 vec_revb built-in functions.
33226 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33228 PR rtl-optimization/78952
33229 * config/i386/i386.md (any_extract): New code iterator.
33230 (*insvqi_2): Use any_extract for source operand.
33231 (*insvqi_3): Use any_shiftrt for source operand.
33233 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33235 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33237 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33239 2017-01-18 Matthias Klose <doko@ubuntu.com>
33241 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33243 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33245 * config/rs6000/altivec.h (vec_bperm): Change #define.
33246 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33247 (altivec_vbpermq2): New define_insn.
33248 (altivec_vbpermd): Likewise.
33249 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33250 function interface.
33251 (VBPERMD): Likewise.
33252 (VBPERM): New polymorphic function interface.
33253 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33254 Add entries for P9V_BUILTIN_VEC_VBPERM.
33255 * doc/extend.texi: Add interfaces for vec_bperm.
33257 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33259 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33260 first letter of error messages.
33261 (s390_resolve_overloaded_builtin): Likewise.
33262 * config/s390/s390.c (s390_expand_builtin): Likewise.
33263 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33264 (s390_valid_target_attribute_inner_p): Likewise.
33265 * config/s390/s390.md ("tabort"): Likewise.
33267 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33269 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33270 (ISA_AVOID_DIV_HILO): New macro.
33271 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33272 (ISA_HAS_DDIV): Likewise.
33274 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33276 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33278 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33280 * doc/invoke.texi (fabi-version): Spelling fix.
33282 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33285 * doc/invoke.texi (fabi-version): Mention mangling fix for
33288 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33291 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33293 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33296 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33297 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33299 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33301 (stack_protect_test): Ditto.
33302 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33303 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33305 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33306 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33307 -mstack-protector-guard-offset=.
33308 (RS/6000 and PowerPC Options): Ditto.
33310 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33312 * config/i386/i386.h (MASK_CLASS_P): New define.
33313 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33314 there are no registers from different register sets also when
33315 mask registers are used. Update function comment.
33316 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33317 to (*k/*r) and (*k/*km) alternatives.
33319 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33321 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33322 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33323 (EH_RETURN_HANDLER_RTX): New define.
33324 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33325 Force frame pointer in EH return functions.
33326 (aarch64_expand_epilogue): Add barrier for eh_return.
33327 (aarch64_final_eh_return_addr): Remove.
33328 (aarch64_eh_return_handler_rtx): New function.
33329 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33331 (aarch64_eh_return_handler_rtx): New prototype.
33333 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33335 * config/rs6000/altivec.h (vec_rlmi): New #define.
33336 (vec_vrlnm): Likewise.
33337 (vec_rlnm): Likewise.
33338 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33339 (UNSPEC_VRLNM): Likewise.
33340 (VIlong): New mode iterator.
33341 (altivec_vrl<VI_char>mi): New define_insn.
33342 (altivec_vrl<VI_char>nm): Likewise.
33343 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33345 (VRLDNM): Likewise.
33346 (RLNM): New polymorphic function entry.
33347 (VRLWMI): New monomorphic function entry.
33348 (VRLDMI): Likewise.
33349 (RLMI): New polymorphic function entry.
33350 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33351 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33352 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33355 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33358 * dwarf2out.c (field_byte_offset): Restore the
33359 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33360 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33361 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33364 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33367 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33369 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33372 * dwarf2out.c (add_data_member_location_attribute): For constant
33373 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33374 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33375 DW_AT_byte_size attributes.
33377 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33379 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33380 after forcing to constant memory when the code model is medium.
33382 2017-01-17 Julia Koval <julia.koval@intel.com>
33385 * config/i386/avx512fintrin.h
33386 (_mm512_i32gather_ps): Change __addr type to void const*.
33387 (_mm512_mask_i32gather_ps): Ditto.
33388 (_mm512_i32gather_pd): Ditto.
33389 (_mm512_mask_i32gather_pd): Ditto.
33390 (_mm512_i64gather_ps): Ditto.
33391 (_mm512_mask_i64gather_ps): Ditto.
33392 (_mm512_i64gather_pd): Ditto.
33393 (_mm512_mask_i64gather_pd): Ditto.
33394 (_mm512_i32gather_epi32): Ditto.
33395 (_mm512_mask_i32gather_epi32): Ditto.
33396 (_mm512_i32gather_epi64): Ditto.
33397 (_mm512_mask_i32gather_epi64): Ditto.
33398 (_mm512_i64gather_epi32): Ditto.
33399 (_mm512_mask_i64gather_epi32): Ditto.
33400 (_mm512_i64gather_epi64): Ditto.
33401 (_mm512_mask_i64gather_epi64): Ditto.
33402 (_mm512_i32scatter_ps): Change __addr type to void*.
33403 (_mm512_mask_i32scatter_ps): Ditto.
33404 (_mm512_i32scatter_pd): Ditto.
33405 (_mm512_mask_i32scatter_pd): Ditto.
33406 (_mm512_i64scatter_ps): Ditto.
33407 (_mm512_mask_i64scatter_ps): Ditto.
33408 (_mm512_i64scatter_pd): Ditto.
33409 (_mm512_mask_i64scatter_pd): Ditto.
33410 (_mm512_i32scatter_epi32): Ditto.
33411 (_mm512_mask_i32scatter_epi32): Ditto.
33412 (_mm512_i32scatter_epi64): Ditto.
33413 (_mm512_mask_i32scatter_epi64): Ditto.
33414 (_mm512_i64scatter_epi32): Ditto.
33415 (_mm512_mask_i64scatter_epi32): Ditto.
33416 (_mm512_i64scatter_epi64): Ditto.
33417 (_mm512_mask_i64scatter_epi64): Ditto.
33418 * config/i386/avx512pfintrin.h
33419 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33420 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33421 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33422 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33423 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33424 (_mm512_prefetch_i32scatter_ps): Ditto.
33425 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33426 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33427 (_mm512_prefetch_i64scatter_pd): Ditto.
33428 (_mm512_prefetch_i64scatter_ps): Ditto.
33429 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33430 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33431 * config/i386/avx512vlintrin.h
33432 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33433 (_mm_mmask_i32gather_ps): Ditto.
33434 (_mm256_mmask_i32gather_pd): Ditto.
33435 (_mm_mmask_i32gather_pd): Ditto.
33436 (_mm256_mmask_i64gather_ps): Ditto.
33437 (_mm_mmask_i64gather_ps): Ditto.
33438 (_mm256_mmask_i64gather_pd): Ditto.
33439 (_mm_mmask_i64gather_pd): Ditto.
33440 (_mm256_mmask_i32gather_epi32): Ditto.
33441 (_mm_mmask_i32gather_epi32): Ditto.
33442 (_mm256_mmask_i32gather_epi64): Ditto.
33443 (_mm_mmask_i32gather_epi64): Ditto.
33444 (_mm256_mmask_i64gather_epi32): Ditto.
33445 (_mm_mmask_i64gather_epi32): Ditto.
33446 (_mm256_mmask_i64gather_epi64): Ditto.
33447 (_mm_mmask_i64gather_epi64): Ditto.
33448 (_mm256_i32scatter_ps): Change __addr type to void*.
33449 (_mm256_mask_i32scatter_ps): Ditto.
33450 (_mm_i32scatter_ps): Ditto.
33451 (_mm_mask_i32scatter_ps): Ditto.
33452 (_mm256_i32scatter_pd): Ditto.
33453 (_mm256_mask_i32scatter_pd): Ditto.
33454 (_mm_i32scatter_pd): Ditto.
33455 (_mm_mask_i32scatter_pd): Ditto.
33456 (_mm256_i64scatter_ps): Ditto.
33457 (_mm256_mask_i64scatter_ps): Ditto.
33458 (_mm_i64scatter_ps): Ditto.
33459 (_mm_mask_i64scatter_ps): Ditto.
33460 (_mm256_i64scatter_pd): Ditto.
33461 (_mm256_mask_i64scatter_pd): Ditto.
33462 (_mm_i64scatter_pd): Ditto.
33463 (_mm_mask_i64scatter_pd): Ditto.
33464 (_mm256_i32scatter_epi32): Ditto.
33465 (_mm256_mask_i32scatter_epi32): Ditto.
33466 (_mm_i32scatter_epi32): Ditto.
33467 (_mm_mask_i32scatter_epi32): Ditto.
33468 (_mm256_i32scatter_epi64): Ditto.
33469 (_mm256_mask_i32scatter_epi64): Ditto.
33470 (_mm_i32scatter_epi64): Ditto.
33471 (_mm_mask_i32scatter_epi64): Ditto.
33472 (_mm256_i64scatter_epi32): Ditto.
33473 (_mm256_mask_i64scatter_epi32): Ditto.
33474 (_mm_i64scatter_epi32): Ditto.
33475 (_mm_mask_i64scatter_epi32): Ditto.
33476 (_mm256_i64scatter_epi64): Ditto.
33477 (_mm256_mask_i64scatter_epi64): Ditto.
33478 (_mm_i64scatter_epi64): Ditto.
33479 (_mm_mask_i64scatter_epi64): Ditto.
33480 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33481 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33482 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33483 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33484 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33485 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33486 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33487 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33488 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33489 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33490 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33491 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33492 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33493 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33494 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33495 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33496 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33497 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33498 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33499 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33500 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33501 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33502 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33503 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33504 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33505 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33506 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33507 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33508 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33509 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33510 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33511 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33512 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33513 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33514 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33515 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33516 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33517 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33518 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33519 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33520 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33521 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33522 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33523 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33524 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33525 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33526 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33527 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33528 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33529 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33530 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33531 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33532 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33533 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33534 definitions accordingly.
33536 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33537 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33540 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33543 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33546 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33547 hard regs for inner regno.
33549 2017-01-17 Martin Liska <mliska@suse.cz>
33552 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33553 assumption and add comment.
33555 2017-01-17 Nathan Sidwell <nathan@acm.org>
33557 * ipa-visibility.c (localize_node): New function, broken out of ...
33558 (function_and_variable_visibility): ... here. Call it.
33560 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33562 PR middle-end/77445
33563 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33564 correctly set frequency of oudgoing edge.
33565 (duplicate_thread_path): Fix profile updating.
33567 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33570 * configure.ac: Add GCC_BASE_VER.
33571 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33572 version from BASE-VER file.
33573 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33574 (gcc.o): Depend on $(BASEVER).
33575 * common.opt (dumpfullversion): New option.
33576 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33577 * doc/invoke.texi: Document -dumpfullversion.
33578 * doc/install.texi: Document --with-gcc-major-version-only.
33579 * configure: Regenerated.
33581 2017-01-17 Richard Biener <rguenther@suse.de>
33583 PR tree-optimization/71433
33584 * tree-vrp.c (register_new_assert_for): Merge same asserts
33585 on all incoming edges.
33586 (process_assert_insertions_for): Handle insertions at the
33589 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33591 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33592 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33594 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33597 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33600 2017-01-17 Alan Modra <amodra@gmail.com>
33603 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33604 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33605 symbolic stack limit when pic.
33607 2017-01-16 Martin Sebor <msebor@redhat.com>
33609 PR tree-optimization/78608
33610 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33612 2017-01-16 Jeff Law <law@redhat.com>
33615 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33616 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33617 for several include directories that may be relative to sysroot.
33618 * config/i386/x-mingw32 (gplus_includedir): Define.
33619 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33620 (native_system_includedir): Likewise.
33621 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33622 override if TARGET_SYSTEM_ROOT is defined.
33623 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33625 PR tree-optimization/79090
33626 PR tree-optimization/33562
33627 PR tree-optimization/61912
33628 PR tree-optimization/77485
33629 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33630 and computed trims into the dump file.
33632 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33634 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33636 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33639 * gimplify.c (gimplify_init_constructor): If want_value and
33640 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33644 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33645 sequence. Formatting fixes.
33646 (doloop_optimize): Formatting fixes.
33649 * gcc.c (debug_level_greater_than_spec_func): New function.
33650 (static_spec_functions): Add debug-level-gt spec function.
33651 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33653 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33654 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33655 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33656 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33657 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33658 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33660 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33662 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33663 QImode fixups to general and mask registers only.
33665 2017-01-16 Carl Love <cel@us.ibm.com>
33667 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33668 for built-in functions
33669 vector signed char vec_nabs (vector signed char)
33670 vector signed short vec_nabs (vector signed short)
33671 vector signed int vec_nabs (vector signed int)
33672 vector signed long long vec_nabs (vector signed long long)
33673 vector float vec_nabs (vector float)
33674 vector double vec_nabs (vector double)
33675 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33677 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33678 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33679 * doc/extend.texi: Update the documentation file for the new built-in
33682 2017-01-16 Martin Sebor <msebor@redhat.com>
33684 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33687 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33689 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33690 UNSPEC_VSX__XXSPLTD to require special splat handling.
33692 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33695 * system.h: Poison strndup.
33697 2017-01-16 Alan Modra <amodra@gmail.com>
33700 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33703 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33705 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33707 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33709 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33710 call recog here. Assert that INSN_CODE (insn) is non-negative.
33712 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33715 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33717 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33718 in the currently scheduled RTL fragment.
33720 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33722 PR rtl-optimization/78751
33723 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33726 2017-01-14 Jeff Law <law@redhat.com>
33728 PR tree-optimization/79090
33729 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33730 variable length stores.
33731 (compute_trims): Delete dead assignment to *trim_tail.
33732 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33735 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33737 PR rtl-optimization/78626
33738 PR rtl-optimization/78727
33739 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33740 of a block, and split such blocks after everything else is finished.
33742 2017-01-14 Alan Modra <amodra@gmail.com>
33745 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33746 target legitimate_combined_insn.
33747 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33748 (rs6000_legitimate_combined_insn): New function.
33749 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33751 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33752 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33753 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33755 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33757 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33759 2017-01-13 Jeff Law <law@redhat.com>
33761 PR tree-optimization/33562
33762 PR tree-optimization/61912
33763 PR tree-optimization/77485
33764 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33766 (delete_dead_assignment): Likewise.
33767 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33768 statement to delete_dead_call and delete_dead_assignment.
33770 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33773 * substring-locations.c (format_warning_va): Strengthen case 1 so
33774 that both endpoints of the substring must be within the format
33775 range for just the substring to be printed.
33777 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33779 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33780 * config/i386/i386.c (ix86_target_string): Add missing options
33781 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33782 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33783 flags_other and ix86_target_other to flags2_other. Display unknown
33785 (ix86_valid_target_attribute_inner_p): Add missing options and
33786 reorder options by implied ISAs, as in ix86_target_string.
33788 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33790 * hash-table.h (hash_table::too_empty_p): New function.
33791 (hash_table::expand): Use it.
33792 (hash_table::traverse): Likewise.
33793 (hash_table::empty_slot): Use sizeof (value_type) instead of
33794 sizeof (PTR) to convert bytes to elements. Shrink the table
33795 if the current size is excessive for the current number of
33798 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33800 * ira-costs.c (record_reg_classes): Break from the inner loop
33801 early once alt_fail is known to be true. Update outer loop
33802 handling accordingly.
33804 2017-01-13 Jeff Law <law@redhat.com>
33806 * tree-ssa-dse.c (decrement_count): New function.
33807 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33808 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33809 when we know the partially dead statement is a mem* function.
33811 PR tree-optimization/61912
33812 PR tree-optimization/77485
33813 * tree-ssa-dse.c: Include expr.h.
33814 (maybe_trim_constructor_store): New function.
33815 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33817 PR tree-optimization/33562
33818 PR tree-optimization/61912
33819 PR tree-optimization/77485
33820 * doc/invoke.texi: Document new dse-max-object-size param.
33821 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33822 * tree-ssa-dse.c: Include params.h.
33823 (dse_store_status): New enum.
33824 (initialize_ao_ref_for_dse): New, partially extracted from
33826 (valid_ao_ref_for_dse, normalize_ref): New.
33827 (setup_live_bytes_from_ref, compute_trims): Likewise.
33828 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33829 (maybe_trim_partially_dead_store): Likewise.
33830 (maybe_trim_complex_store): Likewise.
33831 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33832 Track what bytes live from the original store. Return tri-state
33833 for dead, partially dead or live.
33834 (dse_dom_walker): Add constructor, destructor and new private members.
33835 (delete_dead_call, delete_dead_assignment): New extracted from
33837 (dse_optimize_stmt): Make a member of dse_dom_walker.
33838 Use initialize_ao_ref_for_dse.
33840 PR tree-optimization/33562
33841 PR tree-optimization/61912
33842 PR tree-optimization/77485
33843 * sbitmap.h (bitmap_count_bits): Prototype.
33844 (bitmap_clear_range, bitmap_set_range): Likewise.
33845 * sbitmap.c (bitmap_clear_range): New function.
33846 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33848 2017-01-13 Martin Liska <mliska@suse.cz>
33851 * function.c (set_cfun): Add new argument force.
33852 * function.h (set_cfun): Likewise.
33853 * ipa-inline-transform.c (inline_call): Use the function when
33854 strict alising from is dropped for function we inline to.
33856 2017-01-13 Richard Biener <rguenther@suse.de>
33858 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33859 for dumping GIMPLE INTEGER_CSTs.
33861 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33863 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33864 to 201112L since C++17.
33866 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33869 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33870 if -fsanitize=kernel-address is present.
33872 2017-01-13 Richard Biener <rguenther@suse.de>
33874 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33875 as _Literal ( type ) number in case usual suffixes do not
33876 preserve all information.
33878 2017-01-13 Richard Biener <rguenther@suse.de>
33880 PR tree-optimization/77283
33881 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33882 and ssa-iterators.h.
33883 (is_feasible_trace): Implement a cost model based on joiner
33886 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33889 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33890 char or short to __float128/_Float128 directly.
33892 2017-01-12 Martin Sebor <msebor@redhat.com>
33894 to -Wformat-overflow.
33895 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33896 (min_bytes_remaining): Same.
33897 (get_string_length): Same.
33898 (format_string): Same.
33899 (format_directive): Same.
33901 (pass_sprintf_length::handle_gimple_call): Same.
33903 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33905 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33906 info.nowrite calls with no lhs that can't throw. Return bool
33907 whether gsi_remove has been called or not.
33908 (pass_sprintf_length::handle_gimple_call): Return bool whether
33909 try_substitute_return_value called gsi_remove. Formatting fix.
33910 (pass_sprintf_length::execute): Don't use gsi_remove if
33911 handle_gimple_call returned true.
33914 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33915 be removed due to side-effects, don't remove following barrier nor
33916 turn the successor edge into fallthru edge.
33918 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33921 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33922 element-reversing loads and stores as not swappable.
33924 2017-01-12 Nathan Sidwell <nathan@acm.org>
33925 Nicolai Stange <nicstange@gmail.com>
33927 * combine.c (try_combine): Don't ignore result of overlap checking
33928 loop. Combine overlap & asm check into single loop.
33930 2017-01-12 Richard Biener <rguenther@suse.de>
33932 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33933 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33935 2017-01-12 Richard Biener <rguenther@suse.de>
33937 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33938 and TS_TARGET_OPTION directly derive from TS_BASE.
33939 * tree-core.h (tree_optimization_option): Derive from tree_base.
33940 (tree_target_option): Likewise.
33942 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33944 * config/i386/i386.c (memory_address_length): Increase len
33945 only when rip_relative_addr_p returns false.
33947 2017-01-11 Julia Koval <julia.koval@intel.com>
33949 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33950 (OPTION_MASK_ISA_SGX_SET): New.
33951 (ix86_handle_option): Handle OPT_msgx.
33952 * config.gcc: Added sgxintrin.h.
33953 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33954 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33955 * config/i386/i386.c (ix86_target_string): Add -msgx.
33957 (ix86_option_override_internal): Handle new options.
33958 (ix86_valid_target_attribute_inner_p): Add sgx.
33959 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33960 * config/i386/i386.opt: Add msgx.
33961 * config/i386/sgxintrin.h: New file.
33962 * config/i386/x86intrin.h: Add sgxintrin.h.
33964 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33967 * fold-const.c (maybe_nonzero_address): Return 1 for function
33969 (tree_single_nonzero_warnv_p): Don't handle function local objects
33973 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33976 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33979 * opts.c: Include "spellcheck.h"
33980 (struct string_fragment): New struct.
33981 (struct edit_distance_traits<const string_fragment &>): New
33983 (get_closest_sanitizer_option): New function.
33984 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33986 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33988 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33990 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33991 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33992 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33993 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33994 for initial die_offset if dwarf_split_debug_info.
33995 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33996 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33998 (output_skeleton_debug_sections): Formatting fix. Use
33999 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34000 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34002 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34004 * config/arm/cortex-a53.md: Add bypasses for
34005 cortex_a53_r2f_cvt.
34006 (cortex_a53_r2f): Only use for transfers.
34007 (cortex_a53_f2r): Likewise.
34008 (cortex_a53_r2f_cvt): Add reservation for conversions.
34009 (cortex_a53_f2r_cvt): Likewise.
34011 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34013 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34014 to all inlined functions, change static to extern.
34016 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34019 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34021 (arm_assemble_integer): Likewise.
34023 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34025 * config.gcc: Use new awk script to check CPU, FPU and architecture
34026 parameters for --with-... options.
34027 * config/arm/parsecpu.awk: New file
34028 * config/arm/arm-cpus.in: New file.
34029 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34031 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34033 * config/arm/t-arm: Update dependency rules.
34034 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34035 of processing .def files.
34036 * config/arm/genopt.sh: Deleted.
34037 * config/arm/gentune.sh: Deleted.
34038 * config/arm/arm-cores.def: Deleted.
34039 * config/arm/arm-arches.def: Deleted.
34040 * config/arm/arm-fpus.def: Deleted.
34041 * config/arm/arm-tune.md: Regenerated.
34042 * config/arm/arm-tables.opt: Regenerated.
34043 * config/arm/arm-cpu.h: New generated file.
34044 * config/arm/arm-cpu-data.h: New generated file.
34045 * config/arm/arm-cpu-cdata.h: New generated file.
34047 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34050 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34052 (input_varpool_node): Unpack dynamically_initialized bit.
34054 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34056 PR rtl-optimization/79032
34057 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34058 the alignment of the adjusted memory reference against that of MODE,
34059 instead of the alignment of the original memory reference.
34061 2017-01-11 Martin Jambor <mjambor@suse.cz>
34063 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34065 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34066 decorated functions.
34068 2017-01-11 Richard Biener <rguenther@suse.de>
34070 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34071 set range/nonnull info for PHI results. Do not set it on
34072 stmts marked for removal.
34074 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34076 * expr.c (store_field): In the bitfield case, fetch the return value
34077 from the registers before applying a single big-endian adjustment.
34078 Always do a final load for a BLKmode value not larger than a word.
34080 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34083 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34084 that we correctly handle column numbers greater than
34085 LINE_MAP_MAX_COLUMN_NUMBER.
34087 2017-01-10 Martin Sebor <msebor@redhat.com>
34089 PR middle-end/78245
34090 * gimple-ssa-sprintf.c (get_destination_size): Call
34091 {init,fini}object_sizes.
34092 * tree-object-size.c (addr_object_size): Adjust.
34093 (pass_through_call): Adjust.
34094 (pass_object_sizes::execute): Adjust.
34095 * tree-object-size.h (fini_object_sizes): Declare.
34097 2017-01-10 Martin Sebor <msebor@redhat.com>
34099 PR tree-optimization/78775
34100 * builtins.c (get_size_range): Move...
34101 * calls.c: ...to here.
34102 (alloc_max_size): Accept zero argument.
34103 (operand_signed_p): Remove.
34104 (maybe_warn_alloc_args_overflow): Call get_size_range.
34105 * calls.h (get_size_range): Declare.
34107 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34109 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34110 from TI's devices.csv file as of September 2016.
34111 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34113 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34115 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34116 * doc/invoke.texi: Likewise.
34117 * doc/md.texi: Likewise.
34118 * doc/objc.texi: Likewise.
34120 2017-01-10 Joshua Conner <joshconner@google.com>
34122 * config/arm/fuchsia-elf.h: New file.
34123 * config/fuchsia.h: New file.
34124 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34125 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34127 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34129 2016-01-10 Richard Biener <rguenther@suse.de>
34131 PR tree-optimization/79034
34132 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34133 Propagate out degenerate PHIs in the joiner.
34135 2017-01-10 Martin Liska <mliska@suse.cz>
34137 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34138 (sort_congruence_classes_by_decl_uid): Likewise.
34139 (sort_congruence_class_groups_by_decl_uid): Likewise.
34140 (sem_item_optimizer::merge_classes): Sort class, groups in these
34141 classes and members in the groups by DECL_UID of declarations.
34142 This would make merge operations stable.
34144 2017-01-10 Martin Liska <mliska@suse.cz>
34146 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34147 usage of m_classes_vec.
34148 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34149 (sem_item_optimizer::get_group_by_hash): Likewise.
34150 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34151 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34152 (sem_item_optimizer::verify_classes): Likewise.
34153 (sem_item_optimizer::process_cong_reduction): Likewise.
34154 (sem_item_optimizer::dump_cong_classes): Likewise.
34155 (sem_item_optimizer::merge_classes): Likewise.
34156 * ipa-icf.h (congruence_class_hash): Rename from
34157 congruence_class_group_hash. Remove declaration of m_classes_vec.
34159 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34161 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34162 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34163 * config.gcc: Add avx512vpopcntdqintrin.h.
34164 * config/i386/avx512vpopcntdqintrin.h: New.
34165 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34166 * config/i386/i386-builtin-types.def: Add new types.
34167 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34168 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34169 __builtin_ia32_vpopcountq_v8di_mask): New.
34170 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34171 __AVX512VPOPCNTDQ__.
34172 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34173 (PTA_AVX512VPOPCNTDQ): Define.
34174 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34175 TARGET_AVX512VPOPCNTDQ_P): Define.
34176 * config/i386/i386.opt: Add mavx512vpopcntdq.
34177 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34178 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34180 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34182 PR middle-end/77484
34183 * predict.def (PRED_CALL): Set to 67.
34185 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34187 * expr.c (store_field): In the bitfield case, if the value comes from
34188 a function call and is of an aggregate type returned in registers, do
34189 not modify the field mode; extract the value in all cases if the mode
34190 is BLKmode and the size is not larger than a word.
34192 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34195 * config/i386/cpuid.h: Fix undefined behavior.
34197 2017-01-04 Jeff Law <law@redhat.com>
34199 PR tree-optimization/79007
34200 PR tree-optimization/67955
34201 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34202 conservative for pt.null when flag_non_call_exceptions is on.
34204 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34206 PR translation/79019
34207 PR translation/79020
34208 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34209 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34210 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34212 * config/avr/avr.opt (maccumulate-args): Likewise.
34213 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34214 * common.opt (freport-bug): Likewise.
34215 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34216 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34217 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34218 translatable string.
34219 * config/i386/i386.c (function_value_32): Likewise.
34220 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34221 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34223 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34224 * common/config/msp430/msp430-common.c (msp430_handle_option):
34226 * symtab.c (symtab_node::verify_base): Likewise.
34227 * opts.c (set_debug_level): Likewise.
34228 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34229 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34230 missing whitespace to translatable strings.
34231 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34232 * config/sh/superh.h: Likewise.
34233 * config/i386/xopintrin.h: Likewise.
34234 * config/i386/znver1.md: Likewise.
34235 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34236 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34237 * double-int.h (struct double_int): Likewise.
34238 * double-int.c (div_and_round_double): Likewise.
34239 * wide-int.cc: Likewise.
34240 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34241 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34242 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34243 (crossjumps_occurred): ... this.
34244 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34247 PR tree-optimization/78899
34248 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34249 returning bool return struct loop *, NULL for failure and the new
34251 (versionable_outer_loop_p): Don't version outer loop if it has
34252 dont_vectorized bit set.
34253 (tree_if_conversion): When versioning outer loop, ensure
34254 tree_if_conversion is performed also on the inner loop of the
34255 non-vectorizable outer loop copy.
34256 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34257 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34258 prevent vectorization of it.
34259 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34260 the outer loop vectorization of the non-scalar version is attempted
34261 before vectorization of the inner loop in scalar version. If
34262 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34263 vectorization of its inner loop.
34264 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34265 has 2 inner loops, rename also on edges from bb whose single pred
34266 is outer_loop->header. Fix typo in function comment.
34268 2017-01-09 Martin Sebor <msebor@redhat.com>
34271 * asan.c (asan_emit_stack_protection): Increase local buffer size
34272 to avoid snprintf truncation warning.
34274 2017-01-09 Andrew Pinski <apinski@cavium.com>
34276 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34277 to reference thunderx2t99 for the tuning structure
34278 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34280 (thunderx2t99_extra_costs): This.
34281 * config/aarch64/aarch64-tune.md: Regenerate.
34282 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34283 (vulcan_addrcost_table): This.
34284 (vulcan_regmove_cost): Rename to ...
34285 (thunderx2t99_regmove_cost): This.
34286 (vulcan_vector_cost): Rename to ...
34287 (thunderx2t99_vector_cost): this.
34288 (vulcan_branch_cost): Rename to ...
34289 (thunderx2t99_branch_cost): This.
34290 (vulcan_tunings): Rename to ...
34291 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34292 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34294 2017-01-09 Martin Jambor <mjambor@suse.cz>
34298 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34299 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34300 (propagate_vr_accross_jump_function): Use the above function for all
34301 value range computations for pass-through jump functions and type
34302 converasion from explicit value range values.
34303 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34304 parameters from TYPE_ARG_TYPES.
34305 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34306 (ipa_write_node_info): Stream type of the actual argument.
34307 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34309 2017-01-09 Martin Liska <mliska@suse.cz>
34312 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34313 (lookup_compiler): Do not show error message with have_E.
34315 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34317 PR tree-optimization/78938
34318 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34319 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34320 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34321 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34324 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34326 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34329 2017-01-09 Richard Biener <rguenther@suse.de>
34331 PR tree-optimization/78997
34332 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34333 name condition properly.
34335 2017-01-09 Richard Biener <rguenther@suse.de>
34338 * dwarf2out.c (is_cxx): New overload with context.
34339 (is_naming_typedef_decl): Use it.
34341 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34343 * invoke.texi (Option Summary): Correct spacing in option lists
34344 and add line breaks to fix over-long lines.
34346 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34348 PR middle-end/17660
34350 * extend.texi (Common Variable Attributes): Add xref to GCC
34351 Internals manual to explain mode attribute keywords.
34353 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34356 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34357 and Preprocessor Options.
34358 (Options for Linking): Document -pthread here....
34359 (RS/6000 and PowerPC Options): ...not here.
34360 (Solaris 2 Options): ...or here.
34361 * doc/cppopts.texi: Document -pthread.
34363 2017-01-08 Martin Sebor <msebor@redhat.com>
34365 PR middle-end/77708
34366 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34367 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34368 New member functions.
34369 (format_directive): Used them.
34371 (pass_sprintf_length::handle_gimple_call): Same.
34372 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34373 to avoid truncation for any argument.
34374 (extract_affine_mul): Same.
34375 * tree.c (get_file_function_name): Same.
34377 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34379 PR middle-end/77484
34380 * predict.def (PRED_INDIR_CALL): Set to 86.
34382 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34384 PR preprocessor/54124
34385 * doc/cppopts.texi: Reformat -d subtable to list the full name
34386 of the options. Add cross-reference to the docs for the general
34387 compiler -d options.
34388 * doc/invoke.texi (Developer Options): Add cross-reference to the
34389 preprocessor-specific -d option documentation.
34391 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34393 PR preprocessor/13498
34394 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34395 redudant material, and reflect new command-line options.
34396 (System Headers): Likewise.
34398 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34400 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34401 -isystem, and -idirafter. Copy-edit.
34402 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34403 default for -ftrack-macro-expansion. Delete obsolete and
34404 badly-formatted implementation details about -fdebug-cpp output.
34405 * doc/cppwarnopts.texi: Copy-edit.
34407 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34410 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34411 that the transition from a max line width >= 1<<10 to narrower
34412 lines works correctly.
34414 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34416 * doc/options.texi (PerFunction): New.
34417 * opt-functions.awk (switch_flags): Map both Optimization and
34418 PerFunction to CL_OPTIMIZATION.
34419 * opth-gen.awk: Test for PerFunction flag along with
34421 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34422 it only when the latter is present. Skip those that don't in
34423 the hash function generator.
34424 * common.opt (fvar-tracking): Mark as PerFunction instead of
34426 (fvar-tracking-assignments): Likewise.
34427 (fvar-tracking-assignments-toggle): Likewise.
34428 (fvar-tracking-uninit): Likewise.
34430 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34432 PR translation/79018
34433 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34436 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34439 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34440 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34441 TARGET_LONG_JUMP_TABLE_OFFSETS.
34442 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34443 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34444 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34445 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34446 * config/m68k/m68k.md (tablejump expander): Likewise.
34447 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34448 TARGET_LONG_JUMP_TABLE_OFFSETS.
34449 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34450 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34452 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34453 David Holsgrove <david.holsgrove@xilinx.com>
34455 * common/config/microblaze/microblaze-common.c
34456 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34457 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34459 * config/microblaze/microblaze.c (microblaze_must_save_register)
34460 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34462 (microblaze_eh_return): New function.
34463 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34464 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34465 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34466 * config/microblaze/microblaze.md (eh_return): New pattern.
34468 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34470 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34471 GCC_DIAGNOSTIC_STRINGIFY): Define.
34473 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34475 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34477 * config/arm/arm.md (<mcrr>): New.
34479 * config/arm/arm.c (arm_arch5te): New.
34480 (arm_option_override): Set arm_arch5te.
34481 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34483 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34484 (arm_mcrr_qualifiers): ... this. New.
34485 (MRRC_QUALIFIERS): Define to...
34486 (arm_mrrc_qualifiers): ... this. New.
34487 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34489 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34490 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34491 (MRRCI, mrrc, MRRC): New.
34492 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34493 VUNSPEC_MRRC2): New.
34495 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34497 * config/arm/arm.md (<mcr>): New.
34499 * config/arm/arm.c (arm_coproc_builtin_available): Add
34500 support for mcr, mrc, mcr2 and mrc2.
34501 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34502 (arm_mcr_qualifiers): ... this. New.
34503 (MRC_QUALIFIERS): Define to ...
34504 (arm_mrc_qualifiers): ... this. New.
34505 (MCR_QUALIFIERS): Define to ...
34506 (arm_mcr_qualifiers): ... this. New.
34507 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34509 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34510 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34511 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34512 VUNSPEC_MRC2): New.
34514 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34516 * config/arm/arm.md (*ldc): New.
34520 * config/arm/arm.c (arm_coproc_builtin_available): Add
34521 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34522 (arm_coproc_ldc_stc_legitimate_address): New.
34523 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34524 'qualifier_const_pointer'.
34525 (LDC_QUALIFIERS): Define to...
34526 (arm_ldc_qualifiers): ... this. New.
34527 (STC_QUALIFIERS): Define to...
34528 (arm_stc_qualifiers): ... this. New.
34529 * config/arm/arm-protos.h
34530 (arm_coproc_ldc_stc_legitimate_address): New.
34531 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34532 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34533 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34534 stc2, stcl, stc2l): New.
34535 * config/arm/constraints.md (Uz): New.
34536 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34537 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34538 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34539 VUNSPEC_STC2L): New.
34541 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34543 * config/arm/arm.md (<cdp>): New.
34544 * config/arm/arm.c (neon_const_bounds): Rename this ...
34545 (arm_const_bounds): ... this.
34546 (arm_coproc_builtin_available): New.
34547 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34548 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34549 (CDP_QUALIFIERS): Define to...
34550 (arm_cdp_qualifiers): ... this. New.
34552 (arm_expand_builtin_args): Add case for 6 arguments.
34553 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34554 (arm_const_bounds): ... this.
34555 (arm_coproc_builtin_available): New.
34556 * config/arm/arm_acle.h (__arm_cdp): New.
34558 * config/arm/arm_acle_builtins.def (cdp): New.
34560 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34561 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34562 'arm_const_bounds'.
34563 * config/arm/types.md (coproc): New.
34564 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34565 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34566 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34567 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34569 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34571 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34572 (UBINOP_QUALIFIERS): New.
34574 (acle_builtin_data): New. Change comment.
34575 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34576 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34577 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34578 arm_acle_builtins.def.
34579 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34580 (arm_init_acle_builtins): New.
34581 (CRC32_BUILTIN): Remove.
34582 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34583 crc32cb, crc32ch and crc32cw.
34584 (arm_init_crc32_builtins): Remove.
34585 (arm_init_builtins): Use arm_init_acle_builtins rather
34586 than arm_init_crc32_builtins.
34587 (arm_expand_acle_builtin): New.
34588 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34589 * config/arm/arm_acle_builtins.def: New.
34591 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34593 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34594 (arm_builtin_datum): ... this.
34595 (arm_init_neon_builtin): Rename to ...
34596 (arm_init_builtin): ... this. Add a new parameters PREFIX
34597 and USE_SIG_IN_NAME.
34598 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34599 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34600 'arm_builtin_datum'.
34601 (arm_init_vfp_builtins): Likewise.
34602 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34603 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34604 (arm_expand_neon_args): Rename to ...
34605 (arm_expand_builtin_args): ... this. Rename builtin_arg
34606 enum values and differentiate between ARG_BUILTIN_MEMORY
34607 and ARG_BUILTIN_NEON_MEMORY.
34608 (arm_expand_neon_builtin_1): Rename to ...
34609 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34610 values, arm_expand_builtin_args and add bool parameter NEON.
34611 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34612 (arm_expand_vfp_builtin): Likewise.
34613 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34615 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34617 PR middle-end/77484
34618 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34619 * predict.c (tree_estimate_probability_bb): Reverse direction of
34620 polymorphic call predictor.
34622 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34624 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34625 (determine_pass_name_match): ...this new function and...
34626 (should_skip_pass_p): ...this new function.
34628 2017-01-06 Nathan Sidwell <nathan@acm.org>
34630 * ipa-visibility.c (function_and_variable_visibility): Reformat
34631 comments and long lines. Remove extrneous if.
34632 * symtab.c (symtab_node::make_decl_local): Fix code format.
34633 (symtab_node::set_section_for_node): Fix comment typo.
34635 2017-01-06 Martin Liska <mliska@suse.cz>
34638 * lra-constraints.c: Rename invariant to lra_invariant.
34639 * predict.c (set_even_probabilities): Initialize e to NULL.
34641 2017-01-05 Martin Sebor <msebor@redhat.com>
34643 PR tree-optimization/78910
34644 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34645 (format_integer): Correct off-by-one error in the handling
34646 of precision with negative numbers in signed conversions..
34648 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34650 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34652 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34654 PR tree-optimization/71016
34655 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34656 factor_out_conditional_conversion. Formatting fix.
34657 (factor_out_conditional_conversion): Add cond_stmt argument.
34658 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34659 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34662 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34664 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34665 read-rtl-function.o, and selftest-rtl.o.
34666 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34667 (selftest::aarch64_test_loading_full_dump): New function.
34668 (selftest::aarch64_run_selftests): New function.
34669 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34670 selftest::aarch64_run_selftests.
34671 * config/i386/i386.c
34672 (selftest::ix86_test_loading_dump_fragment_1): New function.
34673 (selftest::ix86_test_loading_call_insn): New function.
34674 (selftest::ix86_test_loading_full_dump): New function.
34675 (selftest::ix86_test_loading_unspec): New function.
34676 (selftest::ix86_run_selftests): Call the new functions.
34677 * emit-rtl.c (maybe_set_max_label_num): New function.
34678 * emit-rtl.h (maybe_set_max_label_num): New decl.
34679 * function.c (instantiate_decls): Guard call to
34680 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34681 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34683 * gensupport.c (gen_reader::gen_reader): Pass "false"
34684 for new "compact" param of rtx_reader.
34685 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34686 rather than an empty string for NULL strings.
34687 * read-md.c: Potentially include config.h rather than bconfig.h.
34688 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34689 (have_error): New global, copied from errors.c.
34690 (md_reader::read_name): Rename to...
34691 (md_reader::read_name_1): ...this, adding "out_loc" param,
34692 and converting "missing name or number" to returning false, rather
34694 (md_reader::read_name): Reimplement in terms of read_name_1.
34695 (md_reader::read_name_or_nil): New function.
34696 (md_reader::read_string): Handle "(nil)" by returning NULL.
34697 (md_reader::md_reader): Add new param "compact".
34698 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34699 (md_reader::read_file): New method.
34700 * read-md.h (md_reader::md_reader): Add new param "compact".
34701 (md_reader::read_file): New method.
34702 (md_reader::is_compact): New accessor.
34703 (md_reader::read_name): Convert return type from void to file_location.
34704 (md_reader::read_name_or_nil): New decl.
34705 (md_reader::read_name_1): New decl.
34706 (md_reader::m_compact): New field.
34707 (noop_reader::noop_reader): Pass "false" for new "compact" param
34709 (rtx_reader::rtx_reader): Add new "compact" param.
34710 (rtx_reader::read_rtx_operand): Make virtual and convert return
34711 type from void to rtx.
34712 (rtx_reader::read_until): New decl.
34713 (rtx_reader::handle_any_trailing_information): New virtual function.
34714 (rtx_reader::postprocess): New virtual function.
34715 (rtx_reader::finalize_string): New virtual function.
34716 (rtx_reader::m_in_call_function_usage): New field.
34717 (rtx_reader::m_reuse_rtx_by_id): New field.
34718 * read-rtl-function.c: New file.
34719 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34720 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34721 (selftest::verify_three_block_rtl_cfg): New decl.
34722 * read-rtl-function.h: New file.
34723 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34724 For host, include function.h, memmodel.h, and emit-rtl.h.
34725 (one_time_initialization): New function.
34726 (struct compact_insn_name): New struct.
34727 (compact_insn_names): New array.
34728 (find_code): Handle insn codes in compact dumps.
34729 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34730 (bind_subst_iter_and_attr): Likewise.
34731 (add_condition_to_string): Likewise.
34732 (add_condition_to_rtx): Likewise.
34733 (apply_attribute_uses): Likewise.
34734 (add_current_iterators): Likewise.
34735 (apply_iterators): Likewise.
34736 (initialize_iterators): Guard usage of apply_subst_iterator with
34737 #ifdef GENERATOR_FILE.
34738 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34739 (md_reader::read_mapping): Likewise.
34740 (add_define_attr_for_define_subst): Likewise.
34741 (add_define_subst_attr): Likewise.
34742 (read_subst_mapping): Likewise.
34743 (check_code_iterator): Likewise.
34744 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34746 (one_time_initialization): New function.
34747 (rtx_reader::read_until): New method.
34748 (read_flags): New function.
34749 (parse_reg_note_name): New function.
34750 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34751 Handle reuse_rtx ids.
34752 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34753 Add parsing support for RTL dumps, mirroring the special-cases in
34754 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34755 values, and calling handle_any_trailing_information.
34756 (rtx_reader::read_rtx_operand): Convert return type from void
34757 to rtx, returning return_rtx. Handle case 'e'. Call
34758 finalize_string on XSTR and XTMPL fields.
34759 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34760 "(nil)" values were omitted. Call the postprocess vfunc on the
34762 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34763 class ctor. Initialize m_in_call_function_usage. Call
34764 one_time_initialization.
34765 * rtl-tests.c (selftest::test_uncond_jump): Call
34766 set_new_first_and_last_insn.
34767 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34768 * selftest-rtl.c: New file.
34769 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34770 (selftest::get_insn_by_uid): New decl.
34771 * selftest-run-tests.c (selftest::run_tests): Call
34772 read_rtl_function_c_tests.
34773 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34774 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34777 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34779 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34780 operands in a special way. Assert that pos+len <= mode precision.
34782 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34784 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34785 3 argument Alias with unlimited for the negative form.
34786 (fno-vect-cost-model): Removed.
34788 2017-01-05 Martin Liska <mliska@suse.cz>
34790 * hsa-gen.c (gen_hsa_divmod): New function.
34791 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34793 2017-01-05 Martin Liska <mliska@suse.cz>
34796 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34799 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34801 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34802 small constant length operands.
34804 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34806 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34807 between loop iterations.
34809 2017-01-05 Martin Liska <mliska@suse.cz>
34812 * gimplify.c (gimplify_decl_expr): Compare to
34813 asan_poisoned_variables instread of checking flags.
34814 (gimplify_target_expr): Likewise.
34815 (gimplify_expr): Likewise.
34816 (gimplify_function_tree): Conditionally initialize
34817 asan_poisoned_variables.
34819 2017-01-04 Jeff Law <law@redhat.com>
34821 PR tree-optimizatin/78812
34822 * rtl.h (contains_mem_rtx_p): Prototype.
34823 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34824 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34825 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34826 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34827 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34829 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34831 * input.c (assert_char_at_range): Default-initialize actual_range.
34833 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34835 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34836 to match the caller.
34838 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34840 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34841 insns after final jump in test to emit dummy move.
34843 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34845 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34846 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34848 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34850 * multiple_target.c (create_dispatcher_calls): Init e_next.
34851 * tree-ssa-loop-split.c (split_loop): Init border.
34852 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34855 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34860 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34861 (altivec_register_operand): Do not return true if the operand
34862 contains a SUBREG mixing SImode and SFmode.
34863 (vsx_register_operand): Likewise.
34864 (vsx_reg_sfsubreg_ok): New predicate.
34865 (vfloat_operand): Do not return true if the operand contains a
34866 SUBREG mixing SImode and SFmode.
34867 (vint_operand): Likewise.
34868 (vlogical_operand): Likewise.
34869 (gpc_reg_operand): Likewise.
34870 (int_reg_operand): Likewise.
34871 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34872 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34873 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34875 (rs6000_emit_move_si_sf_subreg): New helper function.
34876 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34877 fixup SUBREGs involving SImode and SFmode.
34878 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34879 numbers for the new peephole2 optimization.
34880 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34881 the GLIBC math library that do AND/IOR/XOR operations on single
34882 precision floating point.
34883 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34884 target macros to say whether we need to avoid SUBREGs mixing
34886 (TARGET_ALLOW_SF_SUBREG): Likewise.
34887 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34888 (UNSPEC_SI_FROM_SF): Likewise.
34889 (iorxor): Change spacing.
34890 (and_ior_xor): New iterator for AND, IOR, and XOR.
34891 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34892 (movdi_from_sf_zero_ext): Likewise.
34893 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34894 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34895 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34896 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34897 (fms<mode>4): Likewise.
34898 (fnma<mode>4): Likewise.
34899 (fnms<mode>4): Likewise.
34900 (nfma<mode>4): Likewise.
34901 (nfms<mode>4): Likewise.
34903 2017-01-04 Marek Polacek <polacek@redhat.com>
34906 * doc/invoke.texi: Document -Wpointer-compare.
34908 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34910 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34913 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34914 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34917 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34920 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34921 documentation of the powerpc_popcntb_ok attribute.
34922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34923 code to issue warning messages if a requested CPU configuration is
34924 not supported by the binary (assembler and loader) toolchain.
34925 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34926 made to define a built-in function that has been disabled.
34927 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34928 made to define a built-in function that has been disabled.
34929 (altivec_init_builtins): Add comment explaining why definition
34930 of the DST built-in functions is not preceded by an assertion
34931 check. Add assertions to prevent ICE if attempts are made to
34932 define an altivec predicate or an abs* built-in function that has
34934 (htm_init_builtins): Add comment explaining why definition of the
34935 htm built-in functions is not preceded by an assertion check.
34937 2017-01-04 Jeff Law <law@redhat.com>
34939 PR tree-optimizatin/67955
34940 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34941 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34942 the points-to solution does not include pt_null. Use DECL_PT_UID
34945 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34947 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34948 Use gen_int_mode instead of gen_lopwart for const_int operands.
34950 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34952 PR tree-optimization/71563
34953 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34954 out of range value - has low bits known to be zero.
34956 2017-01-04 Alan Modra <amodra@gmail.com>
34958 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34959 * configure: Regenerate.
34960 * config.in: Regenerate.
34962 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34965 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34966 a substring of the message, but strcmp with the whole message. Ifdef
34967 ENABLE_NLS, translate the message first using dgettext.
34969 2017-01-03 Jeff Law <law@redhat.com>
34971 PR tree-optimizatin/78856
34972 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34973 (mark_threaded_blocks): Remove code to truncate thread paths that
34974 cross multiple loop headers. Instead invalidate the cached loop
34975 iteration information and handle case of a thread path walking
34976 into an irreducible region.
34978 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34981 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34982 assertions. Add support for doing the signbit if the IEEE 128-bit
34983 floating point value is in a GPR.
34984 * config/rs6000/rs6000.md (Fsignbit): Delete.
34985 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34986 Update the length attribute if the value is in a GPR.
34987 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34988 the sign or zero extension instruction, since the value is always 0/1.
34989 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34992 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34993 extracting SImode to a GPR register so that we can generate a
34994 store, limit the vector to be in a traditional Altivec register
34995 for the vextuwrx instruction.
34997 2017-01-03 Ian Lance Taylor <iant@google.com>
34999 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35001 2017-01-03 Martin Sebor <msebor@redhat.com>
35003 PR tree-optimization/78696
35004 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35005 precision. Use MPFR for %f for greater fidelity. Correct handling
35007 (pass_sprintf_length::compute_format_length): Set width and precision
35008 specified by asrerisk to void_node for vararg functions.
35009 (try_substitute_return_value): Adjust dump output.
35011 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35013 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35015 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35017 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35018 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35019 -mlra/-mno-lra was passed to the compiler.
35021 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35023 PR rtl-optimization/65618
35024 * emit-rtl.c (try_split): Move initialization of "before" and
35025 "after" to just before the call to emit_insn_after_setloc.
35027 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35029 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35031 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35033 * dwarf2out.c (gen_enumeration_type_die): When
35034 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35036 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35038 PR tree-optimization/78965
35039 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35040 Change first argument from const call_info & to call_info &. For %n
35041 set info.nowrite to false.
35043 PR middle-end/78901
35044 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35045 possibly throwing calls.
35047 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35048 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35049 and fns handling, rather than in a separate case SSA_NAME.
35051 2017-01-02 Jeff Law <law@redhat.com>
35053 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35054 fixes for first_period and second_period variables.
35056 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35059 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35060 (*insvqi_1): New insn pattern.
35061 (*insvqi_1_mem_rex64): Ditto.
35062 (*insvqi_2): Ditto.
35063 (*insvqi_3): Rename from *insvqi.
35065 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35067 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35069 * doc/cfg.texi (Edges): Remove reference to Java.
35070 (Maintaining the CFG): Ditto.
35072 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35074 PR middle-end/77674
35075 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35076 transparent aliases.
35078 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35080 PR middle-end/77484
35081 * predict.def (PRED_CALL): Update hitrate.
35082 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35083 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35084 into direct/indirect/polymorphic variants.
35086 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35088 Update copyright years.
35090 * gcc.c (process_command): Update copyright notice dates.
35091 * gcov-dump.c (print_version): Ditto.
35092 * gcov.c (print_version): Ditto.
35093 * gcov-tool.c (print_version): Ditto.
35094 * gengtype.c (create_file): Ditto.
35095 * doc/cpp.texi: Bump @copying's copyright year.
35096 * doc/cppinternals.texi: Ditto.
35097 * doc/gcc.texi: Ditto.
35098 * doc/gccint.texi: Ditto.
35099 * doc/gcov.texi: Ditto.
35100 * doc/install.texi: Ditto.
35101 * doc/invoke.texi: Ditto.
35103 Copyright (C) 2017 Free Software Foundation, Inc.
35105 Copying and distribution of this file, with or without modification,
35106 are permitted in any medium without royalty provided the copyright
35107 notice and this notice are preserved.