1 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4 from explicit instantiation of debug_helper.
5 * vec.h (DEFINE_DEBUG_VEC): Ditto.
7 2017-11-27 Richard Biener <rguenther@suse.de>
9 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
12 2017-11-27 Richard Biener <rguenther@suse.de>
14 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
15 (type_hash_canon): Also clear int_cst_hash_table entry for
17 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
19 2017-11-27 Tamar Christina <tamar.christina@arm.com>
21 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
22 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
23 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
25 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
27 * hash-map.h (gt_cleare_cache): Avoid UB.
29 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
31 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
32 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
33 (decide_unroll_runtime_iterations): Likewise.
34 (decide_unroll_stupid): Likewise.
36 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
39 * config/i386/i386.md: Add a loop with incssp.
41 2017-11-27 Martin Jambor <mjambor@suse.cz>
43 PR tree-optimization/81248
44 * tree-sra.c (splice_param_accesses): Remove size check.
45 (decide_one_param_reduction): Fix size check.
46 * gimple-pretty-print.c (dump_profile): Silence warning.
47 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
49 2017-11-27 Jakub Jelinek <jakub@redhat.com>
52 * dbxout.c (lastlineno): New variable.
53 (dbx_debug_hooks): Use dbxout_switch_text_section as
54 switch_text_section debug hook.
55 (dbxout_function_end): Switch to current_function_section
56 rather than function_section. If crtl->has_bb_partition,
57 output just one N_FUN, depending on in_cold_section_p.
58 (dbxout_source_line): Remember last lineno in lastlineno.
59 (dbxout_switch_text_section): New function.
60 (dbxout_function_decl): Adjust dbxout_block caller.
61 (dbx_block_with_cold_children): New function.
62 (dbxout_block): Return true if any LBRAC/RBRAC have been
63 emitted. Use dbx_block_with_cold_children at depth == 0
64 in second partition. Add PARENT_BLOCKNUM argument, pass
65 it optionally adjusted to children. Output LBRAC/RBRAC
66 around recursive call only if the block is in the current
67 partition, if not and anything was output, emit empty
69 * final.c (final_scan_insn): Compute cold_function_name
70 before calling switch_text_section debug hook. Call
71 that hook even if dwarf2out_do_frame if not emitting
75 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
78 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
80 PR rtl-optimization/82488
81 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
83 2017-11-26 Julia Koval <julia.koval@intel.com>
85 * config/i386/i386.c (processor_target_table): Add skylake_cost for
87 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
90 2017-11-26 Julia Koval <julia.koval@intel.com>
92 * config/i386/driver-i386.c (host_detect_local_cpu):
93 Detect skylake-avx512.
95 2017-11-26 Julia Koval <julia.koval@intel.com>
97 * config.gcc: Add -march=cannonlake.
98 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
99 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
100 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
101 (PTA_CANNONLAKE): New.
102 (processor_target_table): Add cannonlake.
103 (ix86_option_override_internal): Ditto.
104 (fold_builtin_cpu): Ditto.
105 (get_builtin_code_for_version): Handle cannonlake.
106 (M_INTEL_COREI7_CANNONLAKE): New.
107 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
108 * doc/invoke.texi: Add -march=cannonlake.
110 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
112 * plugin.c (add_new_plugin): Use platform-specific library extensions.
113 (try_init_one_plugin): Alternative implementation for MinGW.
114 * Makefile.in (plugin_implib): New.
115 (gengtype-lex.c): Fix broken AIX workaround.
116 * configure: Regenerate.
117 * doc/plugins.texi: Document support for MinGW.
119 2017-11-25 Jakub Jelinek <jakub@redhat.com>
121 PR rtl-optimization/81553
122 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
123 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
124 is a shift where C1 has different mode than the whole shift, use C1's
125 mode for MULT rather than the shift's mode.
128 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
129 builtins not enabled in the currently selected ISA.
131 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
133 PR tree-optimization/71026
134 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
135 (insert_reciprocals): Change to insert reciprocals before a division
136 by a square and to insert the square of a reciprocal.
137 (execute_cse_reciprocals_1): Change to consider division by a square.
138 (register_division_in): Add importance parameter.
140 2017-11-24 Richard Biener <rguenther@suse.de>
142 PR tree-optimization/82402
143 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
144 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
146 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
148 * match.pd (0-ptr): New transformation.
150 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
153 * ipa-inline.c (inline_small_functions): Set current badnes correctly
154 when skipping checking.
156 2017-11-24 Richard Biener <rguenther@suse.de>
158 PR tree-optimization/83128
159 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
160 (vn_reference_lookup_3): Likewise.
162 2017-11-24 Jakub Jelinek <jakub@redhat.com>
165 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
166 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
169 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
170 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
171 check for the latter with gimple_call_builtin_p. Do not handle
172 BUILT_IN_STPNCPY_CHK which is not a pass through call.
174 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
176 * config/arm/arm_neon.h: Fix pragma GCC push_options before
179 2017-11-23 Julia Koval <julia.koval@intel.com>
181 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
182 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
183 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
184 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
185 _mm512_maskz_expandloadu_epi16): New intrinsics.
186 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
187 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
188 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
189 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
190 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
191 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
192 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
193 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
194 _mm256_maskz_expandloadu_epi8): New intrinsics.
195 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
196 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
197 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
198 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
199 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
200 * config/i386/sse.md (VI248_VLBW): New iterator.
201 (expand<mode>_mask, expand<mode>_maskz): New patterns.
203 2017-11-23 Julia Koval <julia.koval@intel.com>
205 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
206 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
207 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
208 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
209 _mm512_mask_compressstoreu_epi16): New.
210 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
211 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
212 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
213 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
214 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
215 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
216 _mm256_mask_compressstoreu_epi8): New.
217 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
218 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
219 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
220 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
221 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
222 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
223 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
224 __builtin_ia32_compresshi128_mask,
225 __builtin_ia32_compressstoreuqi512_mask,
226 __builtin_ia32_compressstoreuhi512_mask,
227 __builtin_ia32_compressstoreuqi256_mask,
228 __builtin_ia32_compressstoreuqi128_mask,
229 __builtin_ia32_compressstoreuhi256_mask,
230 __builtin_ia32_compressstoreuhi128_mask): New builtins.
231 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
233 (ix86_expand_special_args_builtin): Handle new types.
234 (s4fma_expand): Handle new builtin array.
235 * config/i386/immintrin.h: Include new headers.
236 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
237 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
239 2017-11-23 Jakub Jelinek <jakub@redhat.com>
241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
242 fixes. Declare temp and g variables at the top in order to avoid
243 {} in most of the cases.
245 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
247 * match.pd (ptr-0): New transformation.
249 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
251 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
252 (struct addr_mode_cost_table): New.
253 (struct tune_params): Add field addr_mode_costs.
254 * config/arm/arm.c (generic_addr_mode_costs): New.
255 (arm_slowmul_tune): Initialise addr_mode_costs field.
256 (arm_fastmul_tune): Likewise.
257 (arm_strongarm_tune): Likewise.
258 (arm_xscale_tune): Likewise.
259 (arm_9e_tune): Likewise.
260 (arm_marvell_pj4_tune): Likewise.
261 (arm_v6t2_tune): Likewise.
262 (arm_cortex_tune): Likewise.
263 (arm_cortex_a8_tune): Likewise.
264 (arm_cortex_a7_tune): Likewise.
265 (arm_cortex_a15_tune): Likewise.
266 (arm_cortex_a35_tune): Likewise.
267 (arm_cortex_a53_tune): Likewise.
268 (arm_cortex_a57_tune): Likewise.
269 (arm_exynosm1_tune): Likewise.
270 (arm_xgene1_tune): Likewise.
271 (arm_cortex_a5_tune): Likewise.
272 (arm_cortex_a9_tune): Likewise.
273 (arm_cortex_a12_tune): Likewise.
274 (arm_cortex_a73_tune): Likewise.
275 (arm_v7m_tune): Likewise.
276 (arm_cortex_m7_tune): Likewise.
277 (arm_v6m_tune): Likewise.
278 (arm_fa726te_tune): Likewise.
279 (arm_mem_costs): Use table lookup to calculate cost of addressing
282 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
284 * config/arm/arm.c (arm_mem_costs): New function.
285 (arm_rtx_costs_internal): Use arm_mem_costs.
287 2017-11-23 Mark Wielaard <mark@klomp.org>
289 * dwarf2out.c (init_sections_and_labels): Use generation to create
290 unique ranges_section_label and ranges_base_label. Return generation.
291 (output_rnglists): Add generation argument. Use generation to create
292 unique ranges labels.
293 (dwarf2out_finish): Get generation from init_sections_and_labels
294 and pass generation to output_rnglists.
296 2017-11-23 Mike Stump <mikestump@comcast.net>
297 Eric Botcazou <ebotcazou@adacore.com>
299 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
300 * cfgloop.h (struct loop): Add unroll field.
301 * function.h (struct function): Add has_unroll bitfield.
302 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
303 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
304 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
305 (pass_rtl_unroll_loops::gate): Likewise.
306 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
307 for which loop->unroll==1.
308 (decide_unroll_constant_iterations): Use note for consistency and
309 take loop->unroll into account. Return early if loop->unroll is set.
310 Fix thinko in existing test.
311 (decide_unroll_runtime_iterations): Use note for consistency and
312 take loop->unroll into account.
313 (decide_unroll_stupid): Likewise.
314 * lto-streamer-in.c (input_cfg): Read loop->unroll.
315 * lto-streamer-out.c (output_cfg): Write loop->unroll.
316 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
318 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
319 (print_loop): Print loop->unroll if set.
320 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
321 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
322 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
324 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
325 loop->unroll is set and smaller than the trip count. Otherwise bypass
326 entirely the heuristics if loop->unroll is set. Remove dead note.
327 Fix off-by-one bug in other note.
328 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
329 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
331 (tree_unroll_loops_completely): Make static.
332 (pass_complete_unroll::execute): Use correct type for variable.
333 (pass_complete_unrolli::execute): Fix formatting.
334 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
336 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
338 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
339 enable when TARGET_PREFER_AVX128 is set.
341 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
343 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
345 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
346 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
348 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
350 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
351 frequencies. Fix estimation of aggregate parameters.
353 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
355 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
358 2017-11-23 Tom de Vries <tom@codesourcery.com>
360 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
362 2017-11-23 Tom de Vries <tom@codesourcery.com>
364 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
367 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
370 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
371 sibcall_value_pcrel_fdpic): Use local variable instead of
373 (calli_tbr_rel): Add missing operand 2.
374 (call_valuei_tbr_rel): Add missing operand 3.
376 2017-11-23 Jakub Jelinek <jakub@redhat.com>
379 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
380 bitpos/bitsize covering the whole destination, use store_expr only if
381 the complex mode is the same. Otherwise, use expand_normal and if
382 it returns CONCAT, subreg each part separately instead of trying to
383 subreg the whole result.
385 2017-11-23 Richard Biener <rguenther@suse.de>
387 PR tree-optimization/23094
388 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
389 come here from walking over backedges in the first iteration.
390 (vn_reference_lookup_3): Skip clobbers that store the same value.
392 2017-11-23 Richard Biener <rguenther@suse.de>
394 PR tree-optimization/81403
395 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
396 a block we need a leader relative to.
397 (phi_translate_1): For nary processing require a leader from
398 get_representative_for given we run expression simplification
399 using match-and-simplify. Remove previous fix.
401 2017-11-22 Jeff Law <law@redhat.com>
403 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
404 Use new method allocate_value_range rather than accessing the
405 vrp_value_range_pool data member directly.
406 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
407 to use extract_range_from_stmt method to avoid need for
408 extract_range_from_assignment method.
409 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
410 method rather than setting values_propgated data member directly.
411 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
412 and values propagated data members and extract_range_from_assignment
413 method. Reorder private data members to conform to standards.
414 Add new methods set_lattice_propagation_complete and
415 allocate_value_range.
417 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
419 PR rtl-optimization/83030
420 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
421 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
422 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
423 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
424 CROSSING_JUMP_P flag is already set before setting it.
425 * cfgrtl.c (fixup_partition_crossing): Likewise.
426 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
429 2017-11-22 Jakub Jelinek <jakub@redhat.com>
431 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
432 Handle the case where both arguments are using gen_const_vec_series.
434 2017-11-22 David Malcolm <dmalcolm@redhat.com>
437 * pretty-print.c (pp_format): Move quoting implementation to
438 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
439 to pass address of "quote" local.
440 (pp_begin_quote): New function.
441 (pp_end_quote): New function.
442 * pretty-print.h (printer_fn): Convert penultimate param from bool
444 (pp_begin_quote): New decl.
445 (pp_end_quote): New decl.
446 * tree-diagnostic.c (default_tree_printer): Convert penultimate
447 param from bool to bool *.
448 * tree-diagnostic.h (default_tree_printer): Likewise.
450 2017-11-22 Jeff Law <law@redhat.com>
452 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
453 Initialize vr_values.
454 (evrp_range_analyzer::try_find_new_range): Call methods attached to
455 vr_values via vr_values class instance rather than delegators.
456 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
457 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
458 (evrp_range_analyzer::push_value_range): Likewise.
459 (evrp_range_analyzer::pop_value_range): Likewise.
460 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
461 most delegators. Those remaining are exposed as public interfaces.
462 Make vr_values a pointer and private.
463 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
465 (evrp_range_analyzer::get_vr_value): New method.
466 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
467 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
468 (class evrp_dom_walker): Attach evrp_folder class, initialize
469 it in the ctor. Remove temporary delegators.
470 (evrp_dom_walker::before_dom_children): Call methods in attached
471 evrp_range_analyzer class via class instance pointer. Use
472 free value_range_constant_singleton to remove need for
473 op_with_constant_singleton_value delegator method. Do not
474 create a vrp_prop class instance for every call! Narrow
475 scope of a couple variables.
476 (evrp_dom_walker::cleanup): Call methods in attached
477 evrp_range_analyzer class via class instance pointer.
478 * vr-values.h (class vr_values): Privatize many methods and
481 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
483 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
484 * cfgexpand.c (expand_debug_expr): Remove handling for them.
485 * expr.c (expand_expr_real_2): Likewise.
486 * fold-const.c (const_unop): Likewise.
487 * optabs-tree.c (optab_for_tree_code): Likewise.
488 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
489 * tree-inline.c (estimate_operator_cost): Likewise.
490 * tree-pretty-print.c (dump_generic_node): Likewise.
491 (op_code_prio): Likewise.
492 (op_symbol_code): Likewise.
493 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
494 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
495 * internal-fn.c (direct_internal_fn_optab): New function.
496 (direct_internal_fn_array, direct_internal_fn_supported_p
497 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
498 * fold-const-call.c (fold_const_reduction): New function.
499 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
501 * tree-vect-loop.c: Include internal-fn.h.
502 (reduction_code_for_scalar_code): Rename to...
503 (reduction_fn_for_scalar_code): ...this and return an internal
505 (vect_model_reduction_cost): Take an internal_fn rather than
507 (vect_create_epilog_for_reduction): Likewise. Build calls rather
509 (vectorizable_reduction): Use internal functions rather than tree
510 codes for the reduction operation. Update calls to the functions
512 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
513 Use calls to internal functions rather than REDUC tree codes.
514 * config/aarch64/aarch64-simd.md: Update comment accordingly.
516 2017-11-22 Olivier Hainque <hainque@adacore.com>
518 * config/vxworks.c (vxworks_override_options): Pick default
519 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
520 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
521 DWARF_GNAT_ENCODINGS_DEFAULT.
522 * config/vxworksae.h: Likewise.
524 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
526 PR tree-optimization/83104
527 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
530 2017-11-22 Marek Polacek <polacek@redhat.com>
531 H.J. Lu <hongjiu.lu@intel.com>
532 Jason Merrill <jason@redhat.com>
537 * calls.c (initialize_argument_information): Call
538 warn_parameter_passing_abi target hook.
539 (store_one_arg): Use 0 for empty record size. Don't push 0 size
541 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
542 * common.opt: Update -fabi-version description.
543 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
544 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
546 (ix86_is_empty_record): New function.
547 (ix86_warn_parameter_passing_abi): New function.
548 (TARGET_EMPTY_RECORD_P): Redefine.
549 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
550 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
551 * doc/tm.texi: Regenerated.
552 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
553 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
554 * dwarf2out.c (get_ultimate_context): Move to tree.c.
555 * explow.c (hard_function_value): Call arg_int_size_in_bytes
556 instead of int_size_in_bytes.
557 * expr.c (copy_blkmode_to_reg): Likewise.
558 * function.c (aggregate_value_p): Return 0 for empty types.
559 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
560 (locate_and_pad_parm): Call arg size_in_bytes instead
562 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
563 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
564 * target.def (empty_record_p, warn_parameter_passing_abi): New target
566 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
567 (std_gimplify_va_arg_expr): Skip empty records. Call
568 arg_size_in_bytes instead size_in_bytes.
569 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
570 * tree-core.h (tree_type_common): Add empty_flag.
571 (tree_decl_common): Update comments.
572 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
574 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
575 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
577 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
578 * tree.c (default_is_empty_type): New function.
579 (default_is_empty_record): New function.
580 (arg_int_size_in_bytes): New function.
581 (arg_size_in_bytes): New function.
582 (get_ultimate_context): New function.
583 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
584 TRANSLATION_UNIT_WARN_EMPTY_P.
585 (default_is_empty_record, arg_int_size_in_bytes,
586 arg_size_in_bytes, get_ultimate_context): Declare.
588 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
590 * config/arm/arm.c (cmse_clear_registers): New function.
591 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
592 code by call to cmse_clear_registers.
593 (cmse_nonsecure_entry_clear_before_return): Likewise.
595 2017-11-22 Tamar Christina <tamar.christina@arm.com>
597 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
598 (vdot_s32, vdotq_s32): New.
599 (vdot_lane_u32, vdotq_lane_u32): New.
600 (vdot_lane_s32, vdotq_lane_s32): New.
603 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
606 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
607 for unsigned values with fewer HWIs than the precision.
608 (test_overflow): New function.
609 (wide_int_cc_tests): Call it.
611 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
613 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
614 has the same mode class as Pmode.
616 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
618 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
619 padding_bits_to_clear_ptr.
620 (cmse_nonsecure_entry_clear_before_return): Likewise.
622 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
624 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
625 auto_sbitap instead of integer bitfield to control register needing
628 2017-11-22 Jakub Jelinek <jakub@redhat.com>
630 PR tree-optimization/83044
631 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
632 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
633 that need the upper bound. Subtract offset from
634 get_addr_base_and_unit_offset only if positive and subtract it
635 before division by eltsize rather than after it.
638 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
639 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
640 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
644 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
645 expand_binop with *mul_widen_optab, make sure at least one of the
646 operands doesn't have VOIDmode.
649 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
651 PR rtl-optimization/82044
652 PR tree-optimization/82042
653 * dse.c (record_store): Check for overflow.
654 (check_mem_read_rtx): Properly check for overflow if width == -1, call
655 add_wild_read instead of clear_rhs_from_active_local_stores on
656 overflow and log it into dump_file.
658 2017-11-22 Richard Biener <rguenther@suse.de>
660 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
661 fake edges to exit when looking for a place to insert.
662 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
664 (insert_into_preds_of_block): Commit edge insertion immediately,
665 assert that doesn't require new BBs.
666 (fini_pre): Release expressions.
667 (pass_pre::execute): Shuffle things around a bit, if the fn
668 is too large do not compute AVAIL either as this is really the
671 2017-11-22 Richard Biener <rguenther@suse.de>
673 PR tree-optimization/83089
674 * tree-if-conv.c (pass_if_conversion::execute): If anything
675 changed reset SCEV and free the number of iteration estimates.
677 2017-11-21 Martin Sebor <msebor@redhat.com>
679 PR tree-optimization/82945
680 * calls.h (warn_nonstring_bound): Remove unused function.
682 2017-11-21 Martin Sebor <msebor@redhat.com>
684 PR tree-optimization/82945
685 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
686 * calls.h (maybe_warn_nonstring_arg): Declare new function.
687 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
689 (initialize_argument_information): Call maybe_warn_nonstring_arg.
690 * calls.h (get_attr_nonstring_decl): Declare new function.
691 * doc/extend.texi (attribute nonstring): Update.
692 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
693 get_attr_nonstring_decl and handle it.
694 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
695 detection of nul-termination.
696 (strlen_to_stridx): Change to a pointer.
697 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
698 (pass_strlen::execute): Same.
700 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
702 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
703 for the new option -mprefer-vector-width=[none|128|256|512].
704 * config/i386/i386.c (ix86_target_string): remove old style options
705 -mprefer-avx256 and make -mprefer-avx128 as alias.
706 (ix86_option_override_internal): Apply defaults for the
707 -mprefer-vector-width=[128|256] option.
708 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
709 Implement macros to work with -mprefer-vector-width=.
710 * config/i386/i386.opt: Implemented option
711 -mprefer-vector-width=[none|128|256|512].
712 * doc/invoke.texi: Documentation for
713 -mprefer-vector-width=[none|128|256|512].
715 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
717 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
719 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
721 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
723 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
725 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
727 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
728 * expr.c (expand_expr_real_2): Likewise.
729 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
730 fold_binary_loc): Likewise.
731 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
732 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
733 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
734 MINUS_EXPR transformations.
735 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
736 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
737 * tree-inline.c (estimate_operator_cost): Likewise.
738 * tree-pretty-print.c (dump_generic_node, op_code_prio,
739 op_symbol_code): Likewise.
740 * tree-vect-stmts.c (vectorizable_operation): Likewise.
741 * vr-values.c (extract_range_from_binary_expr): Likewise.
742 * varasm.c (initializer_constant_valid_p_1): Likewise.
743 * tree.def: New tree code POINTER_DIFF_EXPR.
745 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
747 * config/i386/i386.md (*bswap<mode>2_movbe): Add
748 integer suffix to movbe mnemonic.
749 (*bswaphi2_movbe): Ditto.
750 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
752 2017-11-21 Jakub Jelinek <jakub@redhat.com>
755 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
756 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
757 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
759 2017-11-21 Martin Liska <mliska@suse.cz>
761 * tree-inline.c (expand_call_inline): Remove not needed
764 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
765 Jakub Jelinek <jakub@redhat.com>
768 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
769 Remove static keyword from f variable.
771 2017-11-21 Jakub Jelinek <jakub@redhat.com>
773 PR tree-optimization/83086
774 * gimple-ssa-store-merging.c
775 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
776 rather than n.base_addr.
778 2017-11-21 Martin Liska <mliska@suse.cz>
780 PR rtl-optimization/82044
781 PR tree-optimization/82042
782 * dse.c (check_mem_read_rtx): Check for overflow.
784 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
786 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
789 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
791 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
792 and its corresponding call arg location note.
794 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
795 Andrew Burgess <andrew.burgess@embecosm.com>
797 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
799 (arc_return_slot_offset): Likewise.
800 (arc_eh_return_address_location): New declaration.
801 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
802 (MUST_SAVE_REGISTER): Add exception handler case.
803 (MUST_SAVE_RETURN_ADDR): Likewise.
804 (arc_frame_pointer_required): Likewise.
805 (arc_frame_pointer_needed): New function.
806 (arc_compute_frame_size): Changed.
807 (arc_expand_prologue): Likewise.
808 (arc_expand_epilogue): Likewise.
809 (arc_initial_elimination_offset): Likewise.
810 (arc_return_slot_offset): Delete.
811 (arc_eh_return_address_location): New function.
812 (arc_builtin_setjmp_frame_value): Likewise.
813 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
814 (EH_RETURN_STACKADJ_RTX): Define.
815 (EH_RETURN_HANDLER_RTX): Likewise.
816 * config/arc/arc.md (eh_return): Delete.
818 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
820 * print-tree.h (debug_vec_tree): Remove prototype.
821 * gdbinit.in (pvt): Remove macro.
823 2017-11-21 Jakub Jelinek <jakub@redhat.com>
825 PR tree-optimization/83047
826 * gimple-ssa-store-merging.c
827 (imm_store_chain_info::output_merged_store): If the loads with the
828 same vuse are in different basic blocks, for load_gsi pick a load
829 location that is dominated by the other loads.
832 * config/i386/i386.c (ix86_memmodel_check): Start
833 -Winvalid-memory-model diagnostics with lowercase letter.
836 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
837 set in_cold_section_p to the partition containing loc_list->first.
838 When seeing loc_list->last_before_switch node, update secname and
839 perform range_across_switch second partition handling only after that.
842 * run-rtl-passes.c: Include debug.h.
843 (run_rtl_passes): Call debug_hooks->assembly_start.
844 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
848 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
849 OPTAB_DIRECT in calls to expand_simple_binop.
851 2017-11-20 David Malcolm <dmalcolm@redhat.com>
854 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
856 2017-11-20 David Malcolm <dmalcolm@redhat.com>
859 * spellcheck.h (best_match::blithely_get_best_candidate): New
862 2017-11-20 Jakub Jelinek <jakub@redhat.com>
864 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
865 with lower case letter.
867 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
869 * config/i386/i386.md (bswaphi2): New expander.
870 (*bswaphi2_movbe): New insn pattern.
871 (bswaphi -> rorhi pepehole2): New peephole pattern.
873 2017-11-20 Jeff Law <law@redhat.com>
875 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
876 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
877 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
878 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
879 gimple-ssa-evrp-analyze.h.
881 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
883 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
885 2017-11-20 Bin Cheng <bin.cheng@arm.com>
887 * tree-predcom.c: Add general comment on Store-Store chains.
888 (split_data_refs_to_components): Postpone clearing eliminate_store_p
890 (get_chain_last_ref_at): Rename into...
891 (get_chain_last_write_at): ...this.
892 (get_chain_last_write_before_load): New function.
893 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
894 CT_STORE_STORE when write reference is added.
895 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
896 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
897 (initialize_root_vars_store_elim_1): Ditto.
898 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
899 definition is created.
900 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
901 chain by replacing it with dominant stored value.
903 2017-11-20 Bin Cheng <bin.cheng@arm.com>
905 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
907 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
909 * vr-values.c (extract_range_from_binary_expr): Use a full range
912 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
914 * config/arm/arm.md (R4_REGNUM): Define constant.
915 (nonsecure_call_internal): Remove r4 clobber.
916 (nonsecure_call_value_internal): Likewise.
917 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
918 clobber and resequence match_operands.
919 (nonsecure_call_value_reg_thumb1_v5): Likewise.
920 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
921 (nonsecure_call_value_reg_thumb2): Likewise.
923 2017-11-20 Jakub Jelinek <jakub@redhat.com>
925 PR tree-optimization/78821
926 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
927 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
928 to the address of the base rather than the base itself.
929 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
930 (find_bswap_or_nop_finalize): New function.
931 (find_bswap_or_nop): Use it.
932 (bswap_replace): Return a tree rather than bool, change first
933 argument from gimple * to gimple_stmt_iterator, allow inserting
934 into an empty sequence, allow ins_stmt to be NULL - then emit
935 all stmts into gsi. Fix up MEM_REF address gimplification.
936 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
937 (struct store_immediate_info): Add N and INS_STMT non-static
939 (store_immediate_info::store_immediate_info): Initialize them
940 from newly added ctor args.
941 (merged_store_group::apply_stores): Formatting fixes. Sort by
943 (stmts_may_clobber_ref_p): For stores call also
944 refs_anti_dependent_p.
945 (gather_bswap_load_refs): New function.
946 (imm_store_chain_info::try_coalesce_bswap): New method.
947 (imm_store_chain_info::coalesce_immediate_stores): Use it.
948 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
949 (imm_store_chain_info::output_merged_store): Fail if number of
950 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
951 and NOP_EXPR rhs_code.
952 (pass_store_merging::process_store): Compute n and ins_stmt, if
953 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
954 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
956 (pass_store_merging::execute): Calculate dominators.
958 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
959 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
960 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
961 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
962 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
963 class pass_optimize_bswap, bswap_replace,
964 pass_optimize_bswap::execute): Moved to ...
965 * gimple-ssa-store-merging.c: ... this file.
966 Include optabs-tree.h.
967 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
968 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
969 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
970 anonymous namespace, remove static keywords.
971 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
972 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
974 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
977 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
979 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
981 * vec.h (debug_helper): New function.
982 (DEFINE_DEBUG_VEC): New macro.
983 * hash-set.h (debug_helper): New function.
984 (DEFINE_DEBUG_HASH_SET): New macro.
985 * cfg.c (debug_slim (edge)): New function.
986 Call DEFINE_DEBUG_VEC for edges.
987 Call DEFINE_DEBUG_HASH_SET for edges.
988 * cfghooks.c (debug_slim (basic_block)): New function.
989 Call DEFINE_DEBUG_VEC for basic blocks.
990 Call DEFINE_DEBUG_HASH_SET for basic blocks.
991 * print-tree.c (debug_slim): New function to handle trees.
992 Call DEFINE_DEBUG_VEC for trees.
993 Call DEFINE_DEBUG_HASH_SET for trees.
994 (debug (vec<tree, va_gc>) &): Remove.
995 (debug (<vec<tree, va_gc>) *): Remove.
996 * print-rtl.c (debug_slim): New function to handle const_rtx.
997 Call DEFINE_DEBUG_VEC for rtx_def.
998 Call DEFINE_DEBUG_VEC for rtx_insn.
999 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1000 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1001 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1002 (debug (vec<rtx_insn *> *ptr): Remove.
1003 (debug_insn_vector): Remove.
1004 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1006 2017-11-20 Tom de Vries <tom@codesourcery.com>
1008 PR rtl-optimization/82020
1009 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1010 IF_THEN_ELSE condition.
1012 2017-11-19 Jeff Law <law@redhat.com>
1014 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1015 of degenerates resulting from ignoring an edge.
1017 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1020 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1022 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1025 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1026 for uninitialied counts.
1028 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1031 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1033 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1036 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1038 2017-11-19 Tom de Vries <tom@codesourcery.com>
1040 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1043 2017-11-19 Tom de Vries <tom@codesourcery.com>
1045 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1048 2017-11-19 Tom de Vries <tom@codesourcery.com>
1050 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1053 2017-11-19 Tom de Vries <tom@codesourcery.com>
1055 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1057 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1059 2017-11-19 Tom de Vries <tom@codesourcery.com>
1061 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1063 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1064 semicolon after MCORE_EXPORT_NAME call.
1066 2017-11-19 Tom de Vries <tom@codesourcery.com>
1069 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1070 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1072 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1074 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1075 (cgraph_edge::clone): Cleanup updating of profile.
1076 * ipa-cp.c (update_profiling_info): Likewise.
1077 * ipa-inline-transform.c (inline_transform): Likewise.
1078 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1079 * ipa-split.c (execute_split_functions): Do not split when function
1081 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1082 * profile-count.c (profile_count::dump): Add global0.
1083 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1085 (profile_count::to_sreal_scale): Likewise.
1086 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1087 (profile_count::combine_with_ipa_count): New function.
1088 * profile-count.h (profile_guessed_global0adjusted): New.
1089 (profile_count::adjusted_zero): New.
1090 (profile_count::global0adjusted): New.
1091 (profile_count::combine_with_ipa_count): New.
1092 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1093 correct profile of return block of split functions.
1094 (copy_cfg_body): Remove unused profile_count.
1095 (copy_body): Likewise.
1096 (expand_call_inline): Update.
1097 (tree_function_versioning): Update.
1099 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1101 * hash-set.h (hash_set::empty): New.
1102 * tree-ssa-threadbackward.h: Delete.
1103 * tree-ssa-threadbackward.c (class thread_jumps): New.
1104 Move max_threaded_paths into class.
1105 (fsm_find_thread_path): Remove arguments that are now in class.
1106 (profitable_jump_thread_path): Rename to...
1107 (thread_jumps::profitable_jump_thread_path): ...this.
1108 (convert_and_register_jump_thread_path): Rename to...
1109 (thread_jumps::convert_and_register_current_path): ...this.
1110 (check_subpath_and_update_thread_path): Rename to...
1111 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1112 (register_jump_thread_path_if_profitable): Rename to...
1113 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1114 (handle_phi): Rename to...
1115 (thread_jumps::handle_phi): ...this.
1116 (handle_assignment): Rename to...
1117 (thread_jumps::handle_assignment): ...this.
1118 (fsm_find_control_statement_thread_paths): Rename to...
1119 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1120 (find_jump_threads_backwards): Rename to...
1121 (thread_jumps::find_jump_threads_backwards): ...this.
1122 Initialize path local data.
1123 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1124 from within thread_jumps class.
1125 (pass_early_thread_jumps::execute): Same.
1127 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1129 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1131 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1133 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1134 to canonical location.
1136 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1139 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1141 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1143 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1145 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1146 rcrt1.o%s/grcrt1.o%s for -static-pie.
1148 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1150 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1151 ix86_shift_rotate_cost): Break out from ...
1152 (ix86_rtx_costs): ... here.
1153 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1156 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1158 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1159 when entry block was promoted unlikely.
1160 (estimate_bb_frequencies): Increase frequency scale.
1161 * profile-count.h (profile_count): Export precision info.
1163 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1165 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1166 disturbin profile of entry block.
1168 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1170 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1173 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1175 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1176 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1178 2017-11-17 Jeff Law <law@redhat.com>
1180 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1181 from evrp_dom_walker class. Various methods moved into new class.
1182 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1183 (evrp_range_analyzer::enter): New method.
1184 (evrp_range_analyzer::leave): New method.
1185 (evrp_dom_walker): Remove delegators no longer needed by this class.
1186 Replace vr_values data member with evrp_range_analyzer
1188 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1189 method extracted from evrp_dom_walker::before_dom_children.
1190 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1191 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1193 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1194 Add private copy constructor and move assignment operators.
1195 Privatize methods and class data where trivially possible.
1196 (evrp_dom_walker::cleanup): New function, extracted from
1197 execute_early_vrp. Simplify access to class data.
1199 * vr-values.h (get_output_for_vrp): Prototype.
1200 * vr-values.c (get_output_for_vrp): New function extracted from
1201 vrp_visit_assignment_or_call and extract_range_from_stmt.
1202 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1204 2017-11-17 Luis Machado <luis.machado@linaro.org>
1206 * config/aarch64/aarch64.c
1207 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1208 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1209 tune_params::AUTOPREFETCHER_WEAK.
1211 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1214 * config/arm/arm.c (arm_valid_target_attribute_rec):
1215 Parse "arch=" and "+<ext>".
1216 (arm_valid_target_attribute_tree): Re-init global options.
1217 (arm_option_override): Make non-static.
1218 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1219 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1220 (__ARM_FEATURE_CRC32): Support undef.
1221 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1222 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1224 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1226 * gdbinit.in (break-on-diagnostic): New command.
1228 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1230 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1231 return to indirect jump for EH return if control-flow
1232 protection is enabled. Change explicit 'false' argument in
1233 pro_epilogue_adjust_stack with a value of
1235 * config/i386/i386.md (simple_return_indirect_internal):
1236 Remove SImode restriction to support 64-bit.
1238 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1240 * combine.c (added_notes_insn): New.
1241 (try_combine): Handle added_notes_insn like added_links_insn.
1242 Rewrite return value code.
1243 (distribute_notes): Set added_notes_insn to the earliest insn we added
1246 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1248 PR rtl-optimization/82621
1249 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1250 dest of one of those SETs is unused.
1252 2017-11-17 Richard Biener <rguenther@suse.de>
1255 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1256 * tree-pretty-print.c (dump_generic_node): Handle
1257 annot_expr_parallel_kind.
1258 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1259 * gimplify.c (gimple_boolify): Likewise.
1261 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1263 * config.gcc (extra_headers): Add cet.h for x86 targets.
1264 * config/i386/cet.h: New file.
1265 * doc/install.texi: Add --enable-cet/--disable-cet.
1267 2017-11-17 Richard Biener <rguenther@suse.de>
1269 PR tree-optimization/83017
1270 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1271 (gen_parallel_loop): Properly count iterations.
1272 (parallelize_loops): Handle loop->can_be_parallel independent
1273 of flag_loop_parallelize_all. Make static profitability test match
1275 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1276 * invoke.texi (parloops-min-per-thread): Document.
1278 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1280 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1281 upstreaming review comments.
1283 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1285 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1286 with fast unaligned access.
1287 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1289 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1291 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1292 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1294 2017-11-17 Richard Biener <rguenther@suse.de>
1296 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1297 folding of references.
1299 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1302 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1303 of replacing call to strncmp with corresponding call to strcmp when
1306 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1308 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1309 option prefer-avx256 for skylake-avx512 configuration.
1310 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1311 (get_builtin_code_for_version): Ditto.
1313 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1314 Monk Chiang <sh.chiang04@gmail.com>
1316 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1317 (FIXED_REGISTERS): Reserve more register numbers.
1318 (CALL_USED_REGISTERS): Likewise.
1319 (REG_ALLOC_ORDER): Likewise.
1320 (REG_CLASS_CONTENTS): Likewise.
1321 (REGISTER_NAMES): Likewise.
1323 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1324 Kito Cheng <kito.cheng@gmail.com>
1326 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1328 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1330 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1332 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1333 parameter list for vec_splats.
1335 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1337 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1339 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1341 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1344 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1345 with the required version of automake 1.11.6.
1347 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1349 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1351 (DU_C2_3_power9): Correct reservation combinations.
1352 (FP_DIV_power9, VEC_DIV_power9): New.
1353 (power9-alu): Split out rotate/shift...
1354 (power9-rot): ...to here, correct dispatch resource.
1355 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1357 (power9-fp): Correct latency.
1358 (power9-sdiv): Add div/sqrt resource.
1359 (power9-ddiv): Correct latency, add div/sqrt resource.
1360 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1361 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1363 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1365 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1367 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1368 switch statement mapping KF built-ins to TF built-ins if we don't
1369 have the proper ISA 3.0 assembler support.
1371 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1373 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1374 (lower_emutls_function_body): Do not compute it.
1376 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1378 * ipa-split.c (split_bb_info): Turn time to sreal.
1379 (split_point): Likewise.
1380 (dump_split_point): Likewise.
1381 (fine_split_points): Likewise.
1382 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1384 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1386 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1389 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1391 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1393 (expensive_function_p): Remove useless assert.
1394 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1396 2017-11-16 Martin Sebor <msebor@redhat.com>
1398 PR tree-optimization/82588
1399 PR tree-optimization/82583
1400 * tree-vrp.c (check_array_ref): Handle flexible array members,
1401 string literals, and inner indices.
1402 (search_for_addr_array): Add detail to diagnostics.
1404 2017-11-16 Nathan Sidwell <nathan@acm.org>
1408 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1409 (SET_DECL_ASSEMBLER_NAME): Forward to
1410 overwrite_decl_assembler_name.
1411 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1412 (overwrite_decl_assembler_name): Declare.
1413 * tree.c (overwrite_decl_assembler_name): New.
1414 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1415 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1416 (LANG_HOOKS_INITIALIZER): Add it.
1417 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1418 * langhooks.c (lhd_set_decl_assembler_name): Use
1419 SET_DECL_ASSEMBLER_NAME.
1420 (lhd_overwrite_decl_assembler_name): Default implementation.
1422 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1423 Jackson Woodruff <jackson.woodruff@arm.com>
1425 PR tree-optimization/71026
1426 * match.pd: Canonicalize constant multiplies in division.
1428 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1430 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1431 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1432 Remove OPT_fomit_frame_pointer entry.
1433 * common/config/arc/arc-common.c: Likewise.
1434 * common/config/arm/arm-common.c: Likewise.
1435 * common/config/avr/avr-common.c: Likewise.
1436 * common/config/c6x/c6x-common.c: Likewise.
1437 * common/config/cr16/cr16-common.c: Likewise.
1438 * common/config/cris/cris-common.c: Likewise.
1439 * common/config/epiphany/epiphany-common.c: Likewise.
1440 * common/config/fr30/fr30-common.c: Likewise.
1441 * common/config/frv/frv-common.c: Likewise.
1442 * common/config/ia64/ia64-common.c: Likewise.
1443 * common/config/iq2000/iq2000-common.c: Likewise.
1444 * common/config/lm32/lm32-common.c: Likewise.
1445 * common/config/m32r/m32r-common.c: Likewise.
1446 * common/config/mcore/mcore-common.c: Likewise.
1447 * common/config/microblaze/microblaze-common.c: Likewise.
1448 * common/config/mips/mips-common.c: Likewise.
1449 * common/config/mmix/mmix-common.c: Likewise.
1450 * common/config/mn10300/mn10300-common.c: Likewise.
1451 * common/config/nios2/nios2-common.c: Likewise.
1452 * common/config/pa/pa-common.c: Likewise.
1453 * common/config/pdp11/pdp11-common.c: Likewise.
1454 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1455 * common/config/riscv/riscv-common.c: Likewise.
1456 * common/config/rs6000/rs6000-common.c: Likewise.
1457 * common/config/rx/rx-common.c: Likewise.
1458 * common/config/s390/s390-common.c: Likewise.
1459 * common/config/sh/sh-common.c: Likewise.
1460 * common/config/sparc/sparc-common.c: Likewise.
1461 * common/config/tilegx/tilegx-common.c: Likewise.
1462 * common/config/tilepro/tilepro-common.c: Likewise.
1463 * common/config/v850/v850-common.c: Likewise.
1464 * common/config/visium/visium-common.c: Likewise.
1465 * common/config/xstormy16/xstormy16-common.c: Likewise.
1466 * common/config/xtensa/xtensa-common.c: Likewise.
1467 * invoke.texi (-fomit-frame-pointer): Update documentation.
1469 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1471 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1473 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1475 * cfg.c (scale_bbs_frequencies_int,
1476 cale_bbs_frequencies_gcov_type): Remove.
1477 * cfg.h (scale_bbs_frequencies_int,
1478 cale_bbs_frequencies_gcov_type): Remove.
1480 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1482 * tree-ssa-loop-manip.c
1483 (scale_dominated_blocks_in_loop): Update to profile counts.
1484 (tree_transform_and_unroll_loop): Likewise.
1486 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1488 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1489 scale_bbs_frequencies_int.
1491 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1493 * final.c (compute_alignments): Use counts rather than frequencies.
1495 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1497 * cfgloopanal.c: Include sreal.h
1498 (average_num_loop_insns): Use counts and sreal for accounting.
1500 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1502 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1505 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1506 Kito Cheng <kito.cheng@gmail.com>
1508 * config/nds32/constraints.md: Provide more constraints.
1509 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1510 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1511 support constraints usage.
1513 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1515 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1517 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1518 Kito Cheng <kito.cheng@gmail.com>
1520 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1521 * config/nds32/nds32.opt: Refine the layout.
1522 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1523 TARGET_EXT_STRING): Support new options.
1524 * config/nds32/nds32.h: Likewise.
1525 * config/nds32/nds32.md: Likewise.
1526 * config/nds32/nds32-predicates.c: Likewise.
1527 * config/nds32/constraints.md: Likewise.
1528 * common/config/nds32/nds32-common.c: Likewise.
1530 2017-11-16 Julia Koval <julia.koval@intel.com>
1533 * config/i386/gfniintrin.h: Add sse check.
1534 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1536 2017-11-16 Julia Koval <julia.koval@intel.com>
1538 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1539 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1540 (ix86_handle_option): Handle -mavx512vbmi2.
1541 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1542 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1543 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1544 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1545 (ix86_valid_target_attribute_inner_p): Ditto.
1546 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1547 * config/i386/i386.opt (mavx512vbmi2): New option.
1548 * doc/invoke.texi: Add new option.
1550 2017-11-16 Julia Koval <julia.koval@intel.com>
1552 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1553 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1554 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1555 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1556 _mm512_gf2p8mul_epi8): New intrinsics.
1557 * config/i386/i386-builtin-types.def
1558 (V64QI_FTYPE_V64QI_V64QI): New type.
1559 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1560 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1561 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1562 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1563 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1564 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1566 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1568 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1569 explicitly as a stream of bytes.
1571 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1573 * config/rs6000/altivec.h (vec_xst_be): New #define.
1574 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1575 and externalize from *altivec_vperm_<mode>_internal.
1576 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1578 (XL_BE_V8HI): Likewise.
1579 (XL_BE_V4SI): Likewise.
1580 (XL_BE_V4SI): Likewise.
1581 (XL_BE_V2DI): Likewise.
1582 (XL_BE_V4SF): Likewise.
1583 (XL_BE_V2DF): Likewise.
1584 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1585 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1586 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1587 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1588 VSX_BUILTIN_VEC_XST_BE.
1589 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1590 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1592 (altivec_init_builtins): Replace conditional calls to def_builtin
1593 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1594 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1595 with unconditional calls. Remove calls to def_builtin for
1596 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1597 __builtin_vec_xst_be.
1598 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1599 to define_expand, and add alternate RTL generation for P8.
1600 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1601 vsx_ld_elemrev_v8hi.
1602 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1603 add alternate RTL generation for P8.
1604 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1605 vsx_ld_elemrev_v16qi.
1606 (vsx_st_elemrev_v8hi): Convert define_insn
1607 to define_expand, and add alternate RTL generation for P8.
1608 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1609 vsx_st_elemrev_v8hi.
1610 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1611 add alternate RTL generation for P8.
1612 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1613 vsx_st_elemrev_v16qi.
1615 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1618 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1619 TARGET_AVX512ER check.
1620 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1621 neither -mzeroupper nor -mno-zeroupper is used and
1622 TARGET_EMIT_VZEROUPPER is set.
1623 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1624 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1626 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1628 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1629 folding of vector compares.
1630 (fold_build_vec_cmp): New helper function.
1631 (fold_compare_helper): New helper function.
1632 (builtin_function_type): Add compare builtins to the list of functions
1633 having unsigned arguments. Cosmetic updates to comment indentation.
1634 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1635 the not+eq combination.
1637 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1639 PR tree-optimization/82726
1640 PR tree-optimization/70754
1641 * tree-predcom.c (order_drefs_by_pos): New function.
1642 (combine_chains): Move code setting has_max_use_after to...
1643 (try_combine_chains): ...here. New parameter. Sort combined chains
1644 according to position information.
1645 (tree_predictive_commoning_loop): Update call to above function.
1646 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1648 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1650 PR tree-optimization/82726
1652 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1654 PR tree-optimization/70754
1655 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1656 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1657 combined stmt before it if not NULL.
1658 (combine_chains): Process refs reversely and compute dominance point
1662 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1664 PR tree-optimization/79663
1665 * tree-predcom.c (combine_chains): Process refs in reverse order
1666 only for ZERO length chains, and add explaining comment.
1668 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1670 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1671 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1672 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1674 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1676 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1678 2017-11-15 Martin Liska <mliska@suse.cz>
1680 * tree-cfg.c (pass_warn_function_return::execute):
1681 Compare warn_return_type for greater than zero.
1683 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1687 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1688 to return true on Xeon and not on Xeon Phi.
1689 (ix86_check_avx256_register): Changed to ...
1690 (ix86_check_avx_upper_register): ... this. Add extra check for
1691 VALID_AVX512F_REG_OR_XI_MODE.
1692 (ix86_avx_u128_mode_needed): Changed
1693 ix86_check_avx256_register to ix86_check_avx_upper_register.
1694 (ix86_check_avx256_stores): Changed to ...
1695 (ix86_check_avx_upper_stores): ... this. Changed
1696 ix86_check_avx256_register to ix86_check_avx_upper_register.
1697 (ix86_avx_u128_mode_after): Changed
1698 avx_reg256_found to avx_upper_reg_found. Changed
1699 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1700 (ix86_avx_u128_mode_entry): Changed
1701 ix86_check_avx256_register to ix86_check_avx_upper_register.
1702 (ix86_avx_u128_mode_exit): Ditto.
1703 * config/i386/i386.h: (host_detect_local_cpu): New define.
1705 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1707 * config/arm/xgene1.md (xgene1): Split into automatons
1708 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1709 (xgene1_f_load): Adjust reservations and/or types.
1710 (xgene1_f_store): Likewise.
1711 (xgene1_load_pair): Likewise.
1712 (xgene1_store_pair): Likewise.
1713 (xgene1_fp_load1): Likewise.
1714 (xgene1_load1): Likewise.
1715 (xgene1_store1): Likewise.
1716 (xgene1_move): Likewise.
1717 (xgene1_alu): Likewise.
1718 (xgene1_simd): Likewise.
1719 (xgene1_bfm): Likewise.
1720 (xgene1_neon_load1): Likewise.
1721 (xgene1_neon_store1): Likewise.
1722 (xgene1_neon_logic): Likewise.
1723 (xgene1_neon_st1): Likewise.
1724 (xgene1_neon_ld1r): Likewise.
1725 (xgene1_alu_cond): Added.
1726 (xgene1_shift_reg): Likwise.
1727 (xgene1_bfx): Likewise.
1728 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1730 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1733 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1735 (can_widen_mult_without_libcall): New function.
1736 (expand_mul_overflow): If only checking unsigned mul overflow,
1737 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1738 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1739 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1740 (expand_DIVMOD): Formatting fix.
1741 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1742 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1743 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1745 PR tree-optimization/82977
1746 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1747 constructed temporary to strlen_to_stridx.put.
1749 2017-11-15 Martin Liska <mliska@suse.cz>
1751 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1752 * configure: Regenerate.
1754 2017-11-15 Martin Liska <mliska@suse.cz>
1757 * config/sh/sh-mem.cc: Use proper probability for
1760 2017-11-14 Jeff Law <law@redhat.com>
1762 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1763 the red zone for stack_clash_protection_final_dynamic_probe targets
1764 when the total dynamic stack size is zero bytes.
1766 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1767 blocks is post order.
1769 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1771 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1772 * final.c (rest_of_clean_state): Set it for the
1773 -fcompare-debug dump.
1774 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1775 class when TDF_COMPARE_DEBUG is set.
1777 * dwarf2out.c (gen_producer_string): Discard
1780 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1783 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1784 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1785 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1786 (__TGMATH_REAL_2_3): Remove macros.
1787 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1788 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1790 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1792 (remquo): Define using __TGMATH_REAL_3.
1794 2017-11-14 Jeff Law <law@redhat.com>
1796 * vr-values.c: New file with contents extracted from tree-vrp.c.
1797 * Makefile.in (OBJS): Add vr-values.o
1798 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
1799 (set_value_range, set_and_canonicalize_value_range): Likewise.
1800 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
1801 (value_range_constant_singleton, symbolic_range_p): Likewise.
1802 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
1803 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
1804 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
1805 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
1806 (find_case_label_range, find_case_label_index): Likewise.
1807 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
1808 (range_int_cst_singleton_p, value_inside_range): Likewise.
1809 (get_single_symbol): Likewise.
1810 (switch_update): Move structure definition here.
1811 (to_remove_edges, to_update_switch_stmts): Provide externs.
1812 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
1813 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
1814 (vrp_val_is_min, set_value_range): Likewise.
1815 (set_and_canonicalize_value_range, copy_value_range): Likewise.
1816 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
1817 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
1818 (range_is_nonnull, range_int_cst_p): Likewwise.
1819 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
1820 (get_single_symbol, operand_less_p): Likewise
1821 (compare_values_warnv, compare_values): Likewise.
1822 (value_inside_range, value_range_constant_singleton): Likewise.
1823 (zero_nonzero_bitgs_from_vr): Likewise.
1824 (extract_range_from_binary_expr_1): Likewise.
1825 (overflow_comparison_p): Likewise.
1826 (to_remove_edges, to_update_switch_stmts): Likewise.
1827 (find_case_label-index, find_case_label_range): Likewise.
1828 (switch_update, set_value_range_to_nonnegative): Remove.
1829 (set_value_range_to_truthvalue): Likewise.
1830 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
1831 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1832 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1833 (find_case_label_ranges, test_for_singularity): Likewise.
1834 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1835 (x_vr_values): Move to its remaining use site.
1837 2017-11-10 Jeff Law <law@redhat.com>
1839 * vr-values.h (VR_INITIALIZER): Move #define here.
1840 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1841 * Makefile.in (OBJS): Add tree-evrp.o
1842 * tree-vrp.h (assert_info): Move structure definition here.
1843 (set_value_range_to_varying): Prototype.
1844 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1845 (infer_value_range, register_edge_assert_for): Likewise.
1846 (stmt_interesting_for_vrp): Likewise.
1847 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1848 (set_value_range_to_varying): No longer static.
1849 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1850 (infer_value_range, register_edge_assert_for): Likewise.
1852 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1854 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1855 generate the XXBRD instruction.
1857 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1858 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1859 as being compatible if -mabi=ieeelongdouble.
1860 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1861 to setup float128 built-ins with hardware support.
1862 (BU_FLOAT128_HW_2): Likewise.
1863 (BU_FLOAT128_HW_3): Likewise.
1864 (BU_FLOAT128_HW_VSX_1): Likewise.
1865 (BU_FLOAT128_HW_VSX_2): Likewise.
1866 (scalar_extract_expq): Change float128 built-in functions to
1867 accommodate having both KFmode and TFmode functions. Use the
1868 KFmode variant as the default.
1869 (scalar_extract_sigq): Likewise.
1870 (scalar_test_neg_qp): Likewise.
1871 (scalar_insert_exp_q): Likewise.
1872 (scalar_insert_exp_qp): Likewise.
1873 (scalar_test_data_class_qp): Likewise.
1874 (sqrtf128_round_to_odd): Delete processing the round to odd
1875 built-in functions as special built-in functions, and define them
1876 as float128 built-ins. Use the KFmode variant as the default.
1877 (truncf128_round_to_odd): Likewise.
1878 (addf128_round_to_odd): Likewise.
1879 (subf128_round_to_odd): Likewise.
1880 (mulf128_round_to_odd): Likewise.
1881 (divf128_round_to_odd): Likewise.
1882 (fmaf128_round_to_odd): Likewise.
1883 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1884 support for KFmode and TFmode xststdcqp calls.
1885 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1886 point, switch the built-in handlers for the get/set float128
1887 exponent, get float128 mantissa, float128 test built-ins, and the
1888 float128 round to odd built-in functions. Eliminate creating the
1889 float128 round to odd built-in functions as special built-ins.
1890 (rs6000_init_builtins): Eliminate special creation of the float128
1891 round to odd built-in functions.
1892 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1893 function insns to support both TFmode and KFmode variants.
1894 (xsxsigqp_<mode>): Likewise.
1895 (xsiexpqpf_<mode>): Likewise.
1896 (xsiexpqp_<mode>): Likewise.
1897 (xststdcqp_<mode>): Likewise.
1898 (xststdcnegqp_<mode>): Likewise.
1899 (xststdcqp_<mode>): Likewise.
1901 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1903 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1904 unused path_in_freq_ptr parameter.
1905 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1907 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1909 * ipa-inline.c (edge_badness): Dump sreal frequency.
1910 (compute_inlined_call_time): Match natural implementaiton ...
1911 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1912 forgotten division by CGRAPH_FREQ_BASE.
1914 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1916 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1917 Solaris 11. Update comment.
1918 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1920 * config/sol2.h (STARTFILE_SPEC): Likewise.
1921 * configure: Regenerate.
1923 2017-11-14 Carl Love <cel@us.ibm.com>
1925 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1926 le_ and be_ prefixes to swap* variables. Remove
1927 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1930 2017-11-14 Jason Merrill <jason@redhat.com>
1932 Support GTY((cache)) on hash_map.
1933 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1934 (ggc_cache_remove): Override it instead of ggc_mx.
1935 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1936 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1937 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1938 (simple_cache_map_traits): Override maybe_mx.
1939 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1940 (hash_map): Friend gt_cleare_cache.
1941 (gt_cleare_cache): New.
1942 * tree.h (tree_cache_traits): New hash_map traits class.
1943 (tree_cache_map): New typedef.
1945 2017-11-14 Richard Biener <rguenther@suse.de>
1947 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1948 paramter and handling.
1949 (cleanup_control_flow_bb): Likewise.
1950 (cleanup_control_flow_pre): New helper performing a DFS walk
1951 to call cleanup_control_flow_bb in PRE order.
1952 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1953 via cleanup_control_flow_pre.
1955 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1957 * config/aarch64/aarch64-simd.md
1958 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1959 (*aarch64_simd_bsl<mode>_alt): Likewise.
1960 (aarch64_simd_bsldi_internal): New.
1961 (aarch64_simd_bsldi_alt): Likewise.
1963 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1965 * tracer.c (better_p): Do not compare frequencies.
1966 * reg-stack.c (better_edge): Likewise.
1967 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1970 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1972 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1973 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1974 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1975 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1976 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1977 * predict.c (maybe_hot_frequency_p): Inline to ...
1978 (maybe_hot_count_p): ... here; rewrite to counts.
1979 (counts_to_freqs): Rename to ...
1980 (update_max_bb_count): ... this one.
1981 (expensive_function_p): Use counts.
1982 (estimate_bb_frequencies): Update.
1983 (rebuild_frequencies): Update.
1984 * predict.h (counts_to_freqs): Rename to ...
1985 (update_max_bb_count): ... this one.
1986 * profile.c (compute_branch_probabilities): Add debug info
1987 * tree-inline.c (expand_call_inline): Update debug info.
1988 (optimize_inline_calls): Use update_max_bb_count..
1989 (tree_function_versioning): Use update_max_bb_count..
1990 * value-prof.c (gimple_value_profile_transformations):
1991 Do not use update_max_bb_count.
1993 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1995 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1996 always use frequencies.
1998 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2000 * bb-reorder.c: Remove frequencies from comments.
2001 (better_edge_p): Use profile counts.
2002 (find_traces): Dump profile counts.
2003 (rotate_loop): Use profile counts.
2004 (find_traces_1_round): Likewise.
2005 (connect_better_edge_p): Use counts instead of probabilities for
2007 (copy_bb_p): Drop early check for non-0 frequency.
2008 (sanitize_hot_paths): Update comments.
2010 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2012 * ipa-split.c (struct split_point): Add count.
2013 (consider_split): Do not compute incoming frequency; compute incoming
2014 count and store it to split_point.
2015 (split_function): Set count of the call to split part correctly.
2017 2017-11-13 Carl Love <cel@us.ibm.com>
2019 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2021 2017-11-13 Tom Tromey <tom@tromey.com>
2023 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2025 2017-11-13 Carl Love <cel@us.ibm.com>
2027 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2028 Add support for builtins:
2029 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2030 vector {un,}signed {char,int,short},
2031 vector {un,}signed {char,int,short}) arguments.
2032 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2033 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2034 Add BU_P9V_AV_2 expansions for the builtins.
2035 * config/rs6000/altivec.h (vec_first_match_index,
2036 vec_first_mismatch_index, vec_first_match_or_eos_index,
2037 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2038 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2039 new extern declaration.
2040 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2041 * config/rs6000/vsx.md (first_match_index_<mode>,
2042 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2043 first_mismatch_or_eos_index_<mode>): Add define expand.
2044 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2045 * doc/extend.texi: Update the built-in documenation file for the new
2048 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2050 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2051 into the min/max operations for _Float<N> and _Float<N>X types.
2053 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2056 * dwarf2out.c (do_eh_frame): New static variable.
2057 (dwarf2out_begin_prologue): Set it.
2058 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2060 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2062 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2064 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2066 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2069 * debug.h (dwarf2out_do_eh_frame): Declare.
2070 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2071 (dwarf2out_do_frame): Use it.
2072 (dwarf2out_do_cfi_asm): Likewise.
2073 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2074 (dwarf2out_assembly_start): Likewise.
2075 (dwarf2out_begin_prologue): Fix comment.
2076 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2077 if the target needs either debug or unwind DWARF2 info.
2078 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2079 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2080 -ftrapping-math, -ftrapv and -fwrapv.
2082 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2084 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2085 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2086 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2087 (estimate_edge_size_and_time): Likewise.
2088 (ipa_merge_fn_summary_after_inlining): Likewise.
2089 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2090 (compute_uninlined_call_time): Use sreal_frequency.
2091 (compute_inlined_call_time): Likewise.
2092 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2093 * profile-count.c: Include sreal.h.
2094 (profile_count::to_sreal_scale): New.
2095 * profile-count.h: Forward declare sreal.
2096 (profile_count::to_sreal_scale): Declare.
2098 2017-11-13 Nathan Sidwell <nathan@acm.org>
2100 * diagnostic.c (maybe_line_and_column): New.
2101 (diagnostic_get_location_text): Use it.
2102 (diagnostic_report_current_module): Likewise.
2103 (test_diagnostic_get_location_text): Add tests.
2105 2017-11-13 Luis Machado <luis.machado@linaro.org>
2107 * doc/md.texi (Specifying processor pipeline description): Fix
2108 incorrect latency for the div instruction example.
2110 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2112 PR tree-optimization/78821
2113 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2114 that bit_not_p is the same.
2115 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2116 (split_group): Count precisely bit_not_p bits in each statement.
2117 (invert_op): New function.
2118 (imm_store_chain_info::output_merged_store): Use invert_op to
2119 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2120 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2122 2017-11-13 Martin Liska <mliska@suse.cz>
2124 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2125 (struct source_info): Likewise.
2126 (add_branch_counts): Likewise.
2127 (add_line_counts): Likewise.
2128 (function_summary): Likewise.
2129 (output_intermediate_line): Likewise.
2130 (generate_results): Likewise.
2132 2017-11-13 Martin Liska <mliska@suse.cz>
2134 * gcov.c (struct block_info): Remove typedef for block_t.
2135 (struct line_info): Likewise.
2136 (line_info::has_block): Likewise.
2137 (EXIT_BLOCK): Likewise.
2138 (unblock): Likewise.
2139 (circuit): Likewise.
2140 (get_cycles_count): Likewise.
2141 (process_file): Likewise.
2142 (read_graph_file): Likewise.
2143 (solve_flow_graph): Likewise.
2144 (find_exception_blocks): Likewise.
2145 (add_line_counts): Likewise.
2146 (accumulate_line_info): Likewise.
2147 (output_line_details): Likewise.
2149 2017-11-13 Martin Liska <mliska@suse.cz>
2151 * gcov.c (struct arc_info): Remove typedef for arc_t.
2152 (struct line_info): Likewise.
2153 (add_branch_counts): Likewise.
2154 (output_branch_count): Likewise.
2155 (function_info::~function_info): Likewise.
2156 (circuit): Likewise.
2157 (output_intermediate_line): Likewise.
2158 (read_graph_file): Likewise.
2159 (solve_flow_graph): Likewise.
2160 (find_exception_blocks): Likewise.
2161 (add_line_counts): Likewise.
2162 (accumulate_line_info): Likewise.
2163 (output_line_details): Likewise.
2164 (output_function_details): Likewise.
2166 2017-11-13 Martin Liska <mliska@suse.cz>
2168 * gcov.c (struct function_info): Remove typedef for function_t.
2169 (struct source_info): Likewise.
2170 (source_info::get_functions_at_location): Likewise.
2171 (solve_flow_graph): Likewise.
2172 (find_exception_blocks): Likewise.
2173 (add_line_counts): Likewise.
2174 (output_intermediate_file): Likewise.
2175 (process_file): Likewise.
2176 (generate_results): Likewise.
2177 (release_structures): Likewise.
2178 (read_graph_file): Likewise.
2179 (read_count_file): Likewise.
2180 (accumulate_line_counts): Likewise.
2181 (output_lines): Likewise.
2183 2017-11-13 Martin Liska <mliska@suse.cz>
2185 * gcov.c (function_info::function_info): Remove num_counts
2186 and add vector<gcov_type>.
2187 (function_info::~function_info): Use the vector.
2188 (process_file): Likewise.
2189 (read_graph_file): Likewise.
2190 (read_count_file): Likewise.
2191 (solve_flow_graph): Likewise.
2193 2017-11-13 Martin Liska <mliska@suse.cz>
2195 * gcov.c (function_info::is_artificial): New function.
2196 (process_file): Erase all artificial early.
2197 (generate_results): Skip as all artificial are already
2200 2017-11-13 Martin Liska <mliska@suse.cz>
2202 * gcov.c (read_graph_file): Store to global vector of functions.
2203 (read_count_file): Iterate the vector.
2204 (process_file): Likewise.
2205 (generate_results): Likewise.
2206 (release_structures): Likewise.
2208 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2210 PR tree-optimization/82954
2211 * gimple-ssa-store-merging.c
2212 (imm_store_chain_info::coalesce_immediate_stores): If
2213 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2215 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2217 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2218 Upddate call to ENDIAN_LANE_N.
2219 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2220 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2221 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2222 and use aarch64_endian_lane_rtx.
2223 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2225 2017-11-12 Tom de Vries <tom@codesourcery.com>
2227 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2229 2017-11-12 Tom de Vries <tom@codesourcery.com>
2231 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2233 2017-11-12 Tom de Vries <tom@codesourcery.com>
2235 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2237 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2238 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2239 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2241 2017-11-11 Martin Sebor <msebor@redhat.com>
2244 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2247 * prefic.c (translate_name): Replace strncpy with memcpy to
2248 avoid -Wstringop-truncation.
2250 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2252 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2254 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2256 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2258 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2260 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2263 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2265 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2266 enable generating XXBRH if the value is in a vector register.
2267 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2268 value is in a vector register.
2269 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2270 register to register bswap64's instead of doing the GPR sequence
2271 used on previous machines.
2272 (bswapdi2_xxbrd): New insn.
2273 (bswapdi2_reg): Disallow on ISA 3.0.
2274 (register to register bswap64 splitter): Do not split the insn on
2275 ISA 3.0 systems that use XXBRD.
2277 2017-11-10 Martin Sebor <msebor@redhat.com>
2280 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2282 (find_subframework_file): Same.
2284 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2286 * auto-profile.c (afdo_indirect_call): Drop frequency.
2287 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2288 (cgraph_node::create_edge): Drop frequency argument.
2289 (cgraph_node::create_indirect_edge): Drop frequency argument.
2290 (cgraph_edge::make_speculative): Drop frequency arguments.
2291 (cgraph_edge::resolve_speculation): Do not update frequencies
2292 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2293 (cgraph_node::dump): Check consistency in IPA mode.
2294 (cgraph_edge::maybe_hot_p): Use IPA counter.
2295 (cgraph_edge::verify_count_and_frequency): Rename to ...
2296 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2297 (cgraph_node::verify_node): Update.
2298 * cgraph.h (struct cgraph_edge): Drop frequency.
2299 (cgraph_edge::frequency): New function.
2300 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2302 (cgraph_edge::rebuild_edges): Likewise.
2303 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2304 (duplicate_thunk_for_node): Do not pass frequency.
2305 (cgraph_node::create_clone): Scale only counts.
2306 (cgraph_node::create_virtual_clone): Do not pass frequency.
2307 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2308 (cgraph_node::create_version_clone): Do not pass frequency.
2309 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2310 (cgraph_node::expand_thunk): Do not pass frequency.
2311 (cgraph_node::create_wrapper): Do not pass frequency.
2312 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2314 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2315 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2316 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2317 (gather_caller_stats): Use frequency function.
2318 (ipcp_cloning_candidate_p): Use frequency function.
2319 (ipcp_propagate_stage): Use frequency function.
2320 (get_info_about_necessary_edges): Use frequency function.
2321 (update_profiling_info): Update only IPA profile.
2322 (update_specialized_profile): Use frequency functoin.
2323 (perhaps_add_new_callers): Update only IPA profile.
2324 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2325 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2326 (dump_ipa_call_summary): Use frequency function.
2327 (estimate_edge_size_and_time): Use frequency function.
2328 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2329 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2330 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2331 (update_noncloned_counts): ... ths one; scale counts only.
2332 (clone_inlined_nodes): Do not scale frequency.
2333 (inline_call): Do not pass frequency.
2334 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2335 (compute_inlined_call_time): Use IPA profile.
2336 (want_inline_small_function_p): Use IPA profile.
2337 (want_inline_self_recursive_call_p): Use IPA profile.
2338 (edge_badness): Use IPA profile.
2339 (lookup_recursive_calls): Use IPA profile.
2340 (recursive_inlining): Do not pass frequency.
2341 (resolve_noninline_speculation): Do not update frequency.
2342 (inline_small_functions): Collect max of IPA profile.
2343 (dump_overall_stats): Dump IPA porfile.
2344 (dump_inline_stats): Dump IPA porfile.
2345 (ipa_inline): Collect IPA stats.
2346 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2347 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2348 (ipa_propagate_frequency): Use frequency function.
2349 (ipa_profile): Cleanup.
2350 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2351 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2352 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2353 (input_node): Do not stream frequency.
2354 (input_edge): Do not stream frequency.
2355 (merge_profile_summaries): Scale only IPA profiles.
2356 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2357 * predict.c (drop_profile): Do not recompute frequency.
2358 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2359 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2360 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2361 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2362 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2363 * tree-inline.c (copy_bb): Do not scale frequency.
2364 (expand_call_inline): Do not scale frequency.
2365 (tree_function_versioning): Do not scale frequency.
2366 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2368 2017-11-10 Julia Koval <julia.koval@intel.com>
2370 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2371 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2372 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2373 (_mm256_maskz_gf2p8affine_epi64_epi8)
2374 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2375 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2376 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2377 (__builtin_ia32_vgf2p8affineqb_v32qi)
2378 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2379 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2381 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2385 (arm_option_override): Refactor.
2386 (arm_option_reconfigure_globals): New.
2387 (arm_options_perform_arch_sanity_checks): New.
2388 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2390 (arm_options_perform_arch_sanity_checks): Likewise
2392 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2394 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2395 (power9-qpmul): New.
2396 * rs6000/rs6000.md ("type" attr): Add qmul.
2397 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2398 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2399 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2401 2017-11-10 Martin Sebor <msebor@redhat.com>
2404 * builtins.c (compute_objsize): Handle arrays that
2405 compute_builtin_object_size likes to fail for. Make extern.
2406 * builtins.h (compute_objsize): Declare.
2407 (check_strncpy_sizes): New function.
2408 (expand_builtin_strncpy): Call check_strncpy_sizes.
2409 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2410 -Wstringop-truncation.
2411 (gimple_fold_builtin_strncat): Same.
2412 * gimple.c (gimple_build_call_from_tree): Set call location.
2413 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2414 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2415 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2416 (handle_builtin_strlen): Use strlen_to_stridx.
2417 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2419 Use strlen_to_stridx.
2420 (pass_strlen::execute): Release strlen_to_stridx.
2421 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2422 (-Wstringop-truncation): Document new option.
2424 2017-11-10 Martin Liska <mliska@suse.cz>
2426 PR gcov-profile/82702
2427 * gcov.c (main): Handle intermediate files in a different
2429 (get_gcov_intermediate_filename): New function.
2430 (output_gcov_file): Remove support of intermediate files.
2431 (generate_results): Allocate intermediate file.
2432 (release_structures): Clean-up properly fn_end.
2433 (output_intermediate_file): Start iterating with line 1.
2435 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2437 PR tree-optimization/82929
2438 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2439 ops_swapped_p non-static data member.
2440 (store_immediate_info::store_immediate_info): Clear it.
2441 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2442 ops set ops_swapped_p.
2443 (count_multiple_uses): Handle ops_swapped_p.
2445 2017-11-10 Martin Liska <mliska@suse.cz>
2447 * coverage.c (coverage_init): Stream information about
2448 support of has_unexecuted_blocks.
2449 * doc/gcov.texi: Document that.
2450 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2451 * gcov.c (read_graph_file): Likewise.
2452 (output_line_beginning): Fix a small issue with
2455 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2457 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2458 reference of trivial component.
2460 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2463 * gimple-ssa-store-merging.c
2464 (pass_store_merging::terminate_all_aliasing_chains): For
2465 gimple_store_p stmts also call refs_output_dependent_p.
2467 PR rtl-optimization/82913
2468 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2471 2017-11-09 Jeff Law <law@redhat.com>
2473 * vr-values.h: New file with vr_values class.
2474 * tree-vrp.c: Include vr-values.h
2475 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2476 data objects into the vr_values class.
2477 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2478 (get_value_range): Make it a member function within vr_values class.
2479 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2480 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2481 (op_with_constant_singleton_value_range): Likewise.
2482 (extract_range_for_var_from_comparison_expr): Likewise.
2483 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2484 (extract_range_from_binary_expr): Likewise.
2485 (extract_range_from_unary_expr): Likewise.
2486 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2487 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2488 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2489 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2490 (compare_name_with_value, compare_names): Likewise.
2491 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2492 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2493 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2494 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2495 (extract_range_from_phi_node): Likewise.
2496 (simplify_truth_ops_using_ranges): Likewise.
2497 (simplify_div_or_mod_using_ranges): Likewise.
2498 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2499 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2500 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2501 (simplify_float_conversion_using_ranges): Likewise.
2502 (simplify_internal_call_using_ranges): Likewise.
2503 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2504 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2505 poitner via x_vr_values for calls into gimple folder.
2506 (vrp_initialize_lattice): Make this the vr_values ctor.
2507 (vrp_free_lattice): Make this the vr_values dtor.
2508 (set_vr_value): New function.
2509 (class vrp_prop): Add vr_values data member. Add various member
2510 functions as well as member functions that delegate to vr_values.
2511 (check_array_ref): Make a member function within vrp_prop class.
2512 (search_for_addr_array, vrp_initialize): Likewise.
2513 (vrp_finalize): Likewise. Revamp to avoid direct access to
2514 vr_value, values_propagated, etc.
2515 (check_array_bounds): Extract vrp_prop class instance pointer from
2516 walk info structure. Use it to call member functions.
2517 (check_all_array_refs): Make a member function within vrp_prop class.
2518 Smuggle class instance pointer via walk info structure.
2519 (x_vr_values): New local static.
2520 (vrp_valueize): Use x_vr_values to get class instance.
2521 (vr_valueize_1): Likewise.
2522 (class vrp_folder): Add vr_values data member. Add various member
2523 functions as well as member functions that delegate to vr_values.
2524 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2525 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2526 class instance from vr_values. Use it to call member functions.
2527 (vrp_dom_walker): Add vr_values data member.
2528 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2529 instance via x_vr_values.
2530 (identify_jump_threads): Accept vr_values as argument. Store
2531 it into the walker structure.
2532 (evrp_dom_walker): Add vr_values class data member. Add various
2534 (evrp_dom_walker::try_find_new_range): Use vr_values data
2535 member to access the memory allocator.
2536 (evrp_dom_walker::before_dom_children): Store vr_values class
2537 instance into the vrp_folder class.
2538 (evrp_dom_walker::push_value_range): Rework to avoid direct
2539 access to num_vr_values and vr_value.
2540 (evrp_dom_walker::pop_value_range): Likewise.
2541 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2542 Use vr_values to get to dump_all_value_ranges member function.
2543 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2544 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2545 Pass vr_values class instance down to identify_jump_threads.
2546 Remove call to vrp_free_lattice.
2547 (debug_all_value_ranges): Remove.
2549 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2550 (vrp_folder): Likewise.
2552 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2553 Get it from the existing bitmap instead.
2554 (vrp_intersect_ranges_1): Likewise.
2556 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2558 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2560 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2561 set bit_not_p to it.
2562 (imm_store_chain_info::coalesce_immediate_stores): Break group
2563 if bit_not_p is different.
2564 (count_multiple_uses, split_group,
2565 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2566 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2567 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2568 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2569 to store_immediate_info ctor.
2571 2017-11-09 Jim Wilson <jimw@sifive.com>
2573 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2574 (scan_prog_file): Likewise.
2576 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2578 * bb-reorder.c (max_entry_frequency): Remove.
2579 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2580 connect_traces, push_to_next_round_p): Remove prototypes.
2581 (find_traces_1_round): Use counts only.
2582 (push_to_next_round_p): Likewise.
2583 (find_traces): Likewise.
2584 (rotate_loop): Likewise.
2585 (find_traces_1_round): Likewise.
2586 (connect_traces): Likewise.
2587 (edge_order): Likewise.
2589 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2591 * config/arm/arm.c (output_return_instruction): Add comments to
2592 indicate requirement for cmse_nonsecure_entry return to account
2593 for the size of clearing instruction output here.
2594 (thumb_exit): Likewise.
2595 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2596 return in hardfloat mode.
2598 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2600 * config/rs6000/rs6000.c (machine_function): Add a bool,
2601 "toc_is_wrapped_separately".
2602 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2603 if it wasn't explicitly set or unset, we are optimizing for speed, and
2604 doing separate shrink-wrapping.
2605 (rs6000_get_separate_components): Enable the TOC component if
2606 saving the TOC register in the prologue.
2607 (rs6000_components_for_bb): Handle the TOC component.
2608 (rs6000_emit_prologue_components): Store the TOC register where needed.
2609 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2610 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2612 2017-11-09 Martin Jambor <mjambor@suse.cz>
2614 * ipa-param-manipulation.c: New file.
2615 * ipa-param-manipulation.h: Likewise.
2616 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2617 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2618 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2619 (ipa_parm_adjustment): Likewise.
2620 (ipa_parm_adjustment_vec): Likewise.
2621 (ipa_get_vector_of_formal_parms): Moved declaration to
2622 ipa-param-manipulation.h.
2623 (ipa_get_vector_of_formal_parm_types): Likewise.
2624 (ipa_modify_formal_parameters): Likewise.
2625 (ipa_modify_call_arguments): Likewise.
2626 (ipa_combine_adjustments): Likewise.
2627 (ipa_dump_param_adjustments): Likewise.
2628 (ipa_modify_expr): Likewise.
2629 (ipa_get_adjustment_candidate): Likewise.
2630 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2631 ipa-param-manipulation.c.
2632 (ipa_get_vector_of_formal_parm_types): Likewise.
2633 (ipa_modify_formal_parameters): Likewise.
2634 (ipa_modify_call_arguments): Likewise.
2635 (ipa_modify_expr): Likewise.
2636 (get_ssa_base_param): Likewise.
2637 (ipa_get_adjustment_candidate): Likewise.
2638 (index_in_adjustments_multiple_times_p): Likewise.
2639 (ipa_combine_adjustments): Likewise.
2640 (ipa_dump_param_adjustments): Likewise.
2641 * tree-sra.c: Also include ipa-param-manipulation.h
2642 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2645 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2646 Alan Hayward <alan.hayward@arm.com>
2647 David Sherwood <david.sherwood@arm.com>
2649 * doc/sourcebuild.texi (vect_masked_store): Document.
2651 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2652 Alan Hayward <alan.hayward@arm.com>
2653 David Sherwood <david.sherwood@arm.com>
2655 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2657 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2658 Alan Hayward <alan.hayward@arm.com>
2659 David Sherwood <david.sherwood@arm.com>
2661 * doc/sourcebuild.texi (vect_variable_length): Document.
2663 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2664 Alan Hayward <alan.hayward@arm.com>
2665 David Sherwood <david.sherwood@arm.com>
2667 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2669 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2676 Alan Hayward <alan.hayward@arm.com>
2677 David Sherwood <david.sherwood@arm.com>
2679 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2680 previously undocumented selectors.
2681 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2683 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2685 * doc/rtl.texi (const_vector): Say that elements can be
2686 const_wide_ints too.
2687 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2688 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2689 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2690 * optabs.c (expand_vector_broadcast): Likewise.
2692 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2693 Alan Hayward <alan.hayward@arm.com>
2694 David Sherwood <david.sherwood@arm.com>
2696 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2697 scaled index even if the unscaled address was invalid.
2698 Don't increase the complexity of using a scale in that case.
2700 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2701 Alan Hayward <alan.hayward@arm.com>
2702 David Sherwood <david.sherwood@arm.com>
2704 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2705 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2706 UNITS_PER_WORD bytes.
2707 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2708 based on REGMODE_NATURAL_SIZE of the inner mode.
2709 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2710 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2711 * expmed.c (lowpart_bit_field_p): Divide the value up into
2712 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2713 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2714 whether something is likely to occupy more than one register.
2716 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2719 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2720 profile_count::adjust_for_ipa_scaling.
2721 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2722 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2724 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2726 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2728 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2729 (split_group): Add total_orig and total_new arguments, estimate the
2730 number of statements related to the store group without store merging
2731 and with store merging.
2732 (imm_store_chain_info::output_merged_store): Adjust split_group
2733 callers, punt if estimated number of statements with store merging
2734 is not smaller than estimated number of statements without it.
2736 (handled_load): Remove has_single_use checks.
2737 (pass_store_merging::process_store): Likewise.
2739 2017-11-09 Richard Biener <rguenther@suse.de>
2741 PR tree-optimization/82902
2742 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2744 2017-11-09 Martin Liska <mliska@suse.cz>
2747 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2750 2017-11-09 Martin Liska <mliska@suse.cz>
2752 PR tree-optimization/82669
2753 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2755 2017-11-09 Martin Liska <mliska@suse.cz>
2757 PR gcov-profile/48463
2758 * coverage.c (coverage_begin_function): Output also end locus
2759 of a function and information whether the function is
2761 * gcov-dump.c (tag_function): Parse and print the information.
2762 * gcov.c (INCLUDE_MAP): Add include.
2763 (INCLUDE_SET): Likewise.
2764 (struct line_info): Move earlier in the source file because
2765 of vector<line_info> in function_info structure.
2766 (line_info::line_info): Likewise.
2767 (line_info::has_block): Likewise.
2768 (struct source_info): Add new member index.
2769 (source_info::get_functions_at_location): New function.
2770 (function_info::group_line_p): New function.
2771 (output_intermediate_line): New function.
2772 (output_intermediate_file): Use the mentioned function.
2773 (struct function_start): New.
2774 (struct function_start_pair_hash): Likewise.
2775 (process_file): Add code that identifies group functions.
2776 Assign lines either to global or function scope.
2777 (generate_results): Skip artificial functions.
2778 (find_source): Assign index for each source file.
2779 (read_graph_file): Read new flag artificial and end_line.
2780 (add_line_counts): Assign it either to global of function scope.
2781 (accumulate_line_counts): Isolate core of the function to
2782 accumulate_line_info and call it for both function and global
2784 (accumulate_line_info): New function.
2785 (output_line_beginning): Fix GNU coding style.
2786 (print_source_line): New function.
2787 (output_line_details): Likewise.
2788 (output_function_details): Likewise.
2789 (output_lines): Iterate both source (global) scope and function
2791 (struct function_line_start_cmp): New class.
2792 * doc/gcov.texi: Reflect changes in documentation.
2794 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2797 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
2798 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
2799 and similarly for not instead of neg.
2801 2017-11-08 Andi Kleen <ak@linux.intel.com>
2803 * config/i386/i386.opt: Add -mforce-indirect-call.
2804 * config/i386/predicates.md: Check for flag_force_indirect_call.
2805 * doc/invoke.texi: Document -mforce-indirect-call
2807 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
2809 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
2811 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
2812 config/riscv/riscv.c (predict.h): New include.
2813 (riscv_slow_unaligned_access_p): No longer static.
2814 (riscv_block_move_straight): Add require.
2815 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2817 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2820 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
2821 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
2822 nonimmediate_operand predicate for operand 1 instead of
2825 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2827 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
2829 * config/aarch64/constraints.md (Uml): New constraint.
2830 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2833 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2835 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2836 of two vec_duplicates into a vec_concat.
2838 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2840 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2841 Simplify vec_merge of vec_duplicate and vec_concat.
2842 * config/aarch64/constraints.md (Utq): New constraint.
2843 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2846 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2848 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2849 Simplify vec_merge of vec_duplicate and const_vector.
2850 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2852 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2853 mode iterator. Update predicate on operand 1 to
2854 handle non-const_vec constants. Delete constraints.
2855 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2857 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2859 PR tree-optimization/78821
2860 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2862 (store_operand_info::store_operand_info): Initialize it to false.
2863 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2864 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2865 store in the group, and if chain_info is non-NULL, to ignore altogether
2867 (compatible_load_p): Fail if bit_not_p does not match.
2868 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2869 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2870 (pass_store_merging::process_store): Adjust
2871 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2872 call terminate_all_aliasing_chains newly when adding a store into
2873 a chain with non-NULL chain_info.
2875 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2877 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2879 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2881 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2883 (aarch64_layout_frame): Initialise emit_frame_chain.
2884 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2885 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2887 2017-11-08 Martin Liska <mliska@suse.cz>
2889 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2890 of gimple_call_internal_p.
2892 2017-11-07 Tom de Vries <tom@codesourcery.com>
2894 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2896 2017-11-07 Tom de Vries <tom@codesourcery.com>
2898 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2901 2017-11-08 Martin Liska <mliska@suse.cz>
2904 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2906 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2908 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2910 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2913 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2914 (*cmp<mode>_ccz_1): New insn with $k alternative.
2917 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2918 RTX_COMM_COMPARE as commutative as well.
2919 (ix86_binary_operator_ok): Formatting fix.
2920 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2921 *<code><mode>3<mask_name><round_saeonly_name>,
2922 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2923 *<s>mul<mode>3_highpart<mask_name>,
2924 *vec_widen_umult_even_v16si<mask_name>,
2925 *vec_widen_umult_even_v8si<mask_name>,
2926 *vec_widen_umult_even_v4si<mask_name>,
2927 *vec_widen_smult_even_v16si<mask_name>,
2928 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2929 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2930 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2931 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2932 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2933 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2934 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2935 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2936 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2937 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2938 ix86_binary_operator_ok. Formatting fixes.
2939 (*<plusminus_insn><mode>3<mask_name><round_name>,
2940 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2943 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2945 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2946 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2947 eq and ne if TARGET_ISEL.
2948 (cmp): New code_iterator.
2949 (UNS, UNSU_, UNSIK): New code_attrs.
2950 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2951 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2953 ("eq<mode>3"): ... this.
2954 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2956 ("ne<mode>3"): ... this.
2958 2017-11-07 Julia Koval <julia.koval@intel.com>
2961 * common/config/i386/i386-common.c
2962 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2963 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2964 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2965 * config/i386/i386.opt: Ditto.
2966 * config/i386/i386.c (ix86_target_string): Ditto.
2967 (ix86_option_override_internal): Ditto.
2968 (ix86_init_mpx_builtins): Move MPX to args2.
2969 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2970 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2971 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2972 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2973 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2974 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2975 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2977 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2980 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2981 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2982 (zero-extendsidi peephole2): Remove peephole.
2984 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2987 * stor-layout.c: Include attribs.h.
2988 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2989 explicit lookup of "aligned" attribute.
2991 2017-11-07 Andrew Waterman <andrew@sifive.com>
2993 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2994 (riscv_expand_block_move): Likewise.
2995 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2997 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2998 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2999 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3000 (riscv_adjust_block_mem): Likewise.
3001 (riscv_block_move_loop): Likewise.
3002 (riscv_expand_block_move): Likewise.
3003 gcc/config/riscv/riscv.md (movmemsi): New pattern.
3005 2017-11-07 Michael Clark <michaeljclark@mac.com>
3007 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3008 (MUSL_DYNAMIC_LINKER): Likewise.
3010 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3012 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3013 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3016 2017-11-07 Richard Biener <rguenther@suse.de>
3018 * match.pd: Fix build.
3020 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3021 Jackson Woodruff <jackson.woodruff@arm.com>
3023 PR tree-optimization/71026
3024 * match.pd: Canonicalize negate in division.
3026 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3029 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3031 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3033 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3034 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3036 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3038 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3039 non-scalar integral types.
3040 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3041 (-(A-B), -(~A)): New transformations.
3043 2017-11-07 Tom de Vries <tom@codesourcery.com>
3045 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3046 semicolon after "do {} while (0)".
3047 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3048 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3049 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3050 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3051 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3053 2017-11-07 Tom de Vries <tom@codesourcery.com>
3055 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3056 after "do {} while (0)".
3057 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3058 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3059 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3060 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3061 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3063 2017-11-07 Tom de Vries <tom@codesourcery.com>
3066 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3068 (arm_rtx_costs_internal): Add missing semicolon after
3069 HANDLE_NARROW_SHIFT_ARITH call.
3071 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3073 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3074 disable isel if it was not set explicitly.
3076 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3078 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3079 (add_type_attribute) likewise.
3081 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3083 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3084 of struct ix86_frame.
3085 (ix86_initial_elimination_offset): Likewise.
3086 (ix86_expand_split_stack_prologue): Likewise.
3088 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3090 * tree-vrp.h (enum value_range_type): Update stale comment.
3092 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3093 Alan Hayward <alan.hayward@arm.com>
3094 David Sherwood <david.sherwood@arm.com>
3096 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3097 (aarch64_expand_vec_perm_const): Take the number of units too.
3098 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3099 (aarch64_expand_vec_perm_const): Likewise.
3100 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3101 (vec_perm<mode>): Update accordingly.
3103 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3104 Alan Hayward <alan.hayward@arm.com>
3105 David Sherwood <david.sherwood@arm.com>
3107 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3108 Take the number of units too.
3109 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3110 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3111 but check for a vector mode before rather than after the call.
3112 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3113 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3114 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3115 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3116 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3117 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3118 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3119 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3120 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3121 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3122 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3123 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3124 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3125 (aarch64_sqdmull2_n<mode>): Update accordingly.
3127 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3132 the number of units too.
3133 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3134 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3135 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3136 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3137 (vec_store_lanesxi<mode>): Update accordingly.
3139 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3140 Alan Hayward <alan.hayward@arm.com>
3141 David Sherwood <david.sherwood@arm.com>
3143 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3144 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3145 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3146 of units rather than the mode.
3147 * config/aarch64/iterators.md (nunits): New mode attribute.
3148 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3149 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3150 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3151 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3152 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3153 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3154 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3155 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3156 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3157 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3158 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3159 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3160 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3161 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3162 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3163 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3164 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3165 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3166 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3167 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3168 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3169 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3170 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3171 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3172 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3173 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3174 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3175 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3176 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3177 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3178 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3179 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3180 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3181 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3182 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3183 (aarch64_simd_vec_setv2di): Likewise.
3185 2017-11-06 Carl Love <cel@us.ibm.com>
3187 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3189 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3190 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3192 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3194 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3195 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3196 Add power 8 macro expansions.
3197 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3198 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3199 power 8 instructions. (VSX_XXBR): Add iterator.
3201 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3203 * config/arm/arm.md (predicable_short_it): Change default to "no",
3204 improve documentation, remove uses that are identical to the default.
3205 (enabled_for_depr_it): Rename to enabled_for_short_it.
3206 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3207 Remove default uses.
3208 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3209 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3210 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3211 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3213 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3216 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3217 float128 helper macros, which are no longer used after deleting
3218 the old 'q' built-in functions, and moving the round to odd
3219 built-in functions to being special built-in functions.
3220 (BU_FLOAT128_2): Likewise.
3221 (BU_FLOAT128_1_HW): Likewise.
3222 (BU_FLOAT128_2_HW): Likewise.
3223 (BU_FLOAT128_3_HW): Likewise.
3224 (FABSQ): Delete old 'q' built-in functions.
3225 (COPYSIGNQ): Likewise.
3226 (SQRTF128_ODD): Move round to odd built-in functions to be
3227 special built-in functions, so that we can handle
3228 -mabi=ieeelongdouble.
3229 (TRUNCF128_ODD): Likewise.
3230 (ADDF128_ODD): Likewise.
3231 (SUBF128_ODD): Likewise.
3232 (MULF128_ODD): Likewise.
3233 (DIVF128_ODD): Likewise.
3234 (FMAF128_ODD): Likewise.
3235 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3236 built-in names to 'f128'.
3237 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3238 old 'q' built-in functions, as the machine independent code for
3239 'f128' built-in functions handles this.
3240 (rs6000_expand_builtin): Add expansion for float128 round to odd
3241 functions, keying off on -mabi=ieeelongdouble of whether to use
3242 the KFmode or TFmode variant.
3243 (rs6000_init_builtins): Initialize the _Float128 round to odd
3245 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3246 _Float128 'q' built-in functions are now mapped into the new
3247 'f128' built-in functions.
3249 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3251 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3252 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3254 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3257 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3258 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3259 * toplev.c: Include "ipa-fnsummary.h".
3260 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3262 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3264 PR tree-optimization/82838
3265 * gimple-ssa-store-merging.c
3266 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3267 on a separate gimple_seq which is then appended to seq.
3269 2017-11-06 Jeff Law <law@redhat.com>
3272 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3273 (get_probe_interval): New functions.
3274 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3275 (ix86_adjust_stack_and_probe): Likewise.
3276 (output_adjust_stack_and_probe): Likewise.
3277 (ix86_emit_probe_stack_range): Likewise.
3278 (ix86_expand_prologue): Likewise.
3280 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3282 PR tree-optimization/82816
3283 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3284 if the modes of the two types are the same.
3285 (convert_plusminus_to_widen): Likewise.
3287 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3289 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3291 (usadv16qi): New define_expand.
3292 (usadv8hi): New define_expand.
3294 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3297 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3299 (inline_transform): Likewise.
3300 * predict.c (counts_to_freqs): Remove useless conditional.
3301 * profile-count.h (profile_count::apply_scale): Move sanity check.
3302 * tree-inline.c (copy_bb): Always scale.
3303 (copy_cfg_body): Likewise.
3305 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3308 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3310 (*cmp_ite0): Add enabled_for_depr_it attribute.
3311 (*cmp_ite1): Likewise.
3313 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3315 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3318 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3320 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3321 return the value by pointer.
3322 (extract_range_from_multiplicative_op_1): Update accordingly.
3323 Return as soon as an operation fails.
3325 2017-11-05 Tom de Vries <tom@codesourcery.com>
3328 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3329 (DEF_SANITIZER_BUILTIN): ... here.
3330 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3331 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3333 2017-11-05 Tom de Vries <tom@codesourcery.com>
3336 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3338 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3339 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3340 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3342 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3343 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3344 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3346 2017-11-05 Tom de Vries <tom@codesourcery.com>
3349 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3352 2017-11-04 Michael Clark <michaeljclark@mac.com>
3354 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3355 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3357 (*addsi3_extended): Likewise.
3358 (*addsi3_extended2): Likewise.
3359 (<optab>si3): Likewise.
3360 (<optab>di3): Likewise.
3361 (<optab><mode>3): Likewise.
3362 (<*optabe>si3_internal): Likewise.
3363 (zero_extendqi<SUPERQI:mode>2): Likewise.
3364 (*add<mode>hi3): Likewise.
3365 (*xor<mode>hi3): Likewise.
3366 (<optab>di3): Likewise.
3367 (*<optab>si3_extend): Likewise.
3368 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3369 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3370 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3372 2017-11-04 Andrew Waterman <andrew@sifive.com>
3374 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3375 TARGET_STRICT_ALIGN based upon -mtune argument.
3377 2017-11-04 Andrew Waterman <andrew@sifive.com>
3379 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3381 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3383 config/i386/i386.c (choose_basereg): Use optional scratch
3384 register and add assertion.
3385 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3386 needed, and don't allocate stack.
3387 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3388 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3389 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3391 2017-11-03 Jeff Law <law@redhat.com>
3393 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3394 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3395 to probe at the start of a noreturn function.
3397 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3399 PR tree-optimization/78821
3400 * gimple-ssa-store-merging.c: Update the file comment.
3401 (MAX_STORE_ALIAS_CHECKS): Define.
3402 (struct store_operand_info): New type.
3403 (store_operand_info::store_operand_info): New constructor.
3404 (struct store_immediate_info): Add rhs_code and ops data members.
3405 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3406 and op1r arguments to the ctor, initialize corresponding data members.
3407 (struct merged_store_group): Add load_align_base and load_align
3409 (merged_store_group::merged_store_group): Initialize them.
3410 (merged_store_group::do_merge): Update them.
3411 (merged_store_group::apply_stores): Pick the constant for
3412 encode_tree_to_bitpos from one of the two operands, or skip
3413 encode_tree_to_bitpos if neither operand is a constant.
3414 (class pass_store_merging): Add process_store method decl. Remove
3415 bool argument from terminate_all_aliasing_chains method decl.
3416 (pass_store_merging::terminate_all_aliasing_chains): Remove
3417 var_offset_p argument and corresponding handling.
3418 (stmts_may_clobber_ref_p): New function.
3419 (compatible_load_p): New function.
3420 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3421 if there is overlap and rhs_code is not INTEGER_CST. For
3422 non-overlapping stores terminate group if rhs is not mergeable.
3423 (get_alias_type_for_stmts): Change first argument from
3424 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3425 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3426 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3428 (get_location_for_stmts): Change first argument from
3429 auto_vec<gimple *> & to vec<gimple *> &.
3430 (struct split_store): Remove orig_stmts data member, add orig_stores.
3431 (split_store::split_store): Create orig_stores rather than orig_stmts.
3432 (find_constituent_stmts): Renamed to ...
3433 (find_constituent_stores): ... this. Change second argument from
3434 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3435 to info structures rather than the statements.
3436 (split_group): Rename ALLOW_UNALIGNED argument to
3437 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3438 it. Adjust find_constituent_stores caller.
3439 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3440 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3441 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3442 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3443 (mem_valid_for_store_merging): New function.
3444 (handled_load): New function.
3445 (pass_store_merging::process_store): New method.
3446 (pass_store_merging::execute): Use process_store method. Adjust
3447 terminate_all_aliasing_chains caller.
3449 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3451 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3452 Return true for more constants, symbols and label references.
3453 (aarch64_valid_floating_const): Remove unused function.
3455 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3458 * config/aarch64/aarch64.c (aarch64_layout_frame):
3459 Undo forcing of LR at bottom of frame.
3461 2017-11-03 Jeff Law <law@redhat.com>
3464 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3465 for int_registers_saved.
3467 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3468 extracted from tree-ssa-dom.c.
3469 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3470 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3471 (record_equivalences_from_incoming_edge): Add additional argument
3472 to single_pred_edge_ignoring_loop_edges call.
3473 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3474 (uncprop_dom_walker::before_dom_children): Add additional argument
3475 to single_pred_edge_ignoring_loop_edges call.
3476 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3477 single_pred_edge_ignoring_loop_edges rather than open coding.
3478 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3480 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3482 * match.pd (-(-A)): Rewrite.
3484 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3486 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3487 (rs6000_emit_int_cmove): New declaration.
3488 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3489 (rs6000_emit_sISEL): Delete.
3490 (rs6000_emit_int_cmove): Make non-static.
3491 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3492 instead of rs6000_emit_sISEL.
3494 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3496 * asan.c (create_cond_insert_point): Maintain profile.
3497 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3499 * basic-block.h (struct basic_block_def): Remove frequency.
3500 (EDGE_FREQUENCY): Use to_frequency
3501 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3503 (find_traces): Update to use to_frequency.
3504 (find_traces_1_round): Likewise; use only IPA counts.
3505 (bb_to_key): Likewise.
3506 (connect_traces): Use IPA counts only.
3507 (copy_bb_p): Update to use to_frequency.
3508 (fix_up_crossing_landing_pad): Likewise.
3509 (sanitize_hot_paths): Likewise.
3510 * bt-load.c (basic_block_freq): Likewise.
3511 * cfg.c (init_flow): Set count_max to uninitialized.
3512 (check_bb_profile): Remove frequencies; check counts.
3513 (dump_bb_info): Do not dump frequencies.
3514 (update_bb_profile_for_threading): Update counts only.
3515 (scale_bbs_frequencies_int): Likewise.
3516 (MAX_SAFE_MULTIPLIER): Remove.
3517 (scale_bbs_frequencies_gcov_type): Update counts only.
3518 (scale_bbs_frequencies_profile_count): Update counts only.
3519 (scale_bbs_frequencies): Update counts only.
3520 * cfg.h (struct control_flow_graph): Add count-max.
3521 (update_bb_profile_for_threading): Update prototype.
3522 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3523 (find_many_sub_basic_blocks): Likewise.
3524 * cfgcleanup.c (try_forward_edges): Likewise.
3525 (try_crossjump_to_edge): Likewise.
3526 * cfgexpand.c (expand_gimple_cond): Likewise.
3527 (expand_gimple_tailcall): Likewise.
3528 (construct_init_block): Likewise.
3529 (construct_exit_block): Likewise.
3530 * cfghooks.c (verify_flow_info): Check consistency of counts.
3531 (dump_bb_for_graph): Do not dump frequencies.
3532 (split_block_1): Do not update frequencies.
3533 (split_edge): Do not update frequencies.
3534 (make_forwarder_block): Do not update frequencies.
3535 (duplicate_block): Do not update frequencies.
3536 (account_profile_record): Do not update frequencies.
3537 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3538 for global heuristics.
3539 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3540 (expected_loop_iterations_unbounded): Use counts only.
3541 * cfgloopmanip.c (scale_loop_profile): Simplify.
3542 (create_empty_loop_on_edge): Simplify
3544 (duplicate_loop_to_header_edge): Simplify
3545 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3546 (update_br_prob_note): Take care of removing note when profile
3548 (relink_block_chain): Do not dump frequency.
3549 (rtl_account_profile_record): Use to_frequency.
3550 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3551 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3552 (cgraph_update_edges_for_call_stmt_node): Likewise.
3553 (cgraph_edge::verify_count_and_frequency): Update.
3554 (cgraph_node::verify_node): Temporarily disable frequency verification.
3555 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3556 to_cgraph_frequency.
3557 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3558 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3560 (cgraph_node::expand_thunk): Update profile.
3561 * except.c (dw2_build_landing_pads): Do not update frequency.
3562 * final.c (compute_alignments): Use to_frequency.
3563 (dump_basic_block_info): Do not dump frequency.
3564 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3565 (dump_gimple_bb_header): Do not dump frequency.
3566 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3568 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3569 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3570 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3571 (init_before_recovery): Do not update frequency.
3572 (sched_create_recovery_edges): Do not update frequency.
3573 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3574 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3575 (ipa_cp_c_finalize): Set max_count to uninitialized.
3576 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3577 (param_change_prob): Use counts.
3578 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3580 * ipa-split.c (consider_split): Use to_frequency.
3581 (split_function): Use to_frequency.
3582 * ira-build.c (loop_compare_func): Likewise.
3583 (mark_loops_for_removal): Likewise.
3584 (mark_all_loops_for_removal): Likewise.
3585 * loop-doloop.c (doloop_modify): Do not update frequency.
3586 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3588 * lto-streamer-in.c (input_function): Update count_max.
3589 * omp-expand.c (expand_omp_taskreg): Update count_max.
3590 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3591 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3592 (maybe_hot_count_p): Use ipa counts only.
3593 (maybe_hot_bb_p): Simplify.
3594 (maybe_hot_edge_p): Simplify.
3595 (probably_never_executed): Do not take frequency argument.
3596 (probably_never_executed_bb_p): Do not pass frequency.
3597 (probably_never_executed_edge_p): Likewise.
3598 (combine_predictions_for_bb): Check that profile is nonzero.
3599 (propagate_freq): Do not set frequency.
3600 (drop_profile): Simplify.
3601 (counts_to_freqs): Simplify.
3602 (expensive_function_p): Use to_frequency.
3603 (propagate_unlikely_bbs_forward): Simplify.
3604 (determine_unlikely_bbs): Simplify.
3605 (estimate_bb_frequencies): Add hack to silence graphite issues.
3606 (compute_function_frequency): Use ipa counts.
3607 (pass_profile::execute): Update.
3608 (rebuild_frequencies): Use counts only.
3609 (force_edge_cold): Use counts only.
3610 * profile-count.c (profile_count::dump): Dump new count types.
3611 (profile_count::differs_from_p): Check compatiblity.
3612 (profile_count::to_frequency): New function.
3613 (profile_count::to_cgraph_frequency): New function.
3614 * profile-count.h (struct function): Declare.
3615 (enum profile_quality): Add profile_guessed_local and
3616 profile_guessed_global0.
3617 (class profile_proability): Decrease number of bits to 29;
3618 update from_reg_br_prob_note and to_reg_br_prob_note.
3619 (class profile_count: Update comment; decrease number of bits
3620 to 61. Check compatibility.
3621 (profile_count::compatible_p): New private member function.
3622 (profile_count::ipa_p): New member function.
3623 (profile_count::operator<): Handle global zero correctly.
3624 (profile_count::operator>): Handle global zero correctly.
3625 (profile_count::operator<=): Handle global zero correctly.
3626 (profile_count::operator>=): Handle global zero correctly.
3627 (profile_count::nonzero_p): New member function.
3628 (profile_count::force_nonzero): New member function.
3629 (profile_count::max): New member function.
3630 (profile_count::apply_scale): Handle IPA scalling.
3631 (profile_count::guessed_local): New member function.
3632 (profile_count::global0): New member function.
3633 (profile_count::ipa): New member function.
3634 (profile_count::to_frequency): Declare.
3635 (profile_count::to_cgraph_frequency): Declare.
3636 * profile.c (OVERLAP_BASE): Delete.
3637 (compute_frequency_overlap): Delete.
3638 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3639 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3640 * sched-ebb.c (rank): Use counts only.
3641 * shrink-wrap.c (handle_simple_exit): Use counts only.
3642 (try_shrink_wrapping): Use counts only.
3643 (place_prologue_for_one_component): Use counts only.
3644 * tracer.c (find_best_predecessor): Use to_frequency.
3645 (find_trace): Use to_frequency.
3646 (tail_duplicate): Use to_frequency.
3647 * trans-mem.c (expand_transaction): Do not update frequency.
3648 * tree-call-cdce.c: Do not update frequency.
3649 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3650 (gimple_merge_blocks): Likewise.
3651 (gimple_split_edge): Likewise.
3652 (gimple_duplicate_sese_region): Likewise.
3653 (gimple_duplicate_sese_tail): Likewise.
3654 (move_sese_region_to_fn): Likewise.
3655 (gimple_account_profile_record): Likewise.
3656 (insert_cond_bb): Likewise.
3657 * tree-complex.c (expand_complex_div_wide): Likewise.
3658 * tree-eh.c (lower_resx): Update profile.
3659 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3661 (initialize_cfun): Do not initialize frequencies
3662 (freqs_to_counts): Delete.
3663 (copy_cfg_body): Ignore count parameter.
3664 (copy_body): Update.
3665 (expand_call_inline): Update count_max.
3666 (optimize_inline_calls): Update count_max.
3667 (tree_function_versioning): Update count_max.
3668 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3669 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3671 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3672 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3673 (try_peel_loop): Likewise.
3674 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3676 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3677 (tree_transform_and_unroll_loop): Do not use frequencies
3678 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3679 Use reliable prediction only.
3680 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3681 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3682 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3683 probability scaling.
3684 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3685 not update frequency
3686 (any_remaining_duplicated_blocks): Likewise.
3687 (update_profile): Likewise.
3688 (estimated_freqs_path): Delete.
3689 (freqs_to_counts_path): Delete.
3690 (clear_counts_path): Delete.
3691 (ssa_fix_duplicate_block_edges): Likewise.
3692 (duplicate_thread_path): Likewise.
3693 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3694 * tree-tailcall.c (decrease_profile): Do not update frequency.
3695 (eliminate_tail_call): Likewise.
3696 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3697 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3698 (optimize_mask_stores): Likewise.
3699 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3700 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3701 (ubsan_expand_ptr_ifn): Update profile.
3702 * value-prof.c (gimple_ic): Simplify.
3703 * value-prof.h (gimple_ic): Update prototype.
3704 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3705 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3707 (want_inline_self_recursive_call_p): Likewise.
3708 (resolve_noninline_speculation): Only cummulate defined counts.
3709 (inline_small_functions): Use nonzero_p.
3710 (ipa_inline): Do not access freed node.
3712 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3714 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3715 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3717 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3719 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3720 non-legitimate address.
3722 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3724 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3725 (*lt0_<mode>di, *lt0_<mode>si): New.
3727 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3729 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3730 TARGET_PAIRED_FLOAT.
3732 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3733 Jim Wilson <jim.wilson@linaro.org>
3735 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3736 * config/aarch64/aarch64-tune.md: Regenerated.
3737 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3738 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3740 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3742 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3743 (arc_expand_prologue): Restore blink for millicode.
3744 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3746 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3749 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3750 gen_vec_duplicate after forcing the scalar into a register.
3752 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3754 * combine (try_combine): Print the insns input to try_combine to the
3757 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3760 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3761 Remove second argument from aarch64_process_target_attr call.
3762 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3764 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3766 (aarch64_handle_attr_arch): Remove second argument.
3767 (aarch64_handle_attr_cpu): Ditto.
3768 (aarch64_handle_attr_tune): Ditto.
3769 (aarch64_handle_attr_isa_flags): Ditto.
3770 (aarch64_process_one_target_attr): Ditto.
3771 (aarch64_process_target_attr): Ditto.
3772 (aarch64_option_valid_attribute_p): Remove second argument.
3773 on aarch64_process_target_attr call.
3775 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3777 * diagnostic.c: Include "selftest-diagnostic.h".
3778 (selftest::assert_location_text): New function.
3779 (selftest::test_diagnostic_get_location_text): New function.
3780 (selftest::diagnostic_c_tests): Call it.
3782 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3784 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3785 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3786 (class selftest::test_diagnostic_context): Move to...
3787 * selftest-diagnostic.c: New file.
3788 * selftest-diagnostic.h: New file.
3790 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3792 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3793 offset range for FT32B.
3794 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3795 * config/ft32/ft32.md: Add TARGET_NOPM.
3796 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3797 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
3799 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
3801 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
3803 2017-11-02 Jeff Law <law@redhat.com>
3805 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
3806 virtual keyword on FINAL OVERRIDE members.
3808 * tree-ssa-propagate.h (ssa_propagation_engine): Group
3809 virtuals together. Add virtual destructor.
3810 (substitute_and_fold_engine): Similarly.
3812 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3814 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
3816 2017-11-02 Richard Biener <rguenther@suse.de>
3818 PR tree-optimization/82795
3819 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
3821 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3823 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
3825 (gcc_GAS_CHECK_FEATURE): Remove.
3826 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
3827 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
3828 * configure: Regenerate.
3830 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3832 * config/arc/arc.c (hwloop_optimize): Account for empty
3835 2017-11-02 Richard Biener <rguenther@suse.de>
3838 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3839 Truncate ARRAY_REF index and element size.
3841 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3843 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3845 2017-11-01 Jeff Law <law@redhat.com>
3847 * tree-ssa-ccp.c (ccp_folder): New class derived from
3848 substitute_and_fold_engine.
3849 (ccp_folder::get_value): New member function.
3850 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3851 (ccp_fold_stmt): Remove prototype.
3852 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3853 * tree-ssa-copy.c (copy_folder): New class derived from
3854 substitute_and_fold_engine.
3855 (copy_folder::get_value): Renamed from get_value.
3856 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3857 * tree-vrp.c (vrp_folder): New class derived from
3858 substitute_and_fold_engine.
3859 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3860 (vrp_folder::get_value): New member function.
3861 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3862 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3863 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3864 provide a class interface to folder/substitute routines.
3865 (ssa_prop_fold_stmt_fn): Remove typedef.
3866 (ssa_prop_get_value_fn): Likewise.
3867 (subsitute_and_fold): Remove prototype.
3868 (replace_uses_in): Likewise.
3869 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3870 Renamed from replace_uses_in. Call the virtual member function
3871 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3872 (substitute_and_fold_dom_walker): Remove initialization of
3873 data member entries for calbacks. Add substitute_and_fold_engine
3874 member and initialize it.
3875 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3876 member functions for get_value, replace_phi_args_in c
3877 replace_uses_in, and fold_stmt calls.
3878 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3879 substitute_and_fold. Remove assert. Update ctor call.
3881 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3882 (ssa_prop_visit_phi_fn): Likewise.
3883 (class ssa_propagation_engine): New class to provide an interface
3885 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3887 (ssa_prop_visit_phi): Likewise.
3888 (ssa_propagation_engine::simulate_stmt): Moved into class.
3889 Call visit_phi/visit_stmt from the class rather than via
3890 file scoped static variables.
3891 (ssa_propagation_engine::simulate_block): Moved into class.
3892 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3893 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3894 set file scoped statics for the visit_stmt/visit_phi callbacks.
3895 * tree-complex.c (complex_propagate): New class derived from
3896 ssa_propagation_engine.
3897 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3898 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3899 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3901 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3902 ssa_propagation_engine.
3903 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3904 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3905 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3906 * tree-ssa-copy.c (copy_prop): New class derived from
3907 ssa_propagation_engine.
3908 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3909 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3910 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3911 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3912 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3913 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3914 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3916 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3918 PR rtl-optimization/82778
3919 PR rtl-optimization/82597
3920 * compare-elim.c (struct comparison): Add in_a_setter field.
3921 (find_comparison_dom_walker::before_dom_children): Remove killed
3922 bitmap and df_simulate_find_defs call, instead walk the defs.
3923 Compute last_setter and initialize in_a_setter. Merge definitions
3924 with first initialization for a few variables.
3925 (try_validate_parallel): Use insn_invalid_p instead of
3926 recog_memoized. Return insn rather than just the pattern.
3927 (try_merge_compare): Fix up comment. Don't uselessly test if
3928 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3930 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3933 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3934 Alan Hayward <alan.hayward@arm.com>
3935 David Sherwood <david.sherwood@arm.com>
3937 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3938 aarch64_hard_regno_nregs to get the number of registers
3941 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3942 Alan Hayward <alan.hayward@arm.com>
3943 David Sherwood <david.sherwood@arm.com>
3945 * config/aarch64/constraints.md (Upl): Rename to...
3947 * config/aarch64/aarch64.md
3948 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3951 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3952 Alan Hayward <alan.hayward@arm.com>
3953 David Sherwood <david.sherwood@arm.com>
3955 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3956 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3959 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3960 Alan Hayward <alan.hayward@arm.com>
3961 David Sherwood <david.sherwood@arm.com>
3963 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3964 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3965 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3967 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3968 of a single element.
3969 * config/aarch64/iterators.md: Add a comment above the permute
3970 unspecs to say that they are generated directly by
3971 aarch64_expand_vec_perm_const.
3972 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3974 2017-11-01 Nathan Sidwell <nathan@acm.org>
3976 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3978 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3980 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3981 and medany code models, and describe what they do.
3983 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3985 Revert accidental duplicate:
3987 * combine.c (can_change_dest_mode): Reject changes in
3988 REGMODE_NATURAL_SIZE.
3990 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3992 PR rtl-optimization/64682
3993 PR rtl-optimization/69567
3994 PR rtl-optimization/69737
3995 PR rtl-optimization/82683
3996 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3997 register mentioned in the note, drop the note, unless it came from I3,
3998 in which case it should go to I3 again.
4000 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4002 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4003 and return false if not.
4004 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4006 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4008 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4009 range is known to be empty.
4011 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4012 Alan Hayward <alan.hayward@arm.com>
4013 David Sherwood <david.sherwood@arm.com>
4015 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4016 and CONST_VECTOR_NUNITS instead of computing the number of units from
4017 the byte sizes of the vector and element.
4018 (simplify_binary_operation_1): Likewise.
4019 (simplify_const_binary_operation): Likewise.
4020 (simplify_ternary_operation): Likewise.
4022 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4023 Alan Hayward <alan.hayward@arm.com>
4024 David Sherwood <david.sherwood@arm.com>
4026 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4027 (int_mem_offset): ...this new function.
4028 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4029 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4030 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4034 Alan Hayward <alan.hayward@arm.com>
4035 David Sherwood <david.sherwood@arm.com>
4037 * lower-subreg.c (interesting_mode_p): New function.
4038 (compute_costs, find_decomposable_subregs, decompose_register)
4039 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4040 (resolve_clobber, dump_choices): Use it.
4042 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4043 Alan Hayward <alan.hayward@arm.com>
4044 David Sherwood <david.sherwood@arm.com>
4046 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4048 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4049 Alan Hayward <alan.hayward@arm.com>
4050 David Sherwood <david.sherwood@arm.com>
4052 * alias.c (find_base_value, find_base_term): Only process integer
4053 truncations. Check the precision rather than the size.
4055 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4056 Alan Hayward <alan.hayward@arm.com>
4057 David Sherwood <david.sherwood@arm.com>
4059 * machmode.h (is_narrower_int_mode): New function
4060 * optabs.c (expand_float, expand_fix): Use it.
4061 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4063 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4064 Alan Hayward <alan.hayward@arm.com>
4065 David Sherwood <david.sherwood@arm.com>
4067 * rtl.h (narrower_subreg_mode): New function.
4068 * ira-color.c (update_costs_from_allocno): Use it.
4070 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4071 Alan Hayward <alan.hayward@arm.com>
4072 David Sherwood <david.sherwood@arm.com>
4074 * optabs-query.h (convert_optab_p): New function, split out from...
4075 (convert_optab_handler): ...here.
4076 (widening_optab_handler): Delete.
4077 (find_widening_optab_handler): Remove permit_non_widening parameter.
4078 (find_widening_optab_handler_and_mode): Likewise. Provide an
4079 override that operates on mode class wrappers.
4080 * optabs-query.c (widening_optab_handler): Delete.
4081 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4082 parameter. Assert that the two modes are the same class and that
4083 the "from" mode is narrower than the "to" mode. Use
4084 convert_optab_handler instead of widening_optab_handler.
4085 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4086 instead of widening_optab_handler.
4087 * expr.c (expand_expr_real_2): Update calls to
4088 find_widening_optab_handler.
4089 * optabs.c (expand_widen_pattern_expr): Likewise.
4090 (expand_binop_directly): Take the insn_code as a parameter.
4091 (expand_binop): Only call find_widening_optab_handler for
4092 conversion optabs; use optab_handler otherwise. Update calls
4093 to find_widening_optab_handler and expand_binop_directly.
4094 Use convert_optab_handler instead of widening_optab_handler.
4095 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4096 find_widening_optab_handler and use scalar_mode rather than
4098 (convert_plusminus_to_widen): Likewise.
4100 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4101 Alan Hayward <alan.hayward@arm.com>
4102 David Sherwood <david.sherwood@arm.com>
4104 * machmode.h (fixed_size_mode): New class.
4105 * rtl.h (get_pool_mode): Return fixed_size_mode.
4106 * gengtype.c (main): Add fixed_size_mode.
4107 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4108 (get_raw_arg_mode): Likewise.
4109 * doc/tm.texi: Regenerate.
4110 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4111 * targhooks.c (default_get_reg_raw_mode): Likewise.
4112 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4113 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4114 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4115 (msp430_get_raw_result_mode): Likewise.
4116 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4117 * dbxout.c (dbxout_parms): Require fixed-size modes.
4118 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4119 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4120 * omp-low.c (lower_oacc_reductions): Likewise.
4121 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4122 (simplify_subreg): Update accordingly.
4123 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4124 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4125 that aren't fixed-size.
4126 (get_pool_mode): Return a fixed_size_mode.
4127 (output_constant_pool_2): Take a fixed_size_mode.
4129 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4130 Alan Hayward <alan.hayward@arm.com>
4131 David Sherwood <david.sherwood@arm.com>
4133 * doc/rtl.texi (vec_series): Document.
4134 (const): Say that the operand can be a vec_series.
4135 * rtl.def (VEC_SERIES): New rtx code.
4136 * rtl.h (const_vec_series_p_1): Declare.
4137 (const_vec_series_p): New function.
4138 * emit-rtl.h (gen_const_vec_series): Declare.
4139 (gen_vec_series): Likewise.
4140 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4141 (gen_vec_series): Likewise.
4142 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4143 * simplify-rtx.c (simplify_unary_operation): Handle negations
4145 (simplify_binary_operation_series): New function.
4146 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4147 (test_vector_ops_series): New function.
4148 (test_vector_ops): Call it.
4149 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4150 gen_const_vec_series.
4151 (altivec_lvsr): Likewise.
4152 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4154 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4155 Alan Hayward <alan.hayward@arm.com>
4156 David Sherwood <david.sherwood@arm.com>
4158 * doc/rtl.texi (const): Update description of address constants.
4159 Say that vector constants are allowed too.
4160 * common.md (E, F): Use CONSTANT_P instead of checking for
4162 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4163 checking for CONST_VECTOR.
4164 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4166 * expr.c (expand_expr_real_2): Check for vector modes instead
4167 of checking for CONST_VECTOR.
4168 * rtl.h (const_vec_p): New function.
4169 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4170 (unwrap_const_vec_duplicate): Handle them here too.
4172 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4173 David Malcolm <dmalcolm@redhat.com>
4174 Alan Hayward <alan.hayward@arm.com>
4175 David Sherwood <david.sherwood@arm.com>
4177 * rtl.h (vec_duplicate_p): New function.
4178 * selftest-rtl.c (assert_rtx_eq_at): New function.
4179 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4180 (assert_rtx_eq_at): Declare.
4181 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4182 * selftest-run-tests.c (selftest::run_tests): Call it.
4183 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4184 (simplify_unary_operation_1): Recursively handle vector duplicates.
4185 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4187 (simplify_subreg): Handle subregs of vector duplicates.
4188 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4189 (selftest::simplify_rtx_c_tests): New functions.
4191 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4192 Alan Hayward <alan.hayward@arm.com>
4193 David Sherwood <david.sherwood@arm.com>
4195 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4196 (gen_vec_duplicate): Likewise.
4197 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4199 (gen_const_vector): ...here.
4200 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4201 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4202 whose elements are all equal.
4203 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4204 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4205 (simplify_relational_operation): Likewise.
4206 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4208 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4209 (aarch64_expand_vector_init): Likewise.
4210 * config/arm/arm.c (neon_vdup_constant): Likewise.
4211 (neon_expand_vector_init): Likewise.
4212 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4213 (arm_block_set_unaligned_vect): Likewise.
4214 (arm_block_set_aligned_vect): Likewise.
4215 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4216 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4217 (expand_vec_perm_even_odd_pack): Likewise.
4218 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4219 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4220 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4221 gen_const_vec_duplicate.
4222 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4223 * config/mips/mips.c (mips_gen_const_int_vector): Use
4224 gen_const_vec_duplicate.
4225 (mips_expand_vector_init): Use CONST0_RTX.
4226 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4227 (define_split): Use gen_const_vec_duplicate.
4228 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4229 (define_split): Use gen_const_vec_duplicate.
4230 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4231 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4232 * config/spu/spu.c (spu_const): Likewise.
4234 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4235 Alan Hayward <alan.hayward@arm.com>
4236 David Sherwood <david.sherwood@arm.com>
4238 * combine.c (can_change_dest_mode): Reject changes in
4239 REGMODE_NATURAL_SIZE.
4241 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4243 * configure.ac (--enable-libssp): New.
4244 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4245 trying to determine target-specific default. Adjust indentation.
4246 * configure: Regenerated.
4247 * doc/install.texi (Configuration): Expand --disable-libssp
4250 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4252 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4255 2017-10-31 Martin Jambor <mjambor@suse.cz>
4258 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4260 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4262 * auto-profile.c (autofdo_source_profile::read): Use
4263 UNKNOWN_LOCATION rather than 0.
4264 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4265 (warning_at): ...this overload.
4266 (warning_at_rich_loc_n): Rename to...
4267 (warning_n): ...this overload.
4268 (error_at_rich_loc): Rename to...
4269 (error_at): ...this overload.
4270 (pedwarn_at_rich_loc): Rename to...
4271 (pedwarn): ...this overload.
4272 (permerror_at_rich_loc): Rename to...
4273 (permerror): ...this overload.
4274 (inform_at_rich_loc): Rename to...
4275 (inform): ...this overload.
4276 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4277 (diagnostic_n_impl_richloc): Rename to...
4278 (diagnostic_n_impl): ...this rich_location *-based decl.
4279 (inform_at_rich_loc): Rename to...
4280 (inform): ...this, and add an assertion.
4281 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4282 (warning_at_rich_loc): Rename to...
4283 (warning_at): ...this, and add an assertion.
4284 (warning_at_rich_loc_n): Rename to...
4285 (warning_n): ...this, and add an assertion.
4286 (warning_n): Update location_t-based implementation for removal of
4287 location_t-based diagnostic_n_impl.
4288 (pedwarn_at_rich_loc): Rename to...
4289 (pedwarn): ...this, and add an assertion.
4290 (permerror_at_rich_loc): Rename to...
4291 (permerror): ...this, and add an assertion.
4292 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4293 (error_at_rich_loc): Rename to...
4294 (error_at): ...this, and add an assertion.
4295 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4296 (driver::do_spec_on_infiles): Likewise.
4297 * substring-locations.c (format_warning_va): Update for renaming
4298 of inform_at_rich_loc.
4300 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4302 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4303 _Float<N>X built-in functions so that the variant without the
4304 "__builtin_" prefix is only enabled for the GNU C and Objective C
4305 languages when they are in non-strict ANSI/ISO mode.
4306 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4307 * target.def (floatn_builtin_p): Add a target hook to control
4308 whether _Float<N> and _Float<N>X built-in functions without the
4309 "__builtin_" prefix are enabled, and return true for C and
4310 Objective C in the default hook. Include langhooks.h in
4312 * targhooks.h (default_floatn_builtin_p): Likewise.
4313 * targhooks.c (default_floatn_builtin_p): Likewise.
4314 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4315 floatn_builtin_p target hook.
4316 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4318 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4319 Eric Botcazou <ebotcazou@adacore.com>
4321 PR rtl-optimization/81803
4322 * lra-constraints.c (curr_insn_transform): Also reload the whole
4323 register for a strict subreg no wider than a word if this is for
4324 a WORD_REGISTER_OPERATIONS target.
4326 2017-10-31 Jason Merrill <jason@redhat.com>
4328 * gdbinit.in: Skip over inlines from timevar.h.
4330 2017-10-31 Martin Liska <mliska@suse.cz>
4332 * doc/gcov.texi: Document new option.
4333 * gcov.c (print_usage): Likewise print it.
4334 (process_args): Support the argument.
4335 (format_count): New function.
4336 (format_gcov): Use the function.
4338 2017-10-31 Martin Liska <mliska@suse.cz>
4340 * gcov.c (struct name_map): do not use typedef.
4341 Define operator== and operator<.
4342 (name_search): Remove.
4343 (name_sort): Remove.
4344 (main): Do not allocate names.
4345 (process_file): Add vertical space.
4346 (generate_results): Use std::find.
4347 (release_structures): Do not release memory.
4348 (find_source): Use std::find.
4350 2017-10-31 Martin Liska <mliska@suse.cz>
4352 * gcov.c (struct line_info): Remove it's typedef.
4353 (line_info::line_info): Add proper ctor.
4354 (line_info::has_block): Do not use a typedef.
4355 (struct source_info): Do not use typedef.
4356 (circuit): Likewise.
4357 (get_cycles_count): Likewise.
4358 (output_intermediate_file): Iterate via vector iterator.
4359 (add_line_counts): Use std::vector methods.
4360 (accumulate_line_counts): Likewise.
4361 (output_lines): Likewise.
4363 2017-10-31 Martin Liska <mliska@suse.cz>
4365 * gcov.c (struct source_info): Remove typedef.
4366 (source_info::source_info): Add proper ctor.
4367 (accumulate_line_counts): Use struct, not it's typedef.
4368 (output_gcov_file): Likewise.
4369 (output_lines): Likewise.
4370 (main): Do not allocate an array.
4371 (output_intermediate_file): Use size of vector container.
4372 (process_file): Resize the vector.
4373 (generate_results): Do not preallocate, use newly added vector
4375 (release_structures): Do not release sources.
4376 (find_source): Use vector methods.
4377 (add_line_counts): Do not use typedef.
4379 2017-10-31 Martin Liska <mliska@suse.cz>
4381 * doc/gcov.texi: Document that.
4382 * gcov.c (add_line_counts): Mark lines with a non-executed
4384 (output_line_beginning): Handle such lines.
4385 (output_lines): Pass new argument.
4386 (output_intermediate_file): Print it in intermediate format.
4388 2017-10-31 Martin Liska <mliska@suse.cz>
4390 * color-macros.h: New file.
4391 * diagnostic-color.c: Factor out color related to macros to
4393 * doc/gcov.texi: Document -k option.
4394 * gcov.c (INCLUDE_STRING): Include string.h.
4395 (print_usage): Add -k option.
4396 (process_args): Parse it.
4397 (pad_count_string): New function.
4398 (output_line_beginning): Likewise.
4399 (DEFAULT_LINE_START): New macro.
4400 (output_lines): Support color output.
4402 2017-10-31 Martin Liska <mliska@suse.cz>
4404 PR gcov-profile/82633
4405 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4406 their interaction with GCOV infrastructure.
4407 * configure.ac: Add -fkeep-{inline,static}-functions to
4409 * configure: Regenerate.
4411 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4414 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4416 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4419 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4420 into a register if it does not fit into an immediate offset field.
4422 2017-10-31 Olivier Hainque <hainque@adacore.com>
4424 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4426 2017-10-31 Julia Koval <julia.koval@intel.com>
4428 * config.gcc: Add gfniintrin.h.
4429 * config/i386/gfniintrin.h: New.
4430 * config/i386/i386-builtin-types.def
4431 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4432 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4433 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4434 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4435 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4436 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4437 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4438 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4439 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4440 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4441 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4442 * config/i386/immintrin.h: Include gfniintrin.h.
4443 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4445 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4447 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4449 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4451 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4452 (arm_ashldi3_1bit): Remove pattern.
4453 (ashrdi3): Remove shift by 1 expansion.
4454 (arm_ashrdi3_1bit): Remove pattern.
4455 (lshrdi3): Remove shift by 1 expansion.
4456 (arm_lshrdi3_1bit): Remove pattern.
4457 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4458 cost of ashldi3 by 1.
4459 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4460 (<shift>di3_neon): Likewise.
4462 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4464 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4465 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4466 and (*aarch64_simd_mov<VQ:mode>).
4467 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4468 pattern alternative.
4469 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4470 attributes to match pattern alternative.
4472 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4474 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4475 * config/rs6000/emmintrin.h: New file.
4476 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4478 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4480 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4481 * (movdi_vfp_cortexa8): Remove pattern.
4483 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4485 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4487 (Specific, alpha*-dec-osf5.1): Remove.
4488 (Specific, mips-sgi-irix5): Remove.
4489 (Specific, mips-sgi-irix6): Remove.
4491 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4494 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4495 arguments to clear_bit_region_be.
4497 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4499 * gimplify.c: Include memmodel.h.
4501 2017-10-30 Martin Jambor <mjambor@suse.cz>
4503 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4504 condition whether loop should be executed at all.
4506 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4509 gimple folding of vec_madd() intrinsics.
4510 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4511 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4512 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4514 2017-10-30 Richard Biener <rguenther@suse.de>
4516 PR tree-optimization/82762
4518 2017-10-23 Richard Biener <rguenther@suse.de>
4520 PR tree-optimization/82129
4522 2017-08-01 Richard Biener <rguenther@suse.de>
4524 PR tree-optimization/81181
4525 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4526 (compute_antic): ... end of iteration here.
4528 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4530 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4531 -std=iso9899:2017 and -std=gnu17.
4532 * doc/standards.texi (C Language): Document C17 support.
4533 * doc/cpp.texi (Overview): Mention -std=c17.
4534 (Standard Predefined Macros): Document C11 and C17 values of
4535 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4536 * doc/extend.texi (Inline): Do not list individual options for
4537 standards newer than C99.
4538 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4540 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4543 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4545 * asan.c (asan_finish_file): Align asan globals array by shadow
4548 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4551 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4552 (struct store_immediate_info): Add bitregion_start and bitregion_end
4554 (store_immediate_info::store_immediate_info): Add brs and bre
4555 arguments and initialize bitregion_{start,end} from those.
4556 (struct merged_store_group): Add bitregion_start, bitregion_end,
4557 align_base and mask fields. Drop unnecessary struct keyword from
4558 struct store_immediate_info. Add do_merge method.
4559 (clear_bit_region_be): Use memset instead of loop storing zeros.
4560 (merged_store_group::do_merge): New method.
4561 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4562 stores as long as the surrounding bitregions have no gaps.
4563 (merged_store_group::merge_overlapping): Use do_merge.
4564 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4565 is byte aligned, rather than requiring that start and width are
4566 byte aligned. Drop unnecessary struct keyword from
4567 struct store_immediate_info. Allocate and populate also mask array.
4568 Make start of the arrays relative to bitregion_start rather than
4569 start and size them according to bitregion_{end,start} difference.
4570 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4571 struct store_immediate_info.
4572 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4573 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4574 struct keyword from struct store_immediate_info.
4575 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4576 between stores as long as the surrounding bitregions have no gaps.
4578 (struct split_store): Add orig non-static data member.
4579 (split_store::split_store): Initialize orig to false.
4580 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4581 if there is exactly a single original stmt. Change stmts argument
4582 to pointer from reference, if NULL, don't push anything to it. Add
4583 first argument, use it to optimize skipping over orig stmts that
4584 are known to be before bitpos already. Simplify.
4585 (split_group): Return unsigned int count how many stores are or
4586 would be needed rather than a bool. Add allow_unaligned argument.
4587 Change split_stores argument from reference to pointer, if NULL,
4588 only do a dry run computing how many stores would be produced.
4589 Rewritten algorithm to use both alignment and misalign if
4590 !allow_unaligned and handle bitfield stores with gaps.
4591 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4592 from bitregion_start instead of start. Compute allow_unaligned
4593 here, if true, do 2 split_group dry runs to compute which one
4594 produces fewer stores and prefer aligned if equal. Punt if
4595 new count is bigger or equal than original before emitting any
4596 statements, rather than during that. Remove no longer needed
4597 new_ssa_names tracking. Replace num_stmts with
4598 split_stores.length (). Use 32-bit stack allocated entries
4599 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4600 if possible. Handle bitfields with gaps.
4601 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4602 Compute bitregion_{start,end} for the stores and construct
4603 store_immediate_info with that. Formatting fixes.
4605 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4608 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4609 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4611 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4613 * gimplify.c: Include tm_p.h.
4615 * common.opt (gcoff): Re-add as ignored option.
4616 (gcoff1, gcoff2, gcoff3): Likewise.
4618 * Makefile.in (OBJS): Delete sdbout.o.
4619 (GTFILES): Delete $(srcdir)/sdbout.c.
4620 * debug.h: Delete sdb_debug_hooks.
4621 * final.c: Delete sdbout.h include.
4622 (final_scan_insn): Delete SDB_DEBUG check.
4623 (rest_of_clean_state): Likewise.
4624 * output.h: Delete sdb_begin_function_line.
4627 * toplev.c: Delete sdbout.h include.
4628 (process_options): Delete SDB_DEBUG check.
4629 * tree-core.h (tree_type_common): Delete pointer field of
4631 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4632 TYPE_SYMTAB_POINTER.
4633 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4634 (TYPE_SYMTAB_IS_POINTER): Delete.
4635 (TYPE_SYMTAB_IS_DIE): Renumber.
4636 * xcoffout.c: Refer to former sdbout.c file.
4637 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4639 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4640 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4641 (Debugging Options): Delete -gcoff.
4642 (-gstabs): Delete SDB reference.
4644 (-gcoff@var{level}): Delete.
4645 * doc/passes.texi (Debugging information output): Delete SDB and
4646 sdbout.c references.
4647 * doc/tm.texi: Regenerate.
4648 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4649 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4651 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4652 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4653 (SDB_DEBUGGING_INFO): Delete.
4654 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4655 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4656 * target.def (output_source_filename): Delete COFF reference.
4658 * common.opt (gcoff): Delete.
4659 (gxcoff+): Update Negative chain.
4660 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4662 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4663 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4664 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4666 (expand_function_start): Change sdb reference to past tense.
4667 (expand_function_end): Change sdb reference to past tense.
4668 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4669 * opts.c (debug_type_names): Delete coff entry.
4670 (common_handle_option): Delete OPT_gcoff case.
4671 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4673 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4674 * config/cris/cris.h: Delete SDB reference in comment.
4675 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4676 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4677 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4678 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4680 (ix86_expand_prologue): Likewise.
4681 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4682 * config/ia64/ia64.h: Likewise.
4683 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4684 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4686 * config/mmix/mmix.h: Likewise.
4687 * config/nds32/nds32.c: Likewise.
4688 * config/stormy/storym16.h: Likewise.
4689 * config/visium/visium.h: Likewise.
4690 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4692 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4694 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4695 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4696 FRAME_POINTER_REGNUM point at high end of local var area.
4698 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4700 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4701 Move comment around. Do not reset best_edge for a copiable
4702 destination if the copy would cause a partition change.
4703 (better_edge_p): Remove redundant check.
4705 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4707 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4709 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4711 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4712 for math functions that have _Float<N> and _Float<N>X variants.
4713 (mathfn_built_in_2): Add support for math functions that have
4714 _Float<N> and _Float<N>X variants.
4715 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4716 (expand_builtin_mathfn_ternary): Add support for fma with
4717 _Float<N> and _Float<N>X variants.
4718 (expand_builtin): Likewise.
4719 (fold_builtin_3): Likewise.
4720 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4721 create math function _Float<N> and _Float<N>X variants as external
4723 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4724 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4725 the __builtin_ prefix and if not strict ansi, without the prefix.
4726 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4727 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4728 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4729 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4730 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4731 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4732 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4733 function signatures for fma _Float<N> and _Float<N>X variants.
4734 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4735 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4736 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4737 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4738 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4739 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4740 * gencfn-macros.c (print_case_cfn): Add support for math functions
4741 that have _Float<N> and _Float<N>X variants.
4742 (print_define_operator_list): Likewise.
4743 (fltfn_suffixes): Likewise.
4745 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4746 for math functions that have _Float<N> and _Float<N>X variants.
4747 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4748 and _Float<N>X variants.
4749 (COPYSIGN): Likewise.
4752 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4753 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4755 (integer_valued_read_call_p): Likewise.
4756 * fold-const-call.c (fold_const_call_ss): Likewise.
4757 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4758 _Float<N> and _Float<N>X variants.
4759 (fold_const_call_ssss): Add support for fma _Float<N> and
4760 _Float<N>X variants.
4761 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4762 support for copysign and fma _Float<N> and _Float<N>X variants.
4763 (backprop::process_builtin_call_use): Likewise.
4764 * tree-call-cdce.c (can_test_argument_range); Add support for
4765 sqrt _Float<N> and _Float<N>X variants.
4766 (edom_only_function): Likewise.
4767 (get_no_error_domain): Likewise.
4768 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4769 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4770 copysign _Float<N> and _Float<N>X variants.
4771 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4772 handled by machine independent code.
4773 (FMAF128): Likewise.
4774 * doc/cpp.texi (Common Predefined Macros): Document defining
4775 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4776 fma _Float<N> and _Float<N>X variants.
4778 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4781 * config/i386/i386-modes.def (CCFPU): Remove definition.
4782 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4783 (ix86_cc_modes_compatible): Ditto.
4784 (ix86_expand_carry_flag_compare): Ditto.
4785 (ix86_expand_int_movcc): Ditto.
4786 (ix86_expand_int_addcc): Ditto.
4787 (ix86_reverse_condition): Ditto.
4788 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4789 Return true/false for unordered/ordered fp comparisons.
4790 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4791 (ix86_prepare_fp_compare_args): Update for rename.
4792 (ix86_expand_fp_compare): Update for rename. Generate unordered
4793 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4794 (ix86_expand_sse_compare_and_jump): Ditto.
4795 * config/i386/predicates.md (fcmov_comparison_operator):
4796 Remove CCFPU mode handling.
4797 (ix86_comparison_operator): Ditto.
4798 (ix86_carry_flag_operator): Ditto.
4799 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
4800 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
4801 (*cmpu<mode>_cc_i387): Ditto.
4802 (FPCMP): Remove mode iterator.
4803 (unord): Remove mode attribute.
4804 (unord_subst): New define_subst transformation
4805 (unord): New define_subst attribute.
4807 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
4808 (*cmpi<unord>xf_i387): Ditto.
4809 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
4810 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
4811 using unord_subst transformation.
4812 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
4813 (round_saeonly): Also handle CCFP mode.
4814 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
4815 Remove UNSPEC_SAHF unspec handling.
4817 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4819 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
4821 2017-10-27 Jeff Law <law@redhat.com>
4823 * gimple-ssa-sprintf.c: Include domwalk.h.
4824 (class sprintf_dom_walker): New class, derived from dom_walker.
4825 (sprintf_dom_walker::before_dom_children): New function.
4826 (struct call_info): Moved into sprintf_dom_walker class
4827 (compute_formath_length, handle_gimple_call): Likewise.
4828 (sprintf_length::execute): Call the dominator walker rather
4829 than walking the statements.
4831 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4832 gimple statement locations.
4833 (check_array_bounds): Corresponding changes. Get the statement's
4834 location directly from wi->stmt.
4836 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4839 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4841 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4843 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4844 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4846 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4849 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4850 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4851 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4852 maybe_get_pool_constant.
4853 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4856 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4858 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4859 2.26 caveat. Update gas and gld versions.
4860 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4863 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4865 * cgraph.h (set_malloc_flag): Declare.
4866 * cgraph.c (set_malloc_flag_1): New function.
4867 (set_malloc_flag): Likewise.
4868 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4869 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4871 (read_ipa_call_summary): Add support for reading is_return_callee.
4872 (write_ipa_call_summary): Stream is_return_callee.
4873 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4874 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4875 ipa-prop.h, ipa-fnsummary.h.
4876 (pure_const_names): Change to static.
4877 (malloc_state_e): Define.
4878 (malloc_state_names): Define.
4879 (funct_state_d): Add field malloc_state.
4880 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4881 (check_retval_uses): New function.
4882 (malloc_candidate_p): Likewise.
4883 (analyze_function): Add support for malloc attribute.
4884 (pure_const_write_summary): Stream malloc_state.
4885 (pure_const_read_summary): Add support for reading malloc_state.
4886 (dump_malloc_lattice): New function.
4887 (propagate_malloc): New function.
4888 (warn_function_malloc): New function.
4889 (ipa_pure_const::execute): Call propagate_malloc and
4890 ipa_free_fn_summary.
4891 (pass_local_pure_const::execute): Add support for malloc attribute.
4892 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4893 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4895 2017-10-27 Martin Liska <mliska@suse.cz>
4897 PR gcov-profile/82457
4898 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4899 for fork-like functions to be properly instrumented.
4901 2017-10-27 Richard Biener <rguenther@suse.de>
4904 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4905 info when we redirected EH.
4907 2017-10-26 Michael Collison <michael.collison@arm.com>
4909 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4911 (<optab>_trunchf<GPI:mode>2: New pattern.
4912 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4913 * config/aarch64/iterators.md (wv): New mode attribute.
4914 (vf, VF): New mode attributes.
4915 (vgp, VGP): New mode attributes.
4916 (s): Update attribute with SImode and DImode prefixes.
4918 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4920 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4921 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4922 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4923 (nios2_option_overide): Initialize it. Don't allow R0-relative
4924 addressing with PIC.
4925 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4926 (nios2_symbolic_constant_p): Likewise.
4927 (nios2_legitimate_address_p): Likewise.
4928 (nios2_r0rel_section_name_p): New.
4929 (nios2_symbol_ref_in_r0rel_data_p): New.
4930 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4931 (r0rel_constant_p): New.
4932 (nios2_print_operand_address): Handle r0rel_constant_p.
4933 (nios2_cdx_narrow_form_p): Likewise.
4934 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4935 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4936 (Nios II Options): Document -mr0rel-sec.
4938 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4940 * config/nios2/nios2.c: Include xregex.h.
4941 (nios2_gprel_sec_regex): New.
4942 (nios2_option_overide): Initialize it. Don't allow GP-relative
4943 addressing with PIC.
4944 (nios2_small_section_name_p): Check for regex match.
4945 * config/nios2/nios2.opt (mgprel-sec=): New option.
4946 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4947 (Nios II Options): Document -mgprel-sec.
4949 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4951 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4953 2017-10-26 Tom de Vries <tom@codesourcery.com>
4955 PR tree-optimization/82707
4956 * gimple.c (gimple_copy): Fix unsharing of
4957 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4959 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4961 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4962 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4963 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4964 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4965 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4966 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4967 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4968 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4969 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4970 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4971 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4972 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4973 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4974 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4975 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4976 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4977 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4979 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4981 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4983 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4984 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4985 warning to rs6000.c. Remove the Undocumented flag, since it has
4987 (-mabi=ibmlongdouble): Likewise.
4988 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4989 already set, set the default format for long double.
4990 (rs6000_debug_reg_global): Print whether long double is IBM or
4992 (rs6000_option_override_internal): Rework setting long double
4993 format. Only warn if the user is changing the long double default
4994 and they did not use -Wno-psabi.
4995 * doc/invoke.texi (PowerPC options): Update the documentation for
4996 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4998 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4999 Alan Hayward <alan.hayward@arm.com>
5000 David Sherwood <david.sherwood@arm.com>
5002 * rtl.h (wider_subreg_mode): New function.
5003 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5004 rather than an unsigned int *.
5005 * ira-color.c (regno_max_ref_width): Replace with...
5006 (regno_max_ref_mode): ...this new variable.
5007 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5008 Use wider_subreg_mode.
5009 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5010 rather than an unsigned int *.
5011 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5012 (process_alt_operands): Likewise.
5013 (invariant_p): Likewise.
5014 * lra-spills.c (assign_mem_slot): Likewise.
5015 (add_pseudo_to_slot): Likewise.
5016 * lra.c (collect_non_operand_hard_regs): Likewise.
5017 (add_regs_to_insn_regno_info): Likewise.
5018 * reload1.c (regno_max_ref_width): Replace with...
5019 (regno_max_ref_mode): ...this new variable.
5020 (reload): Update accordingly. Update call to
5021 ira_sort_regnos_for_alter_reg.
5022 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5023 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5024 (scan_paradoxical_subregs): Likewise.
5026 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5028 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5029 (aarch64_frame): Add emit_frame_chain boolean.
5030 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5031 Move eh_return case to aarch64_layout_frame.
5032 (aarch64_layout_frame): Initialize emit_frame_chain.
5033 (aarch64_expand_prologue): Use emit_frame_chain.
5035 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5037 * config/aarch64/aarch64.c (aarch64_layout_frame):
5038 Ensure LR is always stored at the bottom of the callee-saves.
5039 Remove rarely used frame layout which saves callee-saves at top of
5040 frame, so the store of LR can be used as a valid probe in all cases.
5042 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5044 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5045 Improve unaligned TImode/TFmode base/offset split.
5047 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5048 Alan Hayward <alan.hayward@arm.com>
5049 David Sherwood <david.sherwood@arm.com>
5051 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5052 * combine.c (find_single_use_1): Likewise.
5053 (expand_field_assignment): Likewise.
5054 (move_deaths): Likewise.
5055 * lra-constraints.c (simplify_operand_subreg): Likewise.
5056 (curr_insn_transform): Likewise.
5057 * lra.c (collect_non_operand_hard_regs): Likewise.
5058 (add_regs_to_insn_regno_info): Likewise.
5059 * rtlanal.c (reg_referenced_p): Likewise.
5060 (covers_regno_no_parallel_p): Likewise.
5062 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5064 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5065 Don't print any bits outside the precision of the value.
5066 * wide-int.cc (test_printing): Add some new tests.
5068 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5070 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5071 supports -xbrace_comment option.
5072 * configure: Regenerate.
5073 * config.in: Regenerate.
5074 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5075 (ASM_CPU_SPEC): Use it.
5077 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5079 * target.def (static_rtx_alignment): New hook.
5080 * targhooks.h (default_static_rtx_alignment): Declare.
5081 * targhooks.c (default_static_rtx_alignment): New function.
5082 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5083 * doc/tm.texi: Regenerate.
5084 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5085 instead of targetm.constant_alignment. Remove call to
5087 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5088 (cris_preferred_mininum_alignment): New function, split out from...
5089 (cris_constant_alignment): ...here.
5090 (cris_static_rtx_alignment): New function.
5091 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5093 (ix86_constant_alignment): ...here.
5094 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5095 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5096 (mmix_static_rtx_alignment): New function.
5097 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5098 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5100 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5103 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5104 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5106 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5108 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5109 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5110 mark as a sign-extending load.
5111 (local_pic_load_u): Define.
5113 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5116 * fold-const.c (operand_equal_for_comparison_p): Also return true
5117 if ARG0 is a simple variant of ARG1 with narrower precision.
5118 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5120 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5122 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5124 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5125 scatter_static, scatter_per_elt.
5126 * x86-tune-costs.h: Add new cost entries.
5128 2017-10-25 Richard Biener <rguenther@suse.de>
5130 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5131 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5132 class pass_fre): Move to ...
5133 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5134 class pass_fre): ... here and adjust for statistics.
5136 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5139 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5140 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5141 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5142 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5145 2017-10-25 Richard Biener <rguenther@suse.de>
5147 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5148 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5149 eliminate_push_avail, eliminate_insert): Move inside...
5150 (class eliminate_dom_walker): ... this class in preparation
5152 (fini_eliminate): Remove by merging with ...
5153 (eliminate): ... this function. Adjust for class changes.
5154 (pass_pre::execute): Remove fini_eliminate call.
5155 (pass_fre::execute): Likewise.
5157 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5160 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5161 (VPERMI2, VPERMI2I): New mode iterators.
5162 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5163 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5165 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5166 mode iterator. Remove 3 old define_insn patterns.
5167 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5168 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5169 VPERMI2 mode iterator, remove the other two expanders.
5170 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5171 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5172 instructions, remove the other two patterns.
5173 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5174 mode iterator, remove the other two patterns.
5175 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5176 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5177 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5178 and adjust argument order accordingly.
5179 (ix86_expand_vec_perm): Adjust caller.
5180 (expand_vec_perm_1): Likewise.
5181 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5182 (expand_vec_perm_vpermt2_vpshub2): ... this.
5183 (ix86_expand_vec_perm_const_1): Adjust caller.
5184 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5187 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5188 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5189 (vec_shl_<mode>): Remove unused expander.
5190 (avx512bw_<shift_insn><mode>3): New define_insn.
5191 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5192 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5194 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5197 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5200 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5202 PR rtl-optimization/82396
5203 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5204 (autopref_multipass_init): Simplify initialization.
5205 (autopref_rank_data): Simplify sort order.
5206 * gcc/sched-int.h (autopref_multipass_data_): Remove
5207 multi_mem_insn_p, min_offset and max_offset.
5209 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5212 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5213 Check special value of flag_omit_frame_pointer.
5214 (aarch64_can_eliminate): Likewise.
5215 (aarch64_override_options_after_change_1): Simplify handling of
5216 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5218 2017-10-24 Richard Biener <rguenther@suse.de>
5220 PR tree-optimization/82697
5221 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5222 zero for conditional load and unconditional store.
5224 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5226 * doc/install.texi: Document bootstrap-cet.
5228 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5231 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5232 ENDBR instruction at function entrance if function is only
5235 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5238 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5239 patterns to better describe from which operation the CF is computed.
5240 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5241 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5242 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5243 is 0, use _0 suffixed expanders instead of emitting a comparison
5246 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5248 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5249 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5251 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5254 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5255 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5256 * loop-iv.c (iv_get_reaching_def): Likewise.
5257 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5258 variable is promoted and the partition contains undefined values.
5260 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5262 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5264 (nios2_address_cost): Define.
5265 (nios2_legitimize_address): Recognize (exp + constant) directly.
5266 (TARGET_ADDRESS_COST): Define.
5268 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5270 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5271 (nios2_symbolic_memory_operand_p): Declare.
5272 (nios2_split_large_constant): Declare.
5273 (nios2_split_symbolic_memory_operand): Declare.
5274 * config/nios2/nios2.c: Adjust includes.
5275 (nios2_symbolic_constant_allowed): New.
5276 (nios2_symbolic_constant_p): New.
5277 (nios2_plus_symbolic_constant_p): New.
5278 (nios2_valid_addr_expr_p): Recognize addresses involving
5280 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5281 (nios2_symbolic_memory_operand_p): New.
5282 (nios2_large_constant_p): New.
5283 (nios2_split_large_constant): New.
5284 (nios2_split_plus_large_constant): New.
5285 (nios2_split_symbolic_memory_operand): New.
5286 (nios2_legitimize_address): Code refactoring. Handle addresses
5287 involving symbolic constants.
5288 (nios2_emit_move_sequence): Likewise.
5289 (nios2_print_operand): Improve error output.
5290 (nios2_print_operand_address): Handle LO_SUM.
5291 (nios2_cdx_narrow_form_p): Likewise.
5292 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5293 operands involving symbolic constants.
5294 (movhi_internal, movsi_internal): Likewise.
5295 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5296 (extendhisi2, extendqi<mode>2): Likewise.
5298 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5300 * tree-pass.h (PROP_rtl_split_insns): Define.
5301 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5303 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5305 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5307 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5310 * target.def (const_not_ok_for_debug_p): Default to
5311 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5312 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5313 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5314 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5315 which targetm.const_not_ok_for_debug_p returned true.
5316 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5318 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5320 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5321 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5322 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5324 (ix86_const_not_ok_for_debug_p): New function.
5325 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5326 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5328 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5331 * system.h: Conditionally include "unique-ptr.h" if
5332 INCLUDE_UNIQUE_PTR is defined.
5333 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5334 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5336 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5338 * config/rl78/rl78.md: New define_expand "subdi3".
5340 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5343 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5344 DF_REF_INSN if DF_REF_INSN_INFO is false.
5346 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5348 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5349 xmm_move instead of sse_move.
5350 (sse_store_index): New function.
5351 (ix86_register_move_cost): Be more sensible about mismatch stall;
5352 model AVX moves correctly; make difference between sse->integer and
5354 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5355 moves; make difference between SSE and AVX.
5356 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5357 and zmm_move. Increase size of sse load and store tables;
5358 add unaligned load and store tables; add ssemmx_to_integer.
5359 * x86-tune-costs.h: Update all entries according to real
5360 move latencies from Agner Fog's manual and chip documentation.
5362 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5365 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5366 * config/i386/constraints.md (Wf): New constraint.
5367 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5368 (cmp<dwi>_doubleword): Removed.
5369 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5370 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5371 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5372 expand with cmp<dwi>_doubleword. For LTU and GEU use
5373 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5375 * common.opt (gcolumn-info): Enable by default.
5376 * doc/invoke.texi (gcolumn-info): Document new default.
5378 2017-10-23 Richard Biener <rguenther@suse.de>
5380 PR tree-optimization/82672
5381 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5382 Fold the stmt if we propagated into it.
5384 2017-10-23 Richard Biener <rguenther@suse.de>
5386 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5387 (bitmap_remove_expr_from_set): ... this. All callers call this
5388 for non-constant values.
5389 (bitmap_set_subtract): Rename to...
5390 (bitmap_set_subtract_expressions): ... this. Adjust and
5392 (bitmap_set_contains_value): Remove superfluous check.
5393 (bitmap_set_replace_value): Inline into single caller ...
5394 (bitmap_value_replace_in_set): ... here and simplify.
5395 (dependent_clean): Merge into ...
5396 (clean): ... this using an overload. Adjust.
5397 (prune_clobbered_mems): Adjust.
5398 (compute_antic_aux): Likewise.
5399 (compute_partial_antic_aux): Likewise.
5401 2017-10-23 Richard Biener <rguenther@suse.de>
5403 PR tree-optimization/82129
5405 2017-08-01 Richard Biener <rguenther@suse.de>
5407 PR tree-optimization/81181
5408 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5409 (compute_antic): ... end of iteration here.
5411 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5413 * target.def (starting_frame_offset): New hook.
5414 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5415 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5416 * doc/tm.texi.in: Regenerate.
5417 * hooks.h (hook_hwi_void_0): Declare.
5418 * hooks.c (hook_hwi_void_0): New function.
5419 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5420 STARTING_FRAME_OFFSET.
5421 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5422 * reload1.c (reload): Likewise.
5423 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5424 instead of STARTING_FRAME_OFFSET.
5425 * function.c (try_fit_stack_local): Likewise.
5426 (assign_stack_local_1): Likewise
5427 (instantiate_virtual_regs): Likewise.
5428 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5429 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5430 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5431 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5432 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5433 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5434 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5435 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5436 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5437 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5438 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5439 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5440 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5441 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5442 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5443 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5444 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5445 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5446 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5447 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5448 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5449 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5450 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5451 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5452 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5453 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5454 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5455 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5456 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5457 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5458 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5459 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5460 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5461 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5462 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5463 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5464 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5465 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5466 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5467 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5468 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5469 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5470 return a HOST_WIDE_INT.
5471 (avr_builtin_setjmp_frame_value): Use it instead of
5472 STARTING_FRAME_OFFSET.
5473 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5474 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5475 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5477 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5478 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5479 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5480 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5481 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5482 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5483 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5484 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5485 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5486 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5487 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5488 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5490 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5491 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5492 * config/mips/mips.c (mips_compute_frame_info): Refer to
5493 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5494 (mips_starting_frame_offset): New function.
5495 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5496 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5497 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5498 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5499 and return a HOST_WIDE_INT.
5500 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5501 (mmix_initial_elimination_offset): Refer to
5502 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5503 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5504 * config/pa/pa.c (pa_starting_frame_offset): New function.
5505 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5506 (pa_expand_prologue): Likewise.
5507 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5508 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5509 !FRAME_GROWS_DOWNWARD handling to...
5510 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5511 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5512 !FRAME_GROWS_DOWNWARD handling to...
5513 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5514 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5515 !FRAME_GROWS_DOWNWARD handling to...
5516 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5517 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5519 (rs6000_starting_frame_offset): New function.
5520 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5521 !FRAME_GROWS_DOWNWARD handling to...
5522 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5523 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5524 !FRAME_GROWS_DOWNWARD handling to...
5525 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5526 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5527 !FRAME_GROWS_DOWNWARD handling to...
5528 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5529 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5530 (rs6000_starting_frame_offset): New function.
5531 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5532 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5533 * config/vax/vax.c (vax_starting_frame_offset): New function.
5534 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5535 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5536 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5537 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5538 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5539 * system.h (STARTING_FRAME_OFFSET): Poison.
5541 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5543 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5544 SCALAR_TYPE_MODE instead of TYPE_MODE.
5546 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5547 Alan Hayward <alan.hayward@arm.com>
5548 David Sherwood <david.sherwood@arm.com>
5550 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5552 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5553 Alan Hayward <alan.hayward@arm.com>
5554 David Sherwood <david.sherwood@arm.com>
5556 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5558 2017-10-23 Richard Biener <rguenther@suse.de>
5560 PR tree-optimization/82129
5561 * tree-ssa-pre.c (bitmap_set_and): Remove.
5562 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5563 canonicalizing expressions in the set to those with lowest
5564 ID rather than taking that from the first edge.
5566 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5568 * combine.c (rtx_equal_for_field_assignment_p): Use
5569 byte_lowpart_offset.
5571 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5572 Alan Hayward <alan.hayward@arm.com>
5573 David Sherwood <david.sherwood@arm.com>
5575 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5576 to a SUBREG_PROMOTED_VAR.
5578 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5579 Alan Hayward <alan.hayward@arm.com>
5580 David Sherwood <david.sherwood@arm.com>
5582 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5583 (expand_debug_source_expr): Likewise.
5584 * combine.c (combine_simplify_rtx): Likewise.
5585 * cse.c (fold_rtx): Likewise.
5586 * optabs.c (expand_float): Likewise.
5587 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5588 (simplify_binary_operation_1): Likewise.
5590 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5591 Alan Hayward <alan.hayward@arm.com>
5592 David Sherwood <david.sherwood@arm.com>
5594 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5595 (record_promoted_value): Likewise.
5596 * expr.c (expand_expr_real_2): Likewise.
5597 * ree.c (update_reg_equal_equiv_notes): Likewise.
5598 (combine_set_extension): Likewise.
5599 * rtlanal.c (low_bitmask_len): Likewise.
5600 * simplify-rtx.c (neg_const_int): Likewise.
5601 (simplify_binary_operation_1): Likewise.
5603 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5604 Alan Hayward <alan.hayward@arm.com>
5605 David Sherwood <david.sherwood@arm.com>
5607 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5608 * regcprop.c (maybe_mode_change): Likewise.
5609 * reload1.c (alter_reg): Likewise.
5611 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5613 * inchash.h (inchash::hash::add_wide_int): New function.
5614 * lto-streamer-out.c (hash_tree): Use it.
5616 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5618 * inchash.h (inchash::hash::add_wide_int): Rename to...
5619 (inchash::hash::add_hwi): ...this.
5620 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5621 (polymorphic_call_target_hasher::hash): Likewise.
5622 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5623 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5624 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5625 * lto-streamer-out.c (hash_tree): Likewise.
5626 * optc-save-gen.awk: Likewise.
5627 * tree.c (add_expr): Likewise.
5629 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5632 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5633 for ordered inequality comparisons even with TARGET_IEEE_FP.
5635 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5638 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5639 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5640 Expand with cmp<dwi>_doubleword.
5642 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5644 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5645 List CET intrinsics.
5646 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5647 specific to -fcf-protection option.
5649 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5651 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5652 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5653 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5654 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5655 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5656 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5657 (extra_objs): Add cet.o for Linux/x86 targets.
5658 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5659 * config/i386/cet.c: New file.
5660 * config/i386/cetintrin.h: Likewise.
5661 * config/i386/t-cet: Likewise.
5662 * config/i386/cpuid.h (bit_SHSTK): New.
5663 (bit_IBT): Likewise.
5664 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5665 pass IBT and SHSTK bits.
5666 * config/i386/i386-builtin-types.def
5667 (VOID_FTYPE_UNSIGNED_PVOID): New.
5668 (VOID_FTYPE_UINT64_PVOID): Likewise.
5669 * config/i386/i386-builtin.def: Add CET intrinsics.
5670 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5671 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5672 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5673 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5675 * config/i386/i386.c (rest_of_insert_endbranch): New.
5676 (pass_data_insert_endbranch): Likewise.
5677 (pass_insert_endbranch): Likewise.
5678 (make_pass_insert_endbranch): Likewise.
5679 (ix86_notrack_prefixed_insn_p): Likewise.
5680 (ix86_target_string): Add -mibt, -mshstk flags.
5681 (ix86_option_override_internal): Add flag_cf_protection
5683 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5684 (ix86_print_operand): Add 'notrack' prefix output.
5685 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5686 (ix86_expand_builtin): Expand CET intrinsics.
5687 (x86_output_mi_thunk): Add 'endbranch' instruction.
5688 * config/i386/i386.h (TARGET_IBT): New.
5689 (TARGET_IBT_P): Likewise.
5690 (TARGET_SHSTK): Likewise.
5691 (TARGET_SHSTK_P): Likewise.
5692 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5693 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5694 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5695 (builtin_setjmp_setup): New pattern.
5696 (builtin_longjmp): Likewise.
5697 (rdssp<mode>): Likewise.
5698 (incssp<mode>): Likewise.
5699 (saveprevssp): Likewise.
5700 (rstorssp): Likewise.
5701 (wrss<mode>): Likewise.
5702 (wruss<mode>): Likewise.
5703 (setssbsy): Likewise.
5704 (clrssbsy): Likewise.
5705 (nop_endbr): Likewise.
5706 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5708 * config/i386/immintrin.h: Include <cetintrin.h>.
5709 * config/i386/linux-common.h
5710 (file_end_indicate_exec_stack_and_cet): New prototype.
5711 (TARGET_ASM_FILE_END): New.
5713 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5715 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5716 latencies instead of having separate table; make difference between
5717 integer and float costs.
5718 * i386.h (processor_costs): Remove scalar_stmt_cost,
5719 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5720 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5722 * x86-tune-costs.h: Remove entries which has been removed in
5723 procesor_costs from all tables; make cond_taken_branch_cost
5724 and cond_not_taken_branch_cost COST_N_INSNS based.
5726 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5728 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5730 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5732 * config/i386/i386.md (isa): Remove fma_avx512f.
5733 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5734 <avx512>_fmadd_<mode>_mask3<round_name>,
5735 <avx512>_fmsub_<mode>_mask<round_name>,
5736 <avx512>_fmsub_<mode>_mask3<round_name>,
5737 <avx512>_fnmadd_<mode>_mask<round_name>,
5738 <avx512>_fnmadd_<mode>_mask3<round_name>,
5739 <avx512>_fnmsub_<mode>_mask<round_name>,
5740 <avx512>_fnmsub_<mode>_mask3<round_name>,
5741 <avx512>_fmaddsub_<mode>_mask<round_name>,
5742 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5743 <avx512>_fmsubadd_<mode>_mask<round_name>,
5744 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5745 (*vec_widen_umult_even_v16si<mask_name>,
5746 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5747 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5749 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5751 * extend.texi: Add 'nocf_check' documentation.
5752 * gimple.texi: Add second parameter to
5753 gimple_build_call_from_tree.
5754 * invoke.texi: Add -fcf-protection documentation.
5755 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5757 2017-10-20 Richard Biener <rguenther@suse.de>
5759 PR tree-optimization/82473
5760 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5761 the largest input type.
5763 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5765 * c-attribs.c (handle_nocf_check_attribute): New function.
5766 (c_common_attribute_table): Add 'nocf_check' handling.
5767 * gimple-parser.c: Add second argument NULL to
5768 gimple_build_call_from_tree.
5769 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5770 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5772 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5773 * common.opt: Add fcf-protection flag.
5774 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5775 * flag-types.h: Add enum cf_protection_level.
5776 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5777 Add 'nocf_check' attribute propagation to gimple call.
5778 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5779 (gimple_build_call_from_tree): Update prototype.
5780 (gimple_call_nocf_check_p): New function.
5781 (gimple_call_set_nocf_check): Likewise.
5782 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5783 * ipa-icf.c: Add nocf_check attribute in statement hash.
5784 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5785 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5786 * toplev.c (process_options): Add flag_cf_protection handling.
5788 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5790 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5792 2017-10-20 Richard Biener <rguenther@suse.de>
5794 PR tree-optimization/82603
5795 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5796 remove false predicated stores.
5798 2017-10-20 Richard Biener <rguenther@suse.de>
5800 * graphite-isl-ast-to-gimple.c
5801 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
5802 Remove return value and simplify, dump copied stmt after lhs
5804 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
5805 Reduce dump verbosity.
5806 (gsi_insert_earliest): Likewise.
5807 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
5808 * graphite.c (print_global_statistics): Adjust dumping.
5809 (print_graphite_scop_statistics): Likewise.
5810 (print_graphite_statistics): Do not dump loops here.
5811 (graphite_transform_loops): But here.
5813 2017-10-20 Nicolas Roche <roche@adacore.com>
5815 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
5816 * configure: Regenerate.
5818 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5821 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
5822 functions when optimizing replace GIMPLE_RETURN stmts with
5823 calls to __builtin_unreachable ().
5826 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5827 for -fsanitize=thread link of executables.
5828 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
5829 link of executables.
5832 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5834 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5835 define_insns for logical vector shifts to use VI248_AVX512BW
5836 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5837 condition, useless isa and prefix attributes. Change the first
5838 2 of these define_insns to ...
5839 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5840 define_insn for avx512vl.
5841 (<shift_insn><mode>3): ... and this, new define_insn without
5842 masking for non-avx512vl.
5845 * config/i386/sse.md (*andnot<mode>3,
5846 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5847 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5848 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5849 not applied use empty suffix even for TARGET_AVX512VL.
5850 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5851 is applied, supply evex,evex or evex,evex,evex instead of just
5854 2017-10-20 Julia Koval <julia.koval@intel.com>
5856 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5857 (OPTION_MASK_ISA_GFNI_UNSET): New.
5858 (ix86_handle_option): Handle OPT_mgfni.
5859 * config/i386/cpuid.h (bit_GFNI): New.
5860 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5861 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5862 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5863 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5864 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5865 * config/i386/i386.opt: Add mgfni.
5867 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5869 * config/msp430/msp430.c (msp430_option_override): Disable
5870 -fdelete-null-pointer-checks.
5871 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5873 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5875 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5876 of x87 and SSE instructions.
5878 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5880 * asan.c (create_cond_insert_point): Do not update edge count.
5881 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5882 (afdo_propagate_circuit): Likewise.
5883 (afdo_calculate_branch_prob): Likewise.
5884 (afdo_annotate_cfg): Likewise.
5885 * basic-block.h (struct edge_def): Remove count.
5886 (edge_def::count): New accessor.
5887 * bb-reorder.c (rotate_loop): Update.
5888 (find_traces_1_round): Update.
5889 (connect_traces): Update.
5890 (sanitize_hot_paths): Update.
5891 * cfg.c (unchecked_make_edge): Update.
5892 (make_single_succ_edge): Update.
5893 (check_bb_profile): Update.
5894 (dump_edge_info): Update.
5895 (update_bb_profile_for_threading): Update.
5896 (scale_bbs_frequencies_int): Update.
5897 (scale_bbs_frequencies_gcov_type): Update.
5898 (scale_bbs_frequencies_profile_count): Update.
5899 (scale_bbs_frequencies): Update.
5900 * cfganal.c (connect_infinite_loops_to_exit): Update.
5901 * cfgbuild.c (compute_outgoing_frequencies): Update.
5902 (find_many_sub_basic_blocks): Update.
5903 * cfgcleanup.c (try_forward_edges): Update.
5904 (try_crossjump_to_edge): Update
5905 * cfgexpand.c (expand_gimple_cond): Update
5906 (expand_gimple_tailcall): Update
5907 (construct_exit_block): Update
5908 * cfghooks.c (verify_flow_info): Update
5909 (redirect_edge_succ_nodup): Update
5910 (split_edge): Update
5911 (make_forwarder_block): Update
5912 (duplicate_block): Update
5913 (account_profile_record): Update
5914 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5915 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5916 * cfgloopmanip.c (scale_loop_profile): Update.
5918 (lv_adjust_loop_entry_edge): Update.
5919 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5920 (force_nonfallthru_and_redirect): Update.
5921 (purge_dead_edges): Update.
5922 (rtl_flow_call_edges_add): Update.
5923 * cgraphunit.c (init_lowered_empty_function): Update.
5924 (cgraph_node::expand_thunk): Update.
5925 * gimple-pretty-print.c (dump_probability): Update.
5926 (dump_edge_probability): Update.
5927 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5928 * haifa-sched.c (sched_create_recovery_edges): Update.
5929 * hsa-gen.c (convert_switch_statements): Update.
5930 * ifcvt.c (dead_or_predicable): Update.
5931 * ipa-inline-transform.c (inline_transform): Update.
5932 * ipa-split.c (split_function): Update.
5933 * ipa-utils.c (ipa_merge_profiles): Update.
5934 * loop-doloop.c (add_test): Update.
5935 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5936 * lto-streamer-in.c (input_cfg): Update.
5937 (input_function): Update.
5938 * lto-streamer-out.c (output_cfg): Update.
5939 * modulo-sched.c (sms_schedule): Update.
5940 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5941 * predict.c (maybe_hot_edge_p): Update.
5942 (unlikely_executed_edge_p): Update.
5943 (probably_never_executed_edge_p): Update.
5944 (dump_prediction): Update.
5945 (drop_profile): Update.
5946 (propagate_unlikely_bbs_forward): Update.
5947 (determine_unlikely_bbs): Update.
5948 (force_edge_cold): Update.
5949 * profile.c (compute_branch_probabilities): Update.
5950 * reg-stack.c (better_edge): Update.
5951 * shrink-wrap.c (handle_simple_exit): Update.
5952 * tracer.c (better_p): Update.
5953 * trans-mem.c (expand_transaction): Update.
5954 (split_bb_make_tm_edge): Update.
5955 * tree-call-cdce.c: Update.
5956 * tree-cfg.c (gimple_find_sub_bbs): Update.
5957 (gimple_split_edge): Update.
5958 (gimple_duplicate_sese_region): Update.
5959 (gimple_duplicate_sese_tail): Update.
5960 (gimple_flow_call_edges_add): Update.
5961 (insert_cond_bb): Update.
5962 (execute_fixup_cfg): Update.
5963 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5964 * tree-complex.c (expand_complex_div_wide): Update.
5965 * tree-eh.c (lower_resx): Update.
5966 (unsplit_eh): Update.
5967 (cleanup_empty_eh_move_lp): Update.
5968 * tree-inline.c (copy_edges_for_bb): Update.
5969 (freqs_to_counts): Update.
5970 (copy_cfg_body): Update.
5971 * tree-ssa-dce.c (remove_dead_stmt): Update.
5972 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5973 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5974 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5975 (unloop_loops): Update.
5976 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5977 * tree-ssa-loop-split.c (connect_loops): Update.
5978 (split_loop): Update.
5979 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5980 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5981 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5982 * tree-ssa-reassoc.c (branch_fixup): Update.
5983 * tree-ssa-tail-merge.c (replace_block_by): Update.
5984 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5985 (compute_path_counts): Update.
5986 (update_profile): Update.
5987 (recompute_probabilities): Update.
5988 (update_joiner_offpath_counts): Update.
5989 (estimated_freqs_path): Update.
5990 (freqs_to_counts_path): Update.
5991 (clear_counts_path): Update.
5992 (ssa_fix_duplicate_block_edges): Update.
5993 (duplicate_thread_path): Update.
5994 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5995 (case_bit_test_cmp): Update.
5996 (collect_switch_conv_info): Update.
5997 (gen_inbound_check): Update.
5998 (do_jump_if_equal): Update.
5999 (emit_cmp_and_jump_insns): Update.
6000 * tree-tailcall.c (decrease_profile): Update.
6001 (eliminate_tail_call): Update.
6002 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6003 (vect_do_peeling): Update.
6004 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6005 * ubsan.c (ubsan_expand_null_ifn): Update.
6006 (ubsan_expand_ptr_ifn): Update.
6007 * value-prof.c (gimple_divmod_fixed_value): Update.
6008 (gimple_mod_pow2): Update.
6009 (gimple_mod_subtract): Update.
6010 (gimple_ic): Update.
6011 (gimple_stringop_fixed_value): Update.
6013 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6016 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6018 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6020 PR rtl-optimization/82395
6021 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6022 based on non_spilled_static_chain_regno_p.
6024 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6026 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6027 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6028 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6030 2017-10-19 Martin Sebor <msebor@redhat.com>
6032 PR tree-optimization/82596
6033 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6035 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6037 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6038 (get_mem_refs_of_builtin_call): Likewise.
6039 * builtins.c (expand_builtin_apply): Adjust call to
6040 allocate_dynamic_stack_space.
6041 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6042 the third argument to allocate_dynamic_stack_space, otherwise -1.
6043 (expand_builtin): Deal with all alloca variants.
6044 (is_inexpensive_builtin): Likewise.
6045 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6046 * calls.c (special_function_p): Deal with all alloca variants.
6047 (initialize_argument_information): Adjust call to
6048 allocate_dynamic_stack_space.
6049 (expand_call): Likewise.
6050 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6051 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6052 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6053 use it for the stack usage computation.
6054 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6055 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6056 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6057 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6058 (in_loop_p): Remove first argument and useless check.
6059 (pass_walloca::execute): Remove useless test and adjust call to above.
6060 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6061 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6062 (gimplify_call_expr): Deal with all alloca variants.
6063 * hsa-gen.c (gen_hsa_alloca): Likewise.
6064 (gen_hsa_insns_for_call): Likewise.
6065 * ipa-pure-const.c (special_builtin_state): Likewise.
6066 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6067 * tree-object-size.c (alloc_object_size): Likewise.
6068 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6069 (call_may_clobber_ref_p_1): Likewise.
6070 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6071 (ccp_fold_stmt): Likewise.
6072 (optimize_stack_restore): Likewise.
6073 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6074 (mark_all_reaching_defs_necessary_1): Likewise.
6075 (propagate_necessity): Likewise.
6076 (eliminate_unnecessary_stmts): Likewise.
6077 * tree.c (build_common_builtin_nodes): Build
6078 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6079 (build_alloca_call_expr): New function.
6080 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6081 (CASE_BUILT_IN_ALLOCA): Likewise.
6082 (build_alloca_call_expr): Declare.
6083 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6085 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6088 * dwarf2out.c (new_die_raw): New static inline function.
6089 (new_die): Use it to create the DIE.
6090 (add_AT_external_die_ref): Likewise.
6091 (clone_die): Likewise.
6092 (clone_as_declaration): Likewise.
6093 (dwarf2out_vms_debug_main_pointer): Likewise.
6094 (base_type_die): Likewise. Remove early return for corner cases.
6095 Do not call add_pubtype on the DIE here.
6096 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6097 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6098 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6099 native order exists for base types, attach the DIE manually and call
6100 add_pubtype on it. Do not equate a reverse order DIE to the type.
6102 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6104 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6105 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6107 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6109 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6111 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6112 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6113 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6114 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6115 * opts.c (sanitizer_opts): Add builtin.
6116 * ubsan.c (instrument_builtin): New function.
6117 (pass_ubsan::execute): Call it.
6118 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6119 * doc/invoke.texi: Document -fsanitize=builtin.
6121 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6122 builtins, store max (log2 (align), 0) into uchar field instead of
6123 align into uptr field.
6124 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6125 store uchar 0 field instead of uptr 0 field.
6126 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6127 instead of passing one address of struct with 2 locations pass
6128 two addresses of structs with 1 location each.
6129 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6130 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6131 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6132 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6133 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6134 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6135 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6136 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6138 2017-10-19 Martin Liska <mliska@suse.cz>
6141 * file-find.c (remove_prefix): Remove.
6142 * file-find.h (remove_prefix): Likewise.
6143 * gcc-ar.c: Remove smartness of lookup.
6145 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6147 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6148 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6149 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6151 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6154 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6155 (setcc + and to xor + setcc): New peephole2.
6157 2017-10-19 Tom de Vries <tom@codesourcery.com>
6159 * doc/sourcebuild.texi (Test Directives, Variants of
6160 dg-require-support): Add dg-require-stack-size.
6162 2017-10-19 Martin Liska <mliska@suse.cz>
6165 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6166 that have a large alignment.
6167 (gimplify_target_expr): Likewise.
6169 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6171 PR rtl-optimization/82602
6172 * ira.c (rtx_moveable_p): Return false for volatile asm.
6174 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6177 * config/i386/i386-modes.def (CCGZ): New CC mode.
6178 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6179 * config/i386/predicates.md (ix86_comparison_operator):
6181 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6182 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6183 with double-word subtraction.
6184 (put_condition_code): Handle CCGZmode.
6186 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6188 * wide-int.cc (debug (const wide_int &)): New.
6189 (debug (const wide_int *)): New.
6190 (debug (const widest_int &)): New.
6191 (debug (const widest_int *)): New.
6193 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6196 * lra-constraints.c (curr_insn_transform): Use non-input operand
6197 instead of output one for matched reload.
6199 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6201 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6202 (tree-ssa-loop-ivopts.h): New header file.
6203 (struct builtin_info): New fields.
6204 (classify_builtin_1): Compute and record base and offset parts for
6205 memset builtin partition by calling strip_offset.
6206 (offset_cmp, fuse_memset_builtins): New functions.
6207 (finalize_partitions): Fuse adjacent memset partitions by calling
6209 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6210 Expose the interface.
6211 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6213 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6215 PR tree-optimization/82574
6216 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6217 that data reference must be executed exactly once per iteration
6218 against the outermost loop in nest.
6219 (classify_partition): Update call to above function.
6221 2017-10-18 Richard Biener <rguenther@suse.de>
6223 PR tree-optimization/82591
6224 * graphite.c (graphite_transform_loops): Move code gen message
6226 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6227 Here. Handle scop_to_isl_ast failing.
6228 (scop_to_isl_ast): Limit the number of ISL operations.
6230 2017-10-18 Richard Biener <rguenther@suse.de>
6232 * graphite-isl-ast-to-gimple.c
6233 (translate_isl_ast_to_gimple::set_rename): Simplify.
6234 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6235 (graphite_copy_stmts_from_block): ... here.
6236 (copy_bb_and_scalar_dependences): Simplify.
6237 (add_parameters_to_ivs_params): Canonicalize.
6238 (generate_entry_out_of_ssa_copies): Simplify.
6239 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6240 by passing in ISL dimension.
6241 (parameter_index_in_region_1): Rename to ...
6242 (parameter_index_in_region): ... this.
6243 (extract_affine): Adjust assert, pass down parameter index.
6244 (add_param_constraints): Use range-info when available.
6245 (build_scop_context): Adjust.
6246 * sese.c (new_sese_info): Adjust.
6247 (free_sese_info): Likewise.
6248 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6249 Remove unused typedefs.
6250 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6252 2017-10-18 Martin Liska <mliska@suse.cz>
6254 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6256 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6258 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6259 (s390_sched_init): Do not reset s390_sched_state if we entered the
6260 current basic block via a fallthru edge and all others are unlikely.
6262 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6264 * config/s390/s390.c (NUM_SIDES): New variable.
6265 (LONGRUNNING_THRESHOLD): New variable.
6266 (LATENCY_FACTOR): New variable.
6267 (s390_sched_score): Decrease score for long-running instructions on
6269 (s390_sched_variable_issue): Perform bookkeeping for long-running
6272 2017-10-18 Richard Biener <rguenther@suse.de>
6274 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6275 Simplify with removal of the parameter rename map.
6276 (set_rename): Likewise.
6277 (should_copy_to_new_region): Likewise.
6278 (graphite_copy_stmts_from_block): Likewise.
6279 (copy_bb_and_scalar_dependences): Remove initialization of
6280 unused copied_bb_map.
6282 (copy_internal_parameters): Likewise.
6283 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6284 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6285 Use INTEGRAL_TYPE_P.
6286 (parameter_index_in_region_1): Rename to ...
6287 (assign_parameter_index_in_region): ... this. Assert we have
6288 a parameter we handle.
6289 (scan_tree_for_params): Adjust.
6290 * sese.h (parameter_rename_map_t): Remove.
6291 (struct sese_info_t): Remove unused parameter_rename_map and
6292 copied_bb_map members.
6293 * sese.c (new_sese_info): Adjust.
6294 (free_sese_info): Likewise.
6296 2017-10-18 Martin Liska <mliska@suse.cz>
6299 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6300 on an abnormal edge.
6302 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6304 * doc/invoke.texi (ffunction-sections and fdata-sections):
6307 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6309 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6310 the use statement can throw internally.
6312 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6314 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6315 any RTX present on the RHS of a SET.
6316 * compare-elim.c (try_eliminate_compare): Restore comment.
6318 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6320 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6321 may be also called on tcc_type nodes.
6322 * langhooks.c (lhd_tree_size): Likewise.
6324 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6326 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6327 format_warning_at_substring.
6328 (maybe_warn): Convert source_range * param to a location_t. Pass
6329 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6330 (format_directive): Remove code to extract source_ranges and
6331 source_range * in favor of just a location_t.
6332 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6334 * substring-locations.c (format_warning_va): Convert
6335 source_range * param to a location_t.
6336 (format_warning_at_substring): Likewise.
6337 * substring-locations.h (format_warning_va): Likewise.
6338 (format_warning_at_substring): Likewise.
6340 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6342 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6344 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6345 and scatter/gather ops.
6347 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6348 vec_gather_load and vec_scatter_store.
6349 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6350 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6351 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6352 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6353 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6354 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6356 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6358 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6359 Detect FTST instruction and handle its register pops. Only pop
6360 second operand if can_pop_second_op is true.
6361 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6362 set can_pop_second_op to false in the compare_for_stack_reg call.
6364 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6365 output_fp_compare for stack register operands.
6366 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6367 instructions here. Do not emit stack register pops here. Assert
6368 that FCOMPP pops next to top stack register. Rewrite function.
6370 2017-10-17 Nathan Sidwell <nathan@acm.org>
6373 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6374 use DECL_ASSEMBLER_NAME_RAW.
6377 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6380 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6381 Wilco Dijkstra <wilco.dijkstra@arm.com>
6383 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6384 converstion to Pmode from the buf_addr.
6386 2017-10-17 Richard Biener <rguenther@suse.de>
6388 * graphite-dependences.c (scop_get_reads_and_writes): Change
6389 output parameters to references.
6391 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6393 PR 71026/tree-optimization
6394 * fold-const.c (distribute_real_division): Removed.
6395 (fold_binary_loc): Remove calls to distribute_real_divison.
6397 2017-10-17 Richard Biener <rguenther@suse.de>
6399 * graphite-scop-detection.c
6400 (scop_detection::stmt_has_simple_data_refs_p): Always use
6401 the full nest as region.
6402 (try_generate_gimple_bb): Likewise.
6403 * sese.c (scalar_evolution_in_region): Simplify now that
6404 SCEV can handle instantiation in regions.
6405 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6406 in the non-loop part of a function if requested.
6408 2017-10-17 Richard Biener <rguenther@suse.de>
6410 PR tree-optimization/82563
6411 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6413 (graphite_regenerate_ast_isl): Call it.
6414 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6416 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6418 PR tree-optimization/82549
6419 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6420 Formatting fixes. Instead of calling make_bit_field_ref with negative
6423 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6425 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6426 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6427 _mm_maskz_reduce_ss): New.
6428 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6429 __builtin_ia32_reducess_mask): Ditto..
6430 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6431 * config/i386/sse.md (reduces<mode>): Renamed to ...
6432 (reduces<mode><mask_scalar_name>): ... this.
6433 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6435 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6436 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6438 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6440 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6441 * selftest-run-tests.c (selftest::run_tests): Call
6442 selftest::unique_ptr_tests_cc_tests.
6443 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6444 * unique-ptr-tests.cc: New file.
6446 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6449 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6451 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6452 (make_hard_regno_born, make_hard_regno_dead): Update
6453 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6455 2017-10-16 Jeff Law <law@redhat.com>
6457 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6459 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6461 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6463 2017-10-16 Olivier Hainque <hainque@adacore.com>
6465 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6466 with_cpu if we were configured for an e500v2 target cpu name.
6468 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6470 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6471 * doc/invoke.texi: Document +nodsp as a valid extension for
6474 2017-10-16 Martin Liska <mliska@suse.cz>
6476 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6477 (test_set_range): Likewise.
6478 (test_range_functions): Rename to ...
6479 (test_bit_in_range): ... this.
6480 (sbitmap_c_tests): Add new test.
6482 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6484 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6486 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6487 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6489 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6491 * config/aarch64/aarch64-builtins.c
6492 (aarch64_types_quadopu_lane_qualifiers): New.
6493 (TYPES_QUADOPU_LANE): New.
6494 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6495 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6496 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6497 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6498 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6499 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6500 (Vdottype, DOTPROD): New.
6501 (sur): Add SDOT and UDOT.
6503 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6505 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6506 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6507 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6509 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6510 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6511 Enable TARGET_DOTPROD.
6512 (cortex-a75.cortex-a55): Likewise.
6513 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6515 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6517 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6518 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6520 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6522 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6523 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6524 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6525 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6526 * config/arm/types.md (neon_dot, neon_dot_q): New.
6527 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6529 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6531 * config/arm/arm.h (TARGET_DOTPROD): New.
6532 * config/arm/arm.c (arm_arch_dotprod): New.
6533 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6534 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6535 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6536 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6537 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6538 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6539 * doc/invoke.texi (armv8.2-a): Document dotprod
6541 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6543 * i386.c (ix86_vec_cost): New function.
6544 (ix86_rtx_costs): Handle vector operations better.
6545 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6546 * x86-tune-costs.h: Add new costs to all tables.
6548 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6550 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6552 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6553 divsd, sqrtss and sqrtsd
6554 * x86-tune-costs.h: Add new entries to all costs.
6555 (znver1_cost): Fix to match real instruction latencies.
6557 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6558 Michael Collison <michael.collison@arm.com>
6560 * compare-elim.c: Include emit-rtl.h.
6561 (can_merge_compare_into_arith): New function.
6562 (try_validate_parallel): Likewise.
6563 (try_merge_compare): Likewise.
6564 (try_eliminate_compare): Call the above when no previous clobber
6566 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6569 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6573 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6574 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6578 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6579 to be any operand_equal_p operands. For & (B - 1) require
6580 B to be power of 2. Recognize
6581 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6583 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6586 * optabs.c (expand_memory_blockage): Fix call of
6587 targetm.have_memory_blockage.
6589 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6592 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6593 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6594 objects to extra_objs instead of overwriting it.
6596 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6598 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6599 Use any_fp_register_operand as operand[3] predicate. Simplify
6600 equality test for operands[2] and operands[4] memory location.
6601 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6602 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6603 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6604 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6605 any_fp_register_operand as operand[1] predicate. Simplify
6606 equality test for operands[0] and operands[3] memory location.
6607 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6608 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6609 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6611 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6613 * target-insns.def: Add memory_blockage.
6614 * optabs.c (expand_memory_blockage): New function.
6615 (expand_asm_memory_barrier): Rename ...
6616 (expand_asm_memory_blockage): ... to this.
6617 (expand_mem_thread_fence): Call expand_memory_blockage
6618 instead of expand_asm_memory_barrier.
6619 (expand_mem_singnal_fence): Ditto.
6620 (expand_atomic_load): Ditto.
6621 (expand_atomic_store): Ditto.
6622 * doc/md.texi (Standard Pattern Names For Generation):
6623 Document memory_blockage instruction pattern.
6625 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6627 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6628 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6629 * config/rl78/rl78.md: New define_expand "adddi3".
6631 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6633 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6636 2017-10-13 Jeff Law <law@redhat.com>
6638 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6640 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6643 * internal-fn.c (expand_mul_overflow): If both operands have
6644 the same highpart of -1 or 0 and the topmost bit of lowpart
6645 is different, overflow is if res <= 0 rather than res < 0.
6647 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6649 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6650 TARGET_P9_VECTOR code for unaligned_load case.
6652 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6654 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6656 2017-10-13 Nathan Sidwell <nathan@acm.org>
6658 * tree-core.h (tree_contains_struct): Make bool.
6659 * tree.c (tree_contains_struct): Likewise.
6660 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6661 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6662 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6663 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6664 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6666 2017-10-13 Richard Biener <rguenther@suse.de>
6668 * graphite-isl-ast-to-gimple.c
6669 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6670 parameters and dominance check.
6671 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6672 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6673 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6674 Do not update SSA form here or do intermediate IL verification.
6675 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6676 (graphite_initialize): Remove check on the number of loops in
6677 the function and inline into graphite_transform_loops.
6678 (graphite_finalize): Inline into graphite_transform_loops.
6679 (graphite_transform_loops): Perform SSA update and IL verification
6681 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6683 2017-10-13 Richard Biener <rguenther@suse.de>
6685 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6686 graphite_expression_type_precision): Avoid global constructor
6688 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6689 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6690 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6691 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6692 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6693 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6695 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6698 * config/i386/i386.h (ix86_red_zone_size): New.
6699 * config/i386/i386.md (push peephole2s): Replace
6700 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6702 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6703 Alan Hayward <alan.hayward@arm.com>
6704 David Sherwood <david.sherwood@arm.com>
6706 * combine.c (can_change_dest_mode): Reject changes in
6707 REGMODE_NATURAL_SIZE.
6709 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6710 Alan Hayward <alan.hayward@arm.com>
6711 David Sherwood <david.sherwood@arm.com>
6713 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6714 (expand_debug_source_expr): Likewise.
6715 * combine.c (combine_simplify_rtx): Likewise.
6716 * cse.c (fold_rtx): Likewise.
6717 * fwprop.c (canonicalize_address): Likewise.
6718 * targhooks.c (default_shift_truncation_mask): Likewise.
6720 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6721 Alan Hayward <alan.hayward@arm.com>
6722 David Sherwood <david.sherwood@arm.com>
6724 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6725 (widened_mode): Likewise.
6726 (expand_unop): Likewise.
6727 * ree.c (transform_ifelse): Likewise.
6728 (merge_def_and_ext): Likewise.
6729 (combine_reaching_defs): Likewise.
6730 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6732 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6733 Alan Hayward <alan.hayward@arm.com>
6734 David Sherwood <david.sherwood@arm.com>
6736 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6737 * combine.c (gen_lowpart_for_combine): Likewise.
6738 * dwarf2out.c (rtl_for_decl_location): Likewise.
6739 * final.c (alter_subreg): Likewise.
6740 * rtlhooks.c (gen_lowpart_general): Likewise.
6741 (gen_lowpart_if_possible): Likewise.
6743 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6744 Alan Hayward <alan.hayward@arm.com>
6745 David Sherwood <david.sherwood@arm.com>
6747 * calls.c (expand_call): Use subreg_lowpart_offset.
6748 * cse.c (cse_insn): Likewise.
6749 * regcprop.c (copy_value): Likewise.
6750 (copyprop_hardreg_forward_1): Likewise.
6752 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6755 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6756 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6757 =Q constraints to +Q and into insn condition add check
6758 that operands[0] and operands[1] are equal.
6759 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6760 =Q constraints to +Q and into insn condition add check
6761 that operands[0] is equal to either operands[1] or operands[2].
6764 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6765 instead of handling MINUS_EXPR twice (once for each argument),
6766 canonicalize operand order and handle just once, use rtype where
6767 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6770 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6771 any values of __C while still being pattern recognizable as a simple
6774 2017-10-13 Richard Biener <rguenther@suse.de>
6776 PR tree-optimization/82451
6778 2017-10-02 Richard Biener <rguenther@suse.de>
6780 PR tree-optimization/82355
6781 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6782 a mapping for the enclosing loop but avoid generating one for
6784 (copy_bb_and_scalar_dependences): Remove premature codegen
6785 error on PHIs in blocks duplicated into multiple places.
6786 * graphite-scop-detection.c
6787 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6788 in the region use it as loop and nest to analyze the DR in.
6789 (try_generate_gimple_bb): Likewise.
6790 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6791 (add_loop_constraints): For blocks in a loop not in the region
6792 create a dimension with a single iteration.
6793 * sese.h (gbb_loop_at_index): Remove assert.
6795 * cfgloop.c (loop_preheader_edge): For the loop tree root
6796 return the single successor of the entry block.
6797 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6798 Reset the SCEV hashtable and niters.
6799 * graphite-scop-detection.c
6800 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
6801 assert that we only have POLYNOMIAL_CHREC that vary in loops
6802 contained in the region.
6803 (scop_detection::graphite_can_represent_expr): Adjust.
6804 (scop_detection::stmt_has_simple_data_refs_p): For loops
6805 not in the region set loop to NULL. The nest is now the
6806 entry edge to the region.
6807 (try_generate_gimple_bb): Likewise.
6808 * sese.c (scalar_evolution_in_region): Adjust for
6809 instantiate_scev change.
6810 * tree-data-ref.h (graphite_find_data_references_in_stmt):
6811 Make nest parameter the edge into the region.
6812 (create_data_ref): Likewise.
6813 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
6814 entry edge into a region and adjust instantiate_scev calls.
6815 (create_data_ref): Likewise.
6816 (graphite_find_data_references_in_stmt): Likewise.
6817 (find_data_references_in_stmt): Pass the loop preheader edge
6818 from the nest argument.
6819 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
6820 parameter the edge into the region.
6821 (instantiate_parameters): Use the loop preheader edge as entry.
6822 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
6824 (get_instantiated_value_entry): Make instantiate_below parameter
6825 the edge into the region.
6826 (instantiate_scev_name): Likewise. Adjust dominance checks,
6827 when we cannot use loop-based instantiation instantiate by
6828 walking use-def chains.
6829 (instantiate_scev_poly): Adjust.
6830 (instantiate_scev_binary): Likewise.
6831 (instantiate_scev_convert): Likewise.
6832 (instantiate_scev_not): Likewise.
6833 (instantiate_array_ref): Remove.
6834 (instantiate_scev_3): Likewise.
6835 (instantiate_scev_2): Likewise.
6836 (instantiate_scev_1): Likewise.
6837 (instantiate_scev_r): Do not blindly handle N-operand trees.
6838 Do not instantiate array-refs. Handle all constants and invariants.
6839 (instantiate_scev): Make instantiate_below parameter
6840 the edge into the region.
6841 (resolve_mixers): Use the loop preheader edge for the region
6842 parameter to instantiate_scev_r.
6843 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6845 2017-10-13 Richard Biener <rguenther@suse.de>
6847 PR tree-optimization/82525
6848 * graphite-isl-ast-to-gimple.c
6849 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6851 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6852 Fail code generation when we cannot represent the isl integer.
6853 (binary_op_to_tree): Elide modulo operations that are no-ops
6854 in the type we code generate. Remove now superfluous code
6857 2017-10-13 Richard Biener <rguenther@suse.de>
6859 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6860 (scop_detection::harmful_loop_in_region): Remove premature
6861 IV type restriction.
6862 (scop_detection::graphite_can_represent_scev): We can handle
6863 pointer IVs just fine.
6865 2017-10-13 Alan Modra <amodra@gmail.com>
6867 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6868 "Clobbers and Scratch Registers". Add paragraph on
6869 alternative to clobbers for scratch registers and OpenBLAS
6872 2017-10-13 Alan Modra <amodra@gmail.com>
6874 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6875 example of a memory input for a string of known length. Move
6876 commentary out of table. Add a number of new examples
6877 covering array memory inputs.
6879 2017-10-12 Martin Liska <mliska@suse.cz>
6881 PR tree-optimization/82493
6882 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6883 (test_range_functions): New function.
6884 (sbitmap_c_tests): Likewise.
6885 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6886 * selftest.h (sbitmap_c_tests): New function.
6888 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6890 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6892 * config/rs6000/amo.h: Fix spacing issue.
6894 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6897 * config/i386/i386.md (*ashl<mode>3_mask_1,
6898 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6899 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6902 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6904 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6905 (profile_probability): Set max_probability
6906 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6908 * profile-count.c (profile_probability::differs_from_p): Do not
6909 rely on max_probaiblity == 10000
6911 2017-10-12 Jeff Law <law@redhat.com>
6913 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6916 2017-10-12 Martin Sebor <msebor@redhat.com>
6920 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6921 (handle_alias_pairs): Call it.
6922 * common.opt (-Wattribute-alias): New option.
6923 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6924 * doc/invoke.texi (-Wattribute-alias): Document.
6926 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6929 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6931 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6933 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6934 (make_hard_regno_born, make_hard_regno_dead): Update
6935 bb_killed_pseudos and bb_gen_pseudos.
6937 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6939 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6941 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6943 * config/alpha/alpha.c (alpha_split_conditional_move):
6944 Use std::swap instead of manually swapping.
6945 (alpha_stdarg_optimize_hook): Ditto.
6946 (alpha_canonicalize_comparison): Ditto.
6948 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6950 * tree-loop-distribution.c (struct builtin_info): New struct.
6951 (struct partition): Refactor fields into struct builtin_info.
6952 (partition_free): Free struct builtin_info.
6953 (build_size_arg_loc, build_addr_arg_loc): Delete.
6954 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6955 information from struct builtin_info.
6956 (find_single_drs): New function refactored from classify_partition.
6957 Also moved builtin validity checks to this function.
6958 (compute_access_range, alloc_builtin): New functions.
6959 (classify_builtin_st, classify_builtin_ldst): New functions.
6960 (classify_partition): Refactor code into functions find_single_drs,
6961 classify_builtin_st and classify_builtin_ldst.
6962 (distribute_loop): Don't do runtime alias check when distributing
6964 (find_seed_stmts_for_distribution): New function.
6965 (pass_loop_distribution::execute): Refactor code finding seed
6966 stmts into above function. Support distribution for the innermost
6967 two-level loop nest. Adjust dump information.
6969 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6971 * tree-loop-distribution.c: Adjust the general comment.
6972 (NUM_PARTITION_THRESHOLD): New macro.
6973 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6974 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6975 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6976 in call to build_partition_graph.
6977 (finalize_partitions): New parameter. Make loop distribution more
6978 conservative by fusing more partitions.
6979 (distribute_loop): Don't do runtime alias check in case of loop nest
6981 (find_seed_stmts_for_distribution): New function.
6982 (prepare_perfect_loop_nest): New function.
6983 (pass_loop_distribution::execute): Refactor code finding seed stmts
6984 and loop nest into above functions. Support loop nest distribution.
6985 Adjust dump information accordingly.
6987 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6989 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6990 and set PTYPE_SEQUENTIAL for merged partition.
6992 2017-10-12 Richard Biener <rguenther@suse.de>
6994 PR tree-optimization/69728
6996 2017-09-19 Richard Biener <rguenther@suse.de>
6998 PR tree-optimization/69728
6999 * graphite-sese-to-poly.c (schedule_error): New global.
7000 (add_loop_schedule): Handle empty domain by failing the
7002 (build_original_schedule): Handle schedule_error.
7004 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7005 domain by returning an unchanged schedule.
7007 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7009 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7010 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7012 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7014 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7017 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7020 * expr.c (store_field): Don't optimize away bitsize == 0 store
7021 from CALL_EXPR with addressable return type.
7023 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7025 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7026 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7027 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7028 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7029 TARGET_ISEL instead of TARGET_ISEL<sel>.
7031 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7033 * config/rs6000/rs6000.c
7034 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7036 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7038 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7040 (reg_or_cint_operand): Fix comment.
7041 (reg_or_zero_operand): New predicate.
7042 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7043 * config/rs6000/rs6000.c (output_isel): Delete.
7044 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7045 instead of reg_or_cint_operand. Output instruction directly (not via
7047 (isel_unsigned_<mode>): Ditto.
7048 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7049 gpc_reg_operand. Add an instruction alternative for this. Output
7050 instruction directly.
7051 (*isel_reversed_unsigned_<mode>): Ditto.
7053 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7055 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7056 (TARGET_CANONICALIZE_COMPARISON): Define.
7058 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7061 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7062 Check whether the dest is REG before adding REG_EQUIV note.
7064 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7067 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7069 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7070 (make_hard_regno_born, make_hard_regno_dead): Update
7071 bb_killed_pseudos and bb_gen_pseudos.
7073 2017-10-11 Nathan Sidwell <nathan@acm.org>
7075 * incpath.h (enum incpath_kind): Name enum, prefix values.
7076 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7077 * incpath.c (heads, tails): Use INC_MAX.
7078 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7079 (merge_include_chains, split_quote_chain,
7080 register_include_chains): Update incpath_kind names.
7081 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7082 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7084 (add_framework_path, darwin_register_objc_includes): Likewise.
7085 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7087 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7089 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7090 Do not use float_operator operator predicate.
7091 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7092 * config/i386/predicates.md (float_operator): Remove predicate.
7094 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7096 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7097 (*jccxf_i387): Ditto.
7098 (*jcc<mode>_i387): Ditto.
7099 (*jccu<mode>_i387): Ditto.
7100 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7101 (*jcc_*_i387 splitters): Remove.
7102 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7103 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7104 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7107 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7109 * profile-count.h (slow_safe_scale_64bit): New function.
7110 (safe_scale_64bit): New inline.
7111 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7112 * profile-count.c: Include wide-int.h
7113 (slow_safe_scale_64bit): New.
7115 2017-10-11 Nathan Sidwell <nathan@acm.org>
7117 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7118 HAS_DECL_ASSEMBLER_NAME_P.
7119 * gimple-expr.c (gimple_decl_printable_name: Check
7120 HAS_DECL_ASSEMBLER_NAME_P too.
7121 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7122 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7123 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7124 * passes.c (rest_of_decl_compilation): Check
7125 HAS_DECL_ASSEMBLER_NAME_P too.
7126 * recog.c (verify_changes): Likewise.
7127 * tree-pretty-print.c (dump_decl_name): Likewise.
7128 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7130 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7131 (SET_DECL_ASSEMBLER_NAME): Use it.
7132 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7133 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7134 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7136 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7138 * config.gcc (i386, x86_64): Add extra objects.
7139 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7140 (ix86_min_insn_size): Declare.
7141 (ix86_issue_rate): Declare.
7142 (ix86_adjust_cost): Declare.
7143 (ia32_multipass_dfa_lookahead): Declare.
7144 (ix86_macro_fusion_p): Declare.
7145 (ix86_macro_fusion_pair_p): Declare.
7146 (ix86_bd_has_dispatch): Declare.
7147 (ix86_bd_do_dispatch): Declare.
7148 (ix86_core2i7_init_hooks): Declare.
7149 (ix86_atom_sched_reorder): Declare.
7150 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7151 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7152 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7153 (rip_relative_addr_p): Rename to ...
7154 (ix86_rip_relative_addr_p): ... this one; export.
7155 (memory_address_length): Update.
7156 (ix86_issue_rate): Move to x86-tune-sched.c.
7157 (ix86_flags_dependent): Move to x86-tune-sched.c.
7158 (ix86_agi_dependent): Move to x86-tune-sched.c.
7159 (exact_dependency_1): Move to x86-tune-sched.c.
7160 (exact_store_load_dependency): Move to x86-tune-sched.c.
7161 (ix86_adjust_cost): Move to x86-tune-sched.c.
7162 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7163 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7164 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7165 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7166 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7167 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7168 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7169 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7170 (min_insn_size): Rename to ...
7171 (ix86_min_insn_size): ... this one; export.
7172 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7173 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7174 (core2i7_first_cycle_multipass_backtrack): Move to
7175 x86-tune-sched-core.c.
7176 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7177 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7178 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7179 (ix86_avoid_jump_mispredicts): Update.
7180 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7181 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7182 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7183 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7184 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7185 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7186 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7187 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7188 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7189 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7190 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7191 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7192 (BIG): Move to ix86-tune-sched-bd.c.
7193 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7194 (enum insn_path): Move to ix86-tune-sched-bd.c.
7195 (get_mem_group): Move to ix86-tune-sched-bd.c.
7196 (is_cmp): Move to ix86-tune-sched-bd.c.
7197 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7198 (is_branch): Move to ix86-tune-sched-bd.c.
7199 (is_prefetch): Move to ix86-tune-sched-bd.c.
7200 (init_window): Move to ix86-tune-sched-bd.c.
7201 (allocate_window): Move to ix86-tune-sched-bd.c.
7202 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7203 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7204 (process_end_window): Move to ix86-tune-sched-bd.c.
7205 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7206 (find_constant): Move to ix86-tune-sched-bd.c.
7207 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7208 (has_immediate): Move to ix86-tune-sched-bd.c.
7209 (get_insn_path): Move to ix86-tune-sched-bd.c.
7210 (get_insn_group): Move to ix86-tune-sched-bd.c.
7211 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7212 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7213 (add_insn_window): Move to ix86-tune-sched-bd.c.
7214 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7215 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7216 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7217 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7218 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7219 (do_dispatch): Move to ix86-tune-sched-bd.c.
7220 (has_dispatch): Move to ix86-tune-sched-bd.c.
7221 * config/i386/t-i386: Add new object files.
7222 * config/i386/x86-tune-costs.h: New file.
7223 * config/i386/x86-tune-sched-atom.c: New file.
7224 * config/i386/x86-tune-sched-bd.c: New file.
7225 * config/i386/x86-tune-sched-core.c: New file.
7226 * config/i386/x86-tune-sched.c: New file.
7228 2017-10-11 Liu Hao <lh_mouse@126.com>
7230 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7231 the generic version below instead.
7232 (should_colorize): Recognize Windows consoles as terminals
7234 * pretty-print.c [__MINGW32__] (write_all): New function.
7235 [__MINGW32__] (find_esc_head): Likewise.
7236 [__MINGW32__] (find_esc_terminator): Likewise.
7237 [__MINGW32__] (eat_esc_sequence): Likewise.
7238 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7240 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7243 2017-10-11 Richard Biener <rguenther@suse.de>
7245 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7246 Properly call analyze_scalar_evolution with the loop of the stmt.
7248 2017-10-11 Richard Biener <rguenther@suse.de>
7250 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7251 * tree-core.h (tree_base): Add chrec_var union member.
7252 * tree.h (CHREC_VAR): Remove.
7253 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7254 * tree-chrec.h (build_polynomial_chrec): Adjust.
7255 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7256 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7258 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7260 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7261 * match.pd: ... here.
7262 ((T) X == (T) Y): Relax condition.
7264 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7266 PR tree-optimization/82472
7267 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7269 (break_alias_scc_partitions): Update postorder number.
7271 2017-10-11 Martin Liska <mliska@suse.cz>
7274 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7276 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7278 2017-10-11 Martin Liska <mliska@suse.cz>
7280 * print-rtl.c (print_insn): Move declaration of idbuf
7281 to same scope as name.
7283 2017-10-11 Martin Liska <mliska@suse.cz>
7288 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7291 2017-10-11 Martin Liska <mliska@suse.cz>
7294 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7297 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7299 * config.gcc (default_gnu_indirect_function): Default to yes for
7300 arm*-*-linux* with glibc.
7302 2017-10-11 Richard Biener <rguenther@suse.de>
7304 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7305 default-defs and types we do not want to analyze.
7306 (interpret_loop_phi): Replace unreachable code with an assert.
7307 (compute_scalar_evolution_in_loop): Remove and inline ...
7308 (analyze_scalar_evolution_1): ... here, replacing condition with
7309 what makes the intent clearer. Remove handling of cases
7310 get_scalar_evolution now handles.
7312 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7314 PR rtl-optimization/81434
7315 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7316 comment for main loop. In sched_group_found if, also add checks for
7317 pass and min_cost_group.
7319 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7321 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7322 (rs6000_insn_cost): New function.
7323 * config/rs6000/rs6000.md (cost): New attribute.
7325 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7326 H.J. Lu <hongjiu.lu@intel.com>
7330 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7331 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7332 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7333 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7334 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7335 OPTION_MASK_ISA_MMX and some other bit set require both
7336 mmx and the ISAs without the mmx bit.
7337 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7338 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7339 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7340 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7341 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7342 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7343 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7344 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7345 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7346 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7347 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7348 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7349 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7350 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7351 Add OPTION_MASK_ISA_MMX.
7353 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7355 * config.gcc (armv7*-*-freebsd*): New target.
7356 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7358 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7360 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7361 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7364 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7366 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7367 a template parameter.
7368 (WIDE_INT_REF_FOR): Update accordingly.
7369 * tree.h (wi::int_traits <const_tree>): Delete.
7370 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7371 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7372 (wi::tree_to_wide_ref): New typedef.
7373 (wi::to_wide): New function.
7374 * calls.c (get_size_range): Use wi::to_wide when operating on
7376 * cgraph.c (cgraph_node::create_thunk): Likewise.
7377 * config/i386/i386.c (ix86_data_alignment): Likewise.
7378 (ix86_local_alignment): Likewise.
7379 * dbxout.c (stabstr_O): Likewise.
7380 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7381 * expr.c (const_vector_from_tree): Likewise.
7382 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7383 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7384 (fold_negate_expr_1, int_const_binop_1, const_binop)
7385 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7386 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7387 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7388 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7389 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7390 (fold_not_const, round_up_loc): Likewise.
7391 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7392 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7393 (alloca_call_type): Likewise.
7394 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7395 * godump.c (go_output_typedef): Likewise.
7396 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7397 * internal-fn.c (get_min_precision): Likewise.
7398 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7399 * ipa-polymorphic-call.c
7400 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7401 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7402 (ipa_modify_call_arguments): Likewise.
7403 * match.pd: Likewise.
7404 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7405 * print-tree.c (print_node_brief, print_node): Likewise.
7406 * stmt.c (expand_case): Likewise.
7407 * stor-layout.c (layout_type): Likewise.
7408 * tree-affine.c (tree_to_aff_combination): Likewise.
7409 * tree-cfg.c (group_case_labels_stmt): Likewise.
7410 * tree-data-ref.c (dr_analyze_indices): Likewise.
7411 (prune_runtime_alias_test_list): Likewise.
7412 * tree-dump.c (dequeue_and_dump): Likewise.
7413 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7414 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7415 * tree-pretty-print.c (dump_generic_node): Likewise.
7416 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7417 (simple_iv_with_niters): Likewise.
7418 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7419 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7420 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7421 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7422 (refine_value_range_using_guard, number_of_iterations_ne_max)
7423 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7424 (get_cst_init_from_scev, record_nonwrapping_iv)
7425 (scev_var_range_cant_overflow): Likewise.
7426 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7427 * tree-ssa-pre.c (compute_avail): Likewise.
7428 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7429 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7430 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7431 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7432 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7433 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7434 (dump_case_nodes, try_switch_expansion): Likewise.
7435 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7436 (vect_do_peeling): Likewise.
7437 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7438 * tree-vect-stmts.c (vectorizable_load): Likewise.
7439 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7440 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7441 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7442 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7443 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7444 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7445 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7446 (evrp_dom_walker::before_dom_children): Likewise.
7447 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7448 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7449 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7450 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7451 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7452 (get_range_pos_neg): Likewise.
7453 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7454 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7455 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7456 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7457 * config/avr/avr.c (avr_fold_builtin): Likewise.
7458 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7459 * config/msp430/msp430.c (msp430_attr): Likewise.
7460 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7461 * config/powerpcspe/powerpcspe-c.c
7462 (altivec_resolve_overloaded_builtin): Likewise.
7463 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7464 (rs6000_expand_ternop_builtin): Likewise.
7465 * config/rs6000/rs6000-c.c
7466 (altivec_resolve_overloaded_builtin): Likewise.
7467 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7468 (rs6000_expand_ternop_builtin): Likewise.
7469 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7471 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7473 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7474 when copying loop nest with only one inner loop.
7476 2017-10-10 Richard Biener <rguenther@suse.de>
7478 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7479 blocks if SCEV is active.
7480 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7482 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7483 (scev_initialize): Assert we are not yet initialized.
7485 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7487 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7488 inner loop's exit stmt by making it always exit the loop, otherwise
7489 we would generate an infinite empty loop.
7491 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7493 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7494 renaming variables in new preheader if it's deleted.
7496 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7498 * tree-loop-distribution.c (struct partition): Remove unused field
7499 loops of the structure.
7500 (partition_alloc, partition_free): Ditto.
7501 (build_rdg_partition_for_vertex): Ditto.
7503 2017-10-09 Jeff Law <law@redhat.com>
7505 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7506 return type to match prototype and documentation.
7508 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7510 * config/rs6000/rs6000.c (processor_costs): Move to ...
7511 * config/rs6000/rs6000.h: ... here.
7512 (rs6000_cost): Declare.
7514 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7516 * except.c (setjmp_fn): New global variable.
7517 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7518 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7519 if DONT_USE_BUILTIN_SETJMP is defined.
7521 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7523 * target.def (insn_cost): New hook.
7524 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7525 * doc/tm.texi: Regenerate.
7526 * rtlanal.c (insn_cost): Use the new hook.
7528 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7530 * combine.c (combine_validate_cost): Compute the new insn_cost,
7531 not just pattern_cost.
7532 (try_combine): Adjust comment.
7534 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7536 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7538 * combine.c (uid_insn_cost): Adjust comment.
7539 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7541 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7542 * dse.c (find_shift_sequence): Ditto.
7543 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7544 (bb_valid_for_noce_process_p): Use pattern_cost.
7545 * rtl.h (insn_rtx_cost): Delete.
7546 (pattern_cost): New prototype.
7547 (insn_cost): New prototype.
7548 * rtlanal.c (insn_rtx_cost): Rename to...
7549 (pattern_cost): ... this.
7552 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7554 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7555 (*jcc<mode>_0_r_i387): Ditto.
7556 (*jccxf_r_i387): Ditto.
7557 (*jcc<mode>_r_i387): Ditto.
7558 (*jccu<mode>_r_i387): Ditto.
7559 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7560 (*jcc): Rename from *jcc_1.
7562 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7564 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7565 deferred rescans after the lvx/stvx recombination pre-pass.
7567 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7569 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7570 memory operation instruction support.
7571 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7572 (rs6000-ibm-aix[789]*): Likewise.
7573 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7574 Document new functions.
7576 2017-10-09 Richard Biener <rguenther@suse.de>
7578 PR tree-optimization/82397
7579 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7580 equality only for semantically equal trees.
7582 2017-10-09 Richard Biener <rguenther@suse.de>
7584 PR tree-optimization/82449
7585 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7586 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7587 allow constant addresses.
7588 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7591 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7593 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7596 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7599 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7602 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7605 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7607 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7610 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7611 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7613 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7615 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7616 (WI_BINARY_PREDICATE_RESULT): Likewise.
7617 (wi::binary_traits::operator_result): New type.
7618 (wi::binary_traits::predicate_result): Likewise.
7619 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7620 (generic_wide_int::operator==, generic_wide_int::operator!=)
7621 (generic_wide_int::operator&, generic_wide_int::and_not)
7622 (generic_wide_int::operator|, generic_wide_int::or_not)
7623 (generic_wide_int::operator^, generic_wide_int::operator+
7624 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7626 (operator~, unary operator-, operator==, operator!=, operator&)
7627 (operator|, operator^, operator+, binary operator-, operator*): New
7629 * expr.c (get_inner_reference): Use wi::bit_and_not.
7630 * fold-const.c (fold_binary_loc): Likewise.
7631 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7632 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7633 (bit_value_binop): Likewise.
7634 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7635 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7636 (extract_range_from_binary_expr_1): Likewise.
7637 (masked_increment): Likewise.
7638 (simplify_bit_ops_using_ranges): Likewise.
7640 2017-10-09 Martin Jambor <mjambor@suse.cz>
7643 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7644 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7645 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7647 (hsa_fixup_mov_insn_type): New function.
7648 (hsa_op_with_type::get_in_type): Use it.
7649 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7650 immediates in an assert.
7651 (hsa_op_with_type::extend_int_to_32bit): New method.
7652 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7653 types. Convert to dest type if necessary.
7654 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7655 (reg_for_gimple_ssa): Pass false as min32int to
7656 hsa_type_for_scalar_tree_type.
7657 (gen_hsa_addr): Fixup type when creating addresable temporary.
7658 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7659 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7660 necessary. Call hsa_fixup_mov_insn_type.
7661 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7662 extend operands and convert to dest type if necessary.
7663 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7664 to dest type if necessary.
7665 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7666 if conversion nt necessary and size matches.
7667 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7668 to dest type if necessary.
7669 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7670 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7672 (gen_hsa_clrsb): Likewise.
7673 (gen_hsa_ffs): Likewise.
7674 (gen_hsa_divmod): Extend operands and convert to dest type if
7676 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7678 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7680 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7681 Remove empty default arguments. Use a brace block as output
7683 (conditional return): Ditto.
7685 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7686 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7688 (group_ending_nop): Ditto.
7689 (doloop_end): Ditto.
7690 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7691 (splitters for those): Ditto.
7693 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7695 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7696 a conditional jump (and the compare for it) so that pc_rtx is the
7698 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7699 for the deleted and renamed ctr<mode>_internal[234] patterns.
7700 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7701 Delete second conditional return pattern.
7702 (ctr<mode>_internal2): Delete this second bdnz pattern.
7703 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7704 (ctr<mode>_internal4): Delete this second bdz pattern.
7706 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7708 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7709 (always_initialized_rtx_for_ssa_name_p): New predicate.
7710 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7711 (finish_out_of_ssa): Free new field of SA.
7712 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7713 * tree-ssa-coalesce.c: Include tree-ssa.h.
7714 (get_parm_default_def_partitions): Remove extern keyword.
7715 (get_undefined_value_partitions): New function.
7716 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7717 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7719 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7721 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7723 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7725 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7727 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7728 for targets that preffer 128bit.
7730 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7732 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7734 2017-10-08 Olivier Hainque <hainque@adacore.com>
7736 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7737 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7739 (thumb_set_return_address): Likewise.
7741 2017-10-08 Olivier Hainque <hainque@adacore.com>
7743 * common/config/arm/arm-common.c (arm_except_unwind_info):
7744 Handle DWARF2_UNWIND_INFO.
7746 2017-10-07 Michael Collison <michael.collison@arm.com>
7748 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7751 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7753 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7754 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7755 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7756 defined, force the creation of a new block for a dispatch label.
7758 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7760 * invoke.texi (Wsuggest-attribute=cold): Document.
7761 * common.opt (Wsuggest-attribute=cold): New
7762 * ipa-pure-const.c (warn_function_cold): New function.
7763 * predict.c (compute_function_frequency): Use it.
7764 * predict.h (warn_function_cold): Declare.
7766 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7768 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7771 2017-10-06 Martin Liska <mliska@suse.cz>
7773 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7774 keyword for member functions.
7775 (struct sanopt_tree_couple): New struct.
7776 (struct sanopt_tree_couple_hash): New function.
7777 (struct sanopt_ctx): Add new hash_map.
7778 (has_dominating_ubsan_ptr_check): New function.
7779 (record_ubsan_ptr_check_stmt): Likewise.
7780 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7781 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7782 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7784 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7787 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7788 aarch64_simd_valid_immediate on CONST_VECTORs.
7789 (aarch64_reg_or_bic_imm): Likewise.
7791 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7793 PR rtl-optimization/82396
7794 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7796 2017-10-06 Sebastian Pop <sebpop@gmail.com>
7798 * graphite-dependences.c (scop_get_reads): Move code to...
7799 (scop_get_must_writes): Move code to...
7800 (scop_get_may_writes): Move code to...
7801 (scop_get_reads_and_writes): ... here.
7802 (scop_get_dependences): Call scop_get_reads_and_writes.
7804 2017-10-06 Jakub Jelinek <jakub@redhat.com>
7806 PR tree-optimization/82434
7807 * fold-const.h (can_native_encode_type_p,
7808 can_native_encode_string_p): Remove.
7809 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
7810 don't encode anything, just return what would be otherwise returned.
7811 (native_encode_fixed, native_encode_complex, native_encode_vector):
7813 (native_encode_string): Likewise. Inline by hand
7814 can_native_encode_string_p.
7815 (can_native_encode_type_p): Remove.
7816 (can_native_encode_string_p): Remove.
7817 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
7818 STRING_CSTs using can_native_encode_string_p, test all
7819 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
7820 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
7821 argument from native_encode_expr.
7822 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
7823 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
7826 2017-10-06 Richard Biener <rguenther@suse.de>
7828 PR tree-optimization/82397
7829 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
7830 operand_equal_p but rely on data_ref_compare_tree for detecting
7832 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7833 to match up with dr_group_sort_cmp.
7835 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7838 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7840 * config/s390/s390-builtin-types.def: Regenerate.
7842 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7845 * config/s390/s390-builtin-types.def: Regenerate.
7846 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7847 Change flag from B_VXE to B_VX.
7848 (s390_vec_min_dbl): Remove B_VXE flag.
7850 2017-10-06 Richard Biener <rguenther@suse.de>
7852 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7853 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7854 translate_isl_ast_to_gimple::is_valid_rename,
7855 translate_isl_ast_to_gimple::get_rename,
7856 translate_isl_ast_to_gimple::get_def_bb_for_const,
7857 translate_isl_ast_to_gimple::get_new_name,
7858 translate_isl_ast_to_gimple::collect_all_ssa_names,
7859 translate_isl_ast_to_gimple::copy_loop_phi_args,
7860 translate_isl_ast_to_gimple::collect_all_ssa_names,
7861 translate_isl_ast_to_gimple::copy_loop_phi_args,
7862 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7863 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7864 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7865 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7866 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7867 translate_isl_ast_to_gimple::copy_cond_phi_args,
7868 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7869 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7870 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7871 translate_isl_ast_to_gimple::rename_uses,
7872 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7873 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7874 (set_rename_for_each_def): Likewise.
7875 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7876 here. Handle rewriting SCEV analyzable uses here.
7877 (copy_bb_and_scalar_dependences): Generate code for PHI
7879 (graphite_regenerate_ast_isl): Adjust.
7880 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7881 (add_write, add_read): New functions.
7882 (build_cross_bb_scalars_def): Use it and simplify.
7883 (build_cross_bb_scalars_use): Likewise.
7884 (graphite_find_cross_bb_scalar_vars): Inline into...
7885 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7886 simulating out-of-SSA. Compute liveout and add dependencies.
7887 (build_scops): Force an empty entry block.
7888 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7890 (sese_build_liveouts): Declare.
7891 (sese_trivially_empty_bb_p): Likewise.
7892 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7893 compute liveout and debug_liveout.
7894 (sese_bad_liveouts_use): Remove.
7895 (sese_reset_debug_liveouts_bb): Likewise.
7896 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7897 (sese_build_liveouts): Build liveout and debug_liveout and store
7899 (new_sese_info): Adjust.
7900 (free_sese_info): Likewise.
7901 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7902 do not build liveout here.
7903 (move_sese_in_condition): Adjust region entry.
7904 (scev_analyzable_p): Match up with chrec_apply requirements.
7905 (sese_trivially_empty_bb_p): New.
7906 * tree-into-ssa.c (get_reaching_def): Properly support generating
7907 default-defs for incremental rewrite of anonymous names.
7909 2017-10-06 Richard Biener <rguenther@suse.de>
7911 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7912 precision do not perform modulo reduction.
7914 2017-10-06 Richard Biener <rguenther@suse.de>
7916 PR tree-optimization/82436
7917 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7918 conservatively choose the vectorization factor when checking
7919 whether we can perform the required load permutation.
7920 (vect_transform_slp_perm_load): Assert when we may not fail.
7922 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7924 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7925 message for incompatible -msdata=* and -mcall-* options.
7927 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7929 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7930 rate for post-reload scheduling.
7932 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7934 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7936 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7938 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7939 to improve monte carlo in scimark.
7941 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7943 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7944 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7945 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7946 pentium4_cost, nocona_cost): Set reassociation width to 1.
7947 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7948 width to 2 for fp operations and 1 otherwise.
7949 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7951 (atom_cost): Set reassociation width to 2.
7952 (slm_cost, generic_cost): Set fp reassociation width
7953 to 2 and 1 otherwise.
7954 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7955 (core_cost): Set fp reassociation width to 4 and vector to 2.
7956 (ix86_reassociation_width): Rewrite using cost table; special case
7957 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7958 and TARGET_AVX128_OPTIMAL.
7959 * config/i386/i386.h (processor_costs): Add
7960 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7961 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7962 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7963 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7964 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7965 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7967 2017-10-05 Nathan Sidwell <nathan@acm.org>
7969 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7971 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7973 * config/arm/arm.c (arm_test_fpu_data): New.
7974 (arm_run_selftests): Call arm_test_fpu_data.
7976 2017-10-04 Nathan Sidwell <nathan@acm.org>
7978 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7980 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7982 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7984 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7985 check type for aarch64_simd_valid_immediate.
7986 (aarch64_output_simd_mov_immediate): Update prototype.
7987 (aarch64_simd_valid_immediate): Update prototype.
7988 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7989 support for ORR-immediate.
7990 (and<mode>3): modified pattern to add support for BIC-immediate.
7991 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7992 now checks for valid immediate for BIC and ORR based on new enum
7994 (aarch64_output_simd_mov_immediate): Function now used to output
7995 BIC/ORR imm as well based on new enum argument.
7996 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7998 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7999 (aarch64_reg_or_bic_imm): Likewise.
8001 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8003 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8004 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8006 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8010 PR rtl-optimization/82396
8011 * haifa-sched.c (autopref_multipass_init): Simplify
8013 (autopref_rank_data): Simplify sort order.
8014 * sched-int.h (autopref_multipass_data_): Remove
8015 multi_mem_insn_p, min_offset and max_offset.
8017 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8019 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8021 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8023 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8026 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8028 * doc/sourcebuild.texi: Document vect_long_mult.
8030 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8032 PR tree-optimization/82413
8033 * fold-const.c (build_range_check): Use widest_int when comparing
8034 the maximum ETYPE value with HIGH.
8036 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8038 PR rtl-optimization/82396
8039 * haifa-sched.c (autopref_multipass_init): Simplify
8041 (autopref_rank_data): Simplify sort order.
8042 * sched-int.h (autopref_multipass_data_): Remove
8043 multi_mem_insn_p, min_offset and max_offset.
8045 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8047 PR tree-optimization/82381
8048 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8049 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8052 PR tree-optimization/82374
8053 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8054 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8055 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8056 current_function_decl to the new decl.
8058 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8060 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8061 helper macro for IEEE float128 hardware built-in functions.
8062 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8064 (TRUNCF128_ODD): Likewise.
8065 (ADDF128_ODD): Likewise.
8066 (SUBF128_ODD): Likewise.
8067 (MULF128_ODD): Likewise.
8068 (DIVF128_ODD): Likewise.
8069 (FMAF128_ODD): Likewise.
8070 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8071 UNSPEC_TRUNC_ROUND_TO_ODD.
8072 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8073 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8074 floating point round to odd instructions.
8075 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8076 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8077 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8078 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8079 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8080 (trunc<mode>sf2_hw): Change the truncate with round to odd
8081 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8082 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8083 to odd hardware instructions.
8084 (sub<mode>3_odd): Likewise.
8085 (mul<mode>3_odd): Likewise.
8086 (div<mode>3_odd): Likewise.
8087 (sqrt<mode>2_odd): Likewise.
8088 (fma<mode>4_odd): Likewise.
8089 (fms<mode>4_odd): Likewise.
8090 (nfma<mode>4_odd): Likewise.
8091 (nfms<mode>4_odd): Likewise.
8092 (trunc<mode>df2_odd): Change the truncate with round to odd
8093 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8095 * doc/extend.texi (PowerPC built-in functions): Update documentation
8096 for existing IEEE float128-bit built-in functions. Add built-in
8097 functions that generate the IEEE 128-bit floating point round to
8100 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8102 PR rtl-optimization/77729
8103 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8104 to (X&(C1&~C2))|C2 transformations.
8106 2017-10-03 Martin Jambor <mjambor@suse.cz>
8108 PR tree-optimization/82363
8109 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8110 mismatch, mark lacc written regardless of racc.
8112 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8114 PR tree-optimization/82381
8115 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8116 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8117 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8118 fallthrough into reassoc_stmt_dominates_stmt_p.
8121 * combine.c (combine_instructions): Don't combine in unreachable
8124 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8127 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8128 function to not use the have_cpu variable. Do not set cpu_index,
8129 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8131 (rs6000_valid_attribute_p): Remove duplicate initializations of
8132 old_optimize and func_optimize.
8133 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8134 (rs6000_activate_target_options): Make global.
8135 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8138 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8140 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8141 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8142 Set *poffset to 0 and *psize and *pmax_size to -1 if
8143 *poffset + *psize overflows in HOST_WIDE_INT.
8145 PR tree-optimization/82387
8146 PR tree-optimization/82388
8147 PR tree-optimization/82389
8148 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8149 instead of live_bytes non-NULL.
8151 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8154 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8157 2017-10-02 Richard Biener <rguenther@suse.de>
8159 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8160 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8161 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8163 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8165 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8166 requested precision matches the type's.
8167 * calls.c (alloc_max_size): Calculate the new candidate size as
8168 a widest_int and use wi::to_widest when comparing it with the
8169 current candidate size.
8170 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8171 zero rather than integer_zero_node.
8172 * match.pd: Check for a no-op conversion before using wi::add
8173 rather than after. Use tree_to_uhwi when summing small shift
8174 counts into an unsigned int.
8176 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8177 Alan Hayward <alan.hayward@arm.com>
8178 David Sherwood <david.sherwood@arm.com>
8181 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8182 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8183 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8184 POINTER_AND_FP_REGS.
8186 2017-10-02 Richard Biener <rguenther@suse.de>
8188 PR tree-optimization/82355
8189 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8190 a mapping for the enclosing loop but avoid generating one for
8192 (copy_bb_and_scalar_dependences): Remove premature codegen
8193 error on PHIs in blocks duplicated into multiple places.
8194 * graphite-scop-detection.c
8195 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8196 in the region use it as loop and nest to analyze the DR in.
8197 (try_generate_gimple_bb): Likewise.
8198 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8199 (add_loop_constraints): For blocks in a loop not in the region
8200 create a dimension with a single iteration.
8201 * sese.h (gbb_loop_at_index): Remove assert.
8203 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8205 * omp-expand.c (adjust_context_scope): New function.
8206 (expand_parallel_call): Call adjust_context_scope.
8208 2017-10-01 Jeff Law <law@redhat.com>
8210 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8211 dom_opt_dom_walker class with direct access to private members.
8212 Add comments. Call test_for_singularity.
8213 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8214 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8215 m_dummy_cond anymore.
8216 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8218 (pass_dominator:execute): Build the dummy_cond here and pass it
8219 to the dom_opt_dom_walker ctor.
8220 (test_for_singularity): New function.
8222 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8223 Maya Rashish <coypu@sdf.org>
8225 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8226 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8227 (alpha*-*-netbsd*) Use nbsd_tm_file.
8228 (arm*-*-netbsdelf*) Likewise.
8229 (i[34567]86-*-netbsdelf*) Likewise.
8230 (x86_64-*-netbsd*) Likewise.
8231 (mips*-*-netbsd*) Likewise.
8232 (powerpc-*-netbsd*) Likewise.
8233 (sh*-*-netbsd*) Likewise.
8234 (sparc-*-netbsdelf*) Likewise.
8235 (sparc64-*-netbsd*) Likewise.
8236 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8238 (vax-*-netbsdelf*) Likewise.
8239 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8240 (UINT_FAST8_TYPE) Likewise.
8241 (INT_FAST16_TYPE) Check CHAR_FAST16.
8242 (UINT_FAST16_TYPE) Likewise.
8244 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8247 * config/i386/i386.md
8248 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8249 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8250 *divmodsi4_zext_2): New define_insn_and_split.
8251 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8252 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8253 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8254 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8255 New define_insn_and_split.
8256 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8257 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8258 operands[1] having DImode when mode is SImode.
8260 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8261 always SImode for DIV and MOD in REG_EQUAL notes.
8263 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8266 * match.pd: Fix handling of NaNs in pattern.
8268 2017-09-29 Jeff Law <law@redhat.com>
8270 * sbitmap.c (bitmap_bit_in_range_p): New function.
8271 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8272 * tree-ssa-dse.c (live_bytes_read): New function.
8273 (dse_classify_store): Ignore reads of dead bytes.
8275 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8276 typos and whitespace errors.
8277 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8278 * config/s390/s390.c (s390_emit_prologue): Likewise.
8280 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8283 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8284 with a symbol for LRA.
8286 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8288 PR rtl-optimization/82338
8289 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8291 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8293 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8294 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8295 (qsort_chk): Declare.
8296 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8297 (qsort_chk): New function.
8299 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8301 PR tree-optimization/82337
8302 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8303 phi definition if the PHI result appears in an abnormal PHI.
8304 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8305 basis appears in an abnormal PHI.
8307 2017-09-29 Richard Biener <rguenther@suse.de>
8309 * graphite-isl-ast-to-gimple.c
8310 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8311 (binary_op_to_tree): Use it.
8312 (get_rename_from_scev): Likewise.
8313 (copy_loop_phi_nodes): Likewise.
8314 (copy_bb_and_scalar_dependences): Likewise.
8315 (translate_pending_phi_nodes): Likewise.
8317 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8320 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8321 for movabsq $(i32 << shift), r64.
8323 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8325 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8326 index when encoding %esp as %rsp to avoid 0x67 prefix.
8328 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8330 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8331 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8333 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8335 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8336 Extensions with more than 16 double VFP registers.
8337 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8338 to_clear_mask and all code related to it. Replace the remaining
8339 entry by a sbitmap and adapt code accordingly.
8341 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8343 * brig-builtins.def: Change pure attributes to const.
8345 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8347 * config.gcc (default_gnu_indirect_function): Default to yes for
8348 sparc*-*-linux* with glibc.
8350 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8352 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8353 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8354 when creating .init_array and .fini_array sections with priority
8357 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8360 * config/aarch64/aarch64.c
8361 (aarch64_builtin_support_vector_misalignment): Always return false
8362 when misalignment is unknown.
8364 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8366 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8367 this function to return false if the definition used by the swap
8368 instruction is artificial, or if the memory address from which the
8369 constant value is loaded is not represented by a base address held
8370 in a register or if the base address register is a frame or stack
8371 pointer. Additionally, return false if the base address of the
8372 loaded constant is a SYMBOL_REF but is not considered to be a
8374 (replace_swapped_load_constant): New function.
8375 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8376 loaded constant vector with a load of a swapped constant vector.
8378 2017-09-27 Carl Love <cel@us.ibm.com>
8380 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8381 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8382 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8385 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8387 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8388 first, always call autopref_rank_data otherwise.
8390 2017-09-27 Richard Biener <rguenther@suse.de>
8392 * graphite-scop-detection.c (find_scop_parameters): Move
8393 loop bound handling ...
8394 (gather_bbs::before_dom_children): ... here, avoiding the need
8395 to build scop_info->loop_nest.
8396 (record_loop_in_sese): Remove.
8397 * sese.h (sese_info_t::loop_nest): Remove.
8398 * sese.c (new_sese_info): Do not allocate loop_nest.
8399 (free_sese_info): Do not free loop_nest.
8401 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8404 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8405 lhs from calls if the lhs has addressable type.
8407 2017-09-27 Richard Biener <rguenther@suse.de>
8409 * graphite.h (scop::max_alias_set): New member.
8410 * graphite-scop-detection.c: Remove references to non-existing
8411 --param in comments.
8412 (build_alias_sets): Record the maximum alias set used for drs.
8413 (build_scops): Support zero as unlimited for
8414 --param graphite-max-arrays-per-scop.
8415 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8417 (build_poly_sr_1): ... here. Compute alias set based on the
8418 maximum alias set used for drs rather than
8419 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8421 2017-09-27 Richard Biener <rguenther@suse.de>
8423 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8424 --param loop-block-tile-size=0 to disable tiling.
8426 2017-09-27 Richard Biener <rguenther@suse.de>
8428 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8429 (graphite-max-nb-scop-params): Document special value zero.
8430 * domwalk.h (dom_walker::STOP): New symbolical constant.
8431 (dom_walker::dom_walker): Add optional parameter for bb to
8433 (dom_walker::~dom_walker): Declare.
8434 (dom_walker::before_dom_children): Document STOP return value.
8435 (dom_walker::m_user_bb_to_rpo): New member.
8436 (dom_walker::m_bb_to_rpo): Likewise.
8437 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8438 mapping here if not provided by the user.
8439 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8440 provided by the user.
8441 (dom_walker::STOP): Define.
8442 (dom_walker::walk): Do not compute bb to RPO mapping here.
8443 Support STOP return value from before_dom_children to stop
8445 * graphite-optimize-isl.c (optimize_isl): If the schedule
8446 is the same still generate code if -fgraphite-identity
8447 or -floop-parallelize-all are given.
8448 * graphite-scop-detection.c: Include cfganal.h.
8449 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8451 (gather_bbs::before_dom_children): Return STOP for BBs
8453 (build_scops): Compute bb to RPO mapping and pass it to
8454 the domwalk. Treat --param graphite-max-nb-scop-params=0
8455 as not limiting the number of params.
8456 * graphite.c (graphite_initialize): Remove limit on the
8457 number of basic-blocks in a function.
8458 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8459 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8460 default value of 10.
8462 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8464 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8465 Adjust code to eliminate needing to do the shift right 32-bits
8466 operation after XSCVDPSPN.
8468 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8470 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8471 ((X / Y) != 0 -> X >= Y): Likewise.
8473 2017-09-26 Carl Love <cel@us.ibm.com>
8475 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8476 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8477 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8478 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8479 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8480 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8481 definitions and overloading.
8482 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8483 statement for P9V_BUILTIN_XST_LEN_R.
8484 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8485 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8486 define_expand and define_insn for the instructions and builtins.
8487 * doc/extend.texi: Update the built-in documentation file for the new
8489 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8490 define_insn for the instructions
8492 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8495 * gcc/config/netbsd-protos.h: New file.
8496 * gcc/config/netbsd.c: New file.
8497 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8498 * gcc/config/t-netbsd: New file.
8499 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8500 (tmake_file) Add t-netbsd.
8501 (extra_objs) Add netbsd.o.
8503 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8507 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8509 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8511 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8512 sign extension from a vector register to a GPR by doing a 32-bit
8513 direct move and then an EXTSW.
8514 (extendsi<mode>2 splitter): Likewise.
8515 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8516 right or vector extract after doing XSCVDPSPN. Use
8517 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8519 (movdi_from_sf_zero_ext): Likewise.
8520 (reload_gpr_from_vsxsf): Likewise.
8521 (p8_mfvsrd_4_disf): Delete, no longer used.
8522 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8523 then needing to move the SFmode to a GPR to use the XSCVDPSP
8524 instruction instead of FRSP and XSCVDPSPN.
8525 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8526 it is adjacent to the other XSCVSPDP insns.
8527 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8528 SFmode to be in traditional Altivec registers.
8529 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8530 (vsx_xscvspdpn): Likewise.
8531 (vsx_xscvspdpn_scalar): Likewise.
8533 2017-09-26 Martin Jambor <mjambor@suse.cz>
8535 * tree-sra.c (compare_access_positions): Put integral types first,
8536 stabilize sorting of integral types, remove conditions putting
8537 non-full-precision integers last.
8538 (sort_and_splice_var_accesses): Disable scalarization if a
8539 non-integert would be represented by a non-full-precision integer.
8541 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8543 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8544 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8545 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8546 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8547 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8548 conditionals inside the function instead of around it. Call
8549 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8550 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8552 2017-09-26 Richard Biener <rguenther@suse.de>
8554 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8556 (scop_detection::build_scop_breadth): ... this. Removed.
8557 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8558 (scop_detection::harmful_stmt_in_bb): Likewise.
8559 (scop_detection::graphite_can_represent_stmt): Likewise.
8560 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8561 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8562 (scop_detection::can_represent_loop_1): ... this. Removed.
8563 (scop_detection::harmful_loop_in_region): Simplify after inlining
8564 the above and remove more quadraticness.
8565 (build_scops): Adjust.
8566 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8569 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8572 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8573 REGNO (base) == SP_REG if base is a REG.
8576 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8577 if it is different SSA_NAME.
8578 (optimize_range_tests_cmp_bitwise): New function.
8579 (optimize_range_tests): Call it.
8581 2017-09-26 Richard Biener <rguenther@suse.de>
8583 PR tree-optimization/82321
8584 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8585 for the def being inside the loop.
8587 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8589 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8591 * config/s390/s390-builtins.def: Fix constraint on op4.
8593 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8595 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8596 independent expanders.
8597 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8598 ("vec_ordered", "vec_unordered"): New expanders.
8600 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8602 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8605 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8607 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8608 vec_unpacks_lo_v16qi.
8609 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8611 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8613 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8614 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8615 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8617 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8619 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8621 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8623 ("*vec_slb<mode>"): New insn pattern.
8624 ("vec_shr_<mode>"): New expander.
8625 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8626 and force the shift count operand to V16QImode.
8627 ("vec_srb<mode>"): Set shift count mode to V16QI.
8629 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8631 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8632 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8633 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8635 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8638 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8640 2017-09-26 Richard Biener <rguenther@suse.de>
8642 PR tree-optimization/82320
8643 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8646 2017-09-25 Jeff Law <law@redhat.com>
8648 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8649 prototype for new argument.
8650 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8651 mostly extracted from rs6000_emit_allocate_stack.
8652 (rs6000_emit_probe_stack_range_stack_clash): New function.
8653 (rs6000_emit_allocate_stack): Call
8654 rs6000_emit_probe_stack_range_stack_clash as needed.
8655 (rs6000_emit_probe_stack_range): Add additional argument
8656 to call to gen_probe_stack_range{si,di}.
8657 (output_probe_stack_range): New.
8658 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8659 (output_probe_stack_range_stack_clash): New.
8660 (rs6000_emit_prologue): Emit notes into dump file as requested.
8661 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8662 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8663 Add additional operand and pass it to output_probe_stack_range.
8665 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8667 PR tree-optimization/82163
8668 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8669 (checking_verify_loop_closed_ssa): New parameter.
8670 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8671 (check_loop_closed_ssa_stmt): Delete.
8672 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8673 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8674 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8677 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8679 * brig-builtins.def: Treat HSAIL barrier builtins as
8680 setjmp/longjump style functions.
8682 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8684 * target.def (constant_alignment): New hook.
8685 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8686 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8687 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8688 * doc/tm.texi: Regenerate.
8689 * targhooks.h (default_constant_alignment): Declare.
8690 (constant_alignment_word_strings): Likewise.
8691 * targhooks.c (default_constant_alignment): New function.
8692 (constant_alignment_word_strings): Likewise.
8693 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8694 instead of CONSTANT_ALIGNMENT.
8695 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8696 (force_const_mem): Likewise.
8697 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8698 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8699 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8700 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8701 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8703 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8704 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8705 constant_alignment_word_strings.
8706 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8707 (CONSTANT_ALIGNMENT): Likewise.
8708 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8709 (arm_constant_alignment): New function.
8710 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8711 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8712 constant_alignment_word_strings.
8713 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8714 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8715 constant_alignment_word_strings.
8716 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8717 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8718 (cris_constant_alignment): New function.
8719 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8720 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8721 (epiphany_constant_alignment): New function.
8722 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8723 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8724 constant_alignment_word_strings.
8725 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8726 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8727 constant_alignment_word_strings.
8728 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8729 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8730 constant_alignment_word_strings.
8731 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8732 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8733 * config/i386/i386.c (ix86_constant_alignment): Make static.
8734 Use the same interface as the target hook.
8735 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8736 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8737 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8738 constant_alignment_word_strings.
8739 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8740 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8741 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8742 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8743 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8744 constant_alignment_word_strings.
8745 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8746 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8747 constant_alignment_word_strings.
8748 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8749 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8750 constant_alignment_word_strings.
8751 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8752 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8754 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8755 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8756 * config/mips/mips.c (mips_constant_alignment): New function.
8757 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8758 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8759 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8760 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8761 (mmix_constant_alignment): Make static. Use the same interface
8763 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8764 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8765 constant_alignment_word_strings.
8766 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8767 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8768 constant_alignment_word_strings.
8769 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8770 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8771 constant_alignment_word_strings.
8772 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8773 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8774 (rs6000_constant_alignment): New function.
8775 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8776 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8777 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8778 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8779 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8780 (rs6000_constant_alignment): New function.
8781 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8782 * config/s390/s390.c (s390_constant_alignment): New function.
8783 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8784 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8785 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8786 constant_alignment_word_strings.
8787 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8788 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8789 (sparc_constant_alignment): New function.
8790 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8791 * config/spu/spu.c (spu_constant_alignment): New function.
8792 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8793 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8794 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8795 constant_alignment_word_strings.
8796 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8797 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8798 constant_alignment_word_strings.
8799 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
8800 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8801 constant_alignment_word_strings.
8802 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
8803 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8804 (visium_constant_alignment): New function.
8805 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
8806 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8807 (xtensa_constant_alignment): New function.
8808 * system.h (CONSTANT_ALIGNMENT): Poison.
8810 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
8812 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8813 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
8814 (rs6000_builtin_valid_without_lhs): New helper function.
8815 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8816 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
8818 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8820 * target.h (vec_perm_indices): Use unsigned short rather than
8822 (auto_vec_perm_indices): Likewise.
8823 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
8824 Use unsigned int rather than unsigned char.
8825 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
8827 2017-09-25 Richard Biener <rguenther@suse.de>
8829 * cfgloop.h (sort_sibling_loops): Declare.
8830 * cfgloop.c (sort_sibling_loops_cmp): New helper.
8831 (sort_sibling_loops): New function sorting the sibling loop list
8833 * graphite.c (graphite_transform_loops): Sort sibling loops.
8835 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8837 * target.def (vec_perm_const_ok): Change sel parameter to
8839 * optabs-query.c (can_vec_perm_p): Update accordingly.
8840 * doc/tm.texi: Regenerate.
8841 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8842 auto_vec_perm_indices and remove separate nelt field.
8843 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8844 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8845 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8846 (aarch64_expand_vec_perm_const): Update accordingly.
8847 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8848 to vec_perm_indices.
8849 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8850 auto_vec_perm_indices and remove separate nelt field.
8851 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8852 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8853 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8855 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8856 to vec_perm_indices.
8857 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8858 sel to vec_perm_indices.
8859 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8860 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8861 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8863 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8866 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8869 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8870 on the FUNCTION_DECL function context if it has a DIE that is a
8873 2017-09-25 Richard Biener <rguenther@suse.de>
8875 PR tree-optimization/82285
8876 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8879 2017-09-25 Tom de Vries <tom@codesourcery.com>
8883 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8886 2017-09-25 Richard Biener <rguenther@suse.de>
8888 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8889 ISL errors other than isl_error_quota happen. Dump if the
8890 schedule is the same.
8891 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8892 errors instead of aborting inside ISL.
8894 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8897 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8898 of libgcc_eh for m64.
8899 * config/i386/darwin64.h: Likewise.
8901 2017-09-25 Richard Biener <rguenther@suse.de>
8904 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8905 attribute for incomplete types nor twice for complete ones.
8907 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8910 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8911 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8914 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8917 * config/i386/i386.opt (mprefer-avx256): Use
8918 ix86_target_flags variable.
8919 * config/i386/i386.c (ix86_target_string): Move
8920 -mprefer-avx256 to flag2_opts.
8922 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8925 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8926 and x != -1 | y != -1 into (x & y) != -1.
8928 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8930 * config.gcc: Add new case statement to set
8931 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8932 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8933 s390x-*-linux* case statements. Added aarch64 to the list of
8934 supported architectures.
8936 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8938 PR tree-optimization/82289
8939 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8940 STMT_VINFO_RELEVANT_P.
8942 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8943 Alan Hayward <alan.hayward@arm.com>
8944 David Sherwood <david.sherwood@arm.com>
8946 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8947 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8948 (extract_range_from_binary_expr_1): Don't call
8949 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8951 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8952 Alan Hayward <alan.hayward@arm.com>
8953 David Sherwood <david.sherwood@arm.com>
8955 * target.def (preferred_vector_alignment): New hook.
8956 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8958 * doc/tm.texi: Regenerate.
8959 * targhooks.h (default_preferred_vector_alignment): Declare.
8960 * targhooks.c (default_preferred_vector_alignment): New function.
8961 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8963 (DR_TARGET_ALIGNMENT): New macro.
8964 (aligned_access_p): Update commentary.
8965 (vect_known_alignment_in_bytes): New function.
8966 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8968 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8969 Calculate the misalignment based on the target alignment rather than
8971 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8972 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8973 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8974 the target alignment, rather than masking the element misalignment
8975 with the number of elements in a vector. Also use the target
8976 alignment when calculating the maximum number of peels.
8977 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8978 instead of TYPE_ALIGN_UNIT.
8979 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8980 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8981 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8982 (vect_create_data_ref_ptr): Likewise.
8983 (vect_setup_realignment): Realign by ANDing with
8984 -DR_TARGET_MISALIGNMENT.
8985 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8986 the number of peels based on DR_TARGET_ALIGNMENT.
8987 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8988 with the guaranteed alignment boundary when deciding whether
8990 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8991 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8992 (ensure_base_align): Remove stmt_info parameter. Get the
8993 target base alignment from DR_TARGET_ALIGNMENT.
8994 (vectorizable_store): Update call accordingly. Interpret
8995 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8997 (vectorizable_load): Likewise.
8999 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9000 Alan Hayward <alan.hayward@arm.com>
9001 David Sherwood <david.sherwood@arm.com>
9003 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9004 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9005 (vect_enhance_data_refs_alignment): Likewise.
9007 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9009 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9010 error. Only quit immediately if parsing is complete.
9011 (BEGIN): Initialize fatal_err and parse_done.
9012 (begin fpu, end fpu): Check number of arguments.
9013 (begin arch, end arch): Likewise.
9014 (begin cpu, end cpu): Likewise.
9015 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9016 (optalias): Likewise.
9018 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9020 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9021 * config/arm/arm-isa.h: Delete. Move definitions to ...
9022 * arm-cpus.in: ... here. Use new feature and fgroup values.
9023 * config/arm/arm.c (arm_option_override): Use lower case for feature
9025 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9026 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9027 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9029 (print_isa_bits_for): New function.
9030 (gen_isa): New function.
9031 (gen_comm_data): Use print_isa_bits_for.
9032 (define feature): New keyword.
9033 (define fgroup): New keyword.
9034 * config/arm/t-arm (TM_H): Remove.
9035 (GTM_H): Add arm-isa.h.
9036 (arm-isa.h): Add rule to generate file.
9037 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9038 case for feature bit names.
9040 2017-09-22 Richard Biener <rguenther@suse.de>
9042 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9044 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9045 print of no dependency loops ...
9046 * graphite.c (graphite_transform_loops): ... here.
9047 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9049 (same_close_phi_node, remove_duplicate_close_phi,
9050 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9051 (canonicalize_loop_closed_ssa): ... here and simplify.
9052 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9053 (optimize_isl): Use dump_printf_loc to tell when we stopped
9054 optimizing because of an ISL timeout.
9056 2017-09-22 Richard Biener <rguenther@suse.de>
9058 PR tree-optimization/82291
9059 * tree-if-conv.c (predicate_mem_writes): Make sure to
9060 remove writes in blocks predicated with false.
9062 2017-09-22 Richard Biener <rguenther@suse.de>
9064 * sese.c: Include cfganal.h.
9065 (if_region_set_false_region): Remove.
9066 (create_if_region_on_edge): Likewise.
9067 (move_sese_in_condition): Re-implement without destroying
9070 2017-09-22 Richard Biener <rguenther@suse.de>
9072 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9073 Verify both BBs contain loop PHI nodes before dispatching to
9075 (graphite_regenerate_ast_isl): Do not recompute dominators,
9076 do not verify three times. Restructure for clarity.
9077 * graphite-scop-detection.c (same_close_phi_node,
9078 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9079 defined_in_loop_p, canonicalize_loop_closed_ssa,
9080 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9081 checking and SSA rewrite, move to ...
9082 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9083 (graphite_initialize): Do not pass in ctx, do not reset the
9084 SCEV cache, compute only dominators.
9085 (graphite_transform_loops): Allocate ISL ctx after
9086 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9087 Maintain post-dominators only around build_scops.
9088 * sese.c (if_region_set_false_region): Make static. Free
9089 and recompute dominators.
9090 (move_sese_in_condition): Assert we don't get called with
9091 post-dominators computed.
9092 * sese.h (if_region_set_false_region): Remove.
9094 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9096 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9097 mode attribute for TARGET_AVX512VL.
9099 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9101 * config/i386/i386.opt (mprefer-avx256): New option.
9102 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9104 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9105 for TARGET_PREFER_AVX256.
9106 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9108 2017-09-21 Jeff Law <law@redhat.com>
9110 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9111 Fix dump output if the only stack space is for pushed registers.
9113 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9115 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9118 2017-09-21 Martin Sebor <msebor@redhat.com>
9121 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9122 code (in C++) or code that triggers warnings.
9124 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9126 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9128 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9130 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9131 * sched-rgn.c: Ditto.
9132 * sel-sched-ir.c: Ditto.
9134 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9136 * toplev.h (set_random_seed): Adjust return type.
9137 * toplev.c (init_local_tick): Move eager initialization of random_seed
9138 to get_random_seed. Adjust comment.
9139 (init_random_seed): Inline to get_random_seed, delete.
9140 (get_random_seed): Initialize random_seed lazily.
9141 (set_random_seed): Do not return previous value.
9142 (print_switch_value): Do not call get_random_seed.
9144 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9146 * cgraph.c (delete_function_version): New, broken out from...
9147 (cgraph_node::delete_function_version): ...here. Rename to
9148 cgraph_node::delete_function_version_by_decl. Update all uses.
9149 (cgraph_node::remove): Call delete_function_version.
9151 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9154 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9155 VAR_DECLs to which addressable non-volatile parameters are mapped
9156 and for id->retvar after the return value assignment. Clear
9157 id->retval and id->retbnd after inlining.
9159 2017-09-21 Richard Biener <rguenther@suse.de>
9161 PR tree-optimization/82276
9162 PR tree-optimization/82244
9163 * tree-vrp.c (build_assert_expr_for): Set
9164 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9166 (remove_range_assertions): Revert earlier change.
9168 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9171 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9173 2017-09-21 Richard Biener <rguenther@suse.de>
9175 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9176 Restore valid IL after code generation errors.
9177 * graphite.c (graphite_transform_loops): Diagnose code
9178 generation issues as MSG_MISSED_OPTIMIZATION and continue
9179 with processing SCOPs.
9181 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9182 Alan Hayward <alan.hayward@arm.com>
9183 David Sherwood <david.sherwood@arm.com>
9185 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9186 rather than choosing between plus_constant and gen_rtx_<CODE>.
9187 * expr.c (emit_push_insn): Likewise.
9188 (expand_expr_real_2): Likewise.
9190 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9191 Alan Hayward <alan.hayward@arm.com>
9192 David Sherwood <david.sherwood@arm.com>
9194 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9196 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9197 Alan Hayward <alan.hayward@arm.com>
9198 David Sherwood <david.sherwood@arm.com>
9200 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9201 calling tree_to_uhwi.
9203 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9204 Alan Hayward <alan.hayward@arm.com>
9205 David Sherwood <david.sherwood@arm.com>
9207 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9208 INTEGER_CST rather than a negative test for ADDR_EXPR.
9210 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9211 Alan Hayward <alan.hayward@arm.com>
9212 David Sherwood <david.sherwood@arm.com>
9214 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9215 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9217 2017-09-21 Richard Biener <rguenther@suse.de>
9219 PR tree-optimization/71351
9220 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9221 graphite_create_new_loop_guard): Remove, fold remaining parts
9223 (translate_isl_ast_node_for): ... here and simplify.
9225 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9228 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9229 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9230 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9231 alternative always use QI mode, for -Os imov (=R,R) alternative
9232 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9235 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9236 Jeff Law <law@redhat.com>
9238 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9239 (allocate_stack_space): New function, partially extracted from
9241 (s390_emit_prologue): Track offset to most recent stack probe.
9242 Code to allocate space moved into allocate_stack_space.
9243 Dump actions when no stack is allocated.
9244 (s390_prologue_plus_offset): New function.
9245 (s390_emit_stack_probe): Likewise.
9247 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9249 * common.opt (Wa, Wl, Wp, g, gz=): Add
9251 (gno-column-info): Remove.
9252 (gcolumn-info): Drop RejectNegative.
9254 (gno-record-gcc-switches): Remove.
9255 (grecord-gcc-switches): Drop RejectNegative.
9256 (gno-split-dwarf): Remove.
9257 (gsplit-dwarf): Drop RejectNegative.
9258 (gno-strict-dwarf): Remove.
9259 (gstrict-dwarf): Drop RejectNegative.
9260 * config/darwin.opt (gfull, gused): Add RejectNegative.
9261 * dwarf2out.c (gen_producer_string): Drop
9262 gno-record-gcc-switches handler.
9263 * optc-gen.awk: Add g to prefixes with negative forms.
9264 * opts-common.c (remapping_prefix_p): New.
9265 (find_opt): Check it.
9266 (generate_canonical_option): Test g prefix.
9267 (option_map): Add -gno- mapping.
9268 (add_misspelling_candidates): Check remapping_prefix_p.
9270 2017-09-20 Jeff Law <law@redhat.com>
9272 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9273 thinko in stack clash protection support.
9275 * explow.c (compute_stack_clash_protection_loop_data): Use
9276 CONST_INT_P instead of explicit test. Verify object is a
9277 CONST_INT_P before looking at INTVAL.
9278 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9279 instead of explicit test.
9281 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9284 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9285 address instead of to r1 and r11.
9287 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9289 * config.gcc: Support "knm".
9290 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9291 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9293 * config/i386/i386.c (m_KNM): Define.
9294 (processor_target_table): Add "knm".
9296 (ix86_option_override_internal): Add "knm".
9297 (ix86_issue_rate): Add PROCESSOR_KNM.
9298 (ix86_adjust_cost): Ditto.
9299 (ia32_multipass_dfa_lookahead): Ditto.
9300 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9301 (fold_builtin_cpu): Add M_INTEL_KNM.
9302 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9303 (processor_type): Add PROCESSOR_KNM.
9304 * config/i386/x86-tune.def: Add m_KNM.
9305 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9307 2017-09-20 Richard Biener <rguenther@suse.de>
9309 PR tree-optimization/80213
9310 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9311 are allowed in empty BBs as well.
9312 (canonicalize_loop_closed_ssa): Also look for other complex
9314 (scop_detection::get_sese): Include the loop-closed PHI block
9316 (scop_detection::merge_sese): Remove code adding extra blocks.
9317 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9318 (build_scops): Assert the final returned scop is invalid.
9320 2017-09-20 Richard Biener <rguenther@suse.de>
9322 PR tree-optimization/82264
9323 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9325 (vn_phi_lookup): Likewise.
9326 (vn_phi_insert): Likewise.
9328 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9330 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9331 that fits into uhwi or shwi, add DW_AT_const_value regardless
9332 of early_dwarf without going through RTL, using add_AT_unsigned
9335 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9336 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9337 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9338 (DEBUG_MACRO_SECTION): Likewise.
9339 (DEBUG_LTO_MACRO_SECTION): Likewise.
9340 (DEBUG_STR_DWO_SECTION): Likewise.
9341 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9342 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9343 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9344 (DEBUG_LTO_LINE_STR_SECTION): Define.
9345 (init_sections_and_labels): Initialize debug_line_str_section
9346 variable. Initialize debug_loc_section for -gdwarf-5 to
9347 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9349 2017-09-20 Richard Biener <rguenther@suse.de>
9351 * graphite-sese-to-poly.c (extract_affine): Properly handle
9352 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9354 2017-09-20 Richard Biener <rguenther@suse.de>
9356 PR tree-optimization/81373
9357 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9358 Force SESE live-out defs to be handled even if they are
9361 2017-09-19 Jeff Law <law@redhat.com>
9363 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9364 nothing for stack adjustments with REG_STACK_CHECK.
9365 * sched-deps.c (parse_add_or_inc): Reject insns with
9366 REG_STACK_CHECK from dependency breaking.
9367 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9368 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9369 * reg-notes.def (STACK_CHECK): New note.
9371 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9372 (ix86_expand_prologue): Dump stack clash info as needed.
9373 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9375 * function.c (dump_stack_clash_frame_info): New function.
9376 * function.h (dump_stack_clash_frame_info): Prototype.
9377 (enum stack_clash_probes): New enum.
9379 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9380 flag_stack_clash_protection.
9381 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9382 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9383 (arm_frame_pointer_required): Likewise.
9384 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9385 (ia64_expand_prologue): Likewise.
9386 * config/mips/mips.c (mips_expand_prologue): Likewise.
9387 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9388 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9389 (sparc_flat_expand_prologue): Likewise.
9390 * config/spu/spu.c (spu_expand_prologue): Likewise.
9392 * explow.c: Include "params.h".
9393 (anti_adjust_stack_and_probe_stack_clash): New function.
9394 (get_stack_check_protect): Likewise.
9395 (compute_stack_clash_protection_loop_data): Likewise.
9396 (emit_stack_clash_protection_loop_start): Likewise.
9397 (emit_stack_clash_protection_loop_end): Likewise.
9398 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9399 Use anti_adjust_stack_and_probe_stack_clash.
9400 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9401 (emit_stack_clash_protection_loop_start): Likewise.
9402 (emit_stack_clash_protection_loop_end): Likewise.
9403 * rtl.h (get_stack_check_protect): Prototype.
9404 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9405 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9406 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9408 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9410 * doc/tm.texi: Rebuilt.
9411 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9412 get_stack_check_protect.
9413 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9414 * config/arm/arm.c (arm_expand_prologue): Likewise.
9415 (arm_frame_pointer_required): Likewise.
9416 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9417 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9418 * config/mips/mips.c (mips_expand_prologue): Likewise.
9419 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9420 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9421 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9422 (sparc_flat_expand_prologue): Likewise.
9424 * common.opt (-fstack-clash-protection): New option.
9425 * flag-types.h (enum stack_check_type): Note difference between
9426 -fstack-check= and -fstack-clash-protection.
9427 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9428 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9429 * toplev.c (process_options): Issue warnings/errors for cases
9430 not handled with -fstack-clash-protection.
9431 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9432 (-fstack-check): Note additional problem with -fstack-check=generic.
9433 Note that -fstack-check is primarily for Ada and refer users
9434 to -fstack-clash-protection for stack-clash-protection.
9435 Document new params for stack clash protection.
9437 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9439 * config/i386/i386.c (ix86_split_long_move): Do not handle
9440 address used for LEA in a special way.
9442 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9444 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9446 2017-09-19 Martin Sebor <msebor@redhat.com>
9449 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9450 of incompatible types.
9452 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9454 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9455 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9456 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9457 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9459 2017-09-19 Richard Biener <rguenther@suse.de>
9461 PR tree-optimization/82244
9462 * tree-vrp.c (remove_range_assertions): Do not propagate
9463 a constant to abnormals but replace the assert with a copy.
9465 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9467 PR rtl-optimization/57878
9468 PR rtl-optimization/68988
9469 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9470 avoidance test involving non_reload_pseudos. Move frequency test
9471 below the general fragmentation avoidance test.
9473 2017-09-19 Richard Biener <rguenther@suse.de>
9475 PR tree-optimization/69728
9476 * graphite-sese-to-poly.c (schedule_error): New global.
9477 (add_loop_schedule): Handle empty domain by failing the
9479 (build_original_schedule): Handle schedule_error.
9481 2017-09-19 Richard Biener <rguenther@suse.de>
9483 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9484 Do not iterate to sibling loops but only to siblings of inner
9487 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9490 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9491 that the registers are different.
9493 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9495 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9496 to processor_model and "amdfam17h" to arch_names_table.
9497 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9499 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9502 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9504 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9505 Alan Hayward <alan.hayward@arm.com>
9506 David Sherwood <david.sherwood@arm.com>
9508 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9510 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9512 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9513 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9514 vect_schedule_slp_instance.
9515 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9516 Update call to vect_slp_analyze_node_operations. Simplify return
9518 (vect_slp_analyze_bb_1): Update call accordingly.
9519 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9520 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9521 (vect_schedule_slp): Update call accordingly.
9523 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9524 Alan Hayward <alan.hayward@arm.com>
9525 David Sherwood <david.sherwood@arm.com>
9527 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9528 with types that aren't in fact scalar.
9530 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9532 * tree-vect-slp.c (vect_record_max_nunits): New function,
9534 (vect_build_slp_tree_1): ...here.
9535 (vect_build_slp_tree_2): Call it for phis too.
9537 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9539 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9540 to vect_get_vec_def_for_operand when getting the mask operand.
9542 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9543 Alan Hayward <alan.hayward@arm.com>
9544 David Sherwood <david.sherwood@arm.com>
9546 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9549 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9550 Alan Hayward <alan.hayward@arm.com>
9551 David Sherwood <david.sherwood@arm.com>
9553 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9554 calculation for vector booleans.
9556 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9557 Alan Hayward <alan.hayward@arm.com>
9558 David Sherwood <david.sherwood@arm.com>
9560 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9562 (vect_transform_stmt): ...here.
9563 (vect_analyze_stmt): Use it instead of calling
9564 vectorizable_live_operation directly.
9566 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9568 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9569 non-SIMT targets in acc vector loops.
9571 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9573 * configure.ac: Add arc and check if assembler supports gdwarf2.
9574 * configure: Regenerate.
9576 2017-09-18 Richard Biener <rguenther@suse.de>
9578 PR tree-optimization/82220
9579 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9580 epilogue niters from the min_profitable_iters compute.
9582 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9585 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9586 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9587 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9589 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9592 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9593 switching to a new text section.
9595 2017-09-18 Richard Biener <rguenther@suse.de>
9597 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9599 (build_alias_set): Reject aliases with no access function.
9601 2017-09-18 Richard Biener <rguenther@suse.de>
9603 PR tree-optimization/79622
9604 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9606 (build_cross_bb_scalars_use): Likewise.
9608 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9610 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9612 2017-09-18 Alan Modra <amodra@gmail.com>
9615 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9616 stack_pointer_rtx for count 0. Update comments. Break up
9617 large rtl expression.
9619 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9621 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9622 Increase to 20 bytes.
9623 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9624 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9625 or avx version of the stub.
9627 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9630 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9631 compute the minimum stack alignment. Also update preferred stack
9632 boundary for leaf functions.
9634 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9636 PR tree-optimization/82228
9637 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9640 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9642 * common/config/nds32/nds32-common.c
9643 (nds32_option_optimization_table): Refine formatting.
9644 (nds32_option_optimization_table): Use -fsched-pressure and
9645 -fomit-frame-pointer for specific optimization level.
9647 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9649 * config/nds32/nds32.c: Refine formatting and comments.
9650 * config/nds32/nds32.h: Likewise.
9651 * config/nds32/nds32.md: Likewise.
9652 * config/nds32/nds32-cost.c: Likewise.
9653 * config/nds32/nds32-isr.c: Likewise.
9654 * config/nds32/nds32-md-auxiliary.c: Likewise.
9655 * config/nds32/nds32-multiple.md: Likewise.
9656 * config/nds32/nds32-predicates.c: Likewise.
9658 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9659 Jakub Jelinek <jakub@redhat.com>
9661 Add support for -std=c++2a.
9662 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9664 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9666 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9669 * doc/extend.texi (Common Function Attributes): Add
9670 references to ARM, AArch64, and S/390 specific attributes.
9671 (Function Specific Option Pragmas): Add AArch64 and S/390
9672 to list of back ends that support the target pragma.
9674 2017-09-15 Nathan Sidwell <nathan@acm.org>
9676 * doc/standards.texi: Fix C++17 description. Update URLs for
9679 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9681 * common.opt (Wcast-align=strict): New warning option.
9682 * doc/invoke.texi: Document -Wcast-align=strict.
9684 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9686 * cgraph.h (cgraph_thunk_info): Add comments.
9687 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9688 assert for VIRTUAL_* arguments stricter.
9690 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9692 PR tree-optimization/71026
9693 * match.pd: Move RDIV patterns from fold-const.c
9694 * fold-const.c (distribute_real_division): Removed.
9695 (fold_binary_loc): Remove calls to distribute_real_divison.
9697 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9699 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9700 c++1z and gnu++1z as deprecated. Change other references to
9701 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9702 Change -Wc++1z-compat to -Wc++17-compat.
9703 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9704 * dwarf2out.c (highest_c_language): Handle C++17.
9705 (gen_compile_unit_die): Likewise.
9707 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9709 PR rtl-optimization/82192
9710 * combine.c (make_extraction): Don't look through non-paradoxical
9711 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9714 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9715 Alan Hayward <alan.hayward@arm.com>
9716 David Sherwood <david.sherwood@arm.com>
9718 * target.def (function_arg_offset): New hook.
9719 * targhooks.h (default_function_arg_offset): Declare.
9720 * targhooks.c (default_function_arg_offset): New function.
9721 * function.c (locate_and_pad_parm): Use
9722 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9723 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9724 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9725 * doc/tm.texi: Regenerate.
9726 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9727 * config/spu/spu.c (spu_function_arg_offset): New function.
9728 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9729 * system.h (FUNCTION_ARG_OFFSET): Poison.
9731 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9732 Alan Hayard <alan.hayward@arm.com>
9733 David Sherwood <david.sherwood@arm.com>
9735 * target.def (truly_noop_truncation): New hook.
9736 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9737 than TRULY_NOOP_TRUNCATION.
9738 * hooks.h (hook_bool_uint_uint_true): Declare.
9739 * hooks.c (hook_bool_uint_uint_true): New function.
9740 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9741 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9742 * doc/tm.texi: Regenerate.
9743 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9744 rather than TRULY_NOOP_TRUNCATION in comments.
9745 (simplify_comparison): Likewise.
9746 (record_truncated_value): Likewise.
9747 * expmed.c (extract_bit_field_1): Likewise.
9748 (extract_split_bit_field): Likewise.
9749 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9750 instead of TRULY_NOOP_TRUNCATION.
9751 * function.c (assign_parm_setup_block): Likewise.
9752 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9753 * rtlhooks.c: Include target.h.
9754 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9755 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9756 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9757 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9758 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9759 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9760 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9761 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9762 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9763 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9764 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9765 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9766 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9767 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9768 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9769 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9770 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9771 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9772 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9773 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9774 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9775 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9776 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9777 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9778 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9779 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9780 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9781 rather than TRULY_NOOP_TRUNCATION in comments.
9782 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9783 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9784 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9785 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9786 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9787 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9788 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9789 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9790 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9791 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9792 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9793 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9794 rather than TRULY_NOOP_TRUNCATION in comments.
9795 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9796 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9797 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
9798 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
9799 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
9800 TRULY_NOOP_TRUNCATION condition.
9801 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
9802 (TRULY_NOOP_TRUNCATION): Delete.
9803 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
9804 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
9805 * config/spu/spu.c (spu_truly_noop_truncation): New function.
9806 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9807 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
9808 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
9809 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
9810 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9811 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9812 rather than TRULY_NOOP_TRUNCATION in comments.
9813 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
9814 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
9815 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
9816 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
9817 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
9818 * system.h (TRULY_NOOP_TRUNCATION): Poison.
9820 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
9823 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
9824 (*cmp_ior): Likewise.
9825 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
9826 (*ior_scc_scc_cmp): Likewise.
9827 (*and_scc_scc): Likewise.
9828 (*and_scc_scc_cmp): Likewise.
9830 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9831 Alan Hayard <alan.hayward@arm.com>
9832 David Sherwood <david.sherwood@arm.com>
9834 * target.def (can_change_mode_class): New hook.
9835 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9836 (hard_regno_nregs): Likewise.
9837 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9838 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9839 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9840 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9841 (LOAD_EXTEND_OP): Update accordingly.
9842 * doc/tm.texi: Regenerate.
9843 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9844 CANNOT_CHANGE_MODE_CLASS.
9845 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9846 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9847 * combine.c (simplify_set): Update accordingly.
9848 * emit-rtl.c (validate_subreg): Likewise.
9849 * recog.c (general_operand): Likewise.
9850 * regcprop.c (mode_change_ok): Likewise.
9851 * reload1.c (choose_reload_regs): Likewise.
9852 (inherit_piecemeal_p): Likewise.
9853 * rtlanal.c (simplify_subreg_regno): Likewise.
9854 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9855 instead of CANNOT_CHANGE_MODE_CLASS.
9856 (reload_cse_simplify_operands): Likewise.
9857 * reload.c (push_reload): Use targetm.can_change_mode_class
9858 instead of CANNOT_CHANGE_MODE_CLASS.
9859 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9860 REG_CANNOT_CHANGE_MODE_P.
9861 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9862 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9863 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9864 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9865 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9866 (arm_can_change_mode_class): New function.
9867 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9868 than CANNOT_CHANGE_MODE_CLASS in comments.
9869 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9870 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9871 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9872 (ix86_can_change_mode_class): ...this new function, inverting the
9873 sense of the return value.
9874 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9875 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9876 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9877 (ia64_can_change_mode_class): New function.
9878 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9879 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9880 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9881 (m32c_can_change_mode_class): ...this new function, inverting the
9882 sense of the return value.
9883 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9884 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9885 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9886 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9887 (mips_can_change_mode_class): ...this new function, inverting the
9888 sense of the return value.
9889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9890 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9891 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9892 (msp430_can_change_mode_class): New function.
9893 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9894 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9895 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9896 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9897 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9898 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9899 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9900 (pa_cannot_change_mode_class): Replace with...
9901 (pa_can_change_mode_class): ...this new function, inverting the
9902 sense of the return value.
9903 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9904 than CANNOT_CHANGE_MODE_CLASS in comments.
9905 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9906 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9907 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9908 (pdp11_cannot_change_mode_class): Replace with...
9909 (pdp11_can_change_mode_class): ...this new function, inverting the
9910 sense of the return value.
9911 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9912 * config/powerpcspe/powerpcspe-protos.h
9913 (rs6000_cannot_change_mode_class_ptr): Delete.
9914 * config/powerpcspe/powerpcspe.c
9915 (rs6000_cannot_change_mode_class_ptr): Delete.
9916 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9917 (rs6000_option_override_internal): Assign to
9918 targetm.can_change_mode_class instead of
9919 rs6000_cannot_change_mode_class_ptr.
9920 (rs6000_cannot_change_mode_class): Replace with...
9921 (rs6000_can_change_mode_class): ...this new function, inverting the
9922 sense of the return value.
9923 (rs6000_debug_cannot_change_mode_class): Replace with...
9924 (rs6000_debug_can_change_mode_class): ...this new function.
9925 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9926 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9927 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9928 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9929 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9931 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9932 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9933 (rs6000_option_override_internal): Assign to
9934 targetm.can_change_mode_class instead of
9935 rs6000_cannot_change_mode_class_ptr.
9936 (rs6000_cannot_change_mode_class): Replace with...
9937 (rs6000_can_change_mode_class): ...this new function, inverting the
9938 sense of the return value.
9939 (rs6000_debug_cannot_change_mode_class): Replace with...
9940 (rs6000_debug_can_change_mode_class): ...this new function.
9941 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9942 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9943 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9944 (s390_can_change_mode_class): ...this new function, inverting the
9945 sense of the return value.
9946 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9947 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9948 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9949 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9950 (sh_cannot_change_mode_class): Replace with...
9951 (sh_can_change_mode_class): ...this new function, inverting the
9952 sense of the return value.
9953 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9954 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9955 (sparc_can_change_mode_class): New function.
9956 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9957 * config/spu/spu.c (spu_can_change_mode_class): New function.
9958 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9959 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9960 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9961 (visium_can_change_mode_class): New function.
9962 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9964 2017-09-15 Richard Biener <rguenther@suse.de>
9966 PR tree-optimization/82217
9967 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9968 but not undefined case.
9970 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9973 * postreload.c (reload_cse_simplify_operands): Skip
9974 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9976 2017-09-15 Richard Biener <rguenther@suse.de>
9978 PR tree-optimization/68823
9979 * graphite-scop-detection.c (build_alias_set): If we have a
9980 possible dependence check whether we can handle them by just
9981 looking at the DRs DR_ACCESS_FNs.
9982 (build_scops): If build_alias_set fails, fail the SCOP.
9984 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9986 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9987 to support float128 built-in functions that require the ISA 3.0
9989 (BU_FLOAT128_3_HW): Likewise.
9990 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9992 (FMAF128): Likewise.
9994 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9995 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9996 floating point instructions.
9997 (rs6000_invalid_builtin): Likewise.
9998 (rs6000_builtin_mask_names): Likewise.
9999 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10000 (RS6000_BTM_FLOAT128_HW): Likewise.
10001 (RS6000_BTM_COMMON): Likewise.
10002 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10004 * doc/extend.texi (RS/6000 built-in functions): Document the
10005 IEEE 128-bit floating point square root and fused multiply-add
10006 built-in functions.
10008 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10010 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10011 reg (r2) isn't in the set of registers defined in the prologue.
10013 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10014 Alan Hayward <alan.hayward@arm.com>
10015 David Sherwood <david.sherwood@arm.com>
10017 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10018 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10019 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10020 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10021 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10023 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10024 max_vectorization_factor.
10025 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10027 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10028 Alan Hayward <alan.hayward@arm.com>
10029 David Sherwood <david.sherwood@arm.com>
10031 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10032 (vect_worthwhile_without_simd_p): Declare.
10033 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10034 (vectorizable_reduction): Use it.
10035 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10036 (vectorizable_operation): Likewise.
10038 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10039 Alan Hayward <alan.hayward@arm.com>
10040 David Sherwood <david.sherwood@arm.com>
10042 * tree-vectorizer.h (vect_get_num_copies): New function.
10043 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10044 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10045 (vectorizable_induction): Likewise.
10046 (vectorizable_live_operation): Likewise.
10047 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10048 (vectorizable_bswap): Likewise.
10049 (vectorizable_call): Likewise.
10050 (vectorizable_conversion): Likewise.
10051 (vectorizable_assignment): Likewise.
10052 (vectorizable_shift): Likewise.
10053 (vectorizable_operation): Likewise.
10054 (vectorizable_store): Likewise.
10055 (vectorizable_load): Likewise.
10056 (vectorizable_condition): Likewise.
10057 (vectorizable_comparison): Likewise.
10058 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10060 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10061 Alan Hayward <alan.hayward@arm.com>
10062 David Sherwood <david.sherwood@arm.com>
10064 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10065 of vect_init_vector.
10067 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10068 Alan Hayward <alan.hayward@arm.com>
10069 David Sherwood <david.sherwood@arm.com>
10071 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10072 an inline wrapper that provides a location.
10073 (gimple_build_vector): Likewise.
10074 * gimple-fold.c (gimple_build_vector_from_val): New function.
10075 (gimple_build_vector): Likewise.
10076 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10077 functions to build the initial value. Always return a gimple value.
10078 (get_initial_defs_for_reduction): Likewise. Only compute
10080 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10081 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10082 (vectorizable_induction): Use gimple_build_vector rather than
10085 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10086 Alan Hayward <alan.hayward@arm.com>
10087 David Sherwood <david.sherwood@arm.com>
10089 * target.h (vec_perm_indices): New typedef.
10090 (auto_vec_perm_indices): Likewise.
10091 * optabs-query.h: Include target.h
10092 (can_vec_perm_p): Take a vec_perm_indices *.
10093 * optabs-query.c (can_vec_perm_p): Likewise.
10094 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10095 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10096 * tree-vect-generic.c (lower_vec_perm): Likewise.
10097 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10098 (vect_grouped_load_supported): Likewise.
10099 (vect_shift_permute_load_chain): Likewise.
10100 (vect_permute_store_chain): Use auto_vec_perm_indices.
10101 (vect_permute_load_chain): Likewise.
10102 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10103 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10104 Update uses of can_vec_perm_p.
10105 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10106 mode with a number of elements. Take a vec_perm_indices *.
10107 (vect_create_epilog_for_reduction): Update accordingly.
10108 Use auto_vec_perm_indices.
10109 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10110 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10111 (vect_transform_slp_perm_load): Likewise.
10112 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10113 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10114 (vect_gen_perm_mask_checked): Likewise.
10115 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10116 (vect_gen_perm_mask_checked): Likewise.
10117 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10118 (vectorizable_store): Likewise.
10119 (vectorizable_load): Likewise.
10120 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10121 (vectorizable_bswap): Likewise.
10123 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10124 Alan Hayward <alan.hayward@arm.com>
10125 David Sherwood <david.sherwood@arm.com>
10127 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10128 * tree.c (build_vector): Likewise.
10129 (build_vector_from_ctor): Update accordingly.
10130 (build_vector_from_val): Likewise.
10131 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10132 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10133 * tree-vect-generic.c (add_rshift): Likewise.
10134 (expand_vector_divmod): Likewise.
10135 (optimize_vector_constructor): Likewise.
10136 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10137 (vect_transform_slp_perm_load): Likewise.
10138 (vect_schedule_slp_instance): Likewise.
10139 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10140 (vectorizable_call): Likewise.
10141 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10142 * expmed.c (make_tree): Likewise.
10143 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10144 a vector passed to build_vector.
10145 (fold_convert_const): Likewise.
10146 (exact_inverse): Likewise.
10147 (fold_ternary_loc): Likewise.
10148 (fold_relational_const): Likewise.
10149 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10150 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10151 (const_unop): Likewise. Store the reduction accumulator in a
10152 variable rather than an array.
10153 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10154 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10155 the new vector, rather than constructing it after the input arrays.
10156 (native_interpret_vector): Use auto_vec<tree> when building
10157 a vector passed to build_vector. Add elements in order.
10158 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10159 auto_vec<tree> when building a vector passed to build_vector.
10160 (vect_create_epilog_for_reduction): Likewise.
10161 (vectorizable_induction): Likewise.
10162 (get_initial_def_for_reduction): Likewise. Fix indentation of
10164 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10166 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10167 passed to build_vector.
10169 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10170 Alan Hayward <alan.hayward@arm.com>
10171 David Sherwood <david.sherwood@arm.com>
10173 * tree-core.h (tree_base::u): Add an "nelts" field.
10174 (tree_vector): Use VECTOR_CST_NELTS as the length.
10175 * tree.c (tree_size): Likewise.
10176 (make_vector): Initialize VECTOR_CST_NELTS.
10177 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10178 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10179 TYPE_VECTOR_SUBPARTS.
10180 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10181 as the number of units, setting it from VECTOR_CST_NELTS.
10182 (const_vector_from_tree): Likewise.
10183 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10184 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10185 (fold_negate_expr_1): Likewise.
10186 (fold_convert_const): Likewise.
10187 (const_binop): Likewise. Differentiate the number of output and
10189 (const_unop): Likewise.
10190 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10191 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10192 in cases that did the opposite.
10194 2017-09-14 Richard Biener <rguenther@suse.de>
10196 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10199 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10201 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10203 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10206 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10207 if and where to split a bb, except for splitting before debug insn
10208 sequences followed by non-label real insn. Delete debug insns
10209 in between basic blocks.
10211 * combine.c (make_compound_operation_int): Formatting fixes.
10213 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10214 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10215 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10216 * config/sol2.h (LINK_EH_SPEC): Likewise.
10217 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10218 * config/s390/linux.h (LINK_SPEC): Likewise.
10219 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10220 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10221 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10222 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10223 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10224 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10225 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10226 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10228 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10230 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10232 (ENDFILE_LINUX_SPEC): Likewise.
10233 (LINK_EH_SPEC): Likewise.
10234 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10235 (LINK_OS_LINUX_SPEC32): Likewise.
10236 (LINK_OS_LINUX_SPEC64): Likewise.
10237 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10238 (LINK_OS_LINUX_SPEC): Likewise.
10240 2017-09-13 Martin Liska <mliska@suse.cz>
10242 PR middle-end/82154
10243 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10244 CASE_HIGH is NULL_TREE.
10246 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10247 Alan Hayward <alan.hayward@arm.com>
10248 David Sherwood <david.sherwood@arm.com>
10250 * target.def (secondary_memory_needed): New hook.
10251 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10252 instead of SECONDARY_MEMORY_NEEDED.
10253 (secondary_memory_needed_mode): Likewise.
10254 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10255 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10256 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10257 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10258 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10259 * doc/tm.texi: Regenerate.
10260 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10261 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10262 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10263 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10264 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10265 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10266 mode argument first and change the reg_class arguments to reg_class_t.
10267 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10268 Make static. Update the call to inline_secondary_memory_needed.
10269 (ix86_register_move_cost): Update the call to
10270 inline_secondary_memory_needed.
10271 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10272 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10274 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10275 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10277 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10278 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10279 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10280 and match hook interface. Add comment from mips.h.
10281 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10282 * config/mmix/mmix.md (truncdfsf2): Refer to
10283 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10285 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10286 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10287 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10288 (pa_secondary_memory_needed): New function.
10289 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10290 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10291 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10292 (pdp11_secondary_memory_needed): Make static and match hook interface.
10293 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10294 * config/powerpcspe/powerpcspe-protos.h
10295 (rs6000_secondary_memory_needed_ptr): Delete.
10296 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10298 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10299 (rs6000_option_override_internal): Assign to
10300 targetm.secondary_memory_needed rather than
10301 rs6000_secondary_memory_needed_ptr.
10302 (rs6000_secondary_memory_needed): Match hook interface.
10303 (rs6000_debug_secondary_memory_needed): Likewise.
10304 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10305 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10306 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10307 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10308 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10309 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10311 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10312 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10313 (rs6000_option_override_internal): Assign to
10314 targetm.secondary_memory_needed rather than
10315 rs6000_secondary_memory_needed_ptr.
10316 (rs6000_secondary_memory_needed): Match hook interface.
10317 (rs6000_debug_secondary_memory_needed): Likewise.
10318 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10319 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10320 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10321 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10322 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10323 (sparc_secondary_memory_needed): New function.
10324 * lra-constraints.c (check_and_process_move): Refer to
10325 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10327 (curr_insn_transform): Likewise.
10328 (process_alt_operands): Use targetm.secondary_memory_needed
10329 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10330 (check_secondary_memory_needed_p): Likewise.
10331 (choose_split_class): Likewise.
10332 * reload.c: Unconditionally include code that was previously
10333 conditional on SECONDARY_MEMORY_NEEDED.
10334 (push_secondary_reload): Use targetm.secondary_memory_needed
10335 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10336 (push_reload): Likewise.
10337 * reload1.c: Unconditionally include code that was previously
10338 conditional on SECONDARY_MEMORY_NEEDED.
10339 (choose_reload_regs): Use targetm.secondary_memory_needed
10340 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10341 (gen_reload): Likewise.
10342 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10344 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10345 Alan Hayward <alan.hayward@arm.com>
10346 David Sherwood <david.sherwood@arm.com>
10348 * target.def (secondary_memory_needed_mode): New hook:
10349 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10350 * targhooks.h (default_secondary_memory_needed_mode): New function.
10351 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10352 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10353 * doc/tm.texi: Regenerate.
10354 * lra-constraints.c (check_and_process_move): Use
10355 targetm.secondary_memory_needed_mode instead of
10356 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10357 (curr_insn_transform): Likewise.
10358 * reload.c (get_secondary_mem): Likewise.
10359 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10360 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10362 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10363 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10364 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10365 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10366 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10368 * config/powerpcspe/powerpcspe-protos.h
10369 (rs6000_secondary_memory_needed_mode): Delete.
10370 * config/powerpcspe/powerpcspe.c
10371 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10372 (rs6000_secondary_memory_needed_mode): Make static.
10373 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10374 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10376 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10378 (rs6000_secondary_memory_needed_mode): Make static.
10379 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10380 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10381 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10382 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10383 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10385 (sparc_secondary_memory_needed_mode): New function.
10386 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10388 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10390 * config/aarch64/constraints.md (Umq): New constraint.
10391 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10393 (mov<mode>): Update condition.
10395 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10397 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10398 when bitposition is the same.
10400 2017-09-13 Richard Biener <rguenther@suse.de>
10402 * dwarf2out.c (output_die_symbol): Remove.
10403 (output_die): Do not output a DIEs symbol.
10405 2017-09-13 Richard Biener <rguenther@suse.de>
10407 PR middle-end/82128
10408 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10409 default-def to avoid breaking iterator update with the weird
10410 interaction with cgraph_update_edges_for_call_stmt_node.
10412 2017-09-13 Richard Biener <rguenther@suse.de>
10414 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10415 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10416 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10417 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10419 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10421 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10422 Disable pc relative literal load irrespective of
10423 TARGET_FIX_ERR_A53_84341 for default.
10425 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10427 * config/sparc/sparc.c (output_return): Output the source location of
10428 the insn in the delay slot, if any.
10429 (output_sibcall): Likewise.
10431 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10434 * common.opt (-static-pie): New alias.
10435 (shared): Negate static-pie.
10436 (-no-pie): Update help text.
10438 (static-pie): New option.
10439 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10440 -static-pie support.
10441 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10442 (LINK_EH_SPEC): Likewise.
10443 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10444 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10445 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10446 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10447 (init_gcc_specs): Likewise.
10448 (init_spec): Likewise.
10449 (display_help): Update help message for -pie.
10450 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10453 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10455 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10456 (movdi_aarch64): Likewise.
10457 (movti_aarch64): Likewise.
10459 2017-09-12 Simon Wright <simon@pushface.org>
10462 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10463 calculation of the minor version, always output as 0.
10465 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10468 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10469 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10470 on it early, rather than manual conversion late. For
10471 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10472 instead of performing manual conversion.
10474 2017-09-12 Carl Love <cel@us.ibm.com>
10476 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10477 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10479 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10480 VMULOSW): Add definitions.
10481 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10482 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10483 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10484 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10485 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10487 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10489 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10491 (movti_aarch64): Likewise.
10492 (movdf_aarch64): Likewise.
10493 (movtf_aarch64): Likewise.
10494 (load_pairdi): Likewise.
10495 (store_pairdi): Likewise.
10496 (load_pairdf): Likewise.
10497 (store_pairdf): Likewise.
10498 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10499 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10500 (ldr_got_small_<mode>): Likewise.
10501 (ldr_got_small_28k_<mode>): Likewise.
10502 (ldr_got_tiny): Likewise.
10503 * config/aarch64/iterators.md (ldst_sz): New.
10504 (ldpstp_sz): Likewise.
10505 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10507 (thunderx_load): Split load_8 to load_16.
10508 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10510 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10511 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10512 (xgene1_store_pair): Split store_8 to store_16.
10513 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10514 (falkor_st_0_st_sd): Split store_8 to store_16.
10516 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10518 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10519 and store1/2/3/4 to store_4/8/12/16.
10520 * config/aarch64/aarch64.md: Update for rename.
10521 * config/arm/arm.md: Likewise.: Likewise.
10522 * config/arm/arm.c: Likewise.
10523 * config/arm/thumb1.md: Likewise.
10524 * config/arm/thumb2.md: Likewise.
10525 * config/arm/vfp.md: Likewise.
10526 * config/arm/arm-generic.md: Likewise.
10527 * config/arm/arm1020e.md: Likewise.
10528 * config/arm/arm1026ejs.md: Likewise.
10529 * config/arm/arm1136jfs.md: Likewise.
10530 * config/arm/arm926ejs.md: Likewise.
10531 * config/arm/cortex-a15.md: Likewise.
10532 * config/arm/cortex-a17.md: Likewise.
10533 * config/arm/cortex-a5.md: Likewise.
10534 * config/arm/cortex-a53.md: Likewise.
10535 * config/arm/cortex-a57.md: Likewise.
10536 * config/arm/cortex-a7.md: Likewise.
10537 * config/arm/cortex-a8.md: Likewise.
10538 * config/arm/cortex-a9.md: Likewise.
10539 * config/arm/cortex-m4.md: Likewise.
10540 * config/arm/cortex-m7.md: Likewise.
10541 * config/arm/cortex-r4.md: Likewise.
10542 * config/arm/exynos-m1.md: Likewise.
10543 * config/arm/fa526.md: Likewise.
10544 * config/arm/fa606te.md: Likewise.
10545 * config/arm/fa626te.md: Likewise.
10546 * config/arm/fa726te.md: Likewise.
10547 * config/arm/fmp626.md: Likewise.
10548 * config/arm/iwmmxt.md: Likewise.
10549 * config/arm/ldmstm.md: Likewise.
10550 * config/arm/marvell-pj4.md: Likewise.
10551 * config/arm/xgene1.md: Likewise.
10552 * config/aarch64/thunderx.md: Likewise.
10553 * config/aarch64/thunderx2t99.md: Likewise.
10554 * config/aarch64/falkor.md: Likewise.
10556 2017-09-12 Martin Liska <mliska@suse.cz>
10558 * attribs.c (private_lookup_attribute): New function.
10559 * attribs.h (private_lookup_attribute): Declared here.
10560 (lookup_attribute): Called from this place.
10562 2017-09-12 Richard Biener <rguenther@suse.de>
10564 PR tree-optimization/82157
10565 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10566 stmts with side-effects.
10568 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10569 Alan Hayward <alan.hayward@arm.com>
10570 David Sherwood <david.sherwood@arm.com>
10572 * target.def (hard_regno_nregs): New hook.
10573 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10574 * targhooks.h (default_hard_regno_nregs): Declare.
10575 * targhooks.c (default_hard_regno_nregs): New function.
10576 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10577 (TARGET_HARD_REGNO_NREGS): ...this hook.
10578 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10579 (CLASS_MAX_NREGS): Likewise.
10580 * doc/tm.texi: Regenerate.
10581 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10582 instead of HARD_REGNO_NREGS.
10583 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10584 HARD_REGNO_NREGS in the comment.
10585 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10586 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10587 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10588 Return an unsigned int.
10589 (TARGET_HARD_REGNO_NREGS): Redefine.
10590 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10591 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10592 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10593 (arc_hard_regno_nregs): New function.
10594 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10595 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10596 (arm_hard_regno_nregs): New function.
10597 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10598 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10599 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10600 (TARGET_HARD_REGNO_NREGS): Redefine.
10601 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10602 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10603 (HARD_REGNO_NREGS): Delete.
10604 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10605 (cr16_hard_regno_nregs): New function.
10606 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10607 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10608 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10609 (cris_hard_regno_nregs): New function.
10610 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10611 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10612 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10613 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10614 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10615 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10616 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10617 (frv_hard_regno_nregs): Make static. Take and return an
10619 (frv_class_max_nregs): Remove outdated copy of documentation.
10620 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10621 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10622 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10623 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10624 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10625 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10626 (TARGET_HARD_REGNO_NREGS): Redefine.
10627 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10628 (CLASS_MAX_NREGS): Update comment.
10629 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10630 (ia64_hard_regno_nregs): New function.
10631 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10632 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10633 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10634 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10635 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10637 (m32c_hard_regno_nregs): Likewise. Make static.
10638 (TARGET_HARD_REGNO_NREGS): Redefine.
10639 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10640 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10641 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10642 (m68k_hard_regno_nregs): New function.
10643 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10644 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10645 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10646 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10647 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10648 Take and return an unsigned int.
10649 (TARGET_HARD_REGNO_NREGS): Redefine.
10650 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10651 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10652 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10653 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10654 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10655 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10656 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10657 (msp430_hard_regno_nregs): Make static. Take and return an
10659 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10660 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10661 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10662 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10663 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10664 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10665 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10666 (TARGET_HARD_REGNO_NREGS): Redefine.
10667 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10668 (PA_HARD_REGNO_NREGS): ...this.
10669 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10670 (PA_HARD_REGNO_NREGS): ...this.
10671 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10672 (pa_hard_regno_nregs): New function.
10673 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10674 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10675 (pdp11_hard_regno_nregs): New function.
10676 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10677 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10678 (rs6000_hard_regno_nregs_hook): New function.
10679 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10680 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10681 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10682 Take and return an unsigned int. Move earlier in file.
10683 (TARGET_HARD_REGNO_NREGS): Redefine.
10684 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10685 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10686 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10687 (rl78_hard_regno_nregs): Make static. Take and return an
10689 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10690 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10691 (rs6000_hard_regno_nregs_hook): New function.
10692 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10693 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10694 (TARGET_HARD_REGNO_NREGS): Redefine.
10695 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10696 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10697 instead of HARD_REGNO_NREGS.
10698 (s390_hard_regno_nregs): New function.
10699 (s390_hard_regno_mode_ok): Add comment from s390.h.
10700 (TARGET_HARD_REGNO_NREGS): Redefine.
10701 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10702 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10703 (sh_hard_regno_nregs): New function.
10704 (sh_pass_in_reg_p): Use it.
10705 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10706 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10707 (sparc_hard_regno_nregs): New function.
10708 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10709 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10710 (spu_function_arg_advance): Use it, supplying a valid register number.
10711 (TARGET_HARD_REGNO_NREGS): Redefine.
10712 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10713 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10714 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10715 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10716 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10717 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10718 (CLASS_MAX_NREGS): Remove copy of old documentation.
10719 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10720 (visium_hard_regno_nregs): New function.
10721 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10722 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10723 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10724 xtensa_hard_regno_nregs): New function.
10725 * system.h (HARD_REGNO_NREGS): Poison.
10727 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10729 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10730 hard_regno_nregs instead of HARD_REGNO_NREGS.
10731 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10732 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10733 (c6x_expand_epilogue): Likewise.
10734 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10735 (frv_read_iacc_argument): Likewise.
10736 * config/sh/sh.c: Include regs.h.
10737 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10738 (regs_used): Likewise.
10739 (output_stack_adjust): Likewise.
10740 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10741 * expmed.c: Include regs.h.
10742 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10743 * ree.c: Include regs.h.
10744 (combine_reaching_defs): Use hard_regno_nregs instead of
10746 (add_removable_extension): Likewise.
10748 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10750 * regs.h (hard_regno_nregs): Turn into a function.
10751 (end_hard_regno): Update accordingly.
10752 * caller-save.c (setup_save_areas): Likewise.
10753 (save_call_clobbered_regs): Likewise.
10754 (replace_reg_with_saved_mem): Likewise.
10755 (insert_restore): Likewise.
10756 (insert_save): Likewise.
10757 * combine.c (can_change_dest_mode): Likewise.
10758 (move_deaths): Likewise.
10759 (distribute_notes): Likewise.
10760 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10761 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10762 (rs6000_split_multireg_move): Likewise.
10763 (rs6000_register_move_cost): Likewise.
10764 (rs6000_memory_move_cost): Likewise.
10765 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10766 (rs6000_split_multireg_move): Likewise.
10767 (rs6000_register_move_cost): Likewise.
10768 (rs6000_memory_move_cost): Likewise.
10769 * cselib.c (cselib_reset_table): Likewise.
10770 (cselib_lookup_1): Likewise.
10771 * emit-rtl.c (set_mode_and_regno): Likewise.
10772 * function.c (aggregate_value_p): Likewise.
10773 * ira-color.c (setup_profitable_hard_regs): Likewise.
10774 (check_hard_reg_p): Likewise.
10775 (calculate_saved_nregs): Likewise.
10776 (assign_hard_reg): Likewise.
10777 (improve_allocation): Likewise.
10778 (calculate_spill_cost): Likewise.
10779 * ira-emit.c (modify_move_list): Likewise.
10780 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10781 (ira_hard_reg_in_set_p): Likewise.
10782 * ira.c (setup_reg_mode_hard_regset): Likewise.
10783 (clarify_prohibited_class_mode_regs): Likewise.
10784 (check_allocation): Likewise.
10785 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10786 (lra_setup_reg_renumber): Likewise.
10787 (setup_try_hard_regno_pseudos): Likewise.
10788 (spill_for): Likewise.
10789 (assign_hard_regno): Likewise.
10790 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10791 * lra-constraints.c (in_class_p): Likewise.
10792 (lra_constraint_offset): Likewise.
10793 (simplify_operand_subreg): Likewise.
10794 (lra_constraints): Likewise.
10795 (split_reg): Likewise.
10796 (split_if_necessary): Likewise.
10797 (invariant_p): Likewise.
10798 (inherit_in_ebb): Likewise.
10799 * lra-lives.c (process_bb_lives): Likewise.
10800 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
10801 (get_hard_regs): Likewise.
10802 (do_remat): Likewise.
10803 * lra-spills.c (assign_spill_hard_regs): Likewise.
10804 * mode-switching.c (create_pre_exit): Likewise.
10805 * postreload.c (reload_combine_recognize_pattern): Likewise.
10806 * recog.c (peep2_find_free_register): Likewise.
10807 * regcprop.c (kill_value_regno): Likewise.
10808 (set_value_regno): Likewise.
10809 (copy_value): Likewise.
10810 (maybe_mode_change): Likewise.
10811 (find_oldest_value_reg): Likewise.
10812 (copyprop_hardreg_forward_1): Likewise.
10813 * regrename.c (check_new_reg_p): Likewise.
10814 (regrename_do_replace): Likewise.
10815 * reload.c (push_reload): Likewise.
10816 (combine_reloads): Likewise.
10817 (find_dummy_reload): Likewise.
10818 (operands_match_p): Likewise.
10819 (find_reloads): Likewise.
10820 (find_equiv_reg): Likewise.
10821 (reload_adjust_reg_for_mode): Likewise.
10822 * reload1.c (count_pseudo): Likewise.
10823 (count_spilled_pseudo): Likewise.
10824 (find_reg): Likewise.
10825 (clear_reload_reg_in_use): Likewise.
10826 (free_for_value_p): Likewise.
10827 (allocate_reload_reg): Likewise.
10828 (choose_reload_regs): Likewise.
10829 (reload_adjust_reg_for_temp): Likewise.
10830 (emit_reload_insns): Likewise.
10831 (delete_output_reload): Likewise.
10832 * rtlanal.c (subreg_get_info): Likewise.
10833 * sched-deps.c (sched_analyze_reg): Likewise.
10834 * sel-sched.c (init_regs_for_mode): Likewise.
10835 (mark_unavailable_hard_regs): Likewise.
10836 (choose_best_reg_1): Likewise.
10837 (verify_target_availability): Likewise.
10838 * valtrack.c (dead_debug_insert_temp): Likewise.
10839 * var-tracking.c (track_loc_p): Likewise.
10840 (emit_note_insn_var_location): Likewise.
10841 * varasm.c (make_decl_rtl): Likewise.
10842 * reginfo.c (choose_hard_reg_mode): Likewise.
10843 (init_reg_modes_target): Refer directly to
10844 this_target_regs->x_hard_regno_nregs.
10846 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10848 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10849 instead of hard_regno_nregs.
10851 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10853 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10854 end_hard_regno instead of hard_regno_nregs.
10855 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10856 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10857 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10858 * ira-color.c (improve_allocation): Likewise.
10859 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10860 * lra-lives.c (mark_regno_live): Likewise.
10861 (mark_regno_dead): Likewise.
10862 * lra-remat.c (operand_to_remat): Likewise.
10863 * lra.c (collect_non_operand_hard_regs): Likewise.
10864 * postreload.c (reload_combine_note_store): Likewise.
10865 (move2add_valid_value_p): Likewise.
10866 * reload.c (regno_clobbered_p): Likewise.
10868 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10870 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10872 * config/v850/v850.c (v850_reorg): Likewise.
10873 * reload.c (refers_to_regno_for_reload_p): Likewise.
10874 (find_equiv_reg): Likewise.
10875 * reload1.c (reload_reg_reaches_end_p): Likewise.
10877 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10879 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10881 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10882 * config/arm/arm.c (output_move_neon): Likewise.
10883 (arm_attr_length_move_neon): Likewise.
10884 (neon_split_vcombine): Likewise.
10885 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10886 (c6x_mark_reg_written): Likewise.
10887 (c6x_dwarf_register_span): Likewise.
10888 * config/i386/i386.c (ix86_save_reg): Likewise.
10889 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10890 (rws_access_reg): Likewise.
10891 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10892 * mode-switching.c (create_pre_exit): Likewise.
10893 * ree.c (combine_reaching_defs): Likewise.
10894 (add_removable_extension): Likewise.
10895 * regcprop.c (find_oldest_value_reg): Likewise.
10896 (copyprop_hardreg_forward_1): Likewise.
10897 * reload.c (reload_inner_reg_of_subreg): Likewise.
10898 (push_reload): Likewise.
10899 (combine_reloads): Likewise.
10900 (find_dummy_reload): Likewise.
10901 (reload_adjust_reg_for_mode): Likewise.
10902 * reload1.c (find_reload_regs): Likewise.
10903 (forget_old_reloads_1): Likewise.
10904 (reload_reg_free_for_value_p): Likewise.
10905 (reload_adjust_reg_for_temp): Likewise.
10906 (emit_reload_insns): Likewise.
10907 (delete_output_reload): Likewise.
10908 * sel-sched.c (choose_best_reg_1): Likewise.
10909 (choose_best_pseudo_reg): Likewise.
10911 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10912 Alan Hayward <alan.hayward@arm.com>
10913 David Sherwood <david.sherwood@arm.com>
10915 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10916 * target.def (slow_unaligned_access): New hook.
10917 * targhooks.h (default_slow_unaligned_access): Declare.
10918 * targhooks.c (default_slow_unaligned_access): New function.
10919 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10920 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10921 * doc/tm.texi: Regenerate.
10922 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10923 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10924 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10926 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10927 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10929 (rs6000_slow_unaligned_access): New function.
10930 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10931 (expand_block_compare): Likewise.
10932 (expand_strn_compare): Likewise.
10933 (rs6000_rtx_costs): Likewise.
10934 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10935 (riscv_slow_unaligned_access): Likewise.
10936 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10937 (riscv_slow_unaligned_access_p): ...this and make static.
10938 (riscv_option_override): Update accordingly.
10939 (riscv_slow_unaligned_access): New function.
10940 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10941 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10942 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10943 (rs6000_slow_unaligned_access): New function.
10944 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10945 (rs6000_rtx_costs): Likewise.
10946 * config/rs6000/rs6000-string.c (expand_block_compare)
10947 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10948 of SLOW_UNALIGNED_ACCESS.
10949 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10950 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10951 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10952 of SLOW_UNALIGNED_ACCESS.
10953 * expmed.c (simple_mem_bitfield_p): Likewise.
10954 * expr.c (alignment_for_piecewise_move): Likewise.
10955 (emit_group_load_1): Likewise.
10956 (emit_group_store): Likewise.
10957 (copy_blkmode_from_reg): Likewise.
10958 (emit_push_insn): Likewise.
10959 (expand_assignment): Likewise.
10960 (store_field): Likewise.
10961 (expand_expr_real_1): Likewise.
10962 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10963 * lra-constraints.c (simplify_operand_subreg): Likewise.
10964 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10965 * gimple-ssa-store-merging.c: Likewise in block comment at start
10967 * tree-ssa-strlen.c: Include target.h.
10968 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10969 of SLOW_UNALIGNED_ACCESS.
10970 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10972 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10974 PR rtl-optimization/82185
10975 * expmed.c (emit_store_flag_int): Only test tem if it has been
10978 2017-09-12 Richard Biener <rguenther@suse.de>
10980 PR middle-end/82149
10981 * match.pd ((FTYPE) N CMP CST): Fix typo.
10983 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10985 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10987 (mips_near_type_p): Add 'short_call' attribute as a synonym
10989 * doc/extend.texi (short_call): Document new function attribute.
10991 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10994 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10995 assertion check that in the condition.
10996 (get_atomic_generic_size): Likewise. Before testing if parameter
10997 has pointer type, if it has array type, call for C++
10998 default_conversion to perform array-to-pointer conversion.
11000 2017-09-12 Richard Biener <rguenther@suse.de>
11002 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11003 for operations we cannot scalarize.
11005 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11007 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11008 vectors heap vectors. Clean up comments.
11009 Make visited_bbs a reference.
11010 (profitable_jump_thread_path): Make GC
11011 vectors heap vectors. Clean up comments.
11013 (convert_and_register_jump_thread_path): Make GC vectors heap
11015 (check_subpath_and_update_thread_path): Same. Clean up comments.
11016 Make visited_bbs a reference.
11017 (handle_phi): Abstract common code to to
11018 register_jump_thread_path_if_profitable.
11019 Rename VAR_BB to DEF_BB.
11021 Make GC vectors heap vectors.
11022 Make visited_bbs a reference.
11023 (handle_assignment): Same.
11024 (register_jump_thread_path_if_profitable): New.
11025 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11027 Make GC vectors heap vectors. Clean up comments.
11028 Make visited_bbs a reference.
11029 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11030 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11033 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11036 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11037 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11039 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11041 Revert r251800 and r251799.
11043 2017-09-11 Martin Jambor <mjambor@suse.cz>
11046 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11047 arguments in advance.
11048 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11049 use it to find predecessor edges.
11050 (naive_outof_ssa): Collect vector of predecessors.
11052 2017-09-08 Jason Merrill <jason@redhat.com>
11054 PR c++/70029 - ICE with ref-qualifier and -flto
11055 * langhooks.h (struct lang_hooks_for_types): Add
11056 copy_lang_qualifiers.
11057 * attribs.c (build_type_attribute_qual_variant): Use it.
11058 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11060 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11061 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11063 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11066 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11067 (*mulsi3_sp64): New instruction.
11068 (mulsi3): New expander.
11070 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11072 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11074 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11076 * sancov.c: Include memmodel.h.
11078 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11081 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11084 2017-09-07 Carl Love <cel@us.ibm.com>
11086 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11087 the sldi instruction.
11089 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11091 * sancov.c: Include tm_p.h.
11093 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11096 * output.h (switch_to_other_text_partition): New declaration.
11097 * varasm.c (switch_to_other_text_partition): New function.
11098 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11099 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11100 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11101 to the other text partition before emitting LCL label and switch back
11102 after emitting the word after it.
11104 2017-09-07 Richard Biener <rguenther@suse.de>
11106 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11107 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11108 critical edges here, after loop init.
11109 (pass_data_pre): Remove PROP_no_crit_edges flags.
11110 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11111 for valueization of call args to avoid leaking VN_TOP.
11112 (visit_use): Assert we do not visit default defs.
11113 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11114 Use error_mark_node to more easily detect leaking VN_TOP.
11115 All default-defs are varying, not VN_TOP. Mark them visited.
11116 (run_scc_vn): Make code match comment.
11118 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11120 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11121 OPTION_MASK_FLOAT128_KEYWORD.
11122 (POWERPC_MASKS): Likewise.
11123 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11124 support for the -mfloat128-type option, and make -mfloat128
11125 default on PowerPC Linux systems. Define or undefine
11126 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11127 Define __float128 to be __ieee128 if IEEE 128-bit support is
11128 enabled, or undefine it.
11129 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11130 Delete defining __FLOAT128_TYPE__.
11131 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11132 -mfloat128-type option and make -mfloat128 default on PowerPC
11134 (TARGET_FLOAT128_TYPE): Likewise.
11135 (-mfloat128-type): Likewise.
11136 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11137 Delete the -mfloat128-type option and make -mfloat128 default on
11138 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11139 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11140 128-bit floating point is enabled. Change tests from using
11141 -mfloat128-type to -mfloat128.
11142 (rs6000_mangle_type): Use the correct mangling for the __float128
11143 type even if normal long double is restricted to 64-bits.
11144 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11146 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11147 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11148 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11150 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11151 documentation for -mfloat128.
11153 2017-09-06 Olivier Hainque <hainque@adacore.com>
11155 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11157 2017-09-06 Wish Wu <wishwu007@gmail.com>
11158 Jakub Jelinek <jakub@redhat.com>
11160 * asan.c (initialize_sanitizer_builtins): Add
11161 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11162 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11163 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11164 BT_FN_VOID_UINT64_PTR variables.
11165 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11166 (BT_FN_VOID_UINT16_UINT16): Likewise.
11167 (BT_FN_VOID_UINT32_UINT32): Likewise.
11168 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11169 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11170 (BT_FN_VOID_UINT64_PTR): Likewise.
11171 * common.opt (flag_sanitize_coverage): New variable.
11172 (fsanitize-coverage=trace-pc): Remove.
11173 (fsanitize-coverage=): Add.
11174 * flag-types.h (enum sanitize_coverage_code): New enum.
11175 * fold-const.c (fold_range_test): Disable non-short-circuit
11176 optimization if flag_sanitize_coverage.
11177 (fold_truth_andor): Likewise.
11178 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11179 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11180 (coverage_sanitizer_opts): New array.
11181 (get_closest_sanitizer_option): Add OPTS argument, handle also
11182 OPT_fsanitize_coverage_.
11183 (parse_sanitizer_options): Adjusted to also handle
11184 OPT_fsanitize_coverage_.
11185 (common_handle_option): Add OPT_fsanitize_coverage_.
11186 * sancov.c (instrument_comparison, instrument_switch): New function.
11187 (sancov_pass): Add trace-cmp support.
11188 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11189 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11190 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11191 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11192 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11193 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11194 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11195 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11196 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11197 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11199 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11201 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11202 error. Only quit immediately if parsing is complete.
11203 (BEGIN): Initialize fatal_err and parse_done.
11204 (begin fpu, end fpu): Check number of arguments.
11205 (begin arch, end arch): Likewise.
11206 (begin cpu, end cpu): Likewise.
11207 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11208 (optalias): Likewise.
11210 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11212 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11213 * config/arm/arm-isa.h: Delete. Move definitions to ...
11214 * arm-cpus.in: ... here. Use new feature and fgroup values.
11215 * config/arm/arm.c (arm_option_override): Use lower case for feature
11217 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11218 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11219 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11221 (print_isa_bits_for): New function.
11222 (gen_isa): New function.
11223 (gen_comm_data): Use print_isa_bits_for.
11224 (define feature): New keyword.
11225 (define fgroup): New keyword.
11226 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11227 (arm-isa.h): Add rule to generate file.
11228 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11229 case for feature bit names.
11231 2017-09-06 Richard Biener <rguenther@suse.de>
11233 * tree-ssa-pre.c (NECESSARY): Remove.
11234 (create_expression_by_pieces): Do not touch pass-local flags.
11235 (insert_into_preds_of_block): Likewise.
11236 (do_pre_regular_insertion): Likewise.
11237 (eliminate_insert): Likewise.
11238 (eliminate_dom_walker::before_dom_children): Likewise.
11239 (fini_eliminate): Do not look at inserted_exprs.
11240 (mark_operand_necessary): Remove.
11241 (remove_dead_inserted_code): Replace with simple work-list
11242 algorithm based on inserted_exprs and SSA uses.
11243 (pass_pre::execute): Re-order fini_eliminate and
11244 remove_dead_inserted_code.
11246 2017-09-06 Olivier Hainque <hainque@adacore.com>
11248 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11249 for VxWorks 7. Adjust surrounding comments.
11251 2017-09-06 Richard Biener <rguenther@suse.de>
11253 * gimple-ssa-strength-reduction.c
11254 (find_candidates_dom_walker::before_dom_children): Also allow
11257 2017-09-06 Richard Biener <rguenther@suse.de>
11259 PR tree-optimization/82108
11260 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11261 for gap in the non-permutation SLP case.
11263 2017-09-06 Martin Jambor <mjambor@suse.cz>
11265 PR tree-optimization/82078
11266 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11267 add_access_to_work_queue...
11268 (build_accesses_from_assign): ...here.
11269 (propagate_all_subaccesses): Make sure racc is the group
11270 representative, if there is one.
11272 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11274 PR middle-end/82095
11275 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11278 2017-09-06 Richard Biener <rguenther@suse.de>
11280 * gimple-ssa-strength-reduction.c
11281 (find_candidates_dom_walker::before_doom_children): Use a
11282 type and not a mode check.
11284 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11287 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11289 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11291 2017-09-05 Jeff Law <law@redhat.com>
11293 PR tree-optimization/64910
11294 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11295 cases where we have 3 or more operands.
11297 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11299 PR middle-end/81768
11300 * omp-low.c (lower_omp_for): Recompute tree invariant if
11301 gimple_omp_for_initial/final is ADDR_EXPR.
11303 PR middle-end/81768
11304 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11305 into gimple val before gimplification fo the COND_EXPR.
11307 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11309 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11310 REGION_COPY argument.
11311 (thread_through_all_blocks): Remove unused argument to
11312 duplicate_thread_path.
11314 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11315 Alan Hayward <alan.hayward@arm.com>
11316 David Sherwood <david.sherwood@arm.com>
11318 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11319 Take a scalar_mode rather than a machine_mode.
11320 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11321 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11322 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11323 (aarch64_gen_adjusted_ldpstp): Likewise.
11324 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11326 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11327 Alan Hayward <alan.hayward@arm.com>
11328 David Sherwood <david.sherwood@arm.com>
11330 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11331 Take a scalar_int_mode instead of a machine_mode.
11332 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11333 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11334 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11335 (aarch64_simd_attr_length_rglist): Delete.
11336 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11337 a scalar_int_mode instead of a machine_mode.
11338 (aarch64_add_offset): Likewise.
11339 (aarch64_internal_mov_immediate): Likewise
11340 (aarch64_add_constant_internal): Likewise.
11341 (aarch64_add_constant): Likewise.
11342 (aarch64_movw_imm): Likewise.
11343 (aarch64_rtx_arith_op_extract_p): Likewise.
11344 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11345 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11346 Remove assert that the mode isn't a vector.
11347 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11348 (aarch64_expand_mov_immediate): Update calls after above changes.
11349 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11350 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11351 (aarch64_move_imm): Likewise.
11352 (aarch64_can_const_movi_rtx_p): Likewise.
11353 (aarch64_strip_extend): Likewise.
11354 (aarch64_extr_rtx_p): Likewise.
11355 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11356 a CONST_INT when the mode parameter is VOIDmode.
11357 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11359 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11361 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11362 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11363 (bitwise_type_for_mode): Update accordingly.
11365 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11367 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11368 * stor-layout.c (mode_for_size_tree): Likewise.
11369 (mode_for_array): Update accordingly.
11370 (layout_decl): Likewise.
11371 (compute_record_mode): Likewise. Only set the mode once.
11373 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11375 * target.def (get_mask_mode): Change return type to opt_mode.
11377 * doc/tm.texi: Regenerate.
11378 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11379 * targhooks.c (default_get_mask_mode): Likewise.
11380 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11381 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11382 targetm.get_mask_mode.
11383 * tree.c (build_truth_vector_type): Likewise.
11385 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11387 * machmode.h (mode_for_vector): Return an opt_mode.
11388 * stor-layout.c (mode_for_vector): Likewise.
11389 (mode_for_int_vector): Update accordingly.
11390 (layout_type): Likewise.
11391 * config/i386/i386.c (emit_memmov): Likewise.
11392 (ix86_expand_set_or_movmem): Likewise.
11393 (ix86_expand_vector_init): Likewise.
11394 (ix86_get_mask_mode): Likewise.
11395 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11397 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11398 * expmed.c (extract_bit_field_1): Likewise.
11399 * expr.c (expand_expr_real_2): Likewise.
11400 * optabs-query.c (can_vec_perm_p): Likewise.
11401 (can_vec_mask_load_store_p): Likewise.
11402 * optabs.c (expand_vec_perm): Likewise.
11403 * targhooks.c (default_get_mask_mode): Likewise.
11404 * tree-vect-stmts.c (vectorizable_store): Likewise.
11405 (vectorizable_load): Likewise.
11406 (get_vectype_for_scalar_type_and_size): Likewise.
11408 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11410 * machmode.h (mode_for_int_vector): New function.
11411 * stor-layout.c (mode_for_int_vector): Likewise.
11412 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11413 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11414 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11415 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11416 (s390_expand_vcond): Likewise.
11418 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11420 * machmode.h (opt_machine_mode): New type.
11421 (opt_mode<T>): Allow construction from anything that can be
11423 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11424 (mode_for_size): Return an opt_machine_mode.
11425 * stor-layout.c (mode_for_size): Likewise.
11426 (mode_for_size_tree): Update call accordingly.
11427 (bitwise_mode_for_mode): Likewise.
11428 (make_fract_type): Likewise.
11429 (make_accum_type): Likewise.
11430 * caller-save.c (replace_reg_with_saved_mem): Update call
11432 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11433 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11434 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11435 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11436 * expmed.c (extract_bit_field_1): Likewise.
11437 * reload.c (get_secondary_mem): Likewise.
11438 * varasm.c (assemble_integer): Likewise.
11439 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11442 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11444 * machmode.h (decimal_float_mode_for_size): New function.
11445 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11446 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11447 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11448 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11449 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11450 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11452 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11454 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11455 (get_builtin_sync_mode): Likewise.
11456 (expand_ifn_atomic_compare_exchange): Likewise.
11457 (expand_builtin_atomic_clear): Likewise.
11458 (expand_builtin_atomic_test_and_set): Likewise.
11459 (fold_builtin_atomic_always_lock_free): Likewise.
11460 * calls.c (compute_argument_addresses): Likewise.
11461 (emit_library_call_value_1): Likewise.
11462 (store_one_arg): Likewise.
11463 * combine.c (combine_instructions): Likewise.
11464 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11465 * config/arm/arm.c (arm_function_value): Likewise.
11466 (aapcs_allocate_return_reg): Likewise.
11467 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11468 * config/i386/i386.c (construct_container): Likewise.
11469 (ix86_gimplify_va_arg): Likewise.
11470 (ix86_expand_sse_cmp): Likewise.
11471 (emit_memmov): Likewise.
11472 (emit_memset): Likewise.
11473 (expand_small_movmem_or_setmem): Likewise.
11474 (ix86_expand_pextr): Likewise.
11475 (ix86_expand_pinsr): Likewise.
11476 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11477 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11479 * config/mips/mips.c (mips_function_value_1) Likewise.
11480 (mips_block_move_straight): Likewise.
11481 (mips_expand_ins_as_unaligned_store): Likewise.
11482 * config/powerpcspe/powerpcspe.c
11483 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11484 (rs6000_darwin64_record_arg_flush): Likewise.
11485 * config/rs6000/rs6000.c
11486 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11487 (rs6000_darwin64_record_arg_flush): Likewise.
11488 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11489 (sparc_function_value_1): Likewise.
11490 * config/spu/spu.c (adjust_operand): Likewise.
11491 (spu_emit_branch_or_set): Likewise.
11492 (arith_immediate_p): Likewise.
11493 * emit-rtl.c (gen_lowpart_common): Likewise.
11494 * expr.c (expand_expr_real_1): Likewise.
11495 * function.c (assign_parm_setup_block): Likewise.
11496 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11497 * reload1.c (alter_reg): Likewise.
11498 * stor-layout.c (mode_for_vector): Likewise.
11499 (layout_type): Likewise.
11501 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11503 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11504 (spu_convert_move): Likewise.
11505 * lower-subreg.c (resolve_simple_move): Likewise.
11507 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11510 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11511 define_insn to a define_expand.
11512 (altivec_vsum2sws_direct): New define_insn.
11513 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11515 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11517 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11520 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11525 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11526 realign stack if stack alignment needed is less than incoming
11529 2017-09-05 Marek Polacek <polacek@redhat.com>
11532 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11535 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11537 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11539 2017-09-05 Richard Biener <rguenther@suse.de>
11541 PR tree-optimization/82084
11542 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11544 2017-09-05 Richard Biener <rguenther@suse.de>
11546 PR tree-optimization/82102
11547 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11549 2017-09-05 Martin Liska <mliska@suse.cz>
11551 PR tree-optimization/82032
11552 * tree-cfg.c (generate_range_test): New function.
11553 * tree-cfg.h (generate_range_test): Declared here.
11554 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11555 (cleanup_control_expr_graph): Use it.
11556 * tree-switch-conversion.c (try_switch_expansion): Remove
11558 (emit_case_nodes): Use generate_range_test.
11560 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11563 * config/i386/i386.md (*<btsc><mode>_mask): Add
11564 TARGET_USE_BT to insn constraint.
11565 (*btr<mode>_mask): Ditto.
11567 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11569 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11570 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11572 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11575 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11576 TARGET_NEON and TARGET_IWMMXT.
11577 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11578 TARGET_NEON and TARGET_IWMMXT.
11579 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11581 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11583 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11584 (ix86_rewrite_tls_address): Ditto.
11585 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11586 (ix86_rewrite_tls_address_1): Ditto.
11587 (ix86_rewrite_tls_address): Ditto.
11588 * config/i386/predicates.md (tls_address_pattern): New predicate.
11589 * config/i386/i386.md (TLS address splitter): New splitter.
11591 2017-09-04 Richard Biener <rguenther@suse.de>
11593 PR tree-optimization/82084
11594 * fold-const.h (can_native_encode_string_p): Declare.
11595 * fold-const.c (can_native_encode_string_p): Factor out from ...
11596 (native_encode_string): ... here.
11597 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11598 vectorizing stores from constants we later cannot handle.
11600 2017-09-04 Marek Polacek <polacek@redhat.com>
11603 * doc/invoke.texi: Update -Wtautological-compare documentation.
11605 2017-09-04 Jeff Law <law@redhat.com>
11607 PR tree-optimization/64910
11608 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11609 swap the first and last operand if the last is a constant.
11611 2017-09-04 Marek Polacek <polacek@redhat.com>
11614 * convert.c (do_narrow): When sanitizing signed integer overflows,
11615 bail out for signed types.
11616 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11618 2017-09-04 Richard Biener <rguenther@suse.de>
11620 PR tree-optimization/82060
11621 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11622 Move devirtualization after stmt folding and before EH/AB/noreturn
11623 cleanup to get the stmt refs canonicalized. Use a bool instead
11624 of gimple_modified_p since that doesn't work for NOPs. Schedule
11625 NOPs generated by folding for removal.
11627 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11628 Alan Hayward <alan.hayward@arm.com>
11629 David Sherwood <david.sherwood@arm.com>
11631 * coretypes.h (pad_direction): New enum.
11632 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11633 (FUNCTION_ARG_PADDING): Likewise.
11634 * target.def (function_arg_padding): New hook.
11635 * targhooks.h (default_function_arg_padding): Declare.
11636 * targhooks.c (default_function_arg_padding): New function.
11637 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11638 (TARGET_FUNCTION_ARG_PADDING): ...this.
11639 * doc/tm.texi: Regenerate.
11640 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11641 instead of direction.
11642 (compute_argument_addresses): Likewise.
11643 (load_register_parameters): Likewise.
11644 (emit_library_call_value_1): Likewise.
11645 (store_one_arg): Use targetm.calls.function_arg_padding instead
11646 of FUNCTION_ARG_PADDING.
11647 (must_pass_in_stack_var_size_or_pad): Likewise.
11648 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11649 (emit_group_store): Likewise.
11650 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11651 instead of FUNCTION_ARG_PADDING.
11652 (emit_push_insn): Likewise, and propagate enum change throughout
11654 * function.h (direction): Delete.
11655 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11657 * function.c (assign_parm_find_stack_rtl): Likewise.
11658 (assign_parm_setup_block_p): Likewise.
11659 (assign_parm_setup_block): Likewise.
11660 (gimplify_parameters): Likewise.
11661 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11662 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11664 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11665 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11666 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11667 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11668 (aarch64_function_arg_padding): ...this new function.
11669 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11670 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11671 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11672 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11673 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11674 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11675 (arm_pad_arg_upward): Replace with...
11676 (arm_function_arg_padding): ...this new function.
11677 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11679 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11680 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11681 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11682 (ia64_hpux_function_arg_padding): Replace with...
11683 (ia64_function_arg_padding): ...this new function. Use pad_direction
11684 instead of direction. Check for TARGET_HPUX.
11685 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11686 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11687 (iq2000_function_arg_padding): New function.
11688 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11689 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11690 (mips_function_arg_padding): ...this new function.
11691 (mips_pad_reg_upward): Update accordingly.
11692 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11693 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11694 targetm.calls.function_arg_padding.
11695 (FUNCTION_ARG_PADDING): Delete.
11696 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11697 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11698 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11699 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11700 (nios2_block_reg_padding): Return pad_direction instead of direction.
11701 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11702 instead of direction.
11703 (nios2_function_arg_padding): Likewise. Make static.
11704 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11705 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11706 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11707 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11708 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11709 (pa_function_arg_padding): Make static. Return pad_direction instead
11711 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11712 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11713 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11714 instead of direction. Use targetm.calls.function_arg_padding.
11715 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11716 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11717 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11718 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11719 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11721 (function_arg_padding): Rename to...
11722 (rs6000_function_arg_padding): ...this. Make static. Return
11723 pad_direction instead of direction.
11724 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11725 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11726 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11727 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11728 instead of direction. Use targetm.calls.function_arg_padding.
11729 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11730 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11731 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11732 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11733 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11734 (function_arg_padding): Rename to...
11735 (rs6000_function_arg_padding): ...this. Make static. Return
11736 pad_direction instead of direction.
11737 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11738 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11739 * config/s390/s390.c (s390_function_arg_padding): New function.
11740 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11741 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11742 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11743 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11744 (function_arg_padding): Rename to...
11745 (sparc_function_arg_padding): ...this. Make static. Return
11746 pad_direction instead of direction.
11747 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11748 * config/spu/spu.c (spu_function_arg_padding): New function.
11749 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11750 * system.h (FUNCTION_ARG_PADDING): Poison.
11752 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11753 Alan Hayward <alan.hayward@arm.com>
11754 David Sherwood <david.sherwood@arm.com>
11756 * target.def (modes_tieable_p): New hook.
11757 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11758 (TARGET_MODES_TIEABLE_P): ...this.
11759 * doc/tm.texi.in: Regenerate.
11760 * hooks.h (hook_bool_mode_mode_true): Declare.
11761 * hooks.c (hook_bool_mode_mode_true): New function.
11762 * combine.c (subst): Use targetm.modes_tieable_p instead of
11764 * dse.c (find_shift_sequence): Likewise.
11765 * expmed.c (extract_low_bits): Likewise.
11766 * lower-subreg.c: Include target.h.
11767 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11769 * rtlanal.c (rtx_cost): Likewise.
11770 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11771 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11772 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11773 (TARGET_MODES_TIEABLE_P): Redefine.
11774 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11775 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11776 (TARGET_MODES_TIEABLE_P): Redefine.
11777 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11778 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11779 (arc_modes_tieable_p): New function.
11780 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11781 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11782 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11783 (arm_modes_tieable_p): Make static.
11784 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11785 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11786 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11787 (TARGET_MODES_TIEABLE_P): Redefine.
11788 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11789 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11790 (TARGET_MODES_TIEABLE_P): Redefine.
11791 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11792 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11793 (cr16_modes_tieable_p): New function.
11794 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11795 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11796 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11797 (TRULY_NOOP_TRUNCATION): Update comment.
11798 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
11799 (TRULY_NOOP_TRUNCATION): Update comment.
11800 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
11801 (frv_modes_tieable_p): New function.
11802 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
11803 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
11804 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
11805 (TARGET_MODES_TIEABLE_P): Redefine.
11806 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
11807 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
11808 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
11809 (TARGET_MODES_TIEABLE_P): Redefine.
11810 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
11811 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
11812 (ia64_modes_tieable_p): New function.
11813 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
11814 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
11815 (iq2000_modes_tieable_p): New function.
11816 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
11817 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
11818 (lm32_modes_tieable_p): New function.
11819 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
11820 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
11821 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
11822 (TARGET_MODES_TIEABLE_P): Redefine.
11823 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
11824 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
11825 (m32r_modes_tieable_p): New function.
11826 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
11827 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
11828 (m68k_modes_tieable_p): New function.
11829 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
11830 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
11831 (mcore_modes_tieable_p): New function.
11832 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11833 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11835 (TARGET_MODES_TIEABLE_P): Redefine.
11836 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11837 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11838 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11839 (TARGET_MODES_TIEABLE_P): Redefine.
11840 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11841 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11842 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11843 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11844 (mn10300_modes_tieable_p): ...this and make static.
11845 (TARGET_MODES_TIEABLE_P): Redefine.
11846 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11847 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11848 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11849 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11850 (msp430_modes_tieable_p): Make static.
11851 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11852 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11853 (TARGET_MODES_TIEABLE_P): Redefine.
11854 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11855 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11856 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11857 (TARGET_MODES_TIEABLE_P): Redefine.
11858 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11859 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11860 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11861 (TARGET_MODES_TIEABLE_P): Redefine.
11862 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11863 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11864 (pdp11_modes_tieable_p): New function.
11865 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11866 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11867 (rs6000_modes_tieable_p): New function.
11868 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11869 * config/powerpcspe/powerpcspe.md: Update comment.
11870 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11871 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11872 (TARGET_MODES_TIEABLE_P): Redefine.
11873 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11874 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11875 (rl78_modes_tieable_p): New function.
11876 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11877 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11878 (rs6000_modes_tieable_p): New function.
11879 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11880 * config/rs6000/rs6000.md: Update comment.
11881 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11882 * config/rx/rx.c (rx_modes_tieable_p): New function.
11883 (TARGET_MODES_TIEABLE_P): Redefine.
11884 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11885 * config/s390/s390.c (s390_modes_tieable_p): New function.
11886 (TARGET_MODES_TIEABLE_P): Redefine.
11887 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11888 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11889 (sh_modes_tieable_p): New function.
11890 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11891 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11892 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11893 (sparc_modes_tieable_p): Make static.
11894 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11895 * config/spu/spu.c (spu_modes_tieable_p): New function.
11896 (TARGET_MODES_TIEABLE_P): Redefine.
11897 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11898 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11899 (TARGET_MODES_TIEABLE_P): Redefine.
11900 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11901 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11902 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11903 * config/v850/v850.c (v850_modes_tieable_p): New function.
11904 (TARGET_MODES_TIEABLE_P): Redefine.
11905 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11906 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11907 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11908 (visium_modes_tieable_p): New function.
11909 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11910 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11911 (xtensa_modes_tieable_p): New function.
11912 * system.h (MODES_TIEABLE_P): Poison.
11914 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11915 Alan Hayward <alan.hayward@arm.com>
11916 David Sherwood <david.sherwood@arm.com>
11918 * target.def (hard_regno_mode_ok): New hook.
11919 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11920 (TARGET_HARD_REGNO_MODE_OK): ...this.
11921 * doc/tm.texi.in: Regenerate.
11922 * hooks.h (hook_bool_uint_mode_true): Declare.
11923 * hooks.c (hook_bool_uint_mode_true): New function.
11924 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11925 HARD_REGNO_MODE_OK.
11926 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11927 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11928 instead of HARD_REGNO_MODE_OK.
11929 * caller-save.c: Include target.h.
11930 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11931 HARD_REGNO_MODE_OK.
11932 * combine.c (can_combine_p): Likewise.
11933 (combinable_i3pat): Likewise.
11934 (can_change_dest_mode): Likewise.
11935 * expr.c (init_expr_target): Likewise.
11936 (convert_move): Likewise.
11937 (convert_modes): Likewise.
11938 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11939 (setup_prohibited_mode_move_regs): Likewise.
11940 * ira.h (target_ira): Likewise.
11941 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11942 * lra-constraints.c (process_alt_operands): Likewise.
11943 (split_reg): Likewise.
11944 * recog.c (peep2_find_free_register): Likewise.
11945 * ree.c (combine_reaching_defs): Likewise.
11946 * regcprop.c (maybe_mode_change): Likewise.
11947 * reginfo.c (init_reg_sets_1): Likewise.
11948 (choose_hard_reg_mode): Likewise.
11949 (simplifiable_subregs): Likewise.
11950 * regrename.c (check_new_reg_p): Likewise.
11951 * reload.c (find_valid_class): Likewise.
11952 (find_valid_class_1): Likewise.
11953 (reload_inner_reg_of_subreg): Likewise.
11954 (push_reload): Likewise.
11955 (combine_reloads): Likewise.
11956 (find_dummy_reload): Likewise.
11957 (find_reloads): Likewise.
11958 * reload1.c (find_reg): Likewise.
11959 (set_reload_reg): Likewise.
11960 (allocate_reload_reg): Likewise.
11961 (choose_reload_regs): Likewise.
11962 (reload_adjust_reg_for_temp): Likewise.
11963 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11964 (simplify_subreg_regno): Likewise.
11965 * sel-sched.c (init_regs_for_mode): Likewise.
11966 * varasm.c (make_decl_rtl): Likewise.
11967 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11968 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11969 HARD_REGNO_MODE_OK.
11970 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11971 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11973 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11974 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11976 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11977 (arc_mode_class): Delete.
11978 (HARD_REGNO_MODE_OK): Delete.
11979 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11980 (arc_hard_regno_mode_ok): Rename old array to...
11981 (arc_hard_regno_mode_ok_modes): ...this.
11982 (arc_conditional_register_usage): Update accordingly.
11983 (arc_mode_class): Make static.
11984 (arc_hard_regno_mode_ok): New function.
11985 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11986 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11987 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11988 (arm_hard_regno_mode_ok): Make static.
11989 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11990 HARD_REGNO_MODE_OK.
11991 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11992 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11993 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11996 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11997 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11998 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11999 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12001 * config/bfin/predicates.md (valid_reg_operand): Use
12002 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12003 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12004 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12005 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12006 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12007 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12008 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12009 (cr16_hard_regno_mode_ok): Make static and return a bool.
12010 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12011 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12012 (cris_hard_regno_mode_ok): New function.
12013 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12014 (epiphany_mode_class): Delete.
12015 (HARD_REGNO_MODE_OK): Delete.
12016 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12017 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12018 (hard_regno_mode_ok): Rename to...
12019 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12020 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12021 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12022 HARD_REGNO_MODE_OK.
12023 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12024 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12025 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12026 (frv_hard_regno_mode_ok): Make static and return a bool.
12027 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12028 HARD_REGNO_MODE_OK.
12029 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12030 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12031 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12032 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12035 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12036 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12037 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12040 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12041 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12042 (ia64_hard_regno_mode_ok): New function.
12043 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12044 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12045 (iq2000_hard_regno_mode_ok): New function.
12046 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12047 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12048 (lm32_hard_regno_mode_ok): New function.
12049 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12050 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12051 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12052 instead of HARD_REGNO_MODE_OK.
12053 (m32c_hard_regno_ok): Rename to...
12054 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12055 (m32c_cannot_change_mode_class): Update accordingly.
12056 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12057 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12058 (m32r_mode_class): Delete.
12059 (HARD_REGNO_MODE_OK): Delete.
12060 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12061 (m32r_hard_regno_mode_ok): Rename to...
12062 (m32r_hard_regno_modes): ...this.
12063 (m32r_mode_class): Make static.
12064 (m32r_hard_regno_mode_ok): New function.
12065 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12066 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12067 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12068 (m68k_hard_regno_mode_ok): Make static.
12069 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12070 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12071 (mcore_hard_regno_mode_ok): New function.
12072 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12073 (HARD_REGNO_MODE_OK): Delete.
12074 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12076 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12077 (microblaze_hard_regno_mode_ok): New function.
12078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12079 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12080 (mips_hard_regno_mode_ok): Delete.
12081 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12082 (mips_hard_regno_mode_ok_p): ...this and make static.
12083 (mips_hard_regno_mode_ok_p): Rename to...
12084 (mips_hard_regno_mode_ok_uncached): ...this.
12085 (mips_hard_regno_mode_ok): New function.
12086 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12087 of HARD_REGNO_MODE_OK.
12088 (mips_option_override): Update after above name changes.
12089 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12090 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12091 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12092 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12093 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12094 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12095 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12096 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12097 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12098 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12099 (msp430_hard_regno_mode_ok): Make static and return a bool.
12100 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12101 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12102 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12104 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12105 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12106 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12107 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12108 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12109 (PA_HARD_REGNO_MODE_OK): ...this
12110 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12111 (PA_HARD_REGNO_MODE_OK): ...this.
12112 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12113 (pa_hard_regno_mode_ok): New function.
12114 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12115 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12116 (pdp11_hard_regno_mode_ok): New function.
12117 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12118 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12120 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12122 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12123 (rs6000_hard_regno_mode_ok): Rename to...
12124 (rs6000_hard_regno_mode_ok_uncached): ...this.
12125 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12126 (rs6000_hard_regno_mode_ok): New function.
12127 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12128 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12129 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12130 (riscv_hard_regno_mode_ok): ...this and make static.
12131 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12132 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12133 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12134 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12135 (rl78_hard_regno_mode_ok): Make static and return bool.
12136 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12137 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12139 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12141 (rs6000_hard_regno_mode_ok): Rename to...
12142 (rs6000_hard_regno_mode_ok_uncached): ...this.
12143 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12144 (rs6000_hard_regno_mode_ok): New function.
12145 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12146 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12147 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12148 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12149 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12150 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12152 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12153 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12154 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12155 (sh_hard_regno_mode_ok): Make static.
12156 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12157 instead of HARD_REGNO_MODE_OK.
12158 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12159 (sparc_mode_class): Delete.
12160 (HARD_REGNO_MODE_OK): Delete.
12161 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12162 (hard_regno_mode_classes): Make static.
12163 (sparc_mode_class): Likewise.
12164 (sparc_hard_regno_mode_ok): New function.
12165 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12166 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12167 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12170 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12171 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12172 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12173 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12174 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12175 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12176 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12177 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12178 (visium_hard_regno_mode_ok): New function.
12179 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12180 instead of HARD_REGNO_MODE_OK.
12181 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12182 (HARD_REGNO_MODE_OK): Delete.
12183 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12184 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12185 (xtensa_option_override): Update accordingly.
12186 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12187 (xtensa_hard_regno_mode_ok): New function.
12188 * system.h (HARD_REGNO_MODE_OK): Poison.
12190 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12191 Alan Hayward <alan.hayward@arm.com>
12192 David Sherwood <david.sherwood@arm.com>
12194 * target.def (hard_regno_call_part_clobbered): New hook.
12195 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12196 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12197 * doc/tm.texi: Regenerate.
12198 * hooks.h (hook_bool_uint_mode_false): Declare.
12199 * hooks.c (hook_bool_uint_mode_false): New function.
12200 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12201 * cselib.c (cselib_process_insn): Use
12202 targetm.hard_regno_call_part_clobbered instead of
12203 HARD_REGNO_CALL_PART_CLOBBERED.
12204 * ira-conflicts.c (ira_build_conflicts): Likewise.
12205 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12206 * lra-constraints.c (need_for_call_save_p): Likewise.
12207 * lra-lives.c: Include target.h.
12208 (check_pseudos_live_through_calls): Use
12209 targetm.hard_regno_call_part_clobbered instead of
12210 HARD_REGNO_CALL_PART_CLOBBERED.
12211 * regcprop.c: Include target.h.
12212 (copyprop_hardreg_forward_1): Use
12213 targetm.hard_regno_call_part_clobbered instead of
12214 HARD_REGNO_CALL_PART_CLOBBERED.
12215 * reginfo.c (choose_hard_reg_mode): Likewise.
12216 * regrename.c (check_new_reg_p): Likewise.
12217 * reload.c (find_equiv_reg): Likewise.
12218 * reload1.c (emit_reload_insns): Likewise.
12219 * sched-deps.c (deps_analyze_insn): Likewise.
12220 * sel-sched.c (init_regs_for_mode): Likewise.
12221 (mark_unavailable_hard_regs): Likewise.
12222 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12223 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12224 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12226 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12227 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12228 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12230 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12232 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12233 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12234 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12236 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12237 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12238 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12240 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12241 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12243 * config/powerpcspe/powerpcspe.c
12244 (rs6000_hard_regno_call_part_clobbered): New function.
12245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12246 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12247 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12249 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12250 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12251 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12253 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12254 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12255 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12257 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12258 Alan Hayward <alan.hayward@arm.com>
12259 David Sherwood <david.sherwood@arm.com>
12261 * rtl.h (subreg_memory_offset): Declare.
12262 * emit-rtl.c (subreg_memory_offset): New function.
12263 * expmed.c (store_bit_field_1): Use it.
12264 * expr.c (undefined_operand_subword_p): Likewise.
12265 * simplify-rtx.c (simplify_subreg): Likewise.
12267 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12269 PR rtl-optimization/57448
12272 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12273 using atomic_load pattern.
12274 (expand_atomic_store): Likewise.
12276 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12279 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12280 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12281 handling. Use replace_call_with_value with NULL instead of
12282 gsi_replace, unlink_stmt_vdef and release_defs.
12284 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12287 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12289 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12292 * rtl.h (emit_library_call_value_1): Declare.
12293 (emit_library_call): Replace declaration with a series of overloads.
12294 Remove the parameter count argument.
12295 (emit_library_call_value): Likewise.
12296 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12297 with an "rtx_mode_t *".
12298 (emit_library_call_value): Delete.
12299 (emit_library_call): Likewise.
12300 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12301 (asan_emit_allocas_unpoison): Likewise.
12302 * builtins.c (expand_builtin_powi): Likewise.
12303 (expand_asan_emit_allocas_unpoison): Likewise.
12304 * cfgexpand.c (expand_main_function): Likewise.
12305 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12306 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12307 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12308 * config/arm/arm.c (arm_trampoline_init): Likewise.
12309 (arm_call_tls_get_addr): Likewise.
12310 (arm_expand_divmod_libfunc): Likewise.
12311 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12312 (smulsi3_highpart): Likewise.
12313 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12314 (c6x_expand_compare): Likewise.
12315 (c6x_expand_movmem): Likewise.
12316 * config/frv/frv.c (frv_trampoline_init): Likewise.
12317 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12318 (ix86_expand_divmod_libfunc): Likewise.
12319 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12320 (ia64_expand_compare): Likewise.
12321 (ia64_profile_hook): Likewise.
12322 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12323 (nonlocal_goto): Likewise.
12324 (restore_stack_nonlocal): Likewise.
12325 * config/m32r/m32r.c (block_move_call): Likewise.
12326 (m32r_trampoline_init): Likewise.
12327 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12328 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12329 (m68k_call_m68k_read_tp): Likewise.
12330 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12331 (microblaze_expand_divide): Likewise.
12332 * config/mips/mips.h (mips_args): Likewise.
12333 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12334 (MIPS_ICACHE_SYNC): Likewise.
12335 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12336 (nios2_trampoline_init): Likewise.
12337 * config/pa/pa.c (hppa_tls_call): Likewise.
12338 (pa_trampoline_init): Likewise.
12339 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12340 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12341 (expand_strn_compare): Likewise.
12342 (rs6000_generate_compare): Likewise.
12343 (rs6000_expand_float128_convert): Likewise.
12344 (output_profile_hook): Likewise.
12345 (rs6000_trampoline_init): Likewise.
12346 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12347 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12348 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12349 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12350 (rs6000_generate_compare): Likewise.
12351 (rs6000_expand_float128_convert): Likewise.
12352 (output_profile_hook): Likewise.
12353 (rs6000_trampoline_init): Likewise.
12354 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12355 * config/sh/sh.c (sh_trampoline_init): Likewise.
12356 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12357 (sparc_emit_float_lib_cmp): Likewise.
12358 (sparc32_initialize_trampoline): Likewise.
12359 (sparc64_initialize_trampoline): Likewise.
12360 (sparc_profile_hook): Likewise.
12361 * config/spu/spu.c (ea_load_store): Likewise.
12362 * config/spu/spu.md (floatunssidf2): Likewise.
12363 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12364 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12365 * config/visium/visium.c (expand_block_move_4): Likewise.
12366 (expand_block_move_2): Likewise.
12367 (expand_block_move_1): Likewise.
12368 (expand_block_set_4): Likewise.
12369 (expand_block_set_2): Likewise.
12370 (expand_block_set_1): Likewise.
12371 (visium_trampoline_init): Likewise.
12372 (visium_profile_hook): Likewise.
12373 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12374 (xtensa_setup_frame_addresses): Likewise.
12375 (xtensa_trampoline_init): Likewise.
12376 * except.c (sjlj_emit_function_enter): Likewise.
12377 (sjlj_emit_function_exit): Likewise.
12378 * explow.c (allocate_dynamic_stack_space): Likewise.
12379 (probe_stack_range): Likewise.
12380 * expr.c (convert_mode_scalar): Likewise.
12381 * optabs.c (expand_binop): Likewise.
12382 (expand_twoval_binop_libfunc): Likewise.
12383 (expand_unop): Likewise.
12384 (prepare_cmp_insn): Likewise.
12385 (prepare_float_lib_cmp): Likewise.
12386 (expand_float): Likewise.
12387 (expand_fix): Likewise.
12388 (expand_fixed_convert): Likewise.
12389 (maybe_emit_sync_lock_test_and_set): Likewise.
12390 (expand_atomic_compare_and_swap): Likewise.
12391 (expand_mem_thread_fence): Likewise.
12392 (expand_atomic_fetch_op): Likewise.
12394 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12396 * doc/generic.texi (OpenACC): Adjust URL.
12397 * doc/invoke.texi (C Dialect Options): Ditto.
12399 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12401 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12402 predicate for operand 1. Add (m,<S>) constraint.
12403 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12404 Prevent memory operand 1 with register operand 2.
12406 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12408 PR rtl-optimization/82024
12409 * combine.c (try_combine): If the combination result is a PARALLEL,
12410 and we only need to retain the SET in there that would be placed
12411 at I2, check that we can place that at I3 instead, before doing so.
12413 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12416 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12418 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12419 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12422 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12423 Jeff Law <law@redhat.com>
12425 * varasm.c (bss_initializer_p): Do not put constants into .bss
12426 (categorize_decl_for_section): Handle bss_initializer_p returning
12427 false when DECL_INITIAL is NULL.
12429 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12432 * config/s390/s390.c (s390_can_inline_p): New function.
12434 2017-09-01 Jeff Law <law@redhat.com>
12436 PR tree-optimization/82052
12437 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12438 Always initialize the returned slot after a hash table miss
12439 when INSERT is true.
12441 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12443 * config/s390/s390.md (mem_signal_fence): Remove.
12444 * doc/md.texi (mem_signal_fence): Remove.
12445 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12447 * target-insns.def (mem_signal_fence): Remove.
12449 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12452 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12455 * asan.c (create_odr_indicator): Strip name encoding from assembler
12456 name before appending it after __odr_asan_.
12458 2017-09-01 Martin Liska <mliska@suse.cz>
12460 PR tree-optimization/82059
12461 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12462 frequency only when an edge is redirected.
12464 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12466 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12467 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12468 (arc_conditional_register_usage): Remove ARC600 lp_count
12470 (arc_file_start): Emit Tag_ARC_CPU_variation.
12471 (arc_can_use_doloop_p): New conditions to use ZOLs.
12472 (hwloop_fail): New function.
12473 (hwloop_optimize): Likewise.
12474 (hwloop_pattern_reg): Likewise.
12475 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12476 (arc_reorg_loops): New function, calls reorg_loops.
12477 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12478 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12480 (arc_loop_hazard): Remove function, functionality moved into
12482 (arc_hazard): Remove arc_loop_hazard call.
12483 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12484 into hwloop_optimize.
12485 (arc_label_align): Remove ZOL handling.
12486 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12487 * config/arc/arc.md (doloop_begin): Remove pattern.
12488 (doloop_begin_i): Likewise.
12489 (doloop_end_i): Likewise.
12490 (doloop_fallback): Likewise.
12491 (doloop_fallback_m): Likewise.
12492 (doloop_end): Reimplement expand.
12493 (arc_lp): New pattern for LP instruction.
12494 (loop_end): New pattern.
12495 (loop_fail): Likewise.
12496 (decrement_and_branch_until_zero): Likewise.
12497 * config/arc/arc.opt (mlpc-width): New option.
12498 * doc/invoke.texi (mlpc-width): Document option.
12500 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12502 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12503 (arc_ccfsm_advance): Fix checking for delay slots.
12504 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12506 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12508 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12510 (movhi_insn): Update store instruction constraint which are saving
12511 6-bit short immediates.
12512 (movsi_insn): Consider also short scaled load operations.
12513 (zero_extendhisi2_i): Use Usd constraint instead of T.
12514 (extendhisi2_i): Add q constraint.
12515 (arc_clzsi2): Add type and length attributes.
12516 (arc_ctzsi2): Likewise.
12517 * config/arc/constraints.md (Usc): Update constraint, the
12518 assembler can parse two relocations for a single instruction.
12520 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12522 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12523 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12525 2017-08-31 Olivier Hainque <hainque@adacore.com>
12527 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12528 match as powerpc-wrs-vxworks*.
12530 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12532 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12533 register constraint for by-element operand.
12534 (aarch64_mls_elt_merge<mode>): Likewise.
12536 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12538 * config/arc/arc.c (arc_can_follow_jump): Check for short
12541 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12543 * config.gcc: Use g.opt for arc.
12544 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12545 functionality moved to ...
12546 (legitimate_scaled_address_p): New function, ...here.
12547 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12548 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12549 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12551 (arc_override_options): Handle G option.
12552 (arc_output_pic_addr_const): Correct function definition.
12553 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12554 (arc_decl_anon_ns_mem_p): Delete.
12555 (arc_in_small_data_p): Overhaul this function to take into
12556 consideration the value given via G option.
12557 (arc_rewrite_small_data_1): Renamed and corrected old
12558 arc_rewrite_small_data function.
12559 (arc_rewrite_small_data): New function.
12560 (small_data_pattern): Don't use pic_offset_table_rtx.
12561 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12562 * config/arc/simdext.md (movmisalignv2hi): Use
12563 prepare_move_operands function.
12565 (movmisalign*): Likewise.
12566 * doc/invoke.texi (ARC options): Document -G option.
12568 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12570 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12572 * config/arc/arc.c (arc_print_operand): Output scalled address for
12573 sdata whenever is possible.
12574 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12575 load/stores are available.
12576 (compact_sda_memory_operand): Check for the alignment required by
12577 code density instructions.
12578 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12580 * config/arc/constraints.md (Usd): Update constraint.
12581 (Us0): New constraint.
12582 (Usc): Update constraint.
12584 2017-08-31 Richard Biener <rguenther@suse.de>
12586 PR middle-end/82054
12587 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12588 function only once.
12590 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12592 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12593 Resize type_signature.
12595 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12596 Alan Hayward <alan.hayward@arm.com>
12597 David Sherwood <david.sherwood@arm.com>
12599 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12600 subregs whose inner modes can be stored in GPRs.
12601 (aarch64_classify_index): Likewise.
12603 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12604 Alan Hayward <alan.hayward@arm.com>
12605 David Sherwood <david.sherwood@arm.com>
12607 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12608 (V_INT_EQUIV): ...this.
12609 (v_cmp_result): Rename to...
12610 (v_int_equiv): ...this.
12611 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12612 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12613 (copysign<mode>3): Likewise.
12614 (aarch64_simd_bsl<mode>_internal): Likewise.
12615 (aarch64_simd_bsl<mode>): Likewise.
12616 (vec_cmp<mode><mode>): Likewise.
12617 (vcond<mode><mode>): Likewise.
12618 (vcond<v_cmp_mixed><mode>): Likewise.
12619 (vcondu<mode><v_cmp_mixed>): Likewise.
12620 (aarch64_cm<optab><mode>): Likewise.
12621 (aarch64_cmtst<mode>): Likewise.
12622 (aarch64_fac<optab><mode>): Likewise.
12623 (vec_perm_const<mode>): Likewise.
12624 (vcond_mask_<mode><v_cmp_result>): Rename to...
12625 (vcond_mask_<mode><v_int_equiv>): ...this.
12626 (vec_cmp<mode><v_cmp_result>): Rename to...
12627 (vec_cmp<mode><v_int_equiv>): ...this.
12629 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12633 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12635 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12636 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12637 (UNSPEC_LD4_DREG): New unspecs.
12638 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12639 (aarch64_ld2<mode>_dreg_be): Replace with...
12640 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12642 (aarch64_ld3<mode>_dreg_le)
12643 (aarch64_ld3<mode>_dreg_be): Replace with...
12644 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12646 (aarch64_ld4<mode>_dreg_le)
12647 (aarch64_ld4<mode>_dreg_be): Replace with...
12648 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12651 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12653 PR tree-optimization/81987
12654 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12655 insert an initializer in a location not dominated by the stride
12658 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12660 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12661 on the entire header of the finally block in the fallthru case.
12663 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12665 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12667 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12669 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12670 rs6000_emit_move_from_cr and call renamed function.
12671 (rs6000_emit_prologue): Call renamed functions.
12672 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12673 movesi_from_cr, remove volatile CRs.
12675 2017-08-30 Jon Beniston <jon@beniston.com>
12676 Richard Biener <rguenther@suse.de>
12678 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12679 of VECTOR_MODE_P check.
12680 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12681 element vector types.
12683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12685 * df.h (df_read_modify_subreg_p): Remove in favor of...
12686 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12687 const_rtx instead of an rtx.
12688 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12689 * df-problems.c (df_word_lr_mark_ref): Likewise.
12690 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12691 (mark_pseudo_reg_dead): Likewise.
12692 (mark_ref_dead): Likewise.
12693 * reginfo.c (init_subregs_of_mode): Likewise.
12694 * sched-deps.c (sched_analyze_1): Likewise.
12695 * df-scan.c (df_def_record_1): Likewise.
12696 (df_uses_record): Likewise.
12697 (df_read_modify_subreg_p): Remove in favor of...
12698 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12699 const_rtx instead of an rtx.
12701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12702 Alan Hayward <alan.hayward@arm.com>
12703 David Sherwood <david.sherwood@arm.com>
12705 * rtl.h (partial_subreg_p): New function.
12706 * caller-save.c (save_call_clobbered_regs): Use it.
12707 * calls.c (expand_call): Likewise.
12708 * combine.c (combinable_i3pat): Likewise.
12709 (simplify_set): Likewise.
12710 (make_extraction): Likewise.
12711 (make_compound_operation_int): Likewise.
12712 (gen_lowpart_or_truncate): Likewise.
12713 (force_to_mode): Likewise.
12714 (make_field_assignment): Likewise.
12715 (reg_truncated_to_mode): Likewise.
12716 (record_truncated_value): Likewise.
12717 (move_deaths): Likewise.
12718 * cse.c (record_jump_cond): Likewise.
12719 (cse_insn): Likewise.
12720 * cselib.c (cselib_lookup_1): Likewise.
12721 * expmed.c (extract_bit_field_using_extv): Likewise.
12722 * function.c (assign_parm_setup_reg): Likewise.
12723 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12724 * ira-build.c (create_insn_allocnos): Likewise.
12725 * lra-coalesce.c (merge_pseudos): Likewise.
12726 * lra-constraints.c (match_reload): Likewise.
12727 (simplify_operand_subreg): Likewise.
12728 (curr_insn_transform): Likewise.
12729 * lra-lives.c (process_bb_lives): Likewise.
12730 * lra.c (new_insn_reg): Likewise.
12731 (lra_substitute_pseudo): Likewise.
12732 * regcprop.c (mode_change_ok): Likewise.
12733 (maybe_mode_change): Likewise.
12734 (copyprop_hardreg_forward_1): Likewise.
12735 * reload.c (push_reload): Likewise.
12736 (find_reloads): Likewise.
12737 (find_reloads_subreg_address): Likewise.
12738 * reload1.c (alter_reg): Likewise.
12739 (eliminate_regs_1): Likewise.
12740 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12742 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12744 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12745 back to if statements, including unpack.
12747 2017-08-30 Martin Liska <mliska@suse.cz>
12749 PR inline-asm/82001
12750 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12752 (func_checker::compare_asm_inputs_outputs): ... this function.
12753 (func_checker::compare_gimple_asm): Use the function to compare
12754 also ASM constrains.
12755 * ipa-icf-gimple.h: Rename the function.
12757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12758 Alan Hayward <alan.hayward@arm.com>
12759 David Sherwood <david.sherwood@arm.com>
12761 * coretypes.h (complex_mode): New type.
12762 * gdbhooks.py (build_pretty_printer): Handle it.
12763 * machmode.h (complex_mode): New class.
12764 (complex_mode::includes_p): New function.
12765 (is_complex_int_mode): Likewise.
12766 (is_complex_float_mode): Likewise.
12767 * genmodes.c (get_mode_class): Handle complex mode classes.
12768 * function.c (expand_function_end): Use is_complex_int_mode.
12770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12771 Alan Hayward <alan.hayward@arm.com>
12772 David Sherwood <david.sherwood@arm.com>
12774 * coretypes.h (scalar_mode_pod): New typedef.
12775 * gdbhooks.py (build_pretty_printer): Handle it.
12776 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12777 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12778 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12779 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12780 as_a <scalar_mode>.
12782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12783 Alan Hayward <alan.hayward@arm.com>
12784 David Sherwood <david.sherwood@arm.com>
12786 * machmode.h (mode_for_vector): Take a scalar_mode instead
12788 * stor-layout.c (mode_for_vector): Likewise.
12789 * explow.c (promote_mode): Use as_a <scalar_mode>.
12790 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12793 Alan Hayward <alan.hayward@arm.com>
12794 David Sherwood <david.sherwood@arm.com>
12796 * target.def (preferred_simd_mode): Take a scalar_mode
12797 instead of a machine_mode.
12798 * targhooks.h (default_preferred_simd_mode): Likewise.
12799 * targhooks.c (default_preferred_simd_mode): Likewise.
12800 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
12801 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
12802 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
12803 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
12804 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
12805 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
12806 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
12807 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
12809 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
12810 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
12811 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
12812 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
12813 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
12814 * doc/tm.texi: Regenerate.
12815 * optabs-query.c (can_vec_mask_load_store_p): Return false for
12818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12819 Alan Hayward <alan.hayward@arm.com>
12820 David Sherwood <david.sherwood@arm.com>
12822 * target.def (scalar_mode_supported_p): Take a scalar_mode
12823 instead of a machine_mode.
12824 * targhooks.h (default_scalar_mode_supported_p): Likewise.
12825 * targhooks.c (default_scalar_mode_supported_p): Likewise.
12826 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
12827 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12828 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
12829 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
12830 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
12831 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12832 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12833 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12834 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12835 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12836 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12837 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12839 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12840 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12841 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12842 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12843 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12845 * doc/tm.texi: Regenerate.
12847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12848 Alan Hayward <alan.hayward@arm.com>
12849 David Sherwood <david.sherwood@arm.com>
12851 * coretypes.h (opt_scalar_mode): New typedef.
12852 * gdbhooks.py (build_pretty_printers): Handle it.
12853 * machmode.h (mode_iterator::get_2xwider): Add overload for
12855 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12857 * expr.c (convert_mode_scalar): Likewise.
12858 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12859 * optabs.c (expand_float): Likewise.
12860 (expand_fix): Likewise.
12861 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12864 Alan Hayward <alan.hayward@arm.com>
12865 David Sherwood <david.sherwood@arm.com>
12867 * optabs.c (expand_float): Explicitly check for scalars before
12868 using a branching expansion.
12869 (expand_fix): Likewise.
12871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12872 Alan Hayward <alan.hayward@arm.com>
12873 David Sherwood <david.sherwood@arm.com>
12875 * expr.c (convert_mode): Split scalar handling out into...
12876 (convert_mode_scalar): ...this new function. Treat the modes
12879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12880 Alan Hayward <alan.hayward@arm.com>
12881 David Sherwood <david.sherwood@arm.com>
12883 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12885 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12888 Alan Hayward <alan.hayward@arm.com>
12889 David Sherwood <david.sherwood@arm.com>
12891 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12892 rather than a machine_mode.
12893 (fixed_from_string): Likewise.
12894 (fixed_convert): Likewise.
12895 (fixed_convert_from_int): Likewise.
12896 (fixed_convert_from_real): Likewise.
12897 (real_convert_from_fixed): Likewise.
12898 * fixed-value.c (fixed_from_double_int): Likewise.
12899 (fixed_from_string): Likewise.
12900 (fixed_convert): Likewise.
12901 (fixed_convert_from_int): Likewise.
12902 (fixed_convert_from_real): Likewise.
12903 (real_convert_from_fixed): Likewise.
12904 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12907 Alan Hayward <alan.hayward@arm.com>
12908 David Sherwood <david.sherwood@arm.com>
12910 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12911 of separate mode class checks. Do not allow vector modes here.
12912 (immed_wide_int_const): Use as_a <scalar_mode>.
12913 * explow.c (trunc_int_for_mode): Likewise.
12914 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12915 (wi::shwi): Likewise.
12916 (wi::min_value): Likewise.
12917 (wi::max_value): Likewise.
12918 * dwarf2out.c (loc_descriptor): Likewise.
12919 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12920 for CONST_WIDE_INT.
12922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12923 Alan Hayward <alan.hayward@arm.com>
12924 David Sherwood <david.sherwood@arm.com>
12926 * tree.h (SCALAR_TYPE_MODE): New macro.
12927 * expr.c (expand_expr_addr_expr_1): Use it.
12928 (expand_expr_real_2): Likewise.
12929 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12930 (fold_convert_const_fixed_from_int): Likewise.
12931 (fold_convert_const_fixed_from_real): Likewise.
12932 (native_encode_fixed): Likewise
12933 (native_encode_complex): Likewise
12934 (native_encode_vector): Likewise.
12935 (native_interpret_fixed): Likewise.
12936 (native_interpret_real): Likewise.
12937 (native_interpret_complex): Likewise.
12938 (native_interpret_vector): Likewise.
12939 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12940 (simd_clone_adjust_argument_types): Likewise.
12941 (simd_clone_init_simd_arrays): Likewise.
12942 (simd_clone_adjust): Likewise.
12943 * stor-layout.c (layout_type): Likewise.
12944 * tree.c (build_minus_one_cst): Likewise.
12945 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12946 * tree-inline.c (estimate_move_cost): Likewise.
12947 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12949 (vectorizable_reduction): Likewise.
12950 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12951 (vect_recog_mixed_size_cond_pattern): Likewise.
12952 (check_bool_pattern): Likewise.
12953 (adjust_bool_pattern): Likewise.
12954 (search_type_for_mask_1): Likewise.
12955 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12956 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12957 (vectorizable_load): Likewise.
12958 (vectorizable_store): Likewise.
12959 * ubsan.c (ubsan_encode_value): Likewise.
12960 * varasm.c (output_constant): Likewise.
12962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12963 Alan Hayward <alan.hayward@arm.com>
12964 David Sherwood <david.sherwood@arm.com>
12966 * coretypes.h (scalar_mode): New class.
12967 * machmode.h (scalar_mode): Likewise.
12968 (scalar_mode::includes_p): New function.
12969 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12970 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12971 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12972 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12973 * expmed.c (store_bit_field_1): Likewise.
12974 (extract_bit_field_1): Likewise.
12975 * expr.c (write_complex_part): Likewise.
12976 (read_complex_part): Likewise.
12977 (emit_move_complex_push): Likewise.
12978 (expand_expr_real_2): Likewise.
12979 * function.c (assign_parm_setup_reg): Likewise.
12980 (assign_parms_unsplit_complex): Likewise.
12981 * optabs.c (expand_binop): Likewise.
12982 * rtlanal.c (subreg_get_info): Likewise.
12983 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12984 * varasm.c (output_constant_pool_2): Likewise.
12986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12987 Alan Hayward <alan.hayward@arm.com>
12988 David Sherwood <david.sherwood@arm.com>
12990 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12992 (expmed_mult_highpart_optab): Likewise.
12993 (expmed_mult_highpart): Likewise.
12995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12996 Alan Hayward <alan.hayward@arm.com>
12997 David Sherwood <david.sherwood@arm.com>
12999 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13000 instead of a machine_mode.
13001 (builtin_memset_read_str): Likewise.
13002 * builtins.c (c_readstr): Likewise.
13003 (builtin_memcpy_read_str): Likewise.
13004 (builtin_strncpy_read_str): Likewise.
13005 (builtin_memset_read_str): Likewise.
13006 (builtin_memset_gen_str): Likewise.
13007 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13008 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13009 instead of a machine_mode.
13010 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13012 (make_extraction): Likewise.
13013 (try_widen_shift_mode): Take and return scalar_int_modes instead
13015 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13016 a scalar_int_mode instead of a machine_mode.
13017 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13018 (avr_addr_space_pointer_mode): Likewise.
13019 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13020 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13021 (msp430_unwind_word_mode): Likewise.
13022 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13023 (spu_addr_space_pointer_mode): Likewise.
13024 (spu_addr_space_address_mode): Likewise.
13025 (spu_libgcc_cmp_return_mode): Likewise.
13026 (spu_libgcc_shift_count_mode): Likewise.
13027 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13028 (rl78_addr_space_pointer_mode): Likewise.
13029 (fl78_unwind_word_mode): Likewise.
13030 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13032 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13033 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13034 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13035 (mips_valid_pointer_mode): Likewise.
13036 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13037 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13038 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13040 (ft32_addr_space_address_mode): Likewise.
13041 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13042 scalar_int_mode instead of a machine_mode.
13043 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13045 (m32c_addr_space_address_mode): Likewise.
13046 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13047 (rs6000_eh_return_filter_mode): Likewise.
13048 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13049 (rs6000_eh_return_filter_mode): Likewise.
13050 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13051 (s390_libgcc_shift_count_mode): Likewise.
13052 (s390_unwind_word_mode): Likewise.
13053 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13055 * target.def (mode_rep_extended): Likewise.
13056 (valid_pointer_mode): Likewise.
13057 (addr_space.valid_pointer_mode): Likewise.
13058 (eh_return_filter_mode): Return a scalar_int_mode rather than
13060 (libgcc_cmp_return_mode): Likewise.
13061 (libgcc_shift_count_mode): Likewise.
13062 (unwind_word_mode): Likewise.
13063 (addr_space.pointer_mode): Likewise.
13064 (addr_space.address_mode): Likewise.
13065 * doc/tm.texi: Regenerate.
13066 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13068 (do_jump): Use scalar_int_mode for local variables.
13069 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13070 rather than a machine_mode.
13071 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13072 (scompare_loc_descriptor_wide): Likewise.
13073 (scompare_loc_descriptor_narrow): Likewise.
13074 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13076 * except.c (sjlj_emit_dispatch_table): Likewise.
13077 (expand_builtin_eh_copy_values): Likewise.
13078 * explow.c (convert_memory_address_addr_space_1): Likewise.
13079 Take a scalar_int_mode rather than a machine_mode.
13080 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13081 than a machine_mode.
13082 (memory_address_addr_space): Use scalar_int_mode for local variables.
13083 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13084 rather than a machine_mode.
13085 * expmed.c (mask_rtx): Likewise.
13086 (init_expmed_one_conv): Likewise.
13087 (expand_mult_highpart_adjust): Likewise.
13088 (extract_high_half): Likewise.
13089 (expmed_mult_highpart_optab): Likewise.
13090 (expmed_mult_highpart): Likewise.
13091 (expand_smod_pow2): Likewise.
13092 (expand_sdiv_pow2): Likewise.
13093 (emit_store_flag_int): Likewise.
13094 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13096 (extract_low_bits): Likewise.
13097 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13099 * expr.c (pieces_addr::adjust): Likewise.
13100 (can_store_by_pieces): Likewise.
13101 (store_by_pieces): Likewise.
13102 (clear_by_pieces_1): Likewise.
13103 (expand_expr_addr_expr_1): Likewise.
13104 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13105 (expand_expr_real_1): Likewise.
13106 (try_casesi): Likewise.
13107 * final.c (shorten_branches): Likewise.
13108 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13109 type of "mode" to machine_mode.
13110 * internal-fn.c (expand_arith_overflow_result_store): Take a
13111 scalar_int_mode rather than a machine_mode.
13112 (expand_mul_overflow): Use scalar_int_mode for local variables.
13113 * loop-doloop.c (doloop_modify): Likewise.
13114 (doloop_optimize): Likewise.
13115 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13116 than a machine_mode.
13117 (expand_doubleword_shift_condmove): Likewise.
13118 (expand_doubleword_shift): Likewise.
13119 (expand_doubleword_clz): Likewise.
13120 (expand_doubleword_popcount): Likewise.
13121 (expand_doubleword_parity): Likewise.
13122 (expand_absneg_bit): Use scalar_int_mode for local variables.
13123 (prepare_float_lib_cmp): Likewise.
13124 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13125 rather than a machine_mode.
13126 (convert_memory_address_addr_space): Likewise.
13127 (get_mode_bounds): Likewise.
13128 (get_address_mode): Return a scalar_int_mode rather than a
13130 * rtlanal.c (get_address_mode): Likewise.
13131 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13132 than a machine_mode.
13133 * targhooks.c (default_mode_rep_extended): Likewise.
13134 (default_valid_pointer_mode): Likewise.
13135 (default_addr_space_valid_pointer_mode): Likewise.
13136 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13137 than a machine_mode.
13138 (default_libgcc_cmp_return_mode): Likewise.
13139 (default_libgcc_shift_count_mode): Likewise.
13140 (default_unwind_word_mode): Likewise.
13141 (default_addr_space_pointer_mode): Likewise.
13142 (default_addr_space_address_mode): Likewise.
13143 * targhooks.h (default_eh_return_filter_mode): Likewise.
13144 (default_libgcc_cmp_return_mode): Likewise.
13145 (default_libgcc_shift_count_mode): Likewise.
13146 (default_unwind_word_mode): Likewise.
13147 (default_addr_space_pointer_mode): Likewise.
13148 (default_addr_space_address_mode): Likewise.
13149 (default_mode_rep_extended): Take a scalar_int_mode rather than
13151 (default_valid_pointer_mode): Likewise.
13152 (default_addr_space_valid_pointer_mode): Likewise.
13153 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13155 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13156 rather than a machine_mode.
13157 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13158 for local variables.
13159 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13160 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13161 than a machine_mode.
13163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13164 Alan Hayward <alan.hayward@arm.com>
13165 David Sherwood <david.sherwood@arm.com>
13167 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13168 the mode argument to scalar_int_mode.
13169 (do_jump_by_parts_zero_rtx): Likewise.
13170 (do_jump_by_parts_equality_rtx): Likewise.
13171 (do_jump_by_parts_greater): Take a mode argument.
13172 (do_jump_by_parts_equality): Likewise.
13173 (do_jump_1): Update calls accordingly.
13175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13176 Alan Hayward <alan.hayward@arm.com>
13177 David Sherwood <david.sherwood@arm.com>
13179 * is-a.h (safe_dyn_cast): New function.
13180 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13181 (jump_table_for_label): Likewise.
13182 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13183 instead of an rtx_insn *.
13184 (shorten_branches): Use dyn_cast instead of LABEL_P and
13185 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13186 rtx_jump_table_data::get_data_mode.
13187 (final_scan_insn): Likewise.
13189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13190 Alan Hayward <alan.hayward@arm.com>
13191 David Sherwood <david.sherwood@arm.com>
13193 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13194 trying to combine a full-register integer set with a subreg
13197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13198 Alan Hayward <alan.hayward@arm.com>
13199 David Sherwood <david.sherwood@arm.com>
13201 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13202 that is always either address_mode or pointer_mode.
13204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13205 Alan Hayward <alan.hayward@arm.com>
13206 David Sherwood <david.sherwood@arm.com>
13208 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13209 when the two are known to be equal.
13211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13212 Alan Hayward <alan.hayward@arm.com>
13213 David Sherwood <david.sherwood@arm.com>
13215 * simplify-rtx.c (simplify_const_unary_operation): Use
13216 is_a <scalar_int_mode> instead of checking for a nonzero
13217 precision. Forcibly convert op_mode to a scalar_int_mode
13218 in that case. More clearly differentiate the operand and
13219 result modes and use the former when deciding what the value
13220 of a count-bits operation should be. Use is_int_mode instead
13221 of checking for a MODE_INT. Remove redundant check for whether
13222 this mode has a zero precision.
13224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13225 Alan Hayward <alan.hayward@arm.com>
13226 David Sherwood <david.sherwood@arm.com>
13228 * optabs.c (widen_leading): Change the type of the mode argument
13229 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13230 (widen_bswap): Likewise.
13231 (expand_parity): Likewise.
13232 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13233 (expand_ffs): Likewise.
13234 (epand_unop): Check for scalar integer modes before calling the
13237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13238 Alan Hayward <alan.hayward@arm.com>
13239 David Sherwood <david.sherwood@arm.com>
13241 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13243 (expand_expr_real_1): Update call accordingly.
13245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13246 Alan Hayward <alan.hayward@arm.com>
13247 David Sherwood <david.sherwood@arm.com>
13249 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13250 value_mode arguments. Use scalar_int_mode internally.
13251 (store_bit_field_1): Rename the new integer mode from imode
13252 to op0_mode and use it instead of GET_MODE (op0). Update calls
13253 to store_split_bit_field, store_bit_field_using_insv and
13254 store_fixed_bit_field.
13255 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13256 Use scalar_int_mode internally. Use a bit count rather than a mode
13257 when calculating the largest bit size for get_best_mode.
13258 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13259 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13260 Remove assertion that OP0 has a scalar integer mode.
13261 (store_split_bit_field): Add op0_mode and value_mode arguments.
13262 Update calls to extract_fixed_bit_field.
13263 (extract_bit_field_using_extv): Add an op0_mode argument.
13264 Use scalar_int_mode internally.
13265 (extract_bit_field_1): Rename the new integer mode from imode to
13266 op0_mode and use it instead of GET_MODE (op0). Update calls to
13267 extract_split_bit_field, extract_bit_field_using_extv and
13268 extract_fixed_bit_field.
13269 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13270 to extract_split_bit_field and extract_fixed_bit_field_1.
13271 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13272 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13273 on the target mode.
13274 (extract_split_bit_field): Add an op0_mode argument. Update call
13275 to extract_fixed_bit_field.
13277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13278 Alan Hayward <alan.hayward@arm.com>
13279 David Sherwood <david.sherwood@arm.com>
13281 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13282 * explow.c (hard_function_value): Likewise.
13283 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13284 convert_to_mode call outside the loop.
13285 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13286 for the mode iterator. Require the mode specified by max_pieces
13288 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13290 (copy_blkmode_to_reg): Likewise.
13291 (set_storage_via_setmem): Likewise.
13292 * optabs.c (prepare_cmp_insn): Likewise.
13293 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13294 * stor-layout.c (finish_bitfield_representative): Likewise.
13296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13297 Alan Hayward <alan.hayward@arm.com>
13298 David Sherwood <david.sherwood@arm.com>
13300 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13301 * expr.c (convert_move): Use them.
13302 (convert_modes): Likewise.
13303 (store_expr_with_bounds): Likewise.
13305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13306 Alan Hayward <alan.hayward@arm.com>
13307 David Sherwood <david.sherwood@arm.com>
13309 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13310 parameter for the mode of "x". Remove the "known_x", "known_mode"
13311 and "known_ret" arguments. Change the type of the mode argument
13312 to scalar_int_mode.
13313 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13314 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13315 (reg_num_sign_bit_copies_for_combine): Likewise.
13316 * rtlanal.c (nonzero_bits1): Likewise.
13317 (num_sign_bit_copies1): Likewise.
13318 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13319 (reg_num_sign_bit_copies_general): Likewise.
13320 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13321 (reg_nonzero_bits_general): Likewise.
13323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13324 Alan Hayward <alan.hayward@arm.com>
13325 David Sherwood <david.sherwood@arm.com>
13327 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13328 than in subroutines. Return 1 for non-integer modes.
13329 (cached_num_sign_bit_copies): Change the type of the mode parameter
13330 to scalar_int_mode.
13331 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13332 classes. Handle CONST_INT_P first and then check whether X also
13333 has a scalar integer mode. Check the same thing for inner registers
13334 of a SUBREG and for values that are being extended or truncated.
13336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13337 Alan Hayward <alan.hayward@arm.com>
13338 David Sherwood <david.sherwood@arm.com>
13340 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13341 in subroutines. Return the mode mask for non-integer modes.
13342 (cached_nonzero_bits): Change the type of the mode parameter
13343 to scalar_int_mode.
13344 (nonzero_bits1): Likewise. Remove early exit for other mode
13345 classes. Handle CONST_INT_P first and then check whether X
13346 also has a scalar integer mode.
13348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13349 Alan Hayward <alan.hayward@arm.com>
13350 David Sherwood <david.sherwood@arm.com>
13352 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13353 Return a scalar_int_mode and assert that the size is greater than
13355 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13356 (op_by_pieces_d::op_by_pieces_d): Likewise.
13357 (op_by_pieces_d::run): Likewise.
13358 (can_store_by_pieces): Likewise.
13360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13361 Alan Hayward <alan.hayward@arm.com>
13362 David Sherwood <david.sherwood@arm.com>
13364 * combine.c (extract_left_shift): Add a mode argument and update
13366 (make_compound_operation_int): Change the type of the mode parameter
13367 to scalar_int_mode and update the call to extract_left_shift.
13369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13370 Alan Hayward <alan.hayward@arm.com>
13371 David Sherwood <david.sherwood@arm.com>
13373 * combine.c (simplify_and_const_int): Change the type of the mode
13374 parameter to scalar_int_mode.
13375 (simplify_and_const_int_1): Likewise. Update recursive call.
13377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13378 Alan Hayward <alan.hayward@arm.com>
13379 David Sherwood <david.sherwood@arm.com>
13381 * combine.c (simplify_compare_const): Check that the mode is a
13382 scalar_int_mode (rather than VOIDmode) before testing its
13384 (simplify_comparison): Move COMPARISON_P handling out of the
13385 loop and restrict the latter part of the loop to scalar_int_modes.
13386 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13387 and when considering SUBREG_REGs. Use is_int_mode instead of
13388 checking GET_MODE_CLASS against MODE_INT.
13390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13391 Alan Hayward <alan.hayward@arm.com>
13392 David Sherwood <david.sherwood@arm.com>
13394 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13395 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13396 shift_unit_mode and for modes involved in scalar shifts.
13398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13399 Alan Hayward <alan.hayward@arm.com>
13400 David Sherwood <david.sherwood@arm.com>
13402 * combine.c (force_int_to_mode): New function, split out from...
13403 (force_to_mode): ...here. Keep xmode up-to-date and use it
13404 instead of GET_MODE (x).
13406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13407 Alan Hayward <alan.hayward@arm.com>
13408 David Sherwood <david.sherwood@arm.com>
13410 * optabs-query.h (extraction_insn::struct_mode): Change type to
13411 opt_scalar_int_mode and update comment.
13412 (extraction_insn::field_mode): Change type to scalar_int_mode.
13413 (extraction_insn::pos_mode): Likewise.
13414 * combine.c (make_extraction): Update accordingly.
13415 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13416 (get_optab_extraction_insn): Likewise.
13417 * recog.c (simplify_while_replacing): Likewise.
13418 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13419 parameter to opt_scalar_int_mode.
13421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13422 Alan Hayward <alan.hayward@arm.com>
13423 David Sherwood <david.sherwood@arm.com>
13425 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13426 to a scalar_int_mode instead of a machine_mode.
13427 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13428 (get_best_mode): Return a boolean and use a pointer argument to store
13429 the selected mode. Replace the limit mode parameter with a bit limit.
13430 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13431 for the values returned by bit_field_mode_iterator::next_mode.
13432 (store_bit_field): Update call to get_best_mode.
13433 (store_fixed_bit_field): Likewise.
13434 (extract_fixed_bit_field): Likewise.
13435 * expr.c (optimize_bitfield_assignment_op): Likewise.
13436 * fold-const.c (optimize_bit_field_compare): Likewise.
13437 (fold_truth_andor_1): Likewise.
13438 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13439 Update for new type of m_mode.
13440 (get_best_mode): As above.
13442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13443 Alan Hayward <alan.hayward@arm.com>
13444 David Sherwood <david.sherwood@arm.com>
13446 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13447 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13448 (store_bit_field): Check is_a <scalar_int_mode> before calling
13449 strict_volatile_bitfield_p.
13450 (extract_bit_field): Likewise.
13452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13453 Alan Hayward <alan.hayward@arm.com>
13454 David Sherwood <david.sherwood@arm.com>
13456 * target.def (cstore_mode): Return a scalar_int_mode.
13457 * doc/tm.texi: Regenerate.
13458 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13459 * targhooks.h (default_cstore_mode): Likewise.
13460 * targhooks.c (default_cstore_mode): Likewise, using a forced
13462 * expmed.c (emit_cstore): Expect the target of the cstore to be
13465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13466 Alan Hayward <alan.hayward@arm.com>
13467 David Sherwood <david.sherwood@arm.com>
13469 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13471 (niter_desc): Likewise mode.
13472 (iv_analyze): Add a mode parameter.
13474 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13475 and change its type to scalar_int_mode.
13476 * loop-iv.c: Update commentary at head of file.
13477 (iv_constant): Pass the mode paraeter before the rtx it describes
13478 and change its type to scalar_int_mode. Remove VOIDmode handling.
13479 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13480 (iv_extend): Likewise.
13481 (shorten_into_mode): Likewise.
13482 (iv_add): Use scalar_int_mode.
13483 (iv_mult): Likewise.
13484 (iv_shift): Likewise.
13485 (canonicalize_iv_subregs): Likewise.
13486 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13487 it describes and change its mode to scalar_int_mode. Also change
13488 the type of the returned inner_mode to scalar_int_mode.
13489 (get_biv_step): Likewise, turning outer_mode from a pointer
13490 into a direct parameter. Update call to get_biv_step_1.
13491 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13492 iv_constant and get_biv_step.
13493 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13494 and change its type to scalar_int_mode. Don't initialise iv->mode
13495 to VOIDmode and remove later checks for its still being VOIDmode.
13496 Update calls to iv_analyze_op and iv_analyze_expr. Check
13497 is_a <scalar_int_mode> when changing the mode under consideration.
13498 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13499 Update call to iv_analyze_expr.
13500 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13501 inner register is not also a scalar_int_mode. Update call to
13503 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13504 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13505 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13506 separate mode class checks. Update calls to iv_analyze. Remove
13507 fix-up of VOIDmodes after iv_analyze_biv.
13508 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13509 don't have a scalar_int_mode. Update call to biv_p.
13511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13512 Alan Hayward <alan.hayward@arm.com>
13513 David Sherwood <david.sherwood@arm.com>
13515 * cfgexpand.c (convert_debug_memory_address): Use
13516 as_a <scalar_int_mode>.
13517 * combine.c (expand_compound_operation): Likewise.
13518 (make_extraction): Likewise.
13519 (change_zero_ext): Likewise.
13520 (simplify_comparison): Likewise.
13521 * cse.c (cse_insn): Likewise.
13522 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13523 (mem_loc_descriptor): Likewise.
13524 (loc_descriptor): Likewise.
13525 * expmed.c (init_expmed_one_mode): Likewise.
13526 (synth_mult): Likewise.
13527 (emit_store_flag_1): Likewise.
13528 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13529 of a comparison with size.
13530 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13531 (reduce_to_bit_field_precision): Likewise.
13532 * function.c (expand_function_end): Likewise.
13533 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13534 * loop-doloop.c (doloop_modify): Likewise.
13535 * optabs.c (expand_binop): Likewise.
13536 (expand_unop): Likewise.
13537 (expand_copysign_absneg): Likewise.
13538 (prepare_cmp_insn): Likewise.
13539 (maybe_legitimize_operand): Likewise.
13540 * recog.c (const_scalar_int_operand): Likewise.
13541 * rtlanal.c (get_address_mode): Likewise.
13542 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13543 (simplify_cond_clz_ctz): Likewise.
13544 * tree-nested.c (get_nl_goto_field): Likewise.
13545 * tree.c (build_vector_type_for_mode): Likewise.
13546 * var-tracking.c (use_narrower_mode): Likewise.
13548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13549 Alan Hayward <alan.hayward@arm.com>
13550 David Sherwood <david.sherwood@arm.com>
13552 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13553 * builtins.c (expand_builtin_signbit): Use it.
13554 * cfgexpand.c (expand_debug_expr): Likewise.
13555 * dojump.c (do_jump): Likewise.
13556 (do_compare_and_jump): Likewise.
13557 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13558 * expmed.c (make_tree): Likewise.
13559 * expr.c (expand_expr_real_2): Likewise.
13560 (expand_expr_real_1): Likewise.
13561 (try_casesi): Likewise.
13562 * fold-const-call.c (fold_const_call_ss): Likewise.
13563 * fold-const.c (unextend): Likewise.
13564 (extract_muldiv_1): Likewise.
13565 (fold_single_bit_test): Likewise.
13566 (native_encode_int): Likewise.
13567 (native_encode_string): Likewise.
13568 (native_interpret_int): Likewise.
13569 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13570 * internal-fn.c (expand_addsub_overflow): Likewise.
13571 (expand_neg_overflow): Likewise.
13572 (expand_mul_overflow): Likewise.
13573 (expand_arith_overflow): Likewise.
13574 * match.pd: Likewise.
13575 * stor-layout.c (layout_type): Likewise.
13576 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13577 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13578 * tree-ssanames.c (get_range_info): Likewise.
13579 * tree-switch-conversion.c (array_value_type) Likewise.
13580 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13581 (vect_recog_divmod_pattern): Likewise.
13582 (vect_recog_mixed_size_cond_pattern): Likewise.
13583 * tree-vrp.c (extract_range_basic): Likewise.
13584 (simplify_float_conversion_using_ranges): Likewise.
13585 * tree.c (int_fits_type_p): Likewise.
13586 * ubsan.c (instrument_bool_enum_load): Likewise.
13587 * varasm.c (mergeable_string_section): Likewise.
13588 (narrowing_initializer_constant_valid_p): Likewise.
13589 (output_constant): Likewise.
13591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13592 Alan Hayward <alan.hayward@arm.com>
13593 David Sherwood <david.sherwood@arm.com>
13595 * machmode.h (NARROWEST_INT_MODE): New macro.
13596 * expr.c (alignment_for_piecewise_move): Use it instead of
13597 GET_CLASS_NARROWEST_MODE (MODE_INT).
13598 (push_block): Likewise.
13599 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13601 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13604 Alan Hayward <alan.hayward@arm.com>
13605 David Sherwood <david.sherwood@arm.com>
13607 * postreload.c (move2add_valid_value_p): Change the type of the
13608 mode parameter to scalar_int_mode.
13609 (move2add_use_add2_insn): Add a mode parameter and use it instead
13611 (move2add_use_add3_insn): Likewise.
13612 (reload_cse_move2add): Update accordingly.
13614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13615 Alan Hayward <alan.hayward@arm.com>
13616 David Sherwood <david.sherwood@arm.com>
13618 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13620 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13621 * optabs.c (expand_unop): Likewise.
13623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13624 Alan Hayward <alan.hayward@arm.com>
13625 David Sherwood <david.sherwood@arm.com>
13627 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13628 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13629 (popcount_loc_descriptor): Likewise.
13630 (bswap_loc_descriptor): Likewise.
13631 (rotate_loc_descriptor): Likewise.
13632 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13633 calling the functions above.
13635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13636 Alan Hayward <alan.hayward@arm.com>
13637 David Sherwood <david.sherwood@arm.com>
13639 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13641 (try_combine): Likewise.
13642 (simplify_if_then_else): Likewise.
13643 * cse.c (cse_insn): Likewise.
13644 * dwarf2out.c (mem_loc_descriptor): Likewise.
13645 * emit-rtl.c (gen_lowpart_common): Likewise.
13646 * simplify-rtx.c (simplify_truncation): Likewise.
13647 (simplify_binary_operation_1): Likewise.
13648 (simplify_const_relational_operation): Likewise.
13649 (simplify_ternary_operation): Likewise.
13650 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13653 Alan Hayward <alan.hayward@arm.com>
13654 David Sherwood <david.sherwood@arm.com>
13656 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13657 * reload.c (push_reload): Likewise.
13658 (find_reloads): Likewise.
13660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13661 Alan Hayward <alan.hayward@arm.com>
13662 David Sherwood <david.sherwood@arm.com>
13664 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13665 (make_compound_operation_int): Likewise.
13666 (change_zero_ext): Likewise.
13667 * expr.c (convert_move): Likewise.
13668 (convert_modes): Likewise.
13669 * fwprop.c (forward_propagate_subreg): Likewise.
13670 * loop-iv.c (get_biv_step_1): Likewise.
13671 * optabs.c (widen_operand): Likewise.
13672 * postreload.c (move2add_valid_value_p): Likewise.
13673 * recog.c (simplify_while_replacing): Likewise.
13674 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13675 (simplify_binary_operation_1): Likewise. Remove redundant
13676 mode equality check.
13678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13679 Alan Hayward <alan.hayward@arm.com>
13680 David Sherwood <david.sherwood@arm.com>
13682 * combine.c (combine_simplify_rtx): Add checks for
13683 is_a <scalar_int_mode>.
13684 (simplify_if_then_else): Likewise.
13685 (make_field_assignment): Likewise.
13686 (simplify_comparison): Likewise.
13687 * ifcvt.c (noce_try_bitop): Likewise.
13688 * loop-invariant.c (canonicalize_address_mult): Likewise.
13689 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13692 Alan Hayward <alan.hayward@arm.com>
13693 David Sherwood <david.sherwood@arm.com>
13695 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13696 is_a <scalar_int_mode> instead of != BLKmode.
13698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13699 Alan Hayward <alan.hayward@arm.com>
13700 David Sherwood <david.sherwood@arm.com>
13702 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13703 instead of != VOIDmode.
13704 * combine.c (if_then_else_cond): Likewise.
13705 (change_zero_ext): Likewise.
13706 * dwarf2out.c (mem_loc_descriptor): Likewise.
13707 (loc_descriptor): Likewise.
13708 * rtlanal.c (canonicalize_condition): Likewise.
13709 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13712 Alan Hayward <alan.hayward@arm.com>
13713 David Sherwood <david.sherwood@arm.com>
13715 * simplify-rtx.c (simplify_binary_operation_1): Use
13716 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13719 Alan Hayward <alan.hayward@arm.com>
13720 David Sherwood <david.sherwood@arm.com>
13722 * wide-int.h (int_traits<unsigned char>) New class.
13723 (int_traits<unsigned short>) Likewise.
13724 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13725 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13727 * combine.c (set_nonzero_bits_and_sign_copies): Use
13728 is_a <scalar_int_mode>.
13729 (find_split_point): Likewise.
13730 (combine_simplify_rtx): Likewise.
13731 (simplify_logical): Likewise.
13732 (expand_compound_operation): Likewise.
13733 (expand_field_assignment): Likewise.
13734 (make_compound_operation): Likewise.
13735 (extended_count): Likewise.
13736 (change_zero_ext): Likewise.
13737 (simplify_comparison): Likewise.
13738 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13739 (ucompare_loc_descriptor): Likewise.
13740 (minmax_loc_descriptor): Likewise.
13741 (mem_loc_descriptor): Likewise.
13742 (loc_descriptor): Likewise.
13743 * expmed.c (init_expmed_one_mode): Likewise.
13744 * lra-constraints.c (lra_constraint_offset): Likewise.
13745 * optabs.c (prepare_libcall_arg): Likewise.
13746 * postreload.c (move2add_note_store): Likewise.
13747 * reload.c (operands_match_p): Likewise.
13748 * rtl.h (load_extend_op): Likewise.
13749 * rtlhooks.c (gen_lowpart_general): Likewise.
13750 * simplify-rtx.c (simplify_truncation): Likewise.
13751 (simplify_unary_operation_1): Likewise.
13752 (simplify_binary_operation_1): Likewise.
13753 (simplify_const_binary_operation): Likewise.
13754 (simplify_const_relational_operation): Likewise.
13755 (simplify_subreg): Likewise.
13756 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13757 * var-tracking.c (adjust_mems): Likewise.
13758 (prepare_call_arguments): Likewise.
13760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13761 Alan Hayward <alan.hayward@arm.com>
13762 David Sherwood <david.sherwood@arm.com>
13764 * machmode.h (is_int_mode): New fuction.
13765 * combine.c (find_split_point): Use it.
13766 (combine_simplify_rtx): Likewise.
13767 (simplify_if_then_else): Likewise.
13768 (simplify_set): Likewise.
13769 (simplify_shift_const_1): Likewise.
13770 (simplify_comparison): Likewise.
13771 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13772 * cse.c (notreg_cost): Likewise.
13773 (cse_insn): Likewise.
13774 * cselib.c (cselib_lookup_1): Likewise.
13775 * dojump.c (do_jump_1): Likewise.
13776 (do_compare_rtx_and_jump): Likewise.
13777 * dse.c (get_call_args): Likewise.
13778 * dwarf2out.c (rtl_for_decl_init): Likewise.
13779 (native_encode_initializer): Likewise.
13780 * expmed.c (emit_store_flag_1): Likewise.
13781 (emit_store_flag): Likewise.
13782 * expr.c (convert_modes): Likewise.
13783 (store_field): Likewise.
13784 (expand_expr_real_1): Likewise.
13785 * fold-const.c (fold_read_from_constant_string): Likewise.
13786 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13787 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13788 * optabs.c (expand_binop): Likewise.
13789 (expand_unop): Likewise.
13790 (expand_abs_nojump): Likewise.
13791 (expand_one_cmpl_abs_nojump): Likewise.
13792 * simplify-rtx.c (mode_signbit_p): Likewise.
13793 (val_signbit_p): Likewise.
13794 (val_signbit_known_set_p): Likewise.
13795 (val_signbit_known_clear_p): Likewise.
13796 (simplify_relational_operation_1): Likewise.
13797 * tree.c (vector_type_mode): Likewise.
13799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13800 Alan Hayward <alan.hayward@arm.com>
13801 David Sherwood <david.sherwood@arm.com>
13803 * machmode.h (smallest_mode_for_size): Fix formatting.
13804 (smallest_int_mode_for_size): New function.
13805 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
13806 instead of smallest_mode_for_size.
13807 * combine.c (make_extraction): Likewise.
13808 * config/arc/arc.c (arc_expand_movmem): Likewise.
13809 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
13810 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13811 * config/s390/s390.c (s390_expand_insv): Likewise.
13812 * config/sparc/sparc.c (assign_int_registers): Likewise.
13813 * config/spu/spu.c (spu_function_value): Likewise.
13814 (spu_function_arg): Likewise.
13815 * coverage.c (get_gcov_type): Likewise.
13816 (get_gcov_unsigned_t): Likewise.
13817 * dse.c (find_shift_sequence): Likewise.
13818 * expmed.c (store_bit_field_1): Likewise.
13819 * expr.c (convert_move): Likewise.
13820 (store_field): Likewise.
13821 * internal-fn.c (expand_arith_overflow): Likewise.
13822 * optabs-query.c (get_best_extraction_insn): Likewise.
13823 * optabs.c (expand_twoval_binop_libfunc): Likewise.
13824 * stor-layout.c (layout_type): Likewise.
13825 (initialize_sizetypes): Likewise.
13826 * targhooks.c (default_get_mask_mode): Likewise.
13827 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
13829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13830 Alan Hayward <alan.hayward@arm.com>
13831 David Sherwood <david.sherwood@arm.com>
13833 * machmode.h (opt_mode::else_blk): New function.
13834 (int_mode_for_mode): Declare.
13835 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13836 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13838 * cfgexpand.c (expand_debug_expr): Likewise.
13839 * combine.c (gen_lowpart_or_truncate): Likewise.
13840 (gen_lowpart_for_combine): Likewise.
13841 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13842 * config/avr/avr.c (avr_to_int_mode): Likewise.
13843 (avr_out_plus_1): Likewise.
13844 (avr_out_plus): Likewise.
13845 (avr_out_round): Likewise.
13846 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13847 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13848 (s390_expand_vcond): Likewise.
13849 * config/spu/spu.c (spu_split_immediate): Likewise.
13850 (spu_expand_mov): Likewise.
13851 * dse.c (get_stored_val): Likewise.
13852 * expmed.c (store_bit_field_1): Likewise.
13853 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13855 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13856 (extract_low_bits): Likewise.
13857 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13858 handling rather than repeating the check.
13859 (emit_group_store): Likewise.
13860 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13861 * optabs.c (expand_absneg_bit): Likewise.
13862 (expand_copysign_absneg): Likewise.
13863 (expand_copysign_bit): Likewise.
13864 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13865 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13866 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13867 * var-tracking.c (prepare_call_arguments): Likewise.
13868 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13869 int_mode_for_mode instead of mode_for_size.
13870 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13873 Alan Hayward <alan.hayward@arm.com>
13874 David Sherwood <david.sherwood@arm.com>
13876 * machmode.h (int_mode_for_size): New function.
13877 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13878 instead of mode_for_size.
13879 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13881 * combine.c (expand_field_assignment): Use int_mode_for_size
13882 instead of mode_for_size.
13883 (make_extraction): Likewise.
13884 (simplify_shift_const_1): Likewise.
13885 (simplify_comparison): Likewise.
13886 * dojump.c (do_jump): Likewise.
13887 * dwarf2out.c (mem_loc_descriptor): Likewise.
13888 * emit-rtl.c (init_derived_machine_modes): Likewise.
13889 * expmed.c (flip_storage_order): Likewise.
13890 (convert_extracted_bit_field): Likewise.
13891 * expr.c (copy_blkmode_from_reg): Likewise.
13892 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13893 * internal-fn.c (expand_mul_overflow): Likewise.
13894 * lower-subreg.c (simple_move): Likewise.
13895 * optabs-libfuncs.c (init_optabs): Likewise.
13896 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13897 * tree.c (vector_type_mode): Likewise.
13898 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13899 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13900 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13901 * tree-vect-stmts.c (vectorizable_load): Likewise.
13902 (vectorizable_store): Likewise.
13904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13905 Alan Hayward <alan.hayward@arm.com>
13906 David Sherwood <david.sherwood@arm.com>
13908 * coretypes.h (pod_mode): New type.
13909 (scalar_int_mode_pod): New typedef.
13910 * machmode.h (pod_mode): New class.
13911 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13912 * genmodes.c (emit_mode_int_n): Update accordingly.
13913 * lower-subreg.h (target_lower_subreg): Change type to
13914 scalar_int_mode_pod.
13915 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13916 scalar_int_mode_pod.
13918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13919 Alan Hayward <alan.hayward@arm.com>
13920 David Sherwood <david.sherwood@arm.com>
13922 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13923 machine_mode to scalar_int_mode.
13924 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13925 (rs6000_option_override_internal): Remove cast to int.
13926 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13927 machine_mode to scalar_int_mode.
13928 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13929 (rs6000_option_override_internal): Remove cast to int.
13930 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13931 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13933 * config/s390/s390.c (s390_expand_builtin): Likewise.
13934 * coretypes.h (scalar_int_mode): New type.
13935 (opt_scalar_int_mode): New typedef.
13936 * machmode.h (scalar_int_mode): New class.
13937 (scalar_int_mode::includes_p): New function.
13938 (byte_mode): Change type to scalar_int_mode.
13939 (word_mode): Likewise.
13940 (ptr_mode): Likewise.
13941 * emit-rtl.c (byte_mode): Likewise.
13942 (word_mode): Likewise.
13943 (ptr_mode): Likewise.
13944 (init_derived_machine_modes): Update accordingly.
13945 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13946 and MODE_PARTIAL_INT.
13947 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13948 opt_scalar_int_mode.
13950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13951 Alan Hayward <alan.hayward@arm.com>
13952 David Sherwood <david.sherwood@arm.com>
13954 * target.def (libgcc_floating_mode_supported_p): Take a
13956 * doc/tm.texi: Regenerate.
13957 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13959 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13960 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13964 Alan Hayward <alan.hayward@arm.com>
13965 David Sherwood <david.sherwood@arm.com>
13967 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13968 * doc/tm.texi: Regenerate.
13969 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13970 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13971 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13972 * targhooks.h (default_floatn_mode): Likewise.
13973 * targhooks.c (default_floatn_mode): Likewise.
13974 * tree.c (build_common_tree_nodes): Update accordingly.
13976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13977 Alan Hayward <alan.hayward@arm.com>
13978 David Sherwood <david.sherwood@arm.com>
13980 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13981 (mode_iterator::iterate_p): Likewise.
13982 (mode_iterator::get_wider): Likewise.
13983 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13986 Alan Hayward <alan.hayward@arm.com>
13987 David Sherwood <david.sherwood@arm.com>
13989 * coretypes.h (opt_scalar_float_mode): New typedef.
13990 * machmode.h (float_mode_for_size): New function.
13991 * emit-rtl.c (double_mode): Delete.
13992 (init_emit_once): Use float_mode_for_size.
13993 * stor-layout.c (layout_type): Likewise.
13994 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13997 Alan Hayward <alan.hayward@arm.com>
13998 David Sherwood <david.sherwood@arm.com>
14000 * output.h (assemble_real): Take a scalar_float_mode.
14001 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14002 * config/arm/arm.md (consttable_4): Likewise.
14003 (consttable_8): Likewise.
14004 (consttable_16): Likewise.
14005 * config/mips/mips.md (consttable_float): Likewise.
14006 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14007 * varasm.c (assemble_real): Take a scalar_float_mode.
14008 (output_constant_pool_2): Update accordingly.
14009 (output_constant): Likewise.
14011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14012 Alan Hayward <alan.hayward@arm.com>
14013 David Sherwood <david.sherwood@arm.com>
14015 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14016 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14017 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14018 (native_encode_real): Likewise.
14019 (native_interpret_real): Likewise.
14020 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14021 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14024 Alan Hayward <alan.hayward@arm.com>
14025 David Sherwood <david.sherwood@arm.com>
14027 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14028 <scalar_float_mode>. Simplify.
14029 (gen_extend_conv_libfunc): Likewise.
14031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14032 Alan Hayward <alan.hayward@arm.com>
14033 David Sherwood <david.sherwood@arm.com>
14035 * coretypes.h (scalar_float_mode): New type.
14036 * machmode.h (mode_traits::from_int): Use machine_mode if
14037 USE_ENUM_MODES is defined.
14038 (is_a): New function.
14040 (dyn_cast): Likewise.
14041 (scalar_float_mode): New class.
14042 (scalar_float_mode::includes_p): New function.
14043 (is_float_mode): Likewise.
14044 * gdbhooks.py (MachineModePrinter): New class.
14045 (build_pretty_printer): Use it for scalar_float_mode.
14046 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14047 (format_helper::format_helper): Turn into a template.
14048 * genmodes.c (get_mode_class): New function.
14049 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14050 or machine_mode if none.
14051 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14052 as_a <scalar_float_mode>.
14053 * dwarf2out.c (mem_loc_descriptor): Likewise.
14054 (insert_float): Likewise.
14055 (add_const_value_attribute): Likewise.
14056 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14057 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14058 (expand_unop): Update accordingly.
14059 (expand_abs_nojump): Likewise.
14060 (expand_copysign_absneg): Take a scalar_float_mode.
14061 (expand_copysign_bit): Likewise.
14062 (expand_copysign): Update accordingly.
14064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14065 Alan Hayward <alan.hayward@arm.com>
14066 David Sherwood <david.sherwood@arm.com>
14068 * coretypes.h (opt_mode): New class.
14069 * machmode.h (opt_mode): Likewise.
14070 (opt_mode::else_void): New function.
14071 (opt_mode::require): Likewise.
14072 (opt_mode::exists): Likewise.
14073 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14074 (GET_MODE_2XWIDER_MODE): Likewise.
14075 (mode_iterator::get_wider): Update accordingly.
14076 (mode_iterator::get_2xwider): Likewise.
14077 (mode_iterator::get_known_wider): Likewise, turning into a template.
14078 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14079 forcing a wider mode to exist.
14080 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14081 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14082 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14083 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14084 * lower-subreg.c (init_lower_subreg): Likewise.
14085 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14086 on the final iteration.
14087 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14088 a wider mode exists before asking for a move pattern.
14089 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14090 forcing a wider mode to exist.
14091 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14092 returning false if no such mode exists.
14093 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14094 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14095 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14096 Avoid checking for a MODE_INT if we already know the mode is not a
14098 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14099 forcing a wider mode to exist.
14100 (expmed_mult_highpart_optab): Likewise.
14101 (expmed_mult_highpart): Likewise.
14102 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14104 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14105 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14106 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14107 * internal-fn.c (expand_mul_overflow): Update use of
14108 GET_MODE_2XWIDER_MODE.
14109 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14110 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14111 GET_MODE_WIDER_MODE.
14112 (convert_plusminus_to_widen): Likewise.
14113 * tree-switch-conversion.c (array_value_type): Likewise.
14114 * var-tracking.c (emit_note_insn_var_location): Likewise.
14115 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14116 Return false inside rather than outside the loop if no wider mode
14118 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14119 and GET_MODE_2XWIDER_MODE
14120 (can_compare_p): Use else_void.
14121 * gdbhooks.py (OptMachineModePrinter): New class.
14122 (build_pretty_printer): Use it for opt_mode.
14124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14125 Alan Hayward <alan.hayward@arm.com>
14126 David Sherwood <david.sherwood@arm.com>
14128 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14129 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14132 Alan Hayward <alan.hayward@arm.com>
14133 David Sherwood <david.sherwood@arm.com>
14135 * machmode.h (mode_traits): New structure.
14136 (get_narrowest_mode): New function.
14137 (mode_iterator::start): Likewise.
14138 (mode_iterator::iterate_p): Likewise.
14139 (mode_iterator::get_wider): Likewise.
14140 (mode_iterator::get_known_wider): Likewise.
14141 (mode_iterator::get_2xwider): Likewise.
14142 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14143 (FOR_EACH_MODE): Likewise.
14144 (FOR_EACH_MODE_FROM): Likewise.
14145 (FOR_EACH_MODE_UNTIL): Likewise.
14146 (FOR_EACH_WIDER_MODE): Likewise.
14147 (FOR_EACH_2XWIDER_MODE): Likewise.
14148 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14149 * combine.c (simplify_comparison): Likewise
14150 * config/i386/i386.c (type_natural_mode): Likewise.
14151 * cse.c (cse_insn): Likewise.
14152 * dse.c (find_shift_sequence): Likewise.
14153 * emit-rtl.c (init_derived_machine_modes): Likewise.
14154 (init_emit_once): Likewise.
14155 * explow.c (hard_function_value): Likewise.
14156 * expmed.c (extract_fixed_bit_field_1): Likewise.
14157 (extract_bit_field_1): Likewise.
14158 (expand_divmod): Likewise.
14159 (emit_store_flag_1): Likewise.
14160 * expr.c (init_expr_target): Likewise.
14161 (convert_move): Likewise.
14162 (alignment_for_piecewise_move): Likewise.
14163 (widest_int_mode_for_size): Likewise.
14164 (emit_block_move_via_movmem): Likewise.
14165 (copy_blkmode_to_reg): Likewise.
14166 (set_storage_via_setmem): Likewise.
14167 (compress_float_constant): Likewise.
14168 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14169 * optabs-query.c (get_best_extraction_insn): Likewise.
14170 * optabs.c (expand_binop): Likewise.
14171 (expand_twoval_unop): Likewise.
14172 (expand_twoval_binop): Likewise.
14173 (widen_leading): Likewise.
14174 (widen_bswap): Likewise.
14175 (expand_parity): Likewise.
14176 (expand_unop): Likewise.
14177 (prepare_cmp_insn): Likewise.
14178 (prepare_float_lib_cmp): Likewise.
14179 (expand_float): Likewise.
14180 (expand_fix): Likewise.
14181 (expand_sfix_optab): Likewise.
14182 * postreload.c (move2add_use_add2_insn): Likewise.
14183 * reg-stack.c (reg_to_stack): Likewise.
14184 * reginfo.c (choose_hard_reg_mode): Likewise.
14185 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14186 * stor-layout.c (mode_for_size): Likewise.
14187 (smallest_mode_for_size): Likewise.
14188 (mode_for_vector): Likewise.
14189 (finish_bitfield_representative): Likewise.
14190 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14191 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14192 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14193 * var-tracking.c (prepare_call_arguments): Likewise.
14195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14196 Alan Hayward <alan.hayward@arm.com>
14197 David Sherwood <david.sherwood@arm.com>
14199 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14200 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14201 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14202 * machmode.h (mode_size): Move earlier in file.
14203 (mode_precision): Likewise.
14204 (mode_inner): Likewise.
14205 (mode_nunits): Likewise.
14206 (mode_unit_size): Likewise.
14207 (unit_unit_precision): Likewise.
14208 (mode_wider): Likewise.
14209 (mode_2xwider): Likewise.
14210 (machine_mode): New class.
14211 (mode_to_bytes): New function.
14212 (mode_to_bits): Likewise.
14213 (mode_to_precision): Likewise.
14214 (mode_to_inner): Likewise.
14215 (mode_to_unit_size): Likewise.
14216 (mode_to_unit_precision): Likewise.
14217 (mode_to_nunits): Likewise.
14218 (GET_MODE_SIZE): Use mode_to_bytes.
14219 (GET_MODE_BITSIZE): Use mode_to_bits.
14220 (GET_MODE_PRECISION): Use mode_to_precision.
14221 (GET_MODE_INNER): Use mode_to_inner.
14222 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14223 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14224 (GET_MODE_NUNITS): Use mode_to_nunits.
14225 * system.h (ALWAYS_INLINE): New macro.
14226 * config/powerpcspe/powerpcspe-c.c
14227 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14228 int for arg1_mode and arg2_mode.
14230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14231 Alan Hayward <alan.hayward@arm.com>
14232 David Sherwood <david.sherwood@arm.com>
14234 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14235 Prefix mode names with E_ in case statements.
14236 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14237 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14238 (aarch64_split_simd_move): Likewise.
14239 (aarch64_gen_storewb_pair): Likewise.
14240 (aarch64_gen_loadwb_pair): Likewise.
14241 (aarch64_gen_store_pair): Likewise.
14242 (aarch64_gen_load_pair): Likewise.
14243 (aarch64_get_condition_code_1): Likewise.
14244 (aarch64_constant_pool_reload_icode): Likewise.
14245 (get_rsqrte_type): Likewise.
14246 (get_rsqrts_type): Likewise.
14247 (get_recpe_type): Likewise.
14248 (get_recps_type): Likewise.
14249 (aarch64_gimplify_va_arg_expr): Likewise.
14250 (aarch64_simd_container_mode): Likewise.
14251 (aarch64_emit_load_exclusive): Likewise.
14252 (aarch64_emit_store_exclusive): Likewise.
14253 (aarch64_expand_compare_and_swap): Likewise.
14254 (aarch64_gen_atomic_cas): Likewise.
14255 (aarch64_emit_bic): Likewise.
14256 (aarch64_emit_atomic_swap): Likewise.
14257 (aarch64_emit_atomic_load_op): Likewise.
14258 (aarch64_evpc_trn): Likewise.
14259 (aarch64_evpc_uzp): Likewise.
14260 (aarch64_evpc_zip): Likewise.
14261 (aarch64_evpc_ext): Likewise.
14262 (aarch64_evpc_rev): Likewise.
14263 (aarch64_evpc_dup): Likewise.
14264 (aarch64_gen_ccmp_first): Likewise.
14265 (aarch64_gen_ccmp_next): Likewise.
14266 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14267 (alpha_emit_xfloating_libcall): Likewise.
14268 (emit_insxl): Likewise.
14269 (alpha_arg_type): Likewise.
14270 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14271 (arc_preferred_simd_mode): Likewise.
14272 (arc_secondary_reload): Likewise.
14273 (get_arc_condition_code): Likewise.
14274 (arc_print_operand): Likewise.
14275 (arc_legitimate_constant_p): Likewise.
14276 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14277 * config/arc/arc.md (casesi_load): Likewise.
14278 (casesi_compact_jump): Likewise.
14279 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14280 (cc_use_register): Likewise.
14281 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14282 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14283 (arm_init_iwmmxt_builtins): Likewise.
14284 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14285 (neon_expand_vector_init): Likewise.
14286 (arm_attr_length_move_neon): Likewise.
14287 (maybe_get_arm_condition_code): Likewise.
14288 (arm_emit_vector_const): Likewise.
14289 (arm_preferred_simd_mode): Likewise.
14290 (arm_output_iwmmxt_tinsr): Likewise.
14291 (thumb1_output_casesi): Likewise.
14292 (thumb2_output_casesi): Likewise.
14293 (arm_emit_load_exclusive): Likewise.
14294 (arm_emit_store_exclusive): Likewise.
14295 (arm_expand_compare_and_swap): Likewise.
14296 (arm_evpc_neon_vuzp): Likewise.
14297 (arm_evpc_neon_vzip): Likewise.
14298 (arm_evpc_neon_vrev): Likewise.
14299 (arm_evpc_neon_vtrn): Likewise.
14300 (arm_evpc_neon_vext): Likewise.
14301 (arm_validize_comparison): Likewise.
14302 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14303 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14304 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14305 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14306 (c6x_preferred_simd_mode): Likewise.
14307 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14308 (epiphany_rtx_costs): Likewise.
14309 * config/epiphany/predicates.md (proper_comparison_operator):
14311 * config/frv/frv.c (condexec_memory_operand): Likewise.
14312 (frv_emit_move): Likewise.
14313 (output_move_single): Likewise.
14314 (output_condmove_single): Likewise.
14315 (frv_hard_regno_mode_ok): Likewise.
14316 (frv_matching_accg_mode): Likewise.
14317 * config/h8300/h8300.c (split_adds_subs): Likewise.
14318 (h8300_rtx_costs): Likewise.
14319 (h8300_print_operand): Likewise.
14320 (compute_mov_length): Likewise.
14321 (output_logical_op): Likewise.
14322 (compute_logical_op_length): Likewise.
14323 (compute_logical_op_cc): Likewise.
14324 (h8300_shift_needs_scratch_p): Likewise.
14325 (output_a_shift): Likewise.
14326 (compute_a_shift_length): Likewise.
14327 (compute_a_shift_cc): Likewise.
14328 (expand_a_rotate): Likewise.
14329 (output_a_rotate): Likewise.
14330 * config/i386/i386.c (classify_argument): Likewise.
14331 (function_arg_advance_32): Likewise.
14332 (function_arg_32): Likewise.
14333 (function_arg_64): Likewise.
14334 (function_value_64): Likewise.
14335 (ix86_gimplify_va_arg): Likewise.
14336 (ix86_legitimate_constant_p): Likewise.
14337 (put_condition_code): Likewise.
14338 (split_double_mode): Likewise.
14339 (ix86_avx256_split_vector_move_misalign): Likewise.
14340 (ix86_expand_vector_logical_operator): Likewise.
14341 (ix86_split_idivmod): Likewise.
14342 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14343 (ix86_build_const_vector): Likewise.
14344 (ix86_build_signbit_mask): Likewise.
14345 (ix86_match_ccmode): Likewise.
14346 (ix86_cc_modes_compatible): Likewise.
14347 (ix86_expand_branch): Likewise.
14348 (ix86_expand_sse_cmp): Likewise.
14349 (ix86_expand_sse_movcc): Likewise.
14350 (ix86_expand_int_sse_cmp): Likewise.
14351 (ix86_expand_vec_perm_vpermi2): Likewise.
14352 (ix86_expand_vec_perm): Likewise.
14353 (ix86_expand_sse_unpack): Likewise.
14354 (ix86_expand_int_addcc): Likewise.
14355 (ix86_split_to_parts): Likewise.
14356 (ix86_vectorize_builtin_gather): Likewise.
14357 (ix86_vectorize_builtin_scatter): Likewise.
14358 (avx_vpermilp_parallel): Likewise.
14359 (inline_memory_move_cost): Likewise.
14360 (ix86_tieable_integer_mode_p): Likewise.
14361 (x86_maybe_negate_const_int): Likewise.
14362 (ix86_expand_vector_init_duplicate): Likewise.
14363 (ix86_expand_vector_init_one_nonzero): Likewise.
14364 (ix86_expand_vector_init_one_var): Likewise.
14365 (ix86_expand_vector_init_concat): Likewise.
14366 (ix86_expand_vector_init_interleave): Likewise.
14367 (ix86_expand_vector_init_general): Likewise.
14368 (ix86_expand_vector_set): Likewise.
14369 (ix86_expand_vector_extract): Likewise.
14370 (emit_reduc_half): Likewise.
14371 (ix86_emit_i387_round): Likewise.
14372 (ix86_mangle_type): Likewise.
14373 (ix86_expand_round_sse4): Likewise.
14374 (expand_vec_perm_blend): Likewise.
14375 (canonicalize_vector_int_perm): Likewise.
14376 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14377 (expand_vec_perm_1): Likewise.
14378 (expand_vec_perm_interleave3): Likewise.
14379 (expand_vec_perm_even_odd_pack): Likewise.
14380 (expand_vec_perm_even_odd_1): Likewise.
14381 (expand_vec_perm_broadcast_1): Likewise.
14382 (ix86_vectorize_vec_perm_const_ok): Likewise.
14383 (ix86_expand_vecop_qihi): Likewise.
14384 (ix86_expand_mul_widen_hilo): Likewise.
14385 (ix86_expand_sse2_abs): Likewise.
14386 (ix86_expand_pextr): Likewise.
14387 (ix86_expand_pinsr): Likewise.
14388 (ix86_preferred_simd_mode): Likewise.
14389 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14390 * config/i386/sse.md (*andnot<mode>3): Likewise.
14391 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14392 (*<code><mode>3): Likewise.
14393 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14394 (ia64_expand_atomic_op): Likewise.
14395 (ia64_arg_type): Likewise.
14396 (ia64_mode_to_int): Likewise.
14397 (ia64_scalar_mode_supported_p): Likewise.
14398 (ia64_vector_mode_supported_p): Likewise.
14399 (expand_vec_perm_broadcast): Likewise.
14400 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14401 (iq2000_function_arg_advance): Likewise.
14402 (iq2000_function_arg): Likewise.
14403 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14404 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14405 (m68k_libcall_value): Likewise.
14406 (m68k_function_value): Likewise.
14407 (sched_attr_op_type): Likewise.
14408 * config/mcore/mcore.c (mcore_output_move): Likewise.
14409 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14411 (microblaze_function_arg): Likewise.
14412 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14413 (mips_print_operand): Likewise.
14414 (mips_mode_ok_for_mov_fmt_p): Likewise.
14415 (mips_vector_mode_supported_p): Likewise.
14416 (mips_preferred_simd_mode): Likewise.
14417 (mips_expand_vpc_loongson_even_odd): Likewise.
14418 (mips_expand_vec_unpack): Likewise.
14419 (mips_expand_vi_broadcast): Likewise.
14420 (mips_expand_vector_init): Likewise.
14421 (mips_expand_vec_reduc): Likewise.
14422 (mips_expand_msa_cmp): Likewise.
14423 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14424 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14425 (cc_flags_for_mode): Likewise.
14426 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14427 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14428 (nds32_output_casesi_pc_relative): Likewise.
14429 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14430 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14431 (nvptx_gen_unpack): Likewise.
14432 (nvptx_gen_pack): Likewise.
14433 (nvptx_gen_shuffle): Likewise.
14434 (nvptx_gen_wcast): Likewise.
14435 (nvptx_preferred_simd_mode): Likewise.
14436 * config/pa/pa.c (pa_secondary_reload): Likewise.
14437 * config/pa/predicates.md (base14_operand): Likewise.
14438 * config/powerpcspe/powerpcspe-c.c
14439 (altivec_resolve_overloaded_builtin): Likewise.
14440 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14442 (rs6000_preferred_simd_mode): Likewise.
14443 (output_vec_const_move): Likewise.
14444 (rs6000_expand_vector_extract): Likewise.
14445 (rs6000_split_vec_extract_var): Likewise.
14446 (reg_offset_addressing_ok_p): Likewise.
14447 (rs6000_legitimate_offset_address_p): Likewise.
14448 (rs6000_legitimize_address): Likewise.
14449 (rs6000_emit_set_const): Likewise.
14450 (rs6000_const_vec): Likewise.
14451 (rs6000_emit_move): Likewise.
14452 (spe_build_register_parallel): Likewise.
14453 (rs6000_darwin64_record_arg_recurse): Likewise.
14454 (swap_selector_for_mode): Likewise.
14455 (spe_init_builtins): Likewise.
14456 (paired_init_builtins): Likewise.
14457 (altivec_init_builtins): Likewise.
14458 (do_load_for_compare): Likewise.
14459 (rs6000_generate_compare): Likewise.
14460 (rs6000_expand_float128_convert): Likewise.
14461 (emit_load_locked): Likewise.
14462 (emit_store_conditional): Likewise.
14463 (rs6000_output_function_epilogue): Likewise.
14464 (rs6000_handle_altivec_attribute): Likewise.
14465 (rs6000_function_value): Likewise.
14466 (emit_fusion_gpr_load): Likewise.
14467 (emit_fusion_p9_load): Likewise.
14468 (emit_fusion_p9_store): Likewise.
14469 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14470 (fusion_gpr_mem_load): Likewise.
14471 (fusion_addis_mem_combo_load): Likewise.
14472 (fusion_addis_mem_combo_store): Likewise.
14473 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14474 (fusion_gpr_mem_load): Likewise.
14475 (fusion_addis_mem_combo_load): Likewise.
14476 (fusion_addis_mem_combo_store): Likewise.
14477 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14479 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14480 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14481 (rs6000_preferred_simd_mode): Likewise.
14482 (output_vec_const_move): Likewise.
14483 (rs6000_expand_vector_extract): Likewise.
14484 (rs6000_split_vec_extract_var): Likewise.
14485 (reg_offset_addressing_ok_p): Likewise.
14486 (rs6000_legitimate_offset_address_p): Likewise.
14487 (rs6000_legitimize_address): Likewise.
14488 (rs6000_emit_set_const): Likewise.
14489 (rs6000_const_vec): Likewise.
14490 (rs6000_emit_move): Likewise.
14491 (rs6000_darwin64_record_arg_recurse): Likewise.
14492 (swap_selector_for_mode): Likewise.
14493 (paired_init_builtins): Likewise.
14494 (altivec_init_builtins): Likewise.
14495 (rs6000_expand_float128_convert): Likewise.
14496 (emit_load_locked): Likewise.
14497 (emit_store_conditional): Likewise.
14498 (rs6000_output_function_epilogue): Likewise.
14499 (rs6000_handle_altivec_attribute): Likewise.
14500 (rs6000_function_value): Likewise.
14501 (emit_fusion_gpr_load): Likewise.
14502 (emit_fusion_p9_load): Likewise.
14503 (emit_fusion_p9_store): Likewise.
14504 * config/rx/rx.c (rx_gen_move_template): Likewise.
14505 (flags_from_mode): Likewise.
14506 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14507 (s390_slb_comparison): Likewise.
14508 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14509 (s390_vector_mode_supported_p): Likewise.
14510 (s390_cc_modes_compatible): Likewise.
14511 (s390_match_ccmode_set): Likewise.
14512 (s390_canonicalize_comparison): Likewise.
14513 (s390_emit_compare_and_swap): Likewise.
14514 (s390_branch_condition_mask): Likewise.
14515 (s390_rtx_costs): Likewise.
14516 (s390_secondary_reload): Likewise.
14517 (__SECONDARY_RELOAD_CASE): Likewise.
14518 (s390_expand_cs): Likewise.
14519 (s390_preferred_simd_mode): Likewise.
14520 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14521 * config/sh/sh.c (sh_print_operand): Likewise.
14522 (dump_table): Likewise.
14523 (sh_secondary_reload): Likewise.
14524 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14525 * config/sh/sh.md (casesi_worker_1): Likewise.
14526 (casesi_worker_2): Likewise.
14527 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14528 (fcc_comparison_operator): Likewise.
14529 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14530 (emit_soft_tfmode_cvt): Likewise.
14531 (sparc_preferred_simd_mode): Likewise.
14532 (output_cbranch): Likewise.
14533 (sparc_print_operand): Likewise.
14534 (sparc_expand_vec_perm_bmask): Likewise.
14535 (vector_init_bshuffle): Likewise.
14536 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14537 (spu_vector_mode_supported_p): Likewise.
14538 (spu_expand_insv): Likewise.
14539 (spu_emit_branch_or_set): Likewise.
14540 (spu_handle_vector_attribute): Likewise.
14541 (spu_builtin_splats): Likewise.
14542 (spu_builtin_extract): Likewise.
14543 (spu_builtin_promote): Likewise.
14544 (spu_expand_sign_extend): Likewise.
14545 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14546 (tilegx_simd_int): Likewise.
14547 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14548 (tilepro_simd_int): Likewise.
14549 * config/v850/v850.c (const_double_split): Likewise.
14550 (v850_print_operand): Likewise.
14551 (ep_memory_offset): Likewise.
14552 * config/vax/vax.c (vax_rtx_costs): Likewise.
14553 (vax_output_int_move): Likewise.
14554 (vax_output_int_add): Likewise.
14555 (vax_output_int_subtract): Likewise.
14556 * config/visium/predicates.md (visium_branch_operator): Likewise.
14557 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14558 (visium_print_operand_address): Likewise.
14559 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14560 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14561 (xtensa_expand_conditional_branch): Likewise.
14562 (xtensa_copy_incoming_a7): Likewise.
14563 (xtensa_output_literal): Likewise.
14564 * dfp.c (decimal_real_maxval): Likewise.
14565 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14568 Alan Hayward <alan.hayward@arm.com>
14569 David Sherwood <david.sherwood@arm.com>
14571 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14572 (mode_nunits_inline): Likewise.
14573 (mode_inner_inline): Likewise.
14574 (mode_unit_size_inline): Likewise.
14575 (mode_unit_precision_inline): Likewise.
14576 (emit_insn_modes_h): Likewise. Also emit a #define of the
14578 (emit_mode_wider): Add an E_ prefix to mode names.
14579 (emit_mode_complex): Likewise.
14580 (emit_mode_inner): Likewise.
14581 (emit_mode_adjustments): Likewise.
14582 (emit_mode_int_n): Likewise.
14583 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14584 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14585 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14586 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14587 (CRC32_BUILTIN, ENTRY): Likewise.
14588 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14589 (aarch64_pop_regs): Likewise.
14590 (aarch64_process_components): Likewise.
14591 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14592 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14593 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14594 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14595 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14596 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14598 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14599 (pcmpistr): Likewise.
14600 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14601 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14602 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14604 * config/rl78/rl78.c (mduc_regs): Likewise.
14605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14606 (htm_expand_builtin): Likewise.
14607 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14608 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14609 * config/xtensa/xtensa.c (print_operand): Likewise.
14610 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14611 (NUM_MODE_VECTOR_INT): Likewise.
14612 * genoutput.c (null_operand): Likewise.
14613 (output_operand_data): Likewise.
14614 * genrecog.c (print_parameter_value): Likewise.
14615 * lra.c (debug_operand_data): Likewise.
14617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14619 * dwarf2out.c (scompare_loc_descriptor_wide)
14620 (scompare_loc_descriptor_narrow): New functions, split out from...
14621 (scompare_loc_descriptor): ...here.
14622 * expmed.c (emit_store_flag_int): New function, split out from...
14623 (emit_store_flag): ...here.
14625 2017-08-30 Richard Biener <rguenther@suse.de>
14627 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14628 (dwarf2out_early_finish): Move setting of AT_pubnames from
14629 early debug output to early finish.
14631 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14633 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14634 and -mdata-region to the assembler.
14636 2017-08-30 Richard Biener <rguenther@suse.de>
14638 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14640 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14641 (dwarf2out_early_global_decl): Only generate a DIE for the
14642 abstract origin if it doesn't already exist or is a declaration DIE.
14643 (resolve_addr): Do not add the linkage name twice when
14644 generating a stub DIE for the DW_TAG_GNU_call_site target.
14646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14648 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14649 Use machine_mode rather than int for arg1_mode.
14651 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14654 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14655 that the second argument of the built-in functions to unpack
14656 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14657 switch statement instead a lot of if statements.
14658 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14659 Allow 64-bit values to be in Altivec registers as well as
14660 traditional floating point registers.
14661 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14663 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14665 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14666 MAX_REGS_PER_ADDRESS == 1.
14668 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14670 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14671 * config/i386/i386.c (ix86_option_override_internal): Simplify
14672 setting of opts->x_flag_entry.
14674 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14675 Jakub Jelinek <jakub@redhat.com>
14676 Richard Biener <rguenther@suse.de>
14678 PR tree-optimization/81503
14679 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14680 folded constant fits in the target type; reorder tests for clarity.
14682 2017-08-29 Martin Liska <mliska@suse.cz>
14684 * passes.def: Include pass_lower_switch.
14685 * stmt.c (dump_case_nodes): Remove and move to
14686 tree-switch-conversion.
14687 (case_values_threshold): Likewise.
14688 (expand_switch_as_decision_tree_p): Likewise.
14689 (emit_case_decision_tree): Likewise.
14690 (expand_case): Likewise.
14691 (balance_case_nodes): Likewise.
14692 (node_has_low_bound): Likewise.
14693 (node_has_high_bound): Likewise.
14694 (node_is_bounded): Likewise.
14695 (emit_case_nodes): Likewise.
14696 (struct simple_case_node): New struct.
14697 (add_case_node): Remove.
14698 (emit_case_dispatch_table): Use vector instead of case_list.
14699 (reset_out_edges_aux): Remove.
14700 (compute_cases_per_edge): Likewise.
14701 (expand_case): Build list of simple_case_node.
14702 (expand_sjlj_dispatch_table): Use it.
14703 * tree-switch-conversion.c (struct case_node): Moved from
14704 stmt.c and adjusted.
14705 (emit_case_nodes): Likewise.
14706 (node_has_low_bound): Likewise.
14707 (node_has_high_bound): Likewise.
14708 (node_is_bounded): Likewise.
14709 (case_values_threshold): Likewise.
14710 (reset_out_edges_aux): Likewise.
14711 (compute_cases_per_edge): Likewise.
14712 (add_case_node): Likewise.
14713 (dump_case_nodes): Likewise.
14714 (balance_case_nodes): Likewise.
14715 (expand_switch_as_decision_tree_p): Likewise.
14716 (emit_jump): Likewise.
14717 (emit_case_decision_tree): Likewise.
14718 (try_switch_expansion): Likewise.
14719 (do_jump_if_equal): Likewise.
14720 (emit_cmp_and_jump_insns): Likewise.
14721 (fix_phi_operands_for_edge): New function.
14722 (record_phi_operand_mapping): Likewise.
14723 (class pass_lower_switch): New pass.
14724 (pass_lower_switch::execute): New function.
14725 (make_pass_lower_switch): Likewise.
14726 (conditional_probability):
14727 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14728 * tree-pass.h: Add make_pass_lower_switch.
14730 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14733 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14736 2017-08-29 Richard Biener <rguenther@suse.de>
14738 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14739 we do not add a DW_AT_inline attribute twice.
14740 (gen_subprogram_die): Remove code setting DW_AT_inline on
14741 DECL_ABSTRACT_P nodes.
14743 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14745 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14746 calls to internal functions.
14747 (gimplify_modify_expr): Likewise.
14748 * tree-call-cdce.c (use_internal_fn): Likewise.
14749 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14750 (convert_to_divmod): Set the nothrow flag.
14751 * tree-if-conv.c (predicate_mem_writes): Likewise.
14752 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14753 (vectorizable_call): Likewise.
14754 (vectorizable_store): Likewise.
14755 (vectorizable_load): Likewise.
14756 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14757 (vect_recog_mask_conversion_pattern): Likewise.
14759 2017-08-29 Martin Liska <mliska@suse.cz>
14762 * gcc.c (driver_handle_option): Add new argument.
14763 * opts-common.c (handle_option): Pass
14764 target_option_override_hook.
14765 * opts-global.c (lang_handle_option): Add new option.
14766 (set_default_handlers): Add new argument.
14767 (decode_options): Likewise.
14768 * opts.c (target_handle_option): Likewise.
14769 (common_handle_option): Call target_option_override_hook.
14770 * opts.h (struct cl_option_handler_func): Add hook for
14771 target option override.
14772 (struct cl_option_handlers): Likewise.
14773 (set_default_handlers): Add new argument.
14774 (decode_options): Likewise.
14775 (common_handle_option): Likewise.
14776 (target_handle_option): Likewise.
14777 * toplev.c (toplev::main): Pass targetm.target_option.override
14780 2017-08-29 Richard Biener <rguenther@suse.de>
14781 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14783 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14784 life to the active subtree.
14786 2017-08-28 Jeff Law <law@redhat.com>
14788 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14789 derive_equivalences.
14790 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14791 Code moved into....
14792 (edge_info::derive_equivalences): New private member function
14794 * tree-ssa-dom.c (class edge_info): Changed from a struct
14795 to a class. Add ctor/dtor, methods and data members.
14796 (edge_info::edge_info): Renamed from allocate_edge_info.
14797 Initialize additional members.
14798 (edge_info::~edge_info): New.
14799 (free_dom_edge_info): Delete the edge info.
14800 (record_edge_info): Use new class & associated member functions.
14801 Tighten forms for testing for edge equivalences.
14802 (record_temporary_equivalences): Iterate over the simple
14803 equivalences rather than assuming there's only one per edge.
14804 (cprop_into_successor_phis): Iterate over the simple
14805 equivalences rather than assuming there's only one per edge.
14806 (optimize_stmt): Use operand_equal_p rather than pointer
14807 equality for mini-DSE code.
14809 2017-08-28 Nathan Sidwell <nathan@acm.org>
14811 * gcc.c (execute): Fold SIGPIPE handling into switch
14812 statement. Adjust internal error message.
14814 2017-08-28 Richard Biener <rguenther@suse.de>
14817 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
14818 Do nothing for removed DIEs.
14820 2017-08-28 Richard Biener <rguenther@suse.de>
14822 PR tree-optimization/81977
14823 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
14826 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
14829 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
14830 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
14831 using targetm.gen_mem_thread_fence.
14833 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14836 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14837 predicate to register_operand. Reorder operands.
14838 (*btr<mode>): Ditto.
14839 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14840 (*btr<mode>_mask): Ditto.
14842 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14844 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14845 * config/rs6000/xmmintrin.h: New file.
14846 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14848 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14851 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14852 parameter and_insn and return it.
14853 (recombine_lvx_pattern): Insert a copy to ensure availability of
14854 the base register of the copied masking operation at the point of
14855 the instruction replacement.
14856 (recombine_stvx_pattern): Likewise.
14858 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14860 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14861 undocumented switches.
14862 (-mpower9-dform-vector): Likewise.
14863 (-mpower9-dform): Likewise.
14864 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14865 comments to delete references to -mpower9-dform* switches.
14866 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14867 Delete reference to -mpower9-dform* switches, test for
14868 -mpower9-vector instead.
14869 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14870 (OTHER_P9_VECTOR_MASKS): Likewise.
14871 (POWERPC_MASKS): Likewise.
14872 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14873 tests against -mpower9-dform* to -mpower9-vector. Delete code
14874 that checked for -mpower9-dform* consistancy with other options.
14875 Add test for -mpower9-misc to enable other power9 switches.
14876 (rs6000_init_hard_regno_mode_ok): Likewise.
14877 (rs6000_option_override_internal): Likewise.
14878 (rs6000_emit_prologue): Likewise.
14879 (rs6000_emit_epilogue): Likewise.
14880 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14881 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14882 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14884 (emit_fusion_p9_store): Likewise.
14885 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14886 resetting these macros if the assembler does not support ISA 3.0
14888 (TARGET_P9_DFORM_VECTOR): Likewise.
14889 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14890 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14892 2017-08-25 Alan Modra <amodra@gmail.com>
14894 PR rtl-optimization/81747
14895 * cse.c (cse_extended_basic_block): Don't attempt to record
14896 equivalences for degenerate conditional branches that branch
14897 to their fall-through.
14899 2017-08-24 Martin Sebor <msebor@redhat.com>
14901 PR middle-end/81908
14902 * gimple-fold.c (size_must_be_zero_p): New function.
14903 (gimple_fold_builtin_memory_op): Call it.
14905 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14907 * config/rs6000/mm_malloc.h: New file.
14909 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14911 PR tree-optimization/81913
14912 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14913 analysis when either IVs in condition can wrap.
14915 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14917 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14918 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14920 2017-08-24 Richard Biener <rguenther@suse.de>
14923 * targhooks.c (default_target_can_inline_p): Properly
14924 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14925 is present and always compare.
14926 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14927 infer -mfpmath=sse from TARGET_SSE_P.
14928 (ix86_can_inline_p): Properly use target_option_default_node when
14929 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14931 2017-08-24 Richard Biener <rguenther@suse.de>
14934 * dwarf2out.c (output_die): Handle flag_generate_offload like
14936 (output_comp_unit): Likewise.
14937 (gen_array_type_die): Likewise.
14938 (dwarf2out_early_finish): Likewise.
14939 (note_variable_value_in_expr): Likewise.
14940 (dwarf2out_finish): Likewise. Adjust assert.
14941 * cgraphunit.c (symbol_table::compile): Move setting of
14942 flag_generate_offload earlier ...
14943 (symbol_table::finalize_compilation_unit): ... here, before
14944 early debug finalization.
14946 2017-08-24 Richard Biener <rguenther@suse.de>
14948 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14949 and ipa-fnsummary.h.
14950 (ix86_can_inline_p): When ix86_fpmath flags do not match
14951 check whether the callee uses FP math at all.
14953 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14955 PR middle-end/81931
14956 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14957 instead of TYPE_PRECISION.
14959 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14960 Alan Hayward <alan.hayward@arm.com>
14961 David Sherwood <david.sherwood@arm.com>
14963 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14965 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14967 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14969 (ix86_frame::stack_realign_allocate): New field.
14970 (struct machine_frame_state): Modify comments.
14971 (machine_frame_state::sp_realigned_fp_end): New field.
14972 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14973 layout calculation.
14974 (sp_valid_at): Add assertion to assure no attempt to access invalid
14975 offset of a realigned stack.
14976 (fp_valid_at): Likewise.
14977 (choose_baseaddr): Modify comments.
14978 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14979 ix86_expand_prologue.
14980 (ix86_expand_prologue): Modify stack realignment and allocation.
14981 (ix86_expand_epilogue): Modify comments.
14982 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14983 avx2_runtime, avx512f, and avx512f_runtime.
14985 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14987 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14988 (mstackrealign): Do not init to -1.
14989 * config/i386/i386.c (ix86_option_override_internal):
14990 Check opts_set, not opts when setting default value of
14991 opts->x_ix86_force_align_arg_pointer.
14993 2017-08-23 Richard Biener <rguenther@suse.de>
14995 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14996 lang_hooks.decl_printable_name.
14997 * print-rtl-function.c (print_rtx_function): Likewise.
14998 * tree-pretty-print.c (dump_function_header): Likewise.
15000 2017-08-23 Richard Biener <rguenther@suse.de>
15003 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15004 -g0 at compile-time.
15006 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15008 PR middle-end/19706
15009 * doc/sourcebuild.texi (Other hardware attributes):
15012 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15014 PR middle-end/19706
15015 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15016 Fix single-use check.
15018 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15020 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15022 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15024 * doc/install.texi: Modify to add more details on running selected
15027 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15029 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15030 is combined with -mabi=ms.
15031 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15034 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15036 PR tree-optimization/81488
15037 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15038 and cached_basis fields.
15039 (MAX_SPREAD): New constant.
15040 (alloc_cand_and_find_basis): Initialize new fields.
15041 (clear_visited): New function.
15042 (create_phi_basis_1): Rename from create_phi_basis, set visited
15043 and cached_basis fields.
15044 (create_phi_basis): New wrapper function.
15045 (phi_add_costs_1): Rename from phi_add_costs, add spread
15046 parameter, set visited field, short-circuit when limits reached.
15047 (phi_add_costs): New wrapper function.
15048 (record_phi_increments_1): Rename from record_phi_increments, set
15050 (record_phi_increments): New wrapper function.
15051 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15052 (phi_incr_cost): New wrapper function.
15053 (all_phi_incrs_profitable_1): Rename from
15054 all_phi_incrs_profitable, set visited field.
15055 (all_phi_incrs_profitable): New wrapper function.
15057 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15058 Alan Hayward <alan.hayward@arm.com>
15059 David Sherwood <david.sherwood@arm.com>
15061 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15062 that takes the outer and inner modes.
15063 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15064 comparison as the canonical test for a paradoxical subreg.
15065 * combine.c (simplify_set): Use paradoxical_subreg_p.
15066 (make_extraction): Likewise.
15067 (force_to_mode): Likewise.
15068 (rtx_equal_for_field_assignment_p): Likewise.
15069 (gen_lowpart_for_combine): Likewise.
15070 (simplify_comparison): Likewise.
15071 * cse.c (equiv_constant): Likewise.
15072 * expmed.c (store_bit_field_1): Likewise.
15073 * final.c (alter_subreg): Likewise.
15074 * fwprop.c (propagate_rtx): Likewise.
15075 (forward_propagate_subreg): Likewise.
15076 * ira-conflicts.c (ira_build_conflicts): Likewise.
15077 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15078 * lra-constraints.c (curr_insn_transform): Likewise.
15079 (split_reg): Likewise.
15080 * lra-eliminations.c (move_plus_up): Likewise.
15081 (lra_eliminate_regs_1): Likewise.
15082 * recog.c (general_operand): Likewise.
15083 * ree.c (combine_reaching_defs): Likewise.
15084 * reload.c (push_reload): Likewise.
15085 (find_reloads): Likewise.
15086 * reload1.c (elimination_effects): Likewise.
15087 (compute_reload_subreg_offset): Likewise.
15088 (choose_reload_regs): Likewise.
15089 * rtlanal.c (subreg_lsb_1): Likewise.
15090 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15091 (simplify_subreg): Likewise.
15092 * var-tracking.c (track_loc_p): Likewise.
15093 * emit-rtl.c (byte_lowpart_offset): Likewise.
15094 (paradoxical_subreg_p): Delete out-of-line definition.
15096 2017-08-22 Jeff Law <law@redhat.com>
15098 PR tree-optimization/81741
15099 PR tree-optimization/71947
15100 * tree-ssa-dom.c: Include tree-inline.h.
15101 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15102 equivalences if one is more expensive to compute than the other.
15103 * tree-ssa-scopedtables.h (class const_or_copies): Make
15104 record_const_or_copy_raw method private.
15105 (class avail_exprs_stack): New method simplify_binary_operation.
15106 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15107 avail_exprs_stack::simplify_binary_operation as needed.
15108 (avail_exprs_stack::simplify_binary_operation): New function.
15110 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15112 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15113 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15114 (DOT_SYMBOLS): Likewise.
15115 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15116 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15117 (RS6000_ABI_NAME): Likewise.
15118 (TARGET_CMODEL): Likewise.
15119 (TOC_SECTION_ASM_OP): Likewise.
15120 (SET_CMODEL): New macro.
15121 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15123 2017-08-22 Richard Biener <rguenther@suse.de>
15125 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15126 to allow for free-lang-data replacements similar to verify_type_variant.
15128 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15129 Alan Hayward <alan.hayward@arm.com>
15130 David Sherwood <david.sherwood@arm.com>
15132 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15133 VOIDmode for the LABEL_REF.
15135 2017-08-22 Richard Biener <rguenther@suse.de>
15137 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15139 2017-08-22 Richard Biener <rguenther@suse.de>
15141 * common.opt (feliminate-dwarf2-dups): Ignore.
15142 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15143 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15144 same_die_p_wrap, compute_section_prefix,
15145 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15146 (comdat_symbol_id, comdat_symbol_number): Likewise.
15147 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15149 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15150 (output_die): Mark unreachable path unreachable.
15151 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15152 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15153 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15154 (dwarf2out_early_finish): Likewise.
15156 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15158 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15160 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15163 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15164 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15165 (avr_attribute_table) <io, io_low, address>: Initialize
15166 .decl_required with true.
15168 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15170 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15171 undocumented debugging options.
15172 (-mvsx-scalar-double): Likewise.
15173 (-mallow-df-permute): Likewise.
15174 (-mvectorize-builtins): Likewise.
15175 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15176 (rs6000_builtin_vectorized_function): Likewise.
15177 (rs6000_builtin_md_vectorized_function): Likewise.
15178 (rs6000_opt_vars): Likewise.
15180 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15183 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15184 (*btrq_imm): Rename from *btrq.
15185 (*btcq_imm): Rename from *btcq.
15186 (btsc): New code attribute.
15187 (*<btsc><mode>): New insn pattern.
15188 (*btr<mode>): Ditto.
15189 (*<btsc><mode>_mask): New insn_and_split pattern.
15190 (*btr<mode>_mask): Ditto.
15192 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15193 Alan Hayward <alan.hayward@arm.com>
15194 David Sherwood <david.sherwood@arm.com>
15196 * function.c (pad_below): Simplify padding calculation.
15198 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15199 Alan Hayward <alan.hayward@arm.com>
15200 David Sherwood <david.sherwood@arm.com>
15202 * target.def (function_prologue): Remove frame size argument.
15203 (function_epilogue): Likewise.
15204 * doc/tm.texi: Regenerate.
15205 * final.c (final_start_function): Update call to function_prologue.
15206 (final_end_function): Update call to function_epilogue.
15207 (default_function_pro_epilogue): Remove frame size argument.
15208 * output.h (default_function_pro_epilogue): Likewise.
15209 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15210 (arm_output_function_prologue): Likewise.
15211 * config/frv/frv.c (frv_function_prologue): Likewise.
15212 (frv_function_epilogue): Likewise.
15213 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15214 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15215 (ia64_output_function_epilogue): Likewise.
15216 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15217 (m32r_output_function_epilogue): Likewise.
15218 * config/microblaze/microblaze.c (microblaze_function_prologue)
15219 (microblaze_function_epilogue): Likewise.
15220 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15221 (mips_output_function_epilogue): Likewise.
15222 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15223 (mmix_target_asm_function_epilogue): Likewise.
15224 * config/msp430/msp430.c (msp430_start_function): Likewise.
15225 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15226 (nds32_asm_function_epilogue): Likewise.
15227 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15228 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15229 (pa_output_function_epilogue): Likewise.
15230 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15231 (rs6000_output_function_epilogue): Likewise.
15232 * config/rl78/rl78.c (rl78_start_function): Likewise.
15233 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15234 (rs6000_output_function_epilogue): Likewise.
15235 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15236 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15237 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15238 (sparc_asm_function_epilogue): Likewise.
15240 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15242 * tree.h (type_has_mode_precision_p): New function.
15243 * convert.c (convert_to_integer_1): Use it.
15244 * expr.c (expand_expr_real_2): Likewise.
15245 (expand_expr_real_1): Likewise.
15246 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15247 * match.pd: Likewise.
15248 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15249 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15250 * tree-tailcall.c (process_assignment): Likewise.
15251 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15252 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15253 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15254 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15255 (vectorizable_assignment): Likewise.
15256 (vectorizable_shift): Likewise.
15257 (vectorizable_operation): Likewise.
15258 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15260 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15262 * match.pd: Add pow (C, x) simplification.
15264 2017-08-21 Richard Biener <rguenther@suse.de>
15266 PR tree-optimization/81900
15267 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15268 for blocks with abnormal predecessors.
15269 (compute_antic): Do not set visited flag prematurely.
15271 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15274 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15276 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15278 * stor-layout.h (vector_type_mode): Move to...
15279 * tree.h (vector_type_mode): ...here.
15280 * stor-layout.c (vector_type_mode): Move to...
15281 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15283 2017-08-21 Richard Biener <rguenther@suse.de>
15285 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15286 register_external_die hooks.
15287 (debug_false_tree_charstarstar_uhwistar): Declare.
15288 (debug_nothing_tree_charstar_uhwi): Likewise.
15289 * debug.c (do_nothing_debug_hooks): Adjust.
15290 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15291 (debug_nothing_tree_charstar_uhwi): Likewise.
15292 * dbxout.c (dbx_debug_hooks): Adjust.
15293 (xcoff_debug_hooks): Likewise.
15294 * sdbout.c (sdb_debug_hooks): Likewise.
15295 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15296 * dwarf2out.c (macinfo_label_base): New global.
15297 (dwarf2out_register_external_die): New function for the
15298 register_external_die hook.
15299 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15300 (dwarf2_debug_hooks): Use them.
15301 (dwarf2_lineno_debug_hooks): Adjust.
15302 (struct die_struct): Add with_offset flag.
15303 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15304 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15305 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15306 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15307 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15308 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15309 defining section names for the early LTO debug variants.
15310 (reset_indirect_string): New helper.
15311 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15312 (print_dw_val): Add support for offsetted symbol references.
15313 (get_ultimate_context): Split out from is_cxx.
15314 (is_cxx): Use get_ultimate_context.
15315 (is_fortran): Add decl overload.
15316 (compute_comp_unit_symbol): Split out worker from
15317 compute_section_prefix.
15318 (compute_section_prefix): Call compute_comp_unit_symbol and
15319 set comdat_type_p here.
15320 (output_die): Skip DIE symbol output for the LTO added one.
15321 Handle DIE symbol references with offset.
15322 (output_comp_unit): Guard section name mangling properly.
15323 For LTO debug sections emit a symbol at the section beginning
15324 which we use to refer to its DIEs.
15325 (add_abstract_origin_attribute): For DIEs registered via
15326 dwarf2out_register_external_die directly refer to the early
15327 DIE rather than indirectly through the shadow one we created.
15328 Remove obsolete call to dwarf2out_abstract_function for
15329 non-function/block origins.
15330 (gen_array_type_die): When generating early LTO debug do
15331 not emit DW_AT_string_length.
15332 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15333 late when in LTO. As suggested place a gcc_unreachable for
15334 the DECL_ABSTRACT_P case.
15335 (gen_subprogram_die): Avoid another specification DIE
15336 for early built declarations/definitions for the late LTO case.
15337 (gen_variable_die): Add type references for late duplicated VLA dies
15339 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15340 we have the abstract instance already.
15341 (process_scope_var): Adjust decl DIE contexts in LTO which
15342 first puts them in limbo.
15343 (gen_decl_die): Do not generate type DIEs late apart from
15344 types for VLAs or for decls we do not yet have a DIE. Do not
15345 call dwarf2out_abstract_function late.
15346 (dwarf2out_early_global_decl): Make sure to create DIEs
15347 for abstract instances of a decl first.
15348 (dwarf2out_late_global_decl): Adjust comment.
15349 (output_macinfo_op): With multiple macro sections use
15350 macinfo_label_base to distinguish labels.
15351 (output_macinfo): Likewise. Update macinfo_label_base.
15352 Pass in the line info label.
15353 (note_variable_value_in_expr): When generating LTO resolve
15354 all variable values here by generating DIEs as needed.
15355 (init_sections_and_labels): Add early LTO debug flag parameter
15356 and generate different sections and names if set. Add generation
15357 counter for the labels so we can have multiple of them.
15358 (reset_dies): Helper to allow DIEs to be output multiple times.
15359 (dwarf2out_finish): When outputting DIEs to the fat part of an
15360 LTO object first reset DIEs.
15361 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15362 (modified_type_die): Check for decl_ultimate_origin being self
15364 (gen_type_die_with_usage): Likewise.
15365 (gen_typedef_die): Allow decl_ultimate_origin being self.
15366 (set_decl_abstract_flags): Remove.
15367 (set_block_abstract_flags): Likewise.
15368 (dwarf2out_abstract_function): Treat the early generated DIEs
15369 as the abstract copy and only add DW_AT_inline and
15370 DW_AT_artificial here and call set_decl_origin_self.
15371 If the DIE has an abstract origin don't do anything.
15372 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15373 if we have none yet (Go fails to build one, PR78628).
15374 (variably_modified_type_p): Prevent endless recursion for Ada
15375 cyclic pointer types.
15376 * lto-streamer-in.c: Include debug.h.
15377 (dref_queue): New global.
15378 (lto_read_tree_1): Stream in DIE references.
15379 (lto_input_tree): Register DIE references.
15380 (input_function): Stream DECL_DEBUG_ARGS.
15381 * lto-streamer-out.c: Include debug.h.
15382 (lto_write_tree_1): Output DIE references.
15383 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15384 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15385 (output_function): Stream DECL_DEBUG_ARGS.
15386 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15387 Stream DECL_ABSTRACT_ORIGIN.
15388 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15389 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15390 DECL_CONTEXT for file-scope decls.
15391 * lto-streamer.h (struct dref_entry): Declare.
15392 (dref_queue): Likewise.
15393 * cfgexpand.c (pass_expand::execute): Do not call the
15394 outlining_inline_function hook here.
15395 * lto-wrapper.c (debug_obj): New global.
15396 (tool_cleanup): Unlink it if required.
15397 (debug_objcopy): New function.
15398 (run_gcc): Handle early debug sections in the IL files by
15399 extracting them to separate files, partially linkin them and
15400 feeding the result back as result to the linker.
15401 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15402 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15403 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15404 sections into a separate segment.
15405 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15407 (darwin_asm_dwarf_section): Likewise.
15408 (darwin_asm_output_dwarf_offset): Likewise.
15409 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15411 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15412 Alan Hayward <alan.hayward@arm.com>
15413 David Sherwood <david.sherwood@arm.com>
15415 * read-md.h (md_reader::record_potential_iterator_use): Replace
15416 pointer argument with an rtx and an index.
15417 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15418 (apply_mode_iterator): Likewise.
15419 (apply_code_iterator): Likewise.
15420 (apply_int_iterator): Likewise.
15421 (apply_subst_iterator): Likewise.
15422 (record_iterator_use): Likewise.
15423 (record_attribute_use): Likewise.
15424 (md_reader::record_potential_iterator_use): Likewise. Update calls
15425 to record_iterator_use and apply_iterator.
15426 (iterator_use): Replace ptr with x and index.
15427 (attribute_use): Likewise.
15428 (apply_attribute_uses): Update calls to apply_iterator.
15429 (apply_iterators): Likewise. Update initialization of iterator_use.
15430 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15431 and record_potential_iterator_use.
15432 (rtx_reader::read_rtx_operand): Likewise.
15434 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15435 Alan Hayward <alan.hayward@arm.com>
15436 David Sherwood <david.sherwood@arm.com>
15438 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15441 2017-08-21 Richard Biener <rguenther@suse.de>
15443 PR middle-end/81884
15444 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15445 at struct end conservatively when comparing common bases.
15447 2017-08-21 Richard Biener <rguenther@suse.de>
15449 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15450 (mem_ref_in_stmt): Remove.
15451 (determine_max_movement): Use ref index to get at the reference.
15452 (invariantness_dom_walker::before_dom_children): Deal with
15453 lim data already initialized.
15454 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15456 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15458 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15459 (TARGET_ISA_ROUND): Ditto.
15460 (TARGET_ROUND): Ditto.
15461 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15462 * config/i386/i386.md: Ditto.
15463 * config/i386/sse.md: Ditto.
15464 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15465 with OPTION_MASK_ISA_SSE4_1.
15467 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15470 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15471 __builtin_ia32_lzcnt_u16.
15473 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15476 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15477 (rs6000_set_current_function): Rewrite function to use it.
15479 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15482 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15483 and TYPE_WARN_IF_NOT_ALIGN.
15484 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15485 (handle_warn_if_not_align): New.
15486 (place_union_field): Call handle_warn_if_not_align.
15487 (place_field): Call handle_warn_if_not_align.
15488 Copy TYPE_WARN_IF_NOT_ALIGN.
15489 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15490 (layout_type): Likewise.
15491 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15493 (tree_decl_common): Add warn_if_not_align.
15494 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15495 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15496 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15497 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15498 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15499 * doc/extend.texi: Document warn_if_not_aligned attribute.
15500 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15502 2017-08-17 Martin Liska <mliska@suse.cz>
15505 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15506 (get_data_dependence): Use it as pointer type.
15507 (distribute_loop): Likewise.
15509 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15511 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15512 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15513 (p8_vmrgow_<mode>_direct): New define_insn.
15514 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15515 handle endianness for vmrgew and vmrgow permute patterns.
15517 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15519 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15520 * config/rs6000/rs6000-cpus.def: Remove comment.
15521 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15522 (POWERPC_MASKS): Likewise.
15523 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15524 use of TARGET_VSX_TIMODE.
15525 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15526 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15527 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15528 (rs6000_option_override_internal): Remove dead code.
15529 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15530 (rs6000_legitimize_reload_address): Likewise.
15531 (rs6000_legitimate_address_p): Likewise.
15532 (rs6000_opt_masks): Delete "vsx-timode".
15533 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15534 from function comment.
15535 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15536 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15537 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15539 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15540 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15541 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15542 (VSX_TI): Likewise.
15544 (define_peephole2): Likewise.
15546 2017-08-17 Martin Sebor <msebor@redhat.com>
15549 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15550 past the end of an array.
15551 (test_pp_format): Add test cases.
15553 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15555 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15556 missing ECF_NOTHROW flags.
15558 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15561 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15562 operands residing in integer registers.
15563 (*vsx_le_perm_load_<mode>): Likewise.
15564 (*vsx_le_perm_store_<mode>): Likewise.
15565 (define_peephole2): Add peepholes to optimize the above.
15567 2017-08-17 Marek Polacek <polacek@redhat.com>
15569 PR middle-end/81814
15570 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15571 to mimic what shorten_compare did. Change the return type to bool.
15572 (fold_cond_expr_with_comparison): Update call to
15573 operand_equal_for_comparison_p.
15574 (fold_ternary_loc): Likewise.
15576 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15578 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15580 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15582 2017-08-17 Richard Biener <rguenther@suse.de>
15584 * tree-ssa-structalias.c (solve_graph): When propagating
15585 to successors update the graphs succ edges and avoid duplicate work.
15587 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15590 * config/i386/i386.c (ix86_option_override_internal): Save target
15591 specific options after ix86_stack_protector_guard_reg was changed.
15593 2017-08-17 Richard Biener <rguenther@suse.de>
15595 PR tree-optimization/81827
15596 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15597 (new_var_info): Initialize it conservatively.
15598 (get_call_vi): Mark register vars.
15599 (new_scalar_tmp_constraint_exp): Likewise.
15600 (handle_rhs_call): Likewise.
15601 (handle_const_call): Likewise.
15602 (create_function_info_for): Likewise.
15603 (solve_constraints): Sort varinfos to separate register from
15604 non-register vars to pack points-to solution bitmaps during
15607 2017-08-17 Marek Polacek <polacek@redhat.com>
15609 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15611 2017-08-17 Richard Biener <rguenther@suse.de>
15613 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15614 to true when overflow is undefined and we saturated the result.
15616 2017-08-17 Alan Modra <amodra@gmail.com>
15619 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15620 Don't use store multiple if only one reg needs saving.
15621 (interesting_frame_related_regno): New function.
15622 (rs6000_frame_related): Don't emit frame info for regs that
15624 (rs6000_emit_epilogue): Likewise.
15626 2017-08-16 Nathan Sidwell <nathan@acm.org>
15628 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15629 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15630 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15631 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15632 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15633 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15634 (verify_type): Adjust for TYPE_BINFO move.
15635 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15636 process TYPE_BINFO directly.
15637 (hash_tree): Likewise.
15638 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15640 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15643 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15645 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15647 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15650 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15651 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15652 Add ix86_binary_operator_ok to insn constraint.
15653 (*iordi_1_bts): Ditto.
15654 (*xordi_1_btc): Ditto.
15655 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15656 Update corresponding peephole2 pattern.
15660 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15662 PR tree-optimization/81832
15663 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15664 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15666 2017-08-16 Marek Polacek <polacek@redhat.com>
15669 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15670 regarding size_zero_node.
15672 2017-08-16 Martin Liska <mliska@suse.cz>
15675 * config.gcc: Respect previously set extra_objs in case
15678 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15680 PR tree-optimization/81835
15681 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15682 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15683 not depend on the phi.
15685 2017-08-16 Alan Modra <amodra@gmail.com>
15687 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15690 2017-08-16 Alan Modra <amodra@gmail.com>
15692 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15693 (save_reg_p): ..into this. Update all callers.
15694 (first_reg_to_save): Simplify.
15696 2017-08-16 Alan Modra <amodra@gmail.com>
15698 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15701 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15705 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15706 constant count if that count is less than 32.
15708 2017-08-15 Nathan Sidwell <nathan@acm.org>
15710 * gcc.c (execute): Emit friendlier message if inferior is killed
15711 by an external cause.
15713 2017-08-15 Richard Biener <rguenther@suse.de>
15715 PR tree-optimization/81790
15716 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15717 CONSTRUCTORs from simplifying and VN.
15719 2017-08-14 Martin Sebor <msebor@redhat.com>
15721 * builtin-attrs.def: Add comments.
15723 2017-08-14 Martin Sebor <msebor@redhat.com>
15726 * doc/extend.texi (attribute nonstring): Document new attribute.
15728 2017-08-14 Martin Sebor <msebor@redhat.com>
15731 * tree-diagnostic.c (default_tree_printer): Handle %G.
15732 * gimple-pretty-print.h (percent_G_format): Declare new function.
15733 * gimple-pretty-print.c (percent_G_format): Define.
15734 * tree-pretty-print.c (percent_K_format): Add argument.
15736 2017-08-14 Martin Sebor <msebor@redhat.com>
15738 PR translation/79998
15739 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15740 Remove a stray space.
15742 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15745 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15746 (*iordi_1_bts): Ditto.
15747 (*xordi_1_btc): Ditto.
15749 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15752 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15753 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15755 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15756 quoted strings, and make more translator-friendly.
15757 (darwin_rs6000_override_options): Likewise.
15758 (rs6000_option_override_internal): Likewise.
15759 (rs6000_return_in_memory): Fix overlong line.
15760 (init_cmulative_args): Use quoted strings, and make more
15761 translator-friendly.
15762 (rs6000_pass_by_reference): Fix overlong line.
15763 (def_builtin): Use quoted strings.
15764 (altivec_expand_predicate_builtin): Use quoted strings, and make
15765 more translator-friendly.
15766 (htm_expand_builtin): Use quoted strings.
15767 (cpu_expand_builtin): Use quoted strings, and make more
15768 translator-friendly.
15769 (altivec_expand_builtin): Likewise.
15770 (paired_expand_predicate_builtin): Likewise.
15771 (rs6000_invalid_builtin): Likewise.
15772 (builtin_function_type): Use quoted strings.
15773 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15774 more translator-friendly.
15775 (rs6000_trampoline_init): Likewise.
15776 (rs6000_handle_altivec_attribute): Likewise.
15777 (rs6000_inner_target_options): Use quoted strings.
15778 (rs6000_disable_incompatible_switches): Likewise.
15779 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15780 strings, and make more translator-friendly.
15781 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15783 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15785 PR tree-optimization/81799
15786 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15787 cond_expr to simple gimple operand.
15789 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15791 PR middle-end/46932
15792 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15794 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
15798 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
15799 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
15800 TARGET_GASISR_PROLOGUES.
15801 * config/avr/avr.c (avr_option_override): Same.
15802 (avr_pass_pre_proep::execute): Same.
15804 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
15807 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
15808 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
15810 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
15812 * config/i386/i386.md (*load_tp_<mode>): Redefine as
15813 define_insn_and_split. Split to a memory load from 0 in
15814 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
15815 using PTR mode iterator.
15816 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
15817 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
15818 (*add_tp_<mode>): Redefine as define_insn_and_split.
15819 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
15820 address space. Merge with *add_tp_x32 using PTR mode iterator.
15821 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
15822 Split to an add with a memory load from 0 in
15823 DEFAULT_TLS_SEG_REG address space.
15825 2017-08-12 Andrew Pinski <apinski@cavium.com>
15827 * config/aarch64/aarch64-option-extensions.def (rdma):
15828 Fix feature string to what Linux prints out in /proc/cpuinfo.
15830 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15833 * dwarf2out.c (modified_type_die): For C typedef types that have
15834 an ultimate origin, process the ultimate origin instead of the
15836 (gen_typedef_die): Assert that input DECLs have no ultimate
15838 (gen_type_die_with_usage): For typedef variants that have an
15839 ultimate origin, just call gen_decl_die on the original DECL.
15840 (process_scope_var): Avoid creating DIEs for local typedefs and
15841 concrete static variables.
15843 2017-08-12 Alan Modra <amodra@gmail.com>
15847 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15848 match gnu-user.h startfile.
15849 (ENDFILE_LINUX_SPEC): Similarly.
15851 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15854 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15856 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15858 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15859 * config/aarch64/aarch64.md (mov<mode>): Change.
15860 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15861 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15862 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15864 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15866 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15867 for storage order barriers.
15869 2017-08-11 Martin Liska <mliska@suse.cz>
15872 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15873 variables of void type.
15875 2017-08-11 Martin Liska <mliska@suse.cz>
15877 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15878 TARGET_SUPPORTS_ALIASES.
15879 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15880 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15881 (optimize_weakref): Likewise.
15882 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15883 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15884 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15886 2017-08-11 Martin Liska <mliska@suse.cz>
15889 * config/i386/i386.c (make_resolver_func): Do complete
15890 refactoring of the function.
15892 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15895 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15896 * config/i386/i386.c (ix86_stack_protect_guard): Use
15897 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15898 * doc/invoke.texi (x86 Options): Document
15899 -mstack-protector-guard-symbol= option.
15901 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15903 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15904 * config/i386/i386.c (ix86_split_stack_guard): New function.
15905 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15906 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15907 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15908 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15909 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15910 (split_stack_space_check): Call ix86_split_stack_guard.
15912 2017-08-10 Martin Sebor <msebor@redhat.com>
15914 * print-tree.c (print_node): Print location using the established
15916 Replace spaces with colons.
15917 (debug_raw, debug): Ditto.
15919 2017-08-10 Martin Sebor <msebor@redhat.com>
15922 * pretty-print.c (pp_format): Correct the handling of %s precision.
15924 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15927 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15929 (ix86_finalize_stack_frame_flags): This. Also clear
15930 frame_pointer_needed if -fno-omit-frame-pointer is used without
15932 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15933 with ix86_finalize_stack_frame_flags.
15934 (ix86_expand_epilogue): Likewise.
15935 (ix86_expand_split_stack_prologue): Likewise.
15936 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15938 2017-08-10 Martin Liska <mliska@suse.cz>
15941 * c-attribs.c (handle_target_attribute):
15942 Report warning for an empty string argument of target attribute.
15944 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15947 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15949 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15950 or DECL_NONLOCAL labels.
15951 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15952 or DECL_NONLOCAL labels here.
15954 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15956 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15957 to indicate when early gimple folding has been disabled.
15958 (rs6000_gimple_fold_builtin): Add debug content.
15959 (rs6000_invalid_builtin): Fix whitespace.
15960 (rs6000_expand_builtin): Fix whitespace.
15961 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15963 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15966 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15967 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15969 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15971 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15972 (qdf24xx): Likewise.
15973 * config/aarch64/aarch64.md: Include falkor.md.
15974 * config/aarch64/falkor.md: New.
15976 2017-08-09 Marek Polacek <polacek@redhat.com>
15979 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15980 * diagnostic.c (emit_diagnostic): Add a comment.
15981 (emit_diagnostic_valist): New function.
15983 2017-08-09 Marek Polacek <polacek@redhat.com>
15986 * input.c (make_location): New overload.
15987 * input.h (make_location): Declare.
15989 2017-08-08 Alan Modra <amodra@gmail.com>
15990 H.J. Lu <hongjiu.lu@intel.com>
15993 * gcc.c (NO_PIE_SPEC): Delete.
15994 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15996 (LINK_PIE_SPEC): ..to here.
15997 (LINK_COMMAND_SPEC): Support -no-pie.
15998 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15999 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16001 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16002 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16003 (ENDFILE_CRTEND_SPEC): Similarly.
16005 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16008 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16009 (mstack-protector-guard-offset=): Ditto.
16010 * config/i386/i386.c (ix86_option_override): Handle
16011 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16013 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16014 ix86_stack_protect_guard_offset variables.
16015 (TARGET_STACK_PROTECT_GUARD): Always define.
16016 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16017 and -mstack-protector-guard-offset= options.
16019 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16021 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16022 boundary case for the last candidate.
16024 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16026 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16027 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16029 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16031 PR middle-end/19706
16032 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16033 * config/aarch64/aarch64-builtins.c
16034 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16035 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16036 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16038 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16039 Andrew Pinski <pinskia@gmail.com>
16041 PR middle-end/19706
16042 * internal-fn.def (XORSIGN): New.
16043 * optabs.def (xorsign_optab): New.
16044 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16045 (convert_expand_mult_copysign): New.
16046 (pass_optimize_widening_mul::execute): Call
16047 convert_expand_mult_copysign.
16049 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16051 PR tree-optimization/81354
16052 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16053 Insert on edges rather than explicitly creating landing pads.
16054 (analyze_candidates_and_replace): Commit edge inserts.
16056 2017-08-08 Richard Biener <rguenther@suse.de>
16058 PR middle-end/81719
16059 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16060 (expand_simple_operations): Also look through ADDR_EXPRs with
16061 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16063 2017-08-08 Richard Biener <rguenther@suse.de>
16065 PR tree-optimization/81723
16066 * tree-vect-slp.c (struct bst_traits): New hash traits.
16067 (bst_fail): New global.
16068 (vect_build_slp_tree_2): New worker, split out from ...
16069 (vect_build_slp_tree): ... this now wrapping it with using
16070 bst_fail set to cache SLP tree build fails. Properly handle
16072 (vect_analyze_slp_instance): Allocate and free bst_fail.
16074 2017-08-08 Martin Liska <mliska@suse.cz>
16077 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16078 LABEL_DECLs that can be from a different function.
16080 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16082 PR tree-optimization/81744
16083 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16084 loop's number of iterations.
16086 2017-08-08 Martin Liska <mliska@suse.cz>
16088 * asan.c: Include header files.
16089 * attribs.c (build_decl_attribute_variant): New function moved
16091 (build_type_attribute_qual_variant): Likewise.
16092 (cmp_attrib_identifiers): Likewise.
16093 (simple_cst_list_equal): Likewise.
16094 (omp_declare_simd_clauses_equal): Likewise.
16095 (attribute_value_equal): Likewise.
16096 (comp_type_attributes): Likewise.
16097 (build_type_attribute_variant): Likewise.
16098 (lookup_ident_attribute): Likewise.
16099 (remove_attribute): Likewise.
16100 (merge_attributes): Likewise.
16101 (merge_type_attributes): Likewise.
16102 (merge_decl_attributes): Likewise.
16103 (merge_dllimport_decl_attributes): Likewise.
16104 (handle_dll_attribute): Likewise.
16105 (attribute_list_equal): Likewise.
16106 (attribute_list_contained): Likewise.
16107 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16108 (lookup_attribute_by_prefix): Likewise.
16109 * bb-reorder.c: Include header files.
16110 * builtins.c: Likewise.
16111 * calls.c: Likewise.
16112 * cfgexpand.c: Likewise.
16113 * cgraph.c: Likewise.
16114 * cgraphunit.c: Likewise.
16115 * convert.c: Likewise.
16116 * dwarf2out.c: Likewise.
16117 * final.c: Likewise.
16118 * fold-const.c: Likewise.
16119 * function.c: Likewise.
16120 * gimple-expr.c: Likewise.
16121 * gimple-fold.c: Likewise.
16122 * gimple-pretty-print.c: Likewise.
16123 * gimple.c: Likewise.
16124 * gimplify.c: Likewise.
16125 * hsa-common.c: Likewise.
16126 * hsa-gen.c: Likewise.
16127 * internal-fn.c: Likewise.
16128 * ipa-chkp.c: Likewise.
16129 * ipa-cp.c: Likewise.
16130 * ipa-devirt.c: Likewise.
16131 * ipa-fnsummary.c: Likewise.
16132 * ipa-inline.c: Likewise.
16133 * ipa-visibility.c: Likewise.
16135 * lto-cgraph.c: Likewise.
16136 * omp-expand.c: Likewise.
16137 * omp-general.c: Likewise.
16138 * omp-low.c: Likewise.
16139 * omp-offload.c: Likewise.
16140 * omp-simd-clone.c: Likewise.
16141 * opts-global.c: Likewise.
16142 * passes.c: Likewise.
16143 * predict.c: Likewise.
16144 * sancov.c: Likewise.
16145 * sanopt.c: Likewise.
16146 * symtab.c: Likewise.
16147 * toplev.c: Likewise.
16148 * trans-mem.c: Likewise.
16149 * tree-chkp.c: Likewise.
16150 * tree-eh.c: Likewise.
16151 * tree-into-ssa.c: Likewise.
16152 * tree-object-size.c: Likewise.
16153 * tree-parloops.c: Likewise.
16154 * tree-profile.c: Likewise.
16155 * tree-ssa-ccp.c: Likewise.
16156 * tree-ssa-live.c: Likewise.
16157 * tree-ssa-loop.c: Likewise.
16158 * tree-ssa-sccvn.c: Likewise.
16159 * tree-ssa-structalias.c: Likewise.
16160 * tree-ssa.c: Likewise.
16161 * tree-streamer-in.c: Likewise.
16162 * tree-vectorizer.c: Likewise.
16163 * tree-vrp.c: Likewise.
16164 * tsan.c: Likewise.
16165 * ubsan.c: Likewise.
16166 * varasm.c: Likewise.
16167 * varpool.c: Likewise.
16168 * tree.c: Remove functions moved to attribs.[ch].
16169 * tree.h: Likewise.
16170 * config/aarch64/aarch64.c: Add attrs.h header file.
16171 * config/alpha/alpha.c: Likewise.
16172 * config/arc/arc.c: Likewise.
16173 * config/arm/arm.c: Likewise.
16174 * config/avr/avr.c: Likewise.
16175 * config/bfin/bfin.c: Likewise.
16176 * config/c6x/c6x.c: Likewise.
16177 * config/cr16/cr16.c: Likewise.
16178 * config/cris/cris.c: Likewise.
16179 * config/darwin.c: Likewise.
16180 * config/epiphany/epiphany.c: Likewise.
16181 * config/fr30/fr30.c: Likewise.
16182 * config/frv/frv.c: Likewise.
16183 * config/ft32/ft32.c: Likewise.
16184 * config/h8300/h8300.c: Likewise.
16185 * config/i386/winnt.c: Likewise.
16186 * config/ia64/ia64.c: Likewise.
16187 * config/iq2000/iq2000.c: Likewise.
16188 * config/lm32/lm32.c: Likewise.
16189 * config/m32c/m32c.c: Likewise.
16190 * config/m32r/m32r.c: Likewise.
16191 * config/m68k/m68k.c: Likewise.
16192 * config/mcore/mcore.c: Likewise.
16193 * config/microblaze/microblaze.c: Likewise.
16194 * config/mips/mips.c: Likewise.
16195 * config/mmix/mmix.c: Likewise.
16196 * config/mn10300/mn10300.c: Likewise.
16197 * config/moxie/moxie.c: Likewise.
16198 * config/msp430/msp430.c: Likewise.
16199 * config/nds32/nds32-isr.c: Likewise.
16200 * config/nds32/nds32.c: Likewise.
16201 * config/nios2/nios2.c: Likewise.
16202 * config/nvptx/nvptx.c: Likewise.
16203 * config/pa/pa.c: Likewise.
16204 * config/pdp11/pdp11.c: Likewise.
16205 * config/powerpcspe/powerpcspe.c: Likewise.
16206 * config/riscv/riscv.c: Likewise.
16207 * config/rl78/rl78.c: Likewise.
16208 * config/rx/rx.c: Likewise.
16209 * config/s390/s390.c: Likewise.
16210 * config/sh/sh.c: Likewise.
16211 * config/sol2.c: Likewise.
16212 * config/sparc/sparc.c: Likewise.
16213 * config/spu/spu.c: Likewise.
16214 * config/stormy16/stormy16.c: Likewise.
16215 * config/tilegx/tilegx.c: Likewise.
16216 * config/tilepro/tilepro.c: Likewise.
16217 * config/v850/v850.c: Likewise.
16218 * config/vax/vax.c: Likewise.
16219 * config/visium/visium.c: Likewise.
16220 * config/xtensa/xtensa.c: Likewise.
16222 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16225 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16226 constraints since the -mupper-regs-* switches have been
16228 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16229 into a vector from a double word element that was extracted from
16230 another vector, and eliminate extra XXPERMDI instructions.
16231 (vsx_concat_<mode>_2): Likewise.
16232 (vsx_concat_<mode>_3): Likewise.
16233 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16234 concat to allow optimizing inserts from previous extracts.
16236 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16238 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16239 memory reference to a SSP offset in TLS address space.
16240 (ix86_print_operand) <case '@'>: Remove.
16241 (ix86_print_operand_punct_valid_p): Remove '@' code.
16242 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16243 UNSPEC_SP_TLS_TEST.
16244 (stack_tls_protect_set_<mode>): Remove.
16245 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16246 (stack_tls_protect_test_<mode>): Remove.
16247 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16249 2017-08-07 Olivier Hainque <hainque@adacore.com>
16252 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16254 2017-08-07 Douglas Rupp <rupp@adacore.com>
16256 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16257 variable was referenced as multidir in command.
16259 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16262 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16265 2017-08-07 Martin Liska <mliska@suse.cz>
16267 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16268 * config/rl78/rl78.c: Add include of attribs.h.
16269 * config/sh/sh.c: Likewise.
16270 * config/v850/v850.c: Likewise.
16272 2017-08-07 Tom de Vries <tom@codesourcery.com>
16274 PR middle-end/78266
16275 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16277 2017-08-07 Martin Liska <mliska@suse.cz>
16279 * config/mips/mips.c: Include attribs.h.
16281 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16284 * doc/invoke.texi: Document change in behvaior for -Ofast for
16287 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16289 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16291 (aarch64_pop_regs): Likewise.
16292 (aarch64_gen_load_pair): Likewise.
16293 (aarch64_save_callee_saves): Likewise.
16294 (aarch64_restore_callee_saves): Likewise.
16296 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16298 * config/i386/i386.c: Revert the last change.
16300 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16303 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16305 (ix86_finalize_stack_frame_flags): This. Also clear
16306 frame_pointer_needed if -fno-omit-frame-pointer is used without
16308 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16309 with ix86_finalize_stack_frame_flags.
16310 (ix86_expand_epilogue): Likewise.
16311 (ix86_expand_split_stack_prologue): Likewise.
16313 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16316 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16317 to P_AES for Westmere.
16319 2017-08-07 Jonathan Yong <10walls@gmail.com>
16321 * config/i386/mingw.opt (fset-stack-executable): Removed.
16322 * config/i386/cygming.opt (fset-stack-executable): Moved
16324 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16326 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16328 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16330 2017-08-07 Marek Polacek <polacek@redhat.com>
16332 PR middle-end/81737
16333 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16335 2017-08-07 Martin Liska <mliska@suse.cz>
16337 * attribs.h (canonicalize_attr_name): New function.
16338 (cmp_attribs): Move from c-format.c and adjusted.
16339 (is_attribute_p): Moved from tree.h.
16340 * tree-inline.c: Add new includes.
16341 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16342 (private_is_attribute_p): Remove.
16343 (private_lookup_attribute): Likewise.
16344 (private_lookup_attribute_by_prefix): Simplify.
16345 (remove_attribute): Use is_attribute_p.
16346 * tree.h: Remove removed declarations.
16348 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16350 PR middle-end/81698
16351 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16352 instead of computing it in the function. Formatting fix.
16353 (expand_case): Don't rely on default_edge being the first edge,
16354 clear it if removing it, pass default_edge to
16355 emit_case_dispatch_table.
16356 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16359 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16361 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16362 insn in the function, emit NOP after the insn.
16364 2017-08-06 Tom de Vries <tom@codesourcery.com>
16366 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16369 2017-08-06 Tom de Vries <tom@codesourcery.com>
16371 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16374 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16376 PR tree-optimization/57371
16377 * match.pd: New pattern.
16379 2017-08-04 Marek Polacek <polacek@redhat.com>
16381 PR middle-end/81695
16382 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16383 perform the computation in offset_int.
16385 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16387 PR tree-optimization/81136
16388 * tree-vectorizer.h: Include tree-hash-traits.h.
16389 (vec_base_alignments): New typedef.
16390 (vec_info): Add a base_alignments field.
16391 (vect_record_base_alignments): Declare.
16392 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16394 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16395 (create_data_ref): Add an is_conditional_in_stmt argument.
16396 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16397 the is_conditional_in_stmt field.
16398 (data_ref_loc): Add an is_conditional_in_stmt field.
16399 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16400 (find_data_references_in_stmt): Update call to create_data_ref.
16401 (graphite_find_data_references_in_stmt): Likewise.
16402 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16403 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16404 (vect_record_base_alignment): New function.
16405 (vect_record_base_alignments): Likewise.
16406 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16407 for nested statements even if we fail to compute a misalignment.
16408 Use pooled base alignments for unconditional references.
16409 (vect_find_same_alignment_drs): Compare base addresses instead
16411 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16412 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16414 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16416 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16417 Add an explicit name for the enum. Use auto_vec for slp_instances
16418 and grouped_stores.
16419 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16421 (_bb_vec_info): Add a constructor and destructor.
16422 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16423 (destroy_loop_vec_info): Delete.
16424 (vect_destroy_datarefs): Likewise.
16425 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16426 (vec_info::vec_info): New function.
16427 (vec_info::~vec_info): Likewise.
16428 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16429 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16430 destroy_loop_vec_info.
16431 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16432 (_loop_vec_info::_loop_vec_info): ...this.
16433 (destroy_loop_vec_info): Replace with...
16434 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16435 the stmt_vec_infos. Leave handling of vec_info information to its
16436 destructor. Remove explicit vector releases.
16437 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16438 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16439 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16440 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16441 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16442 (destroy_bb_vec_info): Replace with...
16443 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16444 information to its destructor.
16445 (vect_slp_analyze_bb_1): Use new and delete instead of
16446 new_bb_vec_info and destroy_bb_vec_info.
16447 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16450 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16452 * tree-data-ref.h (subscript): Add access_fn field.
16453 (data_dependence_relation): Add could_be_independent_p.
16454 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16455 (same_access_functions): Move to tree-data-ref.c.
16456 * tree-data-ref.c (ref_contains_union_access_p): New function.
16457 (access_fn_component_p): Likewise.
16458 (access_fn_components_comparable_p): Likewise.
16459 (dr_analyze_indices): Add a reference to access_fn_component_p.
16460 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16462 (constant_access_functions): Likewise.
16463 (add_other_self_distances): Likewise.
16464 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16465 (initialize_data_dependence_relation): Use XCNEW and remove
16466 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16467 of access functions that have the same type. Allow the
16468 subsequence to end with different bases in some circumstances.
16469 Record the chosen access functions in SUB_ACCESS_FN.
16470 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16471 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16472 (subscript_dependence_tester_1): Likewise dra and drb.
16473 (build_classic_dist_vector): Update calls accordingly.
16474 (subscript_dependence_tester): Likewise.
16475 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16476 DDR_COULD_BE_INDEPENDENT_P.
16477 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16478 comp_alias_ddrs instead of may_alias_ddrs.
16479 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16481 (vect_analyze_data_ref_dependence): Use it if
16482 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16483 distance vectors if that fails.
16484 (dependence_distance_ge_vf): New function.
16485 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16486 LOOP_VINFO_MAY_ALIAS_DDRS.
16488 2017-08-04 Richard Biener <rguenther@suse.de>
16490 PR middle-end/81705
16491 * fold-const.c (fold_binary_loc): Properly restrict
16492 minus_var0 && minus_var1 case when associating undefined overflow
16495 2017-08-04 Tom de Vries <tom@codesourcery.com>
16497 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16499 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16501 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16502 Don't start diagnostic messages with a capital letter.
16503 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16505 (rs6000_invalid_builtin): Likewise.
16506 (rs6000_trampoline_init): Likewise.
16508 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16511 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16512 after setting changeable df flags.
16514 2017-08-03 Richard Biener <rguenther@suse.de>
16516 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16517 up if the use is in USE - X.
16519 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16521 * toplev.c (dumpfile.h): New include.
16522 (internal_error_reentered): New static function. Use it...
16523 (internal_error_function): ...here to handle reentered internal_error.
16525 2017-08-03 Richard Biener <rguenther@suse.de>
16527 PR middle-end/81148
16528 * fold-const.c (split_tree): Add minus_var and minus_con
16529 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16530 here but always use minus_*.
16531 (associate_trees): Assert we never associate with MINUS_EXPR
16532 and NULL first operand. Do not recurse for PLUS_EXPR operands
16533 when associating as MINUS_EXPR either.
16534 (fold_binary_loc): Track minus_var and minus_con.
16536 2017-08-03 Tom de Vries <tom@codesourcery.com>
16539 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16540 ACCEL_COMPILER, apply finish_options on
16541 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16543 2017-08-03 Tom de Vries <tom@codesourcery.com>
16546 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16547 function_entry_patch_area_size > 0.
16549 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16552 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16553 instead of 10??LU, perform unit multiplication in wide_int,
16554 don't change alloc_object_size_limit if the limit is larger
16557 PR tree-optimization/81655
16558 PR tree-optimization/81588
16559 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16560 the case when ranges[i].low and high are 1 for unsigned type with
16563 PR middle-end/81052
16564 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16565 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16567 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16569 * tree-vrp.h: Add include guard.
16571 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16574 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16575 (ud2): New insn pattern.
16576 * config/i386/i386.c (ix86_expand_epilogue):
16577 For naked functions, generate ud2 instead of trap insn.
16579 2017-08-02 Marek Polacek <polacek@redhat.com>
16582 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16584 2017-08-02 Tom de Vries <tom@codesourcery.com>
16585 Cesar Philippidis <cesar@codesourcery.com>
16587 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16588 Add missing edge probabilities.
16590 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16592 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16593 Correct endianness.
16595 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16597 PR middle-end/79499
16598 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16599 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16600 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16601 prologue_seq sequences - if any.
16603 2017-08-02 Richard Biener <rguenther@suse.de>
16605 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16606 via vectors if supported, integer extracts via punning if supported
16607 or otherwise vector extracts.
16609 2017-08-02 Richard Biener <rguenther@suse.de>
16611 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16613 (bitmap_insert_into_set): ... this.
16615 2017-08-02 Richard Biener <rguenther@suse.de>
16617 PR tree-optimization/81633
16619 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16621 PR tree-optimization/71752
16622 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16624 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16626 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16627 (machine_function::call_ms2sysv_pad_out): Remove field.
16628 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16629 (ix86_compute_frame_layout): Likewise.
16631 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16634 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16635 attribute with interrupt attribute.
16637 2017-08-01 Andrew Pinski <apinski@cavium.com>
16639 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16640 BIT_INSERT_EXPR's operand 1
16641 to see if the types precision matches.
16643 2017-08-01 Martin Liska <mliska@suse.cz>
16645 PR middle-end/70140
16646 * builtins.c (expand_builtin_memcpy_args): Remove.
16647 (expand_builtin_memcpy): Call newly added function
16648 expand_builtin_memory_copy_args.
16649 (expand_builtin_memcpy_with_bounds): Likewise.
16650 (expand_builtin_mempcpy): Remove last argument.
16651 (expand_builtin_mempcpy_with_bounds): Likewise.
16652 (expand_builtin_memory_copy_args): New function created from
16653 expand_builtin_mempcpy_args with small modifications.
16654 (expand_builtin_mempcpy_args): Remove.
16655 (expand_builtin_stpcpy): Remove unused argument.
16656 (expand_builtin): Likewise.
16657 (expand_builtin_with_bounds): Likewise.
16659 2017-08-01 Martin Liska <mliska@suse.cz>
16662 Make mempcpy more optimal (PR middle-end/70140).
16664 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16668 __builtin_vec_cmpne verify both arguments are compatible vectors
16669 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16670 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16671 move computation of aligned to after checking the argument types.
16675 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16676 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16679 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16680 Doug Rupp <rupp@adacore.com>
16681 Olivier Hainque <hainque@adacore.com>
16683 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16684 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16686 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16687 for TARGET_OS_CPP_BUILTIN.
16688 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16689 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16691 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16692 passing required abi options to the assembler for EABI configurations.
16693 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16694 of .text.hot and .text.unlikely sections for kernel modules when
16695 using ARM style exceptions.
16696 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16697 options. Add EXTRA_CC1_SPEC.
16698 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16700 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16702 (ARM_TARGET2_DWARF_FORMAT): Define.
16703 * config/arm/t-vxworks: Adjust multilib control to removal of the
16704 Diab command line options.
16706 2017-08-01 Martin Liska <mliska@suse.cz>
16708 PR gcov-profile/81561
16709 * gcov.c (unblock): Make unblocking safe as we need to preserve
16710 index correspondence of blocks and block_lists.
16712 2017-08-01 Richard Biener <rguenther@suse.de>
16714 PR tree-optimization/81181
16715 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16716 (compute_antic): ... end of iteration here.
16718 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16720 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16721 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16722 (ftree-slp-vectorize): Likewise.
16723 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16724 can no longer be set independent of flag_tree_loop_vectorize.
16725 * omp-general.c (emp_max_vf): Likewise.
16726 * opts.c (enable_fdo_optimizations): Remove references to
16727 flag_tree_vectorize, these are now implicit.
16728 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16729 and leave it for the options machinery.
16731 2017-08-01 Martin Liska <mliska@suse.cz>
16733 PR middle-end/70140
16734 * builtins.c (expand_builtin_memcpy_args): Remove.
16735 (expand_builtin_memcpy): Call newly added function
16736 expand_builtin_memory_copy_args.
16737 (expand_builtin_memcpy_with_bounds): Likewise.
16738 (expand_builtin_mempcpy): Remove last argument.
16739 (expand_builtin_mempcpy_with_bounds): Likewise.
16740 (expand_builtin_memory_copy_args): New function created from
16741 expand_builtin_mempcpy_args with small modifications.
16742 (expand_builtin_mempcpy_args): Remove.
16743 (expand_builtin_stpcpy): Remove unused argument.
16744 (expand_builtin): Likewise.
16745 (expand_builtin_with_bounds): Likewise.
16747 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16750 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16751 print "ds:" only for immediates in generic address space.
16753 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16756 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16757 (ix86_function_ok_for_sibcall): Return false for naked functions.
16759 2017-08-01 Richard Biener <rguenther@suse.de>
16761 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16762 (compute_antic): Seed worklist with exit block predecessors.
16763 * cfganal.c (dfs_find_deadend): For a cycle return the source
16764 of the edge closing it.
16766 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16768 * config/aarch64/aarch64.c
16769 (aarch64_can_const_movi_rtx_p): Move 0 check.
16771 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16773 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16774 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16776 * match.pd: Ditto in address comparison pattern.
16778 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16780 PR tree-optimization/81627
16781 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16782 closed ssa form for store-store chain.
16784 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16786 PR tree-optimization/81620
16787 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16788 for store-store chain.
16790 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16792 PR tree-optimization/81588
16793 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16794 ranges[i].in_p, invert comparison code ccode. For >/>=,
16795 swap rhs1 and rhs2 and comparison code unconditionally,
16796 for </<= don't do that. Don't swap rhs1/rhs2 again if
16797 ranges[i].in_p, instead invert comparison code ccode if
16798 opcode or oe->rank is BIT_IOR_EXPR.
16801 * optabs.def (vec_extract_optab, vec_init_optab): Change from
16802 a direct optab to conversion optab.
16803 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
16804 with GET_MODE_INNER as last argument instead of optab_handler.
16805 * expmed.c (extract_bit_field_1): Likewise. Use vector from
16806 vector extraction if possible and optab is available.
16807 * expr.c (store_constructor): Use convert_optab_handler instead
16808 of optab_handler. Use vector initialization from smaller
16809 vectors if possible and optab is available.
16810 * tree-vect-stmts.c (vectorizable_load): Likewise.
16811 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16812 now have two modes.
16813 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
16814 of vec_init from half-sized vectors with the same element mode.
16815 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
16816 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
16817 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
16818 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
16819 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
16820 after mode in gen_vec_extract* calls.
16821 (vec_extract<mode>): Renamed to ...
16822 (vec_extract<mode><ssescalarmodelower>): ... this.
16823 (vec_extract<mode><ssehalfvecmodelower>): New expander.
16824 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
16825 element mode after mode in gen_vec_init* calls.
16826 (VEC_INIT_HALF_MODE): New mode iterator.
16827 (vec_init<mode>): Renamed to ...
16828 (vec_init<mode><ssescalarmodelower>): ... this.
16829 (vec_init<mode><ssehalfvecmodelower>): New expander.
16830 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
16831 (vec_extractv2sfsf): ... this.
16832 (vec_initv2sf): Renamed to ...
16833 (vec_initv2sfsf): ... this.
16834 (vec_extractv2si): Renamed to ...
16835 (vec_extractv2sisi): ... this.
16836 (vec_initv2si): Renamed to ...
16837 (vec_initv2sisi): ... this.
16838 (vec_extractv4hi): Renamed to ...
16839 (vec_extractv4hihi): ... this.
16840 (vec_initv4hi): Renamed to ...
16841 (vec_initv4hihi): ... this.
16842 (vec_extractv8qi): Renamed to ...
16843 (vec_extractv8qiqi): ... this.
16844 (vec_initv8qi): Renamed to ...
16845 (vec_initv8qiqi): ... this.
16846 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16847 (vec_init<mode>): Renamed to ...
16848 (vec_init<mode><VEC_base_l>): ... this.
16849 (vec_extract<mode>): Renamed to ...
16850 (vec_extract<mode><VEC_base_l>): ... this.
16851 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16852 (vec_initv2sfsf): ... this.
16853 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16854 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16855 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16856 element mode after mode in gen_vec_init* calls.
16857 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16858 (vec_init<mode><Vel>): ... this.
16859 (vec_extract<mode>): Renamed to ...
16860 (vec_extract<mode><Vel>): ... this.
16861 * config/aarch64/iterators.md (Vel): New mode attribute.
16862 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16863 Add element mode after mode in gen_vec_extract* calls.
16864 * config/s390/vector.md (non_vec_l): New mode attribute.
16865 (vec_extract<mode>): Renamed to ...
16866 (vec_extract<mode><non_vec_l>): ... this.
16867 (vec_init<mode>): Renamed to ...
16868 (vec_init<mode><non_vec_l>): ... this.
16869 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16870 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16872 * config/arm/iterators.md (V_elem_l): New mode attribute.
16873 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16874 (vec_extract<mode><V_elem_l>): ... this.
16875 (vec_extractv2di): Renamed to ...
16876 (vec_extractv2didi): ... this.
16877 (vec_init<mode>): Renamed to ...
16878 (vec_init<mode><V_elem_l>): ... this.
16879 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16880 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16881 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16882 Add element mode after gen_vec_extract* calls.
16883 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16884 (vec_init<mode><unitmode>): ... this.
16885 (vec_extract<mode>): Renamed to ...
16886 (vec_extract<mode><unitmode>): ... this.
16887 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16888 (vec_init<mode><unitmode>): ... this.
16889 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16890 (vec_initv2sfsf): ... this.
16891 (vec_extractv2sf): Renamed to ...
16892 (vec_extractv2sfsf): ... this.
16893 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16894 Add element mode after gen_vec_extract* calls.
16895 * config/mips/mips.md (unitmode): New mode iterator.
16896 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16897 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16898 * config/spu/spu.md (inner_l): New mode attribute.
16899 (vec_init<mode>): Renamed to ...
16900 (vec_init<mode><inner_l>): ... this.
16901 (vec_extract<mode>): Renamed to ...
16902 (vec_extract<mode><inner_l>): ... this.
16903 * config/sparc/sparc.md (veltmode): New mode iterator.
16904 (vec_init<VMALL:mode>): Renamed to ...
16905 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16906 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16907 (vec_initv2sisi): ... this.
16908 (vec_initv2sf): Renamed to ...
16909 (vec_initv2sfsf): ... this.
16910 (vec_extractv2sf): Renamed to ...
16911 (vec_extractv2sfsf): ... this.
16912 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16913 (vec_init<mode>): Renamed to ...
16914 (vec_init<mode><VEC_base_l>): ... this.
16915 (vec_extract<mode>): Renamed to ...
16916 (vec_extract<mode><VEC_base_l>): ... this.
16917 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16918 (vec_initv2sfsf): ... this.
16919 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16920 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16921 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16922 gen_vec_init* calls.
16924 2017-08-01 Richard Biener <rguenther@suse.de>
16926 PR tree-optimization/81297
16927 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16928 TREE_OVERFLOW from INTEGER_CSTs.
16930 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16932 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16934 2017-07-31 Carl Love <cel@us.ibm.com>
16936 * config/rs6000/rs6000-c: Add support for built-in functions
16937 vector signed char vec_xl_be (signed long long, signed char *);
16938 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16939 vector signed int vec_xl_be (signed long long, signed int *);
16940 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16941 vector signed long long vec_xl_be (signed long long, signed long long *);
16942 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16943 vector signed short vec_xl_be (signed long long, signed short *);
16944 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16945 vector double vec_xl_be (signed long long, double *);
16946 vector float vec_xl_be (signed long long, float *);
16947 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16948 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16949 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16951 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16952 (altivec_expand_builtin): Add switch statement to call
16953 altivec_expand_xl_be for each builtin.
16954 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16955 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16956 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16957 __builtin_vsx_le_be_v16qi.
16958 * doc/extend.texi: Update the built-in documentation file for the
16959 new built-in functions.
16961 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16964 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16966 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16968 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16970 * config.gcc: Add z14.
16971 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16972 CPU model numbers for z13s and z14.
16973 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16975 * config/s390/s390-opts.h (enum processor_type): Rename
16976 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16977 * config/s390/s390.c (processor_table): Add field for CPU name to
16978 be passed to Binutils.
16979 (s390_asm_output_machine_for_arch): Use the new field in
16980 processor_table for Binutils.
16981 (s390_expand_builtin): Replace arch12 with z14.
16982 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16983 (s390_get_sched_attrmask): Likewise.
16984 (s390_get_unit_mask): Likewise.
16985 * config/s390/s390.opt: Add z14 to processor_type enum.
16987 2017-07-31 Martin Jambor <mjambor@suse.cz>
16990 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16991 regardless of optimization level.
16993 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16994 Martin Liska <mliska@suse.cz>
16996 * predict.def: Remove old comment and adjust probability.
16997 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16998 PREDICT statements.
17000 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17003 * config/i386/i386.c (ix86_function_naked): New function.
17004 (ix86_can_use_return_insn_p): Return false for naked functions.
17005 (ix86_expand_prologue): Skip prologue for naked functions.
17006 (ix86_expand_epilogue): Skip epilogue for naked functions
17007 and emit trap instruction.
17008 (ix86_warn_func_return): New function.
17009 (ix86_attribute_table): Add "naked" attribute specification.
17010 (TARGET_WARN_FUNC_RETURN): Define.
17011 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17013 2017-07-31 Martin Liska <mliska@suse.cz>
17015 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17016 (dump_gimple_bb_header): Always dump BB info.
17017 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17019 2017-07-31 Martin Liska <mliska@suse.cz>
17022 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17023 also with current_function_decl non-null equality.
17025 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17028 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17029 change type to the element type, instead add eltype variable and
17030 use it where we are interested in the element type.
17032 PR tree-optimization/81603
17033 * ipa-polymorphic-call.c
17034 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17035 offset arithmetic in offset_int, bail out if the resulting bit offset
17036 doesn't fit into shwi.
17038 2017-07-31 Martin Liska <mliska@suse.cz>
17040 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17041 (gimplify_save_expr): Fix comment.
17043 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17046 * config/i386/i386.c (ix86_function_arg): Update arguments for
17048 (ix86_compute_frame_layout): Set the initial stack offset to
17049 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17050 INCOMING_FRAME_SP_OFFSET.
17051 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17052 stack before exception handler returns.
17053 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17054 the 'ERROR_CODE' for exception handler.
17056 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17058 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17059 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17060 (ASM_OUTPUT_REG_POP): Ditto.
17061 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17062 instead of asm_fprintf to output pure string.
17064 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17066 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17067 to imported_module_or_decl hook.
17068 (debug_nothing_tree_tree_tree_bool): Remove.
17069 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17070 * debug.c (do_nothing_debug_hooks): Use
17071 debug_nothing_tree_tree_tree_bool_bool instead of
17072 debug_nothing_tree_tree_tree_bool.
17073 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17074 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17075 * sdbout.c (sdb_debug_hooks): Likewise.
17076 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17077 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17079 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17080 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17081 attribute, don't add anything.
17083 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17085 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17086 (fold_build2_stat_loc): Likewise.
17087 (fold_build3_stat_loc): Likewise.
17088 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17089 (fold_build1_loc): Remove macro.
17090 (fold_build2_loc): Likewise.
17091 (fold_build3_loc): Likewise.
17093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17095 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17096 (gimple_build_debug_bind_source_stat): Likewise.
17097 * gimple.h (gimple_build_debug_bind): Remove macro.
17098 (gimple_build_debug_bind_source): Likewise.
17100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17102 * bitmap.c (bitmap_alloc): Adjust.
17103 (bitmap_gc_alloc): Likewise.
17104 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17106 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17108 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17109 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17110 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17111 (bitmap_gc_alloc_stat): Likewise.
17112 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17114 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17116 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17117 * rtl.h (shallow_copy_rtx): Remove macro.
17119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17121 * emit-rtl.c (gen_raw_REG): Adjust.
17122 * gengenrtl.c (gendef): Likewise.
17123 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17124 * rtl.h (rtx_alloc): Remove macro.
17126 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17128 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17129 (build_tree_list_stat): Likewise.
17130 * tree.h (build_tree_list): Remove macro.
17131 (build_tree_list_vec): Likewise.
17133 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17135 * tree.c (make_vector_stat): Remove _stat from name.
17136 (build_vector_stat): Likewise.
17137 * tree.h (make_vector_stat): Remove macro.
17138 (build_vector_stat): Likewise.
17140 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17142 * tree.h (build_var_debug_value): Remove prototype.
17144 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17146 * tree.c (tree_cons_stat): Remove _stat from name.
17147 * tree.h (tree_cons): Remove macro.
17149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17151 * tree.c (build_vl_exp_stat): Remove _stat from name.
17152 * tree.h (build_vl_exp): Remove macro.
17154 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17156 * tree.c (build_decl_stat): Remove _stat from name.
17157 * tree.h (build_decl): Remove macro.
17159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17161 * gimple.c (gimple_build_with_ops_stat): Adjust.
17162 (gimple_alloc_stat): Remove _stat from name.
17163 * gimple.h (gimple_alloc): Remove macro.
17165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17167 * tree.c (make_tree_vec_stat): Remove _stat from name.
17168 (grow_tree_vec_stat): Likewise.
17169 * tree.h (make_tree_vec_stat): Adjust prototype.
17170 (grow_tree_vec_stat): Likewise.
17171 (make_tree_vec): Remove macro.
17172 (grow_tree_vec): Likewise.
17174 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17176 * fold-const.c (fold_build1_stat_loc): Adjust.
17177 (fold_build2_stat_loc): Likewise.
17178 (fold_build3_stat_loc): Likewise.
17179 * tree.c (build0_stat): Remove _stat from name.
17180 (build1_stat): Likewise.
17181 (build2_stat): Likewise.
17182 (build3_stat): Likewise.
17183 (build4_stat): Likewise.
17184 (build5_stat): Likewise.
17185 * tree.h (build1_loc): Remove macro, and rename _stat function
17187 (build2_loc): Likewise.
17188 (build3_loc): Likewise.
17189 (build4_loc): Likewise.
17190 (build5_loc): Likewise.
17192 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17194 * tree.c (make_int_cst_stat): Remove _stat from name.
17195 * tree.h (make_int_cst_stat): Adjust prototype.
17196 (make_int_cst): Remove macro.
17198 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17200 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17201 * tree.h (make_tree_binfo_stat): Adjust prototype.
17202 (make_tree_binfo): Remove.
17204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17206 * tree.c (copy_node_stat): Rename to copy_node.
17207 (build_distinct_type_copy): Adjust.
17208 * tree.h (copy_node_stat): Adjust prototype.
17209 (copy_node): Remove macro.
17211 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17213 * tree.c (make_node_stat): rename to make_node.
17214 (build_tree_list_stat): Adjust.
17215 (build0_stat): Likewise.
17216 (build2_stat): Likewise.
17217 (build3_stat): Likewise.
17218 (build4_stat): Likewise.
17219 (build5_stat): Likewise.
17220 (build_decl_stat): Likewise.
17221 * tree.h (make_node_stat): Adjust prototype.
17222 (make_node): remove macro.
17224 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17226 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17227 (PPC_FEATURE2_SCV): Likewise.
17228 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17230 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17232 * config/aarch64/aarch64.c
17233 (aarch64_internal_mov_immediate): Add new special pattern.
17234 * config/aarch64/aarch64.md (*movdi_aarch64):
17235 Add reg/32bit const mov case.
17237 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17238 Richard Sandiford <richard.sandiford@linaro.org>
17240 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17241 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17242 Add integer and movi cases.
17243 (movi-split-hf-df-sf split, fp16): New.
17244 (enabled): Added TARGET_FP_F16INST.
17245 * config/aarch64/iterators.md (GPF_HF): New.
17246 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17248 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17250 * config/aarch64/aarch64.c
17251 (aarch64_simd_container_mode): Add prototype.
17252 (aarch64_expand_mov_immediate): Add HI support.
17253 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17254 (aarch64_can_const_movi_rtx_p): New.
17255 (aarch64_preferred_reload_class):
17256 Remove restrictions of using FP registers for certain SIMD operations.
17257 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17258 (aarch64_valid_floating_const): Add integer move validation.
17259 (aarch64_simd_imm_scalar_p): Remove.
17260 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17261 (aarch64_legitimate_constant_p): Expand list of supported cases.
17262 * config/aarch64/aarch64-protos.h
17263 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17264 (aarch64_reinterpret_float_as_int): New.
17265 (aarch64_simd_imm_scalar_p): Remove.
17266 * config/aarch64/constraints.md (Uvi): New.
17267 (Dd): Split into Ds and new Dd.
17268 * config/aarch64/aarch64.md (*movsi_aarch64):
17270 (*movdi_aarch64): Add SIMD mov case.
17272 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17274 * tree-predcom.c: (struct chain): Handle store-store chain in which
17275 stores for elimination only store loop invariant values.
17276 (execute_pred_commoning_chain): Ditto.
17277 (prepare_initializers_chain_store_elim): Ditto.
17278 (prepare_finalizers): Ditto.
17279 (is_inv_store_elimination_chain): New function.
17280 (initialize_root_vars_store_elim_1): New function.
17282 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17284 * tree-predcom.c: Revise general description of the pass.
17285 (enum chain_type): New enum type for store elimination.
17286 (struct chain): New field supporting store elimination.
17287 (struct component): Ditto.
17288 (dump_chain): Dump store-stores chain.
17289 (release_chain): Release resources.
17290 (split_data_refs_to_components): Compute and create component
17291 contains only stores for elimination.
17292 (get_chain_last_ref_at): New function.
17293 (make_invariant_chain): Initialization.
17294 (make_rooted_chain): Specify chain type in parameter and record it.
17295 (add_looparound_copies): Skip for store-stores chain.
17296 (determine_roots_comp): Compute type of chain and pass it to
17298 (initialize_root_vars_store_elim_2): New function.
17299 (finalize_eliminated_stores): New function.
17300 (remove_stmt): Handle store for elimination.
17301 (execute_pred_commoning_chain): Execute predictive commoning on
17302 store-store chains.
17303 (determine_unroll_factor): Skip unroll for store-stores chain.
17304 (prepare_initializers_chain_store_elim): New function.
17305 (prepare_initializers_chain): Hanlde store-store chain.
17306 (prepare_finalizers_chain, prepare_finalizers): New function.
17307 (tree_predictive_commoning_loop): Return integer value indicating
17308 if loop is unrolled or lcssa form is corrupted.
17309 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17311 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17313 * tree-predcom.c (initialize_root): Delete.
17314 (execute_pred_commoning_chain): Initialize root vars and replace
17315 reference of non-combined chain directly, rather than call above
17318 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17320 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17321 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17323 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17325 * tree-predcom.c (struct chain): New field init_seq.
17326 (release_chain): Release init_seq.
17327 (prepare_initializers_chain): Record intialization stmts in above
17329 (insert_init_seqs): New function.
17330 (tree_predictive_commoning_loop): Call insert_init_seqs.
17332 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17334 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17336 2017-07-28 Richard Biener <rguenther@suse.de>
17338 * match.pd: Remove superfluous :c.
17339 * genmatch.c (simplify::id): Add member.
17340 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17342 (current_id): New global.
17343 (dt_node::parent): Move from ...
17344 (dt_operand::parent): ... here. Add for_id member.
17345 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17346 (decision_tree::find_node): Relax order requirement when
17347 merging DT_TRUE nodes to ones inbetween the current simplify
17348 and the one we try to merge with. Add diagnostic whenever
17349 we need to enforce pattern order by not merging.
17350 (decision_tree::insert): Set current_id.
17351 (decision_tree::print_node): Dump parent node and for_id.
17352 (parser::last_id): Add member.
17353 (parser::push_simplify): Assign unique id.
17354 (parser::parser): Initialize last_id.
17356 2017-07-28 Martin Liska <mliska@suse.cz>
17359 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17360 gimple_build_debug_bind.
17362 2017-07-28 Richard Biener <rguenther@suse.de>
17364 PR tree-optimization/81502
17365 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17367 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17369 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17370 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17371 for BIT_FIELD_REF args.
17372 * fold-const.c (make_bit_field_ref): Likewise.
17373 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17375 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17378 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17379 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17380 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17381 Or it into SANITIZER_UNDEFINED.
17382 * ubsan.c: Include gimple-fold.h and varasm.h.
17383 (ubsan_expand_ptr_ifn): New function.
17384 (instrument_pointer_overflow): New function.
17385 (maybe_instrument_pointer_overflow): New function.
17386 (instrument_object_size): Formatting fix.
17387 (pass_ubsan::execute): Call instrument_pointer_overflow
17388 and maybe_instrument_pointer_overflow.
17389 * internal-fn.c (expand_UBSAN_PTR): New function.
17390 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17391 * sanitizer.def (__ubsan_handle_pointer_overflow,
17392 __ubsan_handle_pointer_overflow_abort): New builtins.
17393 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17394 * internal-fn.def (UBSAN_PTR): New internal function.
17395 * opts.c (sanitizer_opts): Add pointer-overflow.
17396 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17397 * fold-const.c (build_range_check): Compute pointer range check in
17398 integral type if pointer arithmetics would be needed. Formatting
17401 2017-07-28 Martin Liska <mliska@suse.cz>
17404 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17405 parameters that are of a variable-length.
17407 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17409 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17411 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17412 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17413 (CRT_CALL_STATIC_FUNCTION): Likewise.
17414 (ASM_DEFAULT_SPEC): New define.
17415 (ASM_SPEC32): Likewise.
17416 (ASM_SPEC64): Likewise.
17417 (ASM_SPEC_COMMON): Likewise.
17418 (ASM_SPEC): Likewise.
17419 (INVALID_64BIT): Likewise.
17420 (LINK_OS_DEFAULT_SPEC): Likewise.
17421 (LINK_OS_SPEC32): Likewise.
17422 (LINK_OS_SPEC64): Likewise.
17423 (POWERPC_LINUX): Likewise.
17424 (PTRDIFF_TYPE): Likewise.
17425 (RESTORE_FP_PREFIX): Likewise.
17426 (RESTORE_FP_SUFFIX): Likewise.
17427 (SAVE_FP_PREFIX): Likewise.
17428 (SAVE_FP_SUFFIX): Likewise.
17429 (SIZE_TYPE): Likewise.
17430 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17431 (TARGET_64BIT): Likewise.
17432 (TARGET_64BIT): Likewise.
17433 (TARGET_AIX): Likewise.
17434 (WCHAR_TYPE_SIZE): Likewise.
17435 (WCHAR_TYPE): Undefine.
17436 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17437 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17438 (CPP_OS_RTEMS_SPEC): Delete.
17439 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17440 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17442 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17444 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17446 PR tree-optimization/81578
17447 * tree-parloops.c (build_new_reduction): Bail out if
17448 reduction_code isn't one of the standard OpenMP reductions.
17449 Move the details printing after that decision.
17451 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17453 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17454 related to reload_in_progress.
17455 (splat_input_operand): Likewise.
17456 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17458 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17460 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17461 (TARGET_INSTANTIATE_DECLS): Likewise.
17462 (legitimate_indexed_address_p): Delete reload_in_progress code.
17463 (rs6000_debug_legitimate_address_p): Likewise.
17464 (rs6000_eliminate_indexed_memrefs): Likewise.
17465 (rs6000_emit_le_vsx_store): Likewise.
17466 (rs6000_emit_move_si_sf_subreg): Likewise.
17467 (rs6000_emit_move): Likewise.
17468 (register_to_reg_type): Likewise.
17469 (rs6000_pre_atomic_barrier): Likewise.
17470 (rs6000_machopic_legitimize_pic_address): Likewise.
17471 (rs6000_allocate_stack_temp): Likewise.
17472 (rs6000_address_for_fpconvert): Likewise.
17473 (rs6000_address_for_altivec): Likewise.
17474 (rs6000_secondary_memory_needed_rtx): Delete function.
17475 (rs6000_check_sdmode): Likewise.
17476 (rs6000_alloc_sdmode_stack_slot): Likewise.
17477 (rs6000_instantiate_decls): Likewise.
17478 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17479 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17480 Delete reload_in_progress.
17481 (*vec_reload_and_plus_<mptrsize>): Likewise.
17482 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17483 (vsx_div_v2di): Likewise.
17484 (vsx_udiv_v2di): Likewise.
17486 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17488 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17489 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17490 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17491 (rs6000_debug_reg_global): Delete print of LRA status.
17492 (rs6000_option_override_internal): Delete dead LRA related code.
17493 (rs6000_lra_p): Delete function.
17494 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17496 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17498 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17499 * config/riscv/rtems.h: New file.
17501 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17502 Sudakshina Das <sudi.das@arm.com>
17504 * config/aarch64/aarch64.md
17505 (define_split for and<mode>3nr_compare): Move
17506 non aarch64_logical_operand to a register.
17507 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17508 register immediate operand to a register.
17509 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17511 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17513 PR middle-end/81564
17514 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17516 2017-07-27 Richard Biener <rguenther@suse.de>
17518 PR tree-optimization/81573
17519 PR tree-optimization/81494
17520 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17521 multi defuse cycle case.
17523 2017-07-27 Richard Biener <rguenther@suse.de>
17525 PR tree-optimization/81571
17526 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17529 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17531 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17532 earlier and only if MASK_FPU is set. Adjust formatting.
17534 2017-07-27 Martin Liska <mliska@suse.cz>
17536 * opt-functions.awk: Add validation of value of Init.
17537 * optc-gen.awk: Pass new argument.
17539 2017-07-27 Martin Liska <mliska@suse.cz>
17541 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17542 Fix wrong condition.
17544 2017-07-27 Martin Liska <mliska@suse.cz>
17546 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17547 BBs and edges seen by autoFDO.
17549 2017-07-27 Richard Biener <rguenther@suse.de>
17551 PR tree-optimization/81502
17552 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17553 with incompatible but same sized type.
17554 (execute_update_addresses_taken): Likewise.
17556 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17558 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17559 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17561 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17564 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17565 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17566 Change s_operand to memory_operand.
17568 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17570 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17571 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17572 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17573 Emit instructions rather than returning an expression. Handle TFmode
17574 and KFmode by casting to TImode.
17575 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17576 (rs6000_emit_le_vsx_store): Likewise.
17577 * config/rs6000/vsx.md (VSX_TI): New iterator.
17578 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17579 (*vsx_le_undo_permute_<mode>): Likewise.
17580 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17581 emit the split sequence.
17582 (*vsx_le_perm_store_<mode>): Likewise.
17584 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17586 PR tree-optimization/81555
17587 PR tree-optimization/81556
17588 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17589 if true, force CHANGED for the recursive invocation.
17590 (reassociate_bb): Remember original length of ops array, pass
17591 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17593 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17594 attributes for noipa attribute. For naked attribute use
17595 lookup_attribute first before lookup_attribute_spec.
17596 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17598 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17599 for functions with noipa attribute.
17600 (cgraph_externally_visible_p): Return true for functions with noipa
17602 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17603 for functions with noipa attribute.
17604 * doc/extend.texi: Document noipa function attribute.
17605 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17606 also for functions with noipa attribute.
17607 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17609 2017-07-26 Andrew Pinski <apinski@cavium.com>
17611 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17612 vec_unalign_load_cost and vec_unalign_store_cost.
17614 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17616 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17617 -mvsx-small-integer option.
17618 (ISA_3_0_MASKS_IEEE): Likewise.
17619 (OTHER_VSX_VECTOR_MASKS): Likewise.
17620 (POWERPC_MASKS): Likewise.
17621 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17622 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17623 code, only testing for DImode being allowed in non-VSX floating
17625 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17626 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17628 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17629 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17630 TARGET_P8_VECTOR test.
17631 (rs6000_secondary_reload_simple_move): Likewise.
17632 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17633 since TARGET_P9_VECTOR was already tested.
17634 (rs6000_opt_masks): Remove -mvsx-small-integer.
17635 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17636 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17638 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17639 test for TARGET_VEXTRACTUB was used, and that uses
17641 (p9 extract splitter): Likewise.
17642 (vsx_extract_<mode>_di_p9): Likewise.
17643 (vsx_extract_<mode>_store_p9): Likewise.
17644 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17645 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17646 the elimination of TARGET_VSX_SMALL_INTEGER.
17647 (vsx_extract_<mode>_p8): Likewise.
17648 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17649 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17650 (vsx_set_<mode>_p9): Likewise.
17651 (vsx_set_v4sf_p9): Likewise.
17652 (vsx_set_v4sf_p9_zero): Likewise.
17653 (vsx_insert_extract_v4sf_p9): Likewise.
17654 (vsx_insert_extract_v4sf_p9_2): Likewise.
17655 * config/rs6000/rs6000.md (sign extend splitter): Change
17656 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17657 (floatsi<mode>2_lfiwax_mem): Likewise.
17658 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17659 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17660 since a test for TARGET_P9_VECTOR was used.
17661 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17662 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17663 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17664 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17665 TARGET_P8_VECTOR test.
17666 (fix_trunc<mode>si2_stfiwx): Likewise.
17667 (fix_trunc<mode>si2_internal): Likewise.
17668 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17669 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17671 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17672 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17673 TARGET_P8_VECTOR test.
17674 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17675 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17676 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17678 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17679 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17680 since a test for TARGET_P9_VECTOR was used.
17681 (splitter for loading small constants): Likewise.
17683 2017-07-26 Andrew Pinski <apinski@cavium.com>
17685 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17688 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17691 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17692 (fp_valid_at): Likewise.
17694 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17696 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17697 (qdf24xx_addrcost_table): Likewise.
17698 (cortexa57_tunings): Update to use generic_branch_cost.
17699 (cortexa72_tunings): Likewise.
17700 (cortexa73_tunings): Likewise.
17701 (qdf24xx_tunings): Likewise.
17703 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17705 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17706 (thunderx2t99_branch_cost): Likewise.
17707 (cortexa35_tunings): Update to use generic_branch_cost.
17708 (cortexa53_tunings): Likewise.
17709 (cortexa57_tunings): Likewise.
17710 (cortexa72_tunings): Likewise.
17711 (cortexa73_tunings): Likewise.
17712 (thunderx2t99_tunings): Likewise.
17714 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17716 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17717 (sparc_option_override): Honour MASK_FSMULD.
17718 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17719 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17720 * config/sparc/sparc.opt (mfsmuld): New option.
17721 * doc/invoke.texi (mfsmuld): Document option.
17723 2017-07-26 Marek Polacek <polacek@redhat.com>
17725 PR middle-end/70992
17726 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17728 2017-07-26 Richard Biener <rguenther@suse.de>
17730 * gimple-match-head.c (do_valueize): Return OP if valueize
17732 (get_def): New helper to get at the def stmt of a SSA name
17733 if valueize allows.
17734 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17735 do_valueize to get at the def stmt.
17736 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17738 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17740 PR middle-end/46932
17741 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17743 2017-07-26 Martin Liska <mliska@suse.cz>
17746 * function.c (expand_function_start): Make expansion of
17747 nonlocal_goto_save_area after parm_birth_insn.
17749 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17751 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17752 from all CPU target flags enable members.
17754 2017-07-26 Richard Biener <rguenther@suse.de>
17756 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17757 (decision_tree::gen): Make 'type' const.
17758 (write_predicate): Likewise.
17760 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17762 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17763 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17764 (rs6000_option_override_internal): Likewise.
17765 (rs6000_expand_vector_set): Likewise.
17766 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17767 (TARGET_UPPER_REGS_SF): Likewise.
17768 (TARGET_UPPER_REGS_DI): Likewise.
17769 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17770 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17771 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17772 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17773 (Splitters for DI constants in Altivec registers): Likewise.
17774 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17775 (vsx_set_v4sf_p9): Likewise.
17776 (vsx_set_v4sf_p9_zero): Likewise.
17777 (vsx_insert_extract_v4sf_p9): Likewise.
17778 (vsx_insert_extract_v4sf_p9_2): Likewise.
17780 2017-07-25 Carl Love <cel@us.ibm.com>
17782 * doc/extend.texi: Update the built-in documentation file for the
17783 existing built-in functions
17784 vector signed char vec_cnttz (vector signed char);
17785 vector unsigned char vec_cnttz (vector unsigned char);
17786 vector signed short vec_cnttz (vector signed short);
17787 vector unsigned short vec_cnttz (vector unsigned short);
17788 vector signed int vec_cnttz (vector signed int);
17789 vector unsigned int vec_cnttz (vector unsigned int);
17790 vector signed long long vec_cnttz (vector signed long long);
17791 vector unsigned long long vec_cnttz (vector unsigned long long);
17793 2017-07-25 Andrew Pinski <apinski@cavium.com>
17795 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17796 accesses where the use is for the first operand of a BIT_INSERT.
17798 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17801 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17802 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17804 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17806 * config/i386/gstabs.h: Delete.
17807 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17809 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
17811 * config/i386/i386.c (ix86_decompose_address): Do not check for
17812 register RTX when looking at index_reg or base_reg.
17813 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
17815 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
17817 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
17818 to update EH info here.
17820 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17822 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
17824 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17826 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
17828 2017-07-25 Torsten Duwe <duwe@suse.de>
17830 * common.opt: Introduce -fpatchable-function-entry
17831 command line option, and its variables function_entry_patch_area_size
17832 and function_entry_patch_area_start.
17833 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17834 including a two-value parser.
17835 * target.def (print_patchable_function_entry): New target hook.
17836 * targhooks.h (default_print_patchable_function_entry): New function.
17837 * targhooks.c (default_print_patchable_function_entry): Likewise.
17838 * toplev.c (process_options): Switch off IPA-RA if
17839 patchable function entries are being generated.
17840 * varasm.c (assemble_start_function): Look at the
17841 patchable-function-entry command line switch and current
17842 function attributes and maybe generate NOP instructions by
17843 calling the print_patchable_function_entry hook.
17844 * doc/extend.texi: Document patchable_function_entry attribute.
17845 * doc/invoke.texi: Document -fpatchable_function_entry
17846 command line option.
17847 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17849 * doc/tm.texi: Re-generate.
17851 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17854 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17855 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17857 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17859 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17861 2017-07-25 Richard Biener <rguenther@suse.de>
17863 PR tree-optimization/81455
17864 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17865 not walk in cycles when looking for guards.
17867 2017-07-25 Richard Biener <rguenther@suse.de>
17869 PR tree-optimization/81529
17870 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17871 when optimizing backedge uses.
17873 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17875 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17877 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17878 to dl_section_ref. On AIX, append an expression to subtract
17879 the size of the section length to dl_section_ref.
17881 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17883 * configure.ac: If any of the config.* scripts fail, exit 1.
17884 * configure: Regenerate.
17886 2017-07-25 Richard Biener <rguenther@suse.de>
17888 PR middle-end/81546
17889 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17890 of immediate uses, be more verbose on errors.
17892 2017-07-25 Richard Biener <rguenther@suse.de>
17894 PR tree-optimization/81510
17895 * tree-vect-loop.c (vect_is_simple_reduction): When the
17896 reduction stmt is not inside the loop bail out.
17898 2017-07-25 Richard Biener <rguenther@suse.de>
17900 PR tree-optimization/81303
17901 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17902 profitability check against LOOP_VINFO_NITERSM1.
17904 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17906 * domwalk.c (cmp_bb_postorder): Simplify.
17907 (sort_bbs_postorder): New function. Use it...
17908 (dom_walker::walk): ...here to optimize common cases.
17910 2017-07-25 Martin Liska <mliska@suse.cz>
17913 * ipa-visibility.c (function_and_variable_visibility): Make the
17914 redirection just on target that supports aliasing.
17915 Fix GNU coding style.
17917 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17920 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17922 * config/arm/rtems.h: Likewise.
17923 * config/bfin/rtems.h: Likewise.
17924 * config/i386/rtemself.h: Likewise.
17925 * config/lm32/rtems.h: Likewise.
17926 * config/m32c/rtems.h: Likewise.
17927 * config/m68k/rtemself.h: Likewise.
17928 * config/microblaze/rtems.h: Likewise.
17929 * config/mips/rtems.h: Likewise.
17930 * config/moxie/rtems.h: Likewise.
17931 * config/nios2/rtems.h: Likewise.
17932 * config/powerpcspe/rtems.h: Likewise.
17933 * config/rs6000/rtems.h: Likewise.
17934 * config/rtems.h: Likewise.
17935 * config/sh/rtems.h: Likewise.
17936 * config/sh/rtemself.h: Likewise.
17937 * config/sparc/rtemself.h: Likewise.
17939 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17942 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17943 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17944 * tree-ssa-structalias.c (alias_get_name): Same.
17946 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17949 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17950 instructions if no du chain is found.
17952 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17954 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17956 2017-07-25 Richard Biener <rguenther@suse.de>
17958 PR middle-end/81505
17959 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17962 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17964 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17965 upper-regs options.
17966 (ISA_2_7_MASKS_SERVER): Likewise.
17967 (ISA_3_0_MASKS_IEEE): Likewise.
17968 (OTHER_P8_VECTOR_MASKS): Likewise.
17969 (OTHER_VSX_VECTOR_MASKS): Likewise.
17970 (POWERPC_MASKS): Likewise.
17971 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17972 duplicate list of options.
17973 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17974 explicit -mupper-regs options.
17975 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17976 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17977 alias for -mupper-regs-df.
17978 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17979 (rs6000_init_hard_regno_mode_ok): Likewise.
17980 (rs6000_option_override_internal): Likewise.
17981 (rs6000_opt_masks): Likewise.
17982 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17983 options in terms of whether -mvsx or -mpower8-vector was used.
17984 (TARGET_UPPER_REGS_DI): Likewise.
17985 (TARGET_UPPER_REGS_SF): Likewise.
17986 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17987 -mupper-regs-* options.
17989 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17991 * passes.c (emergency_dump_function): Print some empty lines and a
17992 header before the RTL dump.
17994 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17996 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17998 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18001 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18002 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18004 2017-07-24 Carl Love <cel@us.ibm.com>
18006 * config/rs6000/rs6000-c.c: Add support for built-in functions
18007 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18008 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18009 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18010 vec_extract_fp_from_shortl): Add defines for the two builtins.
18011 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18012 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18014 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18015 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18016 * doc/extend.texi: Update the built-in documentation file for the
18017 new built-in function.
18019 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18022 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18024 * doc/generic.texi: Likewise.
18025 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18026 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18028 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18030 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18031 (aarch64_mls_elt_merge<mode>): Likewise.
18033 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18035 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18036 having __cxa_atexit.
18038 2017-07-23 Michael Collison <michael.collison@arm.com>
18040 * config/arm/arm.c (arm_option_override): Deprecate
18041 use of -mstructure-size-boundary.
18042 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18043 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18045 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18048 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18049 Reduce cost estimate for direct moves.
18051 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18054 * config/i386/i386.c (ix86_option_override_internal): Disable
18055 BMI, BMI2 and TBM instructions for -m16.
18057 2017-07-21 Carl Love <cel@us.ibm.com>
18059 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18060 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18061 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18063 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18064 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18065 VMULOSW): New enum "unspec" values.
18066 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18067 altivec_vmulosw): New patterns.
18068 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18069 VMULOSW): Add definitions.
18071 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18073 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18074 (qdf24xx): Likewise.
18075 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18076 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18077 (AARCH64_FL_V8_1): Renumber.
18078 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18079 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18080 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18081 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18082 rdma to feature modifiers list.
18084 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18086 PR middle-end/56727
18087 * ipa-visibility (function_and_variable_visibility): Convert
18088 recursive PLT call to direct call if appropriate.
18090 2017-07-21 Andrew Pinski <apinski@cavium.com>
18092 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18093 operand 1 to see if the types precision matches.
18094 * fold-const.c (operand_equal_p): Likewise.
18096 2017-07-21 Richard Biener <rguenther@suse.de>
18098 PR tree-optimization/81303
18099 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18100 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18101 (vect_peeling_hash_get_lowest_cost): Adjust.
18102 (vect_enhance_data_refs_alignment): Likewise. Use
18103 vect_get_peeling_costs_all_drs to compute the penalty for no
18104 peeling to match up costs.
18106 2017-07-21 Richard Biener <rguenther@suse.de>
18108 PR tree-optimization/81500
18109 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18110 we didn't identify a reduction path.
18112 2017-07-21 Tom de Vries <tom@codesourcery.com>
18113 Cesar Philippidis <cesar@codesourcery.com>
18115 PR gcov-profile/81442
18116 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18119 2017-07-21 Tom de Vries <tom@codesourcery.com>
18122 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18124 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18125 nvptx_override_options_after_change.
18127 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18129 * dwarf2out.c (output_file_names): Avoid double testing for
18130 dwarf_version >= 5.
18132 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18134 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18136 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18138 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18140 (try_crossjump_to_edge): Do not punt on partitioned functions.
18142 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18144 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18145 Put all BBs reachable only via paths crossing cold region to cold
18147 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18149 2016-07-21 Richard Biener <rguenther@suse.de>
18151 PR tree-optimization/81303
18152 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18153 into account prologue and epilogue iterations when raising
18154 min_profitable_iters to sth at least covering one vector iteration.
18156 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18158 * config/arm/arm.c (arm_test_cpu_arch_dat):
18161 2017-07-20 Nathan Sidwell <nathan@acm.org>
18163 Remove TYPE_METHODS.
18164 * tree.h (TYPE_METHODS): Delete.
18165 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18166 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18167 (dbxout_type_methods): Scan TYPE_FIELDS.
18168 (dbxout_type): Don't check TYPE_METHODS here.
18169 * function.c (use_register_for_decl): Always ignore register for
18170 class types when not optimizing.
18171 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18172 * tree.c (free_lang_data_in_type): Stitch out member functions and
18173 templates from TYPE_FIELDS.
18174 (build_distinct_type_copy, verify_type_variant,
18175 verify_type): Member fns are on TYPE_FIELDS.
18176 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18177 * tree-pretty-print.c (dump_generic_node): Likewise.
18179 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18182 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18183 V2TImode and V4TImode.
18184 (ix86_expand_vector_extract): Likewise.
18185 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18186 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18187 (ssescalarmode): Handle V4TImode and V2TImode.
18188 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18189 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18190 (VEXTRACTI128_MODE): New mode iterator.
18191 (splitter for *vec_extractv?ti first element): New.
18192 (VEC_INIT_MODE): New mode iterator.
18193 (vec_init<mode>): Consolidate 3 expanders into one using
18194 VEC_INIT_MODE mode iterator.
18196 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18198 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18199 non_spilled_static_chain_regno_p.
18201 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18203 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18205 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18207 * bb-reorder.c (connect_traces): Allow copying of blocks within
18210 2017-07-20 Richard Biener <rguenther@suse.de>
18212 * gimple.h (gimple_phi_result): Add gphi * overload.
18213 (gimple_phi_result_ptr): Likewise.
18214 (gimple_phi_arg): Likewise. Adjust index assert to only
18215 allow actual argument accesses rather than all slots available
18217 (gimple_phi_arg_def): Add gphi * overload.
18218 * tree-phinodes.c (make_phi_node): Initialize only actual
18220 (resize_phi_node): Clear memory not covered by old node,
18221 do not initialize excess argument slots.
18222 (reserve_phi_args_for_new_edge): Initialize new argument slot
18225 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18227 PR tree-optimization/81388
18229 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18231 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18232 by removing computation of may_be_zero.
18234 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18235 Tom de Vries <tom@codesourcery.com>
18237 PR middle-end/81030
18238 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18239 when gimple level profile disagrees with what RTL expander did.
18241 2017-07-20 Richard Biener <rguenther@suse.de>
18243 PR tree-optimization/61171
18244 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18245 (vect_analyze_stmt): Add slp instance parameter.
18246 (vectorizable_reduction): Likewise.
18247 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18248 (vect_is_simple_reduction): Deal with chains not detected
18249 as SLP reduction chain, specifically not properly associated
18250 chains containing a mix of plus/minus.
18251 (get_reduction_op): Remove.
18252 (get_initial_defs_for_reduction): Simplify, pass in whether
18253 this is a reduction chain, pass in the SLP node for the PHIs.
18254 (vect_create_epilog_for_reduction): Get the SLP instance as
18256 (vectorizable_reduction): Get the SLP instance as arg.
18257 During analysis remember the SLP node with the PHIs in the
18258 instance. Simplify getting at the vectorized reduction PHIs.
18259 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18260 through SLP instance.
18261 (vect_slp_analyze_operations): Likewise.
18262 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18263 (vect_transform_stmt): Likewise.
18265 2017-07-20 Tom de Vries <tom@codesourcery.com>
18267 PR tree-optimization/81489
18268 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18269 read of phi arg location to before loop that modifies phi.
18271 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18273 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18276 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18278 PR middle-end/81331
18279 * except.c (execute): Fix ordering issue.
18281 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18283 PR rtl-optimization/81423
18284 * combine.c (make_compound_operation_int): Don't try to optimize
18285 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18287 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18289 PR rtl-optimization/81423
18290 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18291 with a constant that is -1 in the truncated to mode.
18293 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18295 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18296 (determine_unlikely_bbs): ... here.
18297 * predict.h (propagate_unlikely_bbs_forward): Declare.
18298 * cfgexpand.c (pass_expand::execute): Use it.
18299 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18301 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18302 propagate_unlikely_bbs_forward.
18304 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18306 PR middle-end/81331
18307 * except.c (maybe_add_nop_after_section_switch): New function.
18310 2017-07-19 Tom de Vries <tom@codesourcery.com>
18312 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18314 2017-07-19 Tom de Vries <tom@codesourcery.com>
18316 * gimple.h (gimple_phi_arg): Make assert more strict.
18318 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18320 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18321 * config/rs6000/mmintrin.h: New file.
18322 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18324 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18326 PR tree-optimization/81346
18327 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18329 2017-07-19 Tom de Vries <tom@codesourcery.com>
18331 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18333 2017-07-19 Tom de Vries <tom@codesourcery.com>
18335 * config/nvptx/nvptx-modes.def: Add V2DImode.
18336 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18337 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18338 (nvptx_output_mov_insn): Handle lack of mov.b128.
18339 (nvptx_print_operand): Handle 'H' and 'L' codes.
18340 (nvptx_vector_mode_supported): Allow V2DImode.
18341 (nvptx_preferred_simd_mode): New function.
18342 (nvptx_data_alignment): New function.
18343 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18344 nvptx_preferred_simd_mode.
18345 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18347 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18349 2017-07-19 Tom de Vries <tom@codesourcery.com>
18351 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18352 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18353 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18354 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18355 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18356 (mov<VECIM>_insn): New define_insn.
18357 (define_expand "mov<VECIM>): New define_expand.
18359 2017-07-19 Tom de Vries <tom@codesourcery.com>
18361 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18363 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18365 PR tree-optimization/81346
18366 * fold-const.h (fold_div_compare, range_check_type): Declare.
18367 * fold-const.c (range_check_type): New function.
18368 (build_range_check): Use range_check_type.
18369 (fold_div_compare): No longer static, rewritten into
18370 a match.pd helper function.
18371 (fold_comparison): Don't call fold_div_compare here.
18372 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18373 as helper function.
18375 2017-07-19 Nathan Sidwell <nathan@acm.org>
18377 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18378 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18379 * tree.c (find_decls_types_r, verify_type): Use
18380 TYPE_{MIN,MAX}_VALUE_RAW.
18381 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18382 (hash_tree): Likewise.
18383 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18385 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18388 2017-07-18 Tom de Vries <tom@codesourcery.com>
18390 PR middle-end/81464
18391 * omp-expand.c (expand_omp_for_static_chunk): Handle
18392 equal-argument loop exit phi.
18394 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18397 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18398 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18399 operand 2 predicate.
18400 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18401 operand 2 predicate.
18402 (ror,rol -> rorx splitters): Use const_int_operand as
18403 operand 2 predicate.
18405 2017-06-18 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/81410
18408 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18409 the gap in the ! slp_perm SLP case after each group.
18411 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18413 PR middle-end/81463
18414 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18417 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18419 PR middle-end/81462
18420 * predict.c (set_even_probabilities): Cleanup; do not affect
18421 probabilities that are already known.
18422 (combine_predictions_for_bb): Call even when count is set.
18424 2017-07-18 Nathan Sidwell <nathan@acm.org>
18426 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18429 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18432 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18433 optimization for loop niter analysis.
18435 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18438 * config/avr/avr.c (avr_optimize_casesi): Don't use
18439 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18441 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18443 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18444 body_cost_vec from _vect_peel_extended_info.
18445 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18446 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18449 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18451 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18453 2017-07-18 Richard Biener <rguenther@suse.de>
18455 PR tree-optimization/80620
18456 PR tree-optimization/81403
18457 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18458 info when re-using a VN table entry.
18460 2017-07-18 Richard Biener <rguenther@suse.de>
18462 PR tree-optimization/81418
18463 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18464 vectype_in. Verify that with lane-reducing reduction operations
18465 we have a single def-use cycle.
18467 2017-07-17 Carl Love <cel@us.ibm.com>
18469 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18472 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18473 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18475 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18476 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18477 VMULOSW): New enum "unspec" values.
18478 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18479 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18480 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18481 altivec_vmulosw): New patterns.
18482 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18483 VMULOSW): Add definitions.
18484 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18486 * config/alpha/alpha.c: Include predict.h.
18488 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18490 * tree-vrp.c (compare_assert_loc): Fix comparison function
18491 to return predictable results.
18493 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18495 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18497 (subdi3): Likewise.
18498 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18499 * doc/invoke.texi (mexpand-adddi): Update text.
18501 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18503 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18504 that also clobbers the CC register. The old expand code is moved
18506 (*arc_clzsi2): ... here.
18507 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18508 the CC register. The old expand code is moved to ...
18509 (arc_ctzsi2): ... here.
18511 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18513 * config/arc/arc.opt (mindexed-loads): Use initial value
18514 TARGET_INDEXED_LOADS_DEFAULT.
18515 (mauto-modify-reg): Use initial value
18516 TARGET_AUTO_MODIFY_REG_DEFAULT.
18517 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18518 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18519 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18520 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18522 2017-07-17 Martin Liska <mliska@suse.cz>
18525 * opts.c (finish_options): Do not allow -fgnu-tm
18526 w/ -fsanitize={kernel-,}address. Say sorry.
18528 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18531 * tree-loop-distribution.c (classify_partition): Only assert on
18532 numer of iterations.
18533 (merge_dep_scc_partitions): Delete prameter. Update function call.
18534 (distribute_loop): Remove code handling loop with unknown niters.
18535 (pass_loop_distribution::execute): Skip loop with unknown niters.
18537 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18540 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18541 function sort_partitions_by_post_order.
18543 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18545 PR tree-optimization/81374
18546 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18547 the max index of basic blocks, rather than number of basic blocks.
18549 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18551 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18553 (arc_legitimate_pic_operand_p): Likewise.
18554 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18556 (arc_needs_pcl_p): Likewise.
18557 (arc_legitimate_pc_offset_p): Likewise.
18558 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18559 function is also used in constrains.md.
18560 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18561 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18562 PLUS. Only return true/false in known cases, otherwise assert.
18563 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18564 is already called in arc_legitimate_constant_p.
18565 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18567 (LEGITIMATE_PIC_OPERAND_P): Use
18568 arc_raw_symbolic_reference_mentioned_p function.
18569 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18574 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18575 Andrew Burgess <andrew.burgess@embecosm.com>
18577 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18578 (arc_return_address_register): New function.
18579 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18580 (arc_handle_fndecl_attribute): Add naked attribute.
18581 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18582 (TARGET_WARN_FUNC_RETURN): Likewise.
18583 (arc_allocate_stack_slots_for_args): New function.
18584 (arc_warn_func_return): Likewise.
18585 (machine_function): Change type fn_type.
18586 (arc_compute_function_type): Consider new naked function type,
18587 change function return type.
18588 (arc_must_save_register): Adapt to handle new
18589 arc_compute_function_type's return type.
18590 (arc_expand_prologue): Likewise.
18591 (arc_expand_epilogue): Likewise.
18592 (arc_return_address_regs): Delete.
18593 (arc_return_address_register): New function.
18594 (arc_epilogue_uses): Use above function.
18595 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18596 (arc_function_type): Change encoding, add naked type.
18597 (ARC_INTERRUPT_P): Change to handle the new encoding.
18598 (ARC_FAST_INTERRUPT_P): Likewise.
18599 (ARC_NORMAL_P): Define.
18600 (ARC_NAKED_P): Likewise.
18601 (arc_compute_function_type): Delete prototype.
18602 * config/arc/arc.md (in_ret_delay_slot): Use
18603 arc_return_address_register function.
18604 (simple_return): Likewise.
18605 (p_return_i): Likewise.
18607 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18609 PR tree-optimization/81428
18610 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18611 can't be built for those types.
18613 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18615 Remove stuff dead since r239246.
18617 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18618 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18619 (avr_inform_devices): Remove dead stuff.
18621 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18623 * config/arm/arm_neon.h: Fix softp typo.
18625 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18627 PR tree-optimization/81365
18628 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18629 aggregate moves onto bb predecessor edges, make sure there are no
18630 loads that could alias the lhs in between the start of bb and the
18633 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18636 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18637 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18638 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18640 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18642 PR tree-optimization/81396
18643 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18644 (init_symbolic_number): Initialize it to 1.
18645 (perform_symbolic_merge): Add n_ops from both operands into the new
18647 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18648 without base_addr as useless if they need more than one operation.
18649 (bswap_replace): Handle !bswap case for NULL base_addr.
18651 2017-07-17 Tom de Vries <tom@codesourcery.com>
18654 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18657 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18659 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18660 conditional builtin define __FIX_LEON3FT_B2BST.
18662 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18664 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18665 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18668 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18670 PR rtl-optimization/81424
18671 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18672 to remove potential trapping from operands if -fnon-call-exceptions.
18674 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18676 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18677 profile_proability for scalling.
18678 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18680 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18682 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18684 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18686 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18687 fixpoint arithmetics.
18689 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18691 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18692 fixpoint arithmetics.
18694 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18696 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18697 fixpoint arithmetics.
18699 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18701 * profile-count.h (profile_probability::from_reg_br_prob_note,
18702 profile_probability::to_reg_br_prob_note): New functions.
18703 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18704 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18705 * predict.c (probability_reliable_p): Update.
18706 (edge_probability_reliable_p): Update.
18707 (br_prob_note_reliable_p): Update.
18708 (invert_br_probabilities): Update.
18709 (add_reg_br_prob_note): New function.
18710 (combine_predictions_for_insn): Update.
18711 * asan.c (asan_clear_shadow): Update.
18712 * cfgbuild.c (compute_outgoing_frequencies): Update.
18713 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18714 (update_br_prob_note): Update.
18715 (rtl_verify_edges): Update.
18716 (purge_dead_edges): Update.
18717 (fixup_reorder_chain): Update.
18718 * emit-rtl.c (try_split): Update.
18719 * ifcvt.c (cond_exec_process_insns): Update.
18720 (cond_exec_process_if_block): Update.
18721 (dead_or_predicable): Update.
18722 * internal-fn.c (expand_addsub_overflow): Update.
18723 (expand_neg_overflow): Update.
18724 (expand_mul_overflow): Update.
18725 * loop-doloop.c (doloop_modify): Update.
18726 * loop-unroll.c (compare_and_jump_seq): Update.
18727 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18728 * predict.h: Update.
18729 * reorg.c (mostly_true_jump): Update.
18731 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18732 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18733 * config/arc/arc.c: (emit_unlikely_jump): Update.
18734 * config/arm/arm.c: (emit_unlikely_jump): Update.
18735 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18736 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18737 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18738 (ix86_print_operand): Update.
18739 (ix86_split_fp_branch): Update.
18740 (predict_jump): Update.
18741 * config/ia64/ia64.c (ia64_print_operand): Update.
18742 * config/mmix/mmix.c (mmix_print_operand): Update.
18743 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18744 (rs6000_expand_split_stack_prologue): Update.
18745 * config/rs6000/rs6000.c: Update.
18746 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18747 (s390_expand_vec_movstr): Update.
18748 (s390_expand_cs_tdsi): Update.
18749 (s390_expand_split_stack_prologue): Update.
18750 * config/sh/sh.c (sh_print_operand): Update.
18751 (expand_cbranchsi4): Update.
18752 (expand_cbranchdi4): Update.
18753 * config/sparc/sparc.c (output_v9branch): Update.
18754 * config/spu/spu.c (get_branch_target): Update.
18755 (ea_load_store_inline): Update.
18756 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18757 * config/tilepro/tilepro.c: Update.
18759 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18761 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18762 (gimplify_save_expr): Likewise.
18764 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18766 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18768 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18770 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18772 * timevar.def (TV_IPA_FNSUMMARY): Define.
18774 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18776 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18777 to back store errata sensitive sequence from being generated.
18778 (sqrtdf2_fix): Likewise.
18780 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18782 * tree-ssa-threadupdate.c (compute_path_counts,
18783 update_joiner_offpath_counts): Use profile_probability.
18785 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18788 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18790 * config/arm/arm-c.c (arm_cpu_builtins): Define
18791 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18793 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18795 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18796 array entries to represent __ieee128 versions of the
18797 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
18798 scalar_extract_sig, and scalar_insert_exp built-in functions.
18799 (altivec_resolve_overloaded_builtin): Add special case handling
18800 for the __builtin_scalar_insert_exp function, as represented by
18801 the P9V_BUILTIN_VEC_VSIEDP constant.
18802 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
18803 exponent support for __ieee128 argument.
18804 (VSESQP): Add scalar extract signature support for __ieee128
18806 (VSTDCNQP): Add scalar test negative support for __ieee128
18808 (VSIEQP): Add scalar insert exponent support for __int128 argument
18809 with __ieee128 result.
18810 (VSIEQPF): Add scalar insert exponent support for __ieee128
18811 argument with __ieee128 result.
18812 (VSTDCQP): Add scalar test data class support for __ieee128
18814 (VSTDCNQP): Add overload support for scalar test negative with
18815 __ieee128 argument.
18816 (VSTDCQP): Add overload support for scalar test data class
18817 __ieee128 argument.
18818 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
18819 UNSPEC_VSX_SXSIGDP.
18820 (UNSPEC_VSX_SIEXPQP): New constant.
18821 (xsxexpqp): New insn for VSX scalar extract exponent quad
18823 (xsxsigqp): New insn for VSX scalar extract significand quad
18825 (xsiexpqpf): New insn for VSX scalar insert exponent quad
18826 precision with floating point argument.
18827 (xststdcqp): New expand for VSX scalar test data class quad
18829 (xststdcnegqp): New expand for VSX scalar test negative quad
18831 (xststdcqp): New insn to match expansions for VSX scalar test data
18832 class quad precision and VSX scalar test negative quad precision.
18833 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18834 special case operand checking to enforce that second operand of
18835 VSX scalar test data class with quad precision argument is a 7-bit
18837 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18838 prototypes and descriptions of __ieee128 versions of
18839 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18840 scalar_test_data_class, and scalar_test_neg built-in functions.
18842 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18844 PR tree-optimization/81162
18845 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18846 replace a negate with an add.
18848 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18850 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18852 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18854 * config/arm/arm-c.c (arm_cpu_builtins): Define
18855 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18857 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18859 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18860 (armv8-r): Set ARM Cortex-R52 as default CPU.
18861 * config/arm/arm-tables.opt: Regenerate.
18862 * config/arm/arm-tune.md: Regenerate.
18863 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18865 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18866 extension for -mcpu=cortex-r52.
18868 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18870 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18871 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18872 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18873 (fp-armv8): Define it as FP_ARMv8 only.
18874 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18875 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18877 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18878 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18879 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18880 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18881 than TARGET_FPU_ARMV8.
18882 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18883 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18884 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18886 * config/arm/neon.md (neon_vrint): Likewise.
18887 (neon_vcvt): Likewise.
18888 (neon_<fmaxmin_op><mode>): Likewise.
18889 (<fmaxmin><mode>3): Likewise.
18890 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18891 * config/arm/predicates.md (arm_cond_move_operator): Check against
18892 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18894 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18896 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18897 to top of function.
18899 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18901 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18902 loop in comment with memset.
18904 2017-07-14 Martin Liska <mliska@suse.cz>
18906 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18907 * dwarf2out.c (is_java): Remove the function.
18908 (output_pubname): Remove usage of the function.
18909 (lower_bound_default): Remove usage of DW_LANG_Java.
18910 (gen_compile_unit_die): Likewise.
18911 * gcc.c: Remove compiler defaults for .java and .zip files.
18912 * gimple-expr.c (remove_suffix): Change as there's no longer
18913 extension than 4-letter one.
18914 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18915 (gimplify_save_expr): Likewise.
18916 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18917 as it's possible even for other languages than Java.
18918 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18919 * lto-opts.c (lto_write_options): Remove reference to Java.
18920 * opts.c (strip_off_ending): Update file extension handling.
18921 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18922 * tree-eh.c (lower_resx): Likewise.
18923 * tree.c (free_lang_data_in_type): Remove dead code.
18924 (find_decls_types_r): Likewise.
18925 (build_common_builtin_nodes): Remove Java from a comment.
18926 (verify_type): Remove dead code.
18927 * varasm.c (assemble_external): Remove Java from a comment.
18929 2017-07-14 Martin Liska <mliska@suse.cz>
18931 * opts.c (finish_options): Add quotes.
18932 (common_handle_option): Likewise.
18934 2017-07-14 Martin Liska <mliska@suse.cz>
18936 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18937 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18938 Remove N_SO_PASCAL.
18939 * dwarf2out.c (lower_bound_default): Do not handle
18941 (gen_compile_unit_die): Likewise.
18942 * gcc.c: Remove default extension binding for GNU Pascal.
18943 * stmt.c: Remove Pascal language from a comment.
18944 * xcoffout.c: Likewise.
18946 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18949 * diagnostic-show-locus.c (fixit_cmp): New function.
18950 (layout::layout): Sort m_fixit_hints.
18951 (column_range::column_range): Assert that the values are valid.
18952 (struct char_span): New struct.
18953 (correction::overwrite): New method.
18954 (struct source_line): New struct.
18955 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18956 calls in terms of classes source_line and char_span, and
18957 correction::overwrite.
18958 (selftest::test_overlapped_fixit_printing_2): New function.
18959 (selftest::diagnostic_show_locus_c_tests): Call it.
18961 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18963 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18964 early if there is no lhs.
18966 2017-07-13 Martin Liska <mliska@suse.cz>
18968 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18969 (gen_reference_type_die): Likewise.
18970 * stor-layout.c: Remove Pascal-related comment.
18972 2017-07-13 Martin Liska <mliska@suse.cz>
18974 * opts.c (finish_options): Add quotes to error messages.
18975 (parse_sanitizer_options): Likewise.
18977 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18979 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18981 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18983 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18985 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18987 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18989 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18991 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18994 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18995 provides the hardware capability bits, define the macro
18996 __BUILTIN_CPU_SUPPORTS__.
18997 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18998 if GLIBC does not provide the hardware capability bits. Add a
18999 gcc_unreachable call if the built-in cpu function is neither
19000 __builtin_cpu_is nor __builtin_cpu_supports.
19001 (rs6000_get_function_versions_dispatcher): Change the warning
19002 that an old GLIBC is used which does not export the capability
19003 bits to be an error.
19004 * doc/extend.texi (target_clones attribute): Document the
19005 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19006 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19007 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19008 the macros defined by GCC if the newer GLIBC is available.
19010 2017-07-12 Jeff Law <law@redhat.com>
19012 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19013 remaining includes slightly.
19014 * config/riscv/riscv-builtins.c: Include profile-count.h.
19016 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19019 * config/avr/avr.c (avr_set_current_function): In diagnostic
19020 messages: Quote keywords and (parts of) identifiers.
19021 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19024 2017-07-12 Carl Love <cel@us.ibm.com>
19026 * config/rs6000/rs6000-c.c: Add support for built-in functions
19027 vector bool char vec_revb (vector bool char);
19028 vector bool short vec_revb (vector short char);
19029 vector bool int vec_revb (vector bool int);
19030 vector bool long long vec_revb (vector bool long long);
19031 * doc/extend.texi: Update the built-in documentation file for the
19032 new built-in functions.
19034 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19036 * config/s390/s390.md: Remove movcc splitter.
19038 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19040 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19041 load/store on condition.
19043 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19046 * config/avr/avr.c (avr_encode_section_info)
19047 [progmem && !TREE_READONLY]: Error if progmem object needs
19050 2017-07-11 Michael Collison <michael.collison@arm.com>
19052 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19055 2017-07-11 Carl Love <cel@us.ibm.com>
19057 * config/rs6000/rs6000-c.c: Add support for builtins
19058 vector unsigned int vec_parity_lsbb (vector signed int);
19059 vector unsigned int vec_parity_lsbb (vector unsigned int);
19060 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19061 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19062 vector unsigned long long vec_parity_lsbb (vector signed long long);
19063 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19064 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19065 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19066 * doc/extend.texi: Update the built-in documentation file for the
19067 new built-in functions.
19069 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19071 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19072 (layout::m_primary_loc): New field.
19073 (layout::layout): Initialize new field. Move location filtering
19074 logic from here to...
19075 (layout::maybe_add_location_range): ...this new method. Add
19076 support for filtering to just the lines already specified by other
19078 (layout::will_show_line_p): New method.
19079 (gcc_rich_location::add_location_if_nearby): New method.
19080 (selftest::test_add_location_if_nearby): New test function.
19081 (selftest::diagnostic_show_locus_c_tests): Call it.
19082 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19085 2017-07-11 Tom de Vries <tom@codesourcery.com>
19087 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19088 (bb_first_real_insn): New function.
19089 (nvptx_single): Add extra initialization of broadcasted condition
19092 2017-07-11 Nathan Sidwell <nathan@acm.org>
19094 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19096 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19098 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19099 Binutils doc as TEXI will mess them up.
19100 * doc/invoke.texi (AVR Options): Same here.
19102 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19104 * config/sparc/sparc.opt (mfix-ut700): New option.
19105 (mfix-gr712rc): Likewise.
19106 (sparc_fix_b2bst): New variable.
19107 * doc/invoke.texi (SPARC options): Document them.
19108 (ARM options): Fix warnings.
19109 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19110 instructions to prevent sequences that can trigger the store-store
19111 errata for certain LEON3FT processors.
19112 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19113 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19114 * config/sparc/sparc.md (fix_b2bst): New attribute.
19115 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19117 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19120 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19122 (*rsqrtsf2_sse): Ditto.
19124 (div<mode>3): Macroize insn from divdf3 and divsf3
19125 using MODEF mode iterator.
19127 2017-07-10 Martin Sebor <msebor@redhat.com>
19129 PR tree-optimization/80397
19130 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19131 instead of testing for equality to INTEGER_TYPE.
19133 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19135 * config.gcc: Remove uclibc from arc target spec.
19137 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19139 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19141 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19144 * lto-wrapper.c (remove_option): New function.
19145 (merge_and_complain): Merge PIC/PIE options more realistically.
19147 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19149 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19153 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19154 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19155 * config.in: Regenerate.
19156 * configure: Regenerate.
19157 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19158 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19159 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19160 (TARGET_GASISR_PROLOGUES): ...target mask.
19161 * common/config/avr/avr-common.c
19162 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19163 Set -mgas-isr-prologues.
19164 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19165 INSERT_PASS_BEFORE for it.
19166 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19167 * config/avr/avr.c (avr_option_override)
19168 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19169 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19170 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19171 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19172 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19174 (make_avr_pass_pre_proep): New function.
19175 (emit_push_sfr) <treg>: Add argument to function and use it
19176 instead of TMP_REG.
19177 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19178 and set machine->gasisr.yes.
19179 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19180 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19181 __gcc_isr.n_pushed to .L__stack_usage.
19182 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19183 (avr_asm_final_postscan_insn): ...this new static function.
19184 * config/avr/avr.h (machine_function)
19185 <is_no_gccisr, use_L__stack_usage>: New fields.
19186 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19187 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19188 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19189 (gasisr, *gasisr): New expander and insn.
19190 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19191 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19192 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19194 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19196 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19199 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19201 Move jump-tables out of .text again.
19204 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19205 (ASM_OUTPUT_ADDR_VEC): New function.
19206 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19207 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19208 INSN_ADDRESSes as asm comment.
19209 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19210 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19211 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19212 * config/avr/avr.md (*tablejump): Adjust comment.
19213 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19214 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19216 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19217 (avr_output_addr_vec): New proto.
19218 (avr_log_t) <insn_addresses>: New field.
19220 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19223 * config/i386/i386.c (ix86_function_arg_advance): Set
19224 outgoing_args_on_stack to true if there are outgoing arguments
19226 (ix86_function_arg): Likewise.
19227 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19228 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19229 * config/i386/i386.h (machine_function): Add
19230 outgoing_args_on_stack.
19232 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19234 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19235 supporting pthreds.
19236 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19238 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19240 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19241 (REAL_H): Remove $(MACHMODE_H).
19242 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19244 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19245 $(MACHMODE_H) and double-int.h.
19246 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19248 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19251 2017-07-07 Andrew Pinski <apinski@cavium.com>
19253 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19254 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19256 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19258 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19259 Add warning if GCC was not configured to link against a GLIBC that
19260 exports the hardware capability bits.
19261 (make_resolver_func): Make resolver function private and not a
19262 COMDAT function. Create the name with clone_function_name instead
19263 of make_unique_name.
19266 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19267 correct operand in doing the split.
19269 2017-07-07 Carl Love <cel@us.ibm.com>
19271 * config/rs6000/rs6000-c: Add support for built-in function
19272 vector unsigned short vec_pack_to_short_fp32 (vector float,
19274 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19275 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19276 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19277 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19278 (convert_4f32_8i16): Add define_expand.
19279 * doc/extend.texi: Update the built-in documentation file for the
19280 new built-in function.
19282 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19284 * config/sparc/m8.md: New file.
19285 * config/sparc/sparc.md: Include m8.md.
19287 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19289 * config/sparc/sparc.opt: New option -mvis4b.
19290 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19291 (sparc_option_override): Handle VIS4B.
19292 (enum sparc_builtins): Define
19293 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19294 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19295 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19296 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19297 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19298 (check_constant_argument): New function.
19299 (sparc_vis_init_builtins): Define builtins
19300 __builtin_vis_dictunpack{8,16,32},
19301 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19302 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19303 __builtin_vis_fpcmpde{8,16,32}shl and
19304 __builtin_vis_fpcmpur{8,16,32}shl.
19305 (sparc_expand_builtin): Check that the constant operands to
19306 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19307 constant and in range.
19308 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19310 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19311 (SPARC_IMM5_P): Likewise.
19312 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19313 (enabled): Handle vis4b.
19314 (UNSPEC_DICTUNPACK): New unspec.
19315 (UNSPEC_FPCMPSHL): Likewise.
19316 (UNSPEC_FPUCMPSHL): Likewise.
19317 (UNSPEC_FPCMPDESHL): Likewise.
19318 (UNSPEC_FPCMPURSHL): Likewise.
19319 (cpu_feature): New CPU feature `vis4b'.
19320 (dictunpack{8,16,32}): New insns.
19321 (FPCSMODE): New mode iterator.
19322 (fpcscond): New code iterator.
19323 (fpcsucond): Likewise.
19324 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19325 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19326 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19327 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19328 * config/sparc/constraints.md: Define constraints `q' for unsigned
19329 2-bit integer constants and `t' for unsigned 5-bit integer
19331 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19333 (imm5_operand_dictunpack16): Likewise.
19334 (imm5_operand_dictunpack32): Likewise.
19335 (imm2_operand): Likewise.
19336 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19337 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19338 ditunpack* and fpcmp*shl builtins.
19340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19342 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19343 * config.in: Add HAVE_AS_SPARC6 define.
19344 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19346 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19348 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19349 (CPP_CPU_SPEC): Handle m8.
19350 (ASM_CPU_SPEC): Likewise.
19351 * config/sparc/sparc-opts.h (enum processor_type): Add
19353 * config/sparc/sparc.c (m8_costs): New struct.
19354 (sparc_option_override): Handle TARGET_CPU_m8.
19355 (sparc32_initialize_trampoline): Likewise.
19356 (sparc64_initialize_trampoline): Likewise.
19357 (sparc_issue_rate): Likewise.
19358 (sparc_register_move_cost): Likewise.
19359 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19360 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19361 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19362 (CPP_CPU_SPEC): Handle M8.
19363 (ASM_CPU_SPEC): Likewise.
19364 (AS_M8_FLAG): Define.
19365 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19366 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19367 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19369 * configure: Regenerate.
19370 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19373 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19375 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19377 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19378 ("*movdi_insn_sp32"): Do not set v3pipe.
19379 ("*movsi_insn"): Likewise.
19380 ("*movdi_insn_sp64"): Likewise.
19381 ("*movsf_insn"): Likewise.
19382 ("*movdf_insn_sp32"): Likewise.
19383 ("*movdf_insn_sp64"): Likewise.
19384 ("*zero_extendsidi2_insn_sp64"): Likewise.
19385 ("*sign_extendsidi2_insn"): Likewise.
19386 ("*mov<VM32:mode>_insn"): Likewise.
19387 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19388 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19389 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19390 ("<vlop:code><VL:mode>3"): Likewise.
19391 ("*not_<vlop:code><VL:mode>3"): Likewise.
19392 ("*nand<VL:mode>_vis"): Likewise.
19393 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19394 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19395 ("one_cmpl<VL:mode>2"): Likewise.
19396 ("faligndata<VM64:mode>_vis"): Likewise.
19397 ("alignaddrsi_vis"): Likewise.
19398 ("alignaddrdi_vis"): Likweise.
19399 ("alignaddrlsi_vis"): Likewise.
19400 ("alignaddrldi_vis"): Likewise.
19401 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19402 ("bmaskdi_vis"): Likewise.
19403 ("bmasksi_vis"): Likewise.
19404 ("bshuffle<VM64:mode>_vis"): Likewise.
19405 ("cmask8<P:mode>_vis"): Likewise.
19406 ("cmask16<P:mode>_vis"): Likewise.
19407 ("cmask32<P:mode>_vis"): Likewise.
19408 ("pdistn<P:mode>_vis"): Likewise.
19409 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19411 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19413 * config/sparc/sparc.md ("subtype"): New insn attribute.
19414 ("*wrgsr_sp64"): Set insn subtype.
19415 ("*rdgsr_sp64"): Likewise.
19416 ("alignaddrsi_vis"): Likewise.
19417 ("alignaddrdi_vis"): Likewise.
19418 ("alignaddrlsi_vis"): Likewise.
19419 ("alignaddrldi_vis"): Likewise.
19420 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19421 ("fexpand_vis"): Likewise.
19422 ("fpmerge_vis"): Likewise.
19423 ("faligndata<VM64:mode>_vis"): Likewise.
19424 ("bshuffle<VM64:mode>_vis"): Likewise.
19425 ("cmask8<P:mode>_vis"): Likewise.
19426 ("cmask16<P:mode>_vis"): Likewise.
19427 ("cmask32<P:mode>_vis"): Likewise.
19428 ("fchksm16_vis"): Likewise.
19429 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19430 ("fmean16_vis"): Likewise.
19431 ("fp<plusminus_insn>64_vis"): Likewise.
19432 ("<plusminus_insn>v8qi3"): Likewise.
19433 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19434 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19435 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19436 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19437 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19438 ("*movqi_insn"): Likewise.
19439 ("*movhi_insn"): Likewise.
19440 ("*movsi_insn"): Likewise.
19441 ("movsi_pic_gotdata_op"): Likewise.
19442 ("*movdi_insn_sp32"): Likewise.
19443 ("*movdi_insn_sp64"): Likewise.
19444 ("movdi_pic_gotdata_op"): Likewise.
19445 ("*movsf_insn"): Likewise.
19446 ("*movdf_insn_sp32"): Likewise.
19447 ("*movdf_insn_sp64"): Likewise.
19448 ("*zero_extendhisi2_insn"): Likewise.
19449 ("*zero_extendqihi2_insn"): Likewise.
19450 ("*zero_extendqisi2_insn"): Likewise.
19451 ("*zero_extendqidi2_insn"): Likewise.
19452 ("*zero_extendhidi2_insn"): Likewise.
19453 ("*zero_extendsidi2_insn_sp64"): Likewise.
19454 ("ldfsr"): Likewise.
19455 ("prefetch_64"): Likewise.
19456 ("prefetch_32"): Likewise.
19457 ("tie_ld32"): Likewise.
19458 ("tie_ld64"): Likewise.
19459 ("*tldo_ldub_sp32"): Likewise.
19460 ("*tldo_ldub1_sp32"): Likewise.
19461 ("*tldo_ldub2_sp32"): Likewise.
19462 ("*tldo_ldub_sp64"): Likewise.
19463 ("*tldo_ldub1_sp64"): Likewise.
19464 ("*tldo_ldub2_sp64"): Likewise.
19465 ("*tldo_ldub3_sp64"): Likewise.
19466 ("*tldo_lduh_sp32"): Likewise.
19467 ("*tldo_lduh1_sp32"): Likewise.
19468 ("*tldo_lduh_sp64"): Likewise.
19469 ("*tldo_lduh1_sp64"): Likewise.
19470 ("*tldo_lduh2_sp64"): Likewise.
19471 ("*tldo_lduw_sp32"): Likewise.
19472 ("*tldo_lduw_sp64"): Likewise.
19473 ("*tldo_lduw1_sp64"): Likewise.
19474 ("*tldo_ldx_sp64"): Likewise.
19475 ("*mov<VM32:mode>_insn"): Likewise.
19476 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19477 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19479 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19481 * config/sparc/sparc.md ("type"): New insn type viscmp.
19482 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19484 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19485 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19486 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19487 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19489 ("n7_vis_logical_11cycle"): Likewise.
19490 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19491 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19492 * config/sparc/niagara.md ("niag_vis"): Likewise.
19493 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19494 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19496 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19498 * config/sparc/sparc.md: New instruction type `bmask'.
19499 (bmaskdi_vis): Use the `bmask' type.
19500 (bmasksi_vis): Likewise.
19501 * config/sparc/ultra3.md (us3_array): Likewise.
19502 * config/sparc/niagara7.md (n7_array): Likewise.
19503 * config/sparc/niagara4.md (n4_array): Likewise.
19504 * config/sparc/niagara2.md (niag2_vis): Likewise.
19505 (niag3_vis): Likewise.
19506 * config/sparc/niagara.md (niag_vis): Likewise.
19508 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19510 * ipa-comdats.c: Remove optimize check from gate.
19511 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19512 for functions not optimized.
19513 (ipa_fn_summary_read): Skip optimize check.
19514 (ipa_fn_summary_write): Likewise.
19515 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19517 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19519 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19520 (check_callers): Give up on caller which is not optimized.
19521 (inline_small_functions): Likewise.
19522 (ipa_inline): Do not give up when not optimizing.
19523 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19524 away unoptimizes cdtors.
19525 (whole_program_function_and_variable_visibility): Do
19526 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19527 * ipa.c (process_references): Do not check optimize.
19528 (symbol_table::remove_unreachable_nodes): Update optimize check.
19529 (set_writeonly_bit): Update optimize check.
19530 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19531 (pass_ipa_single_use::gate): Remove.
19533 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19535 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19536 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19537 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19538 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19539 permute_load, permute_store, adjust_extract, adjust_splat,
19540 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19541 replace_swap_with_copy, dump_swap_insn_table,
19542 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19543 recombine_lvx_pattern, recombine_stvx_pattern,
19544 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19545 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19546 to file rs6000-p8swap.c.
19547 * config/rs6000/rs6000-p8swap.c: New file.
19548 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19549 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19550 and rs6000*-*-* targets.
19552 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19554 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19556 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19558 * lto-wrapper.c (merge_and_complain): Do not merge
19559 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19560 fsigned_zeros, ftrapping_math, fwrapv.
19561 (append_compiler_options): Do not track these options.
19562 (append_linker_options): Likewie
19564 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19566 * cgraphunit.c (cgraph_node::finalize_function): When
19567 !flag_toplevel_reorde set no_reorder flag.
19568 (varpool_node::finalize_decl): Likewise.
19569 (symbol_table::compile): Drop no toplevel reorder path.
19571 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19573 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19574 edges; zero probability is not better than uninitialized.
19576 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19578 * asan.h (asan_sanitize_allocas_p): Declare.
19579 * asan.c (asan_sanitize_allocas_p): New function.
19580 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19581 (handle_builtin_alloca): Likewise.
19582 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19583 if !asan_sanitize_allocas_p.
19584 * params.def (asan-instrument-allocas): Add new option.
19585 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19586 * opts.c (common_handle_option): Disable allocas sanitization for
19589 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19591 * asan.c: Include gimple-fold.h.
19592 (get_last_alloca_addr): New function.
19593 (handle_builtin_stackrestore): Likewise.
19594 (handle_builtin_alloca): Likewise.
19595 (asan_emit_allocas_unpoison): Likewise.
19596 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19597 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19598 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19599 (instrument_builtin_call): Pass gimple iterator to
19600 get_mem_refs_of_builtin_call.
19601 (last_alloca_addr): New global.
19602 * asan.h (asan_emit_allocas_unpoison): Declare.
19603 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19604 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19605 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19606 if function calls alloca.
19607 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19608 * gimple-fold.h (replace_call_with_value): Declare.
19609 * internal-fn.c: Include asan.h.
19610 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19611 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19613 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19615 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19616 (C_SELFTEST_FLAGS): New.
19617 (CPP_SELFTEST_FLAGS): New.
19618 (SELFTEST_DEPS): New, from deps of s-selftest.
19619 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19620 (CPP_SELFTEST_DEPS): New.
19621 (selftest): Add dependency on s-selftest-c++.
19622 (s-selftest): Rename to...
19623 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19624 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19625 than SELFTEST_FLAGS.
19626 (selftest-gdb): Rename to...
19627 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19629 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19630 (selftest-valgrind): Rename to...
19631 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19633 (selftest-valgrind): Reintroduce as an alias for
19634 selftest-c-valgrind.
19635 (s-selftest-c++): New.
19636 (selftest-c++-gdb): New.
19637 (selftest-c++-valgrind): New.
19639 2017-07-06 Olivier Hainque <hainque@adacore.com>
19641 * gcc.c (process_command): When deciding if undefined variables
19642 should be ignored when processing specs, accept "gcc -v" as well.
19644 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19646 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19647 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19649 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19651 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19652 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19653 * config/arm/arm-tables.opt: Regenerate.
19654 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19656 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19658 2017-07-06 Carl Love <cel@us.ibm.com>
19660 * ChangeLog: Clean up from mid air collision
19662 2017-07-06 Carl Love <cel@us.ibm.com>
19664 * config/rs6000/rs6000-c.c: Add support for built-in functions
19665 vector signed int vec_subc (vector signed int, vector signed int);
19666 vector signed __int128 vec_subc (vector signed __int128,
19667 vector signed __int128);
19668 vector unsigned __int128 vec_subc (vector unsigned __int128,
19669 vector unsigned __int128);
19670 vector signed int vec_sube (vector signed int, vector signed int,
19671 vector signed int);
19672 vector unsigned int vec_sube (vector unsigned int,
19673 vector unsigned int,
19674 vector unsigned int);
19675 vector signed __int128 vec_sube (vector signed __int128,
19676 vector signed __int128,
19677 vector signed__int128);
19678 vector unsigned __int128 vec_sube (vector unsigned __int128,
19679 vector unsigned __int128,
19680 vector unsigned __int128);
19681 vector signed int vec_subec (vector signed int, vector signed int,
19682 vector signed int);
19683 vector unsigned int vec_subec (vector unsigned int,
19684 vector unsigned int,
19685 vector unsigned int);
19686 vector signed __int128 vec_subec (vector signed __int128,
19687 vector signed __int128,
19688 vector signed__int128);
19689 vector unsigned __int128 vec_subec (vector unsigned __int128,
19690 vector unsigned __int128,
19691 vector unsigned __int128);
19692 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19693 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19694 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19695 BU_ALTIVEC_OVERLOAD_X definitions.
19696 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19697 * doc/extend.texi: Update the built-in documentation file for the new
19698 built-in functions.
19700 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19703 * diagnostic-show-locus.c (layout::layout): Use start and finish
19704 spelling location for the start and finish of each range.
19705 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19706 Add unused aspect param.
19707 * input.c (expand_location_1): Add "aspect" param, and use it
19708 to access the correct part of the location.
19709 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19711 (expand_location_to_spelling_point): Likewise.
19712 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19713 param, and pass it to expand_location_1.
19715 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19717 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19718 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19719 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19720 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19721 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19722 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19723 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19724 _mm_maskz_getmant_ss): New intrinsics.
19725 (__builtin_ia32_getexpss128_mask): Changed to ...
19726 __builtin_ia32_getexpss128_round ... this.
19727 (__builtin_ia32_getexpsd128_mask): Changed to ...
19728 __builtin_ia32_getexpsd128_round ... this.
19729 * config/i386/i386-builtin-types.def
19730 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19731 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19732 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19733 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19734 __builtin_ia32_getmantss_mask_round): New builtins.
19735 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19736 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19737 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19738 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19739 * config/i386/sse.md
19740 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19741 avx512f_sgetexp<mode><mask_scalar_name>
19742 <round_saeonly_scalar_name> ... this.
19743 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19744 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19745 vgetexp<ssescalarmodesuffix>
19746 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19747 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19748 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19749 avx512f_vgetmant<mode><mask_scalar_name>
19750 <round_saeonly_scalar_name> ... this.
19751 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19752 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19753 vgetmant<ssescalarmodesuffix>
19754 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19755 %0<mask_scalar_operand4>, %1, %2
19756 <round_saeonly_scalar_mask_op4>, %3} ... this.
19757 * config/i386/subst.md (mask_scalar_operand4,
19758 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19759 round_saeonly_scalar_nimm_predicate): New subst attributes.
19761 2017-07-06 Julia Koval <julia.koval@intel.com>
19763 * config/i386/i386.c (ix86_erase_embedded_rounding):
19764 Remove code for old rounding pattern.
19766 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19768 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19770 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19772 * doc/sourcebuild.texi (Test Directives, Variants of
19773 dg-require-support): Add documentation for dg-require-stack-check.
19775 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19777 * config/i386/subst.md (mask_scalar, round_scalar,
19778 round_saeonly_scalar): New meta-templates.
19779 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19780 round_scalar_mask_operand3, round_scalar_mask_op3,
19781 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19782 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19783 round_saeonly_scalar_constraint,
19784 round_saeonly_scalar_prefix): New subst attribute.
19785 * config/i386/sse.md
19786 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19787 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19788 <round_scalar_name> ... this.
19789 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19790 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19791 <round_scalar_name> ... this.
19792 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19793 <sse>_vm<code><mode>3<mask_scalar_name>
19794 <round_saeonly_scalar_name> ... this.
19795 (v<plusminus_mnemonic><ssescalarmodesuffix>
19796 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19797 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19798 v<plusminus_mnemonic><ssescalarmodesuffix>
19799 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19800 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19801 (v<multdiv_mnemonic><ssescalarmodesuffix>
19802 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19803 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19804 v<multdiv_mnemonic><ssescalarmodesuffix>
19805 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19806 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19807 (v<maxmin_float><ssescalarmodesuffix>
19808 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
19809 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
19810 v<maxmin_float><ssescalarmodesuffix>
19811 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19812 %0<mask_scalar_operand3>, %1, %<iptr>2
19813 <round_saeonly_scalar_mask_op3>} ... this.
19815 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
19817 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
19818 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
19820 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
19821 Alan Hayward <alan.hayward@arm.com>
19822 David Sherwood <david.sherwood@arm.com>
19824 * combine.c (simplify_if_then_else): Remove "enum" before
19826 * compare-elim.c (can_eliminate_compare): Likewise.
19827 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19829 (aarch64_lookup_simd_builtin_type): Likewise.
19830 (aarch64_simd_builtin_type): Likewise.
19831 (aarch64_init_simd_builtin_types): Likewise.
19832 (aarch64_simd_expand_args): Likewise.
19833 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19835 (aarch64_reverse_mask): Likewise.
19836 (aarch64_simd_emit_reg_reg_move): Likewise.
19837 (aarch64_gen_adjusted_ldpstp): Likewise.
19838 (aarch64_ccmp_mode_to_code): Likewise.
19839 (aarch64_operands_ok_for_ldpstp): Likewise.
19840 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19841 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19843 (aarch64_min_divisions_for_recip_mul): Likewise.
19844 (aarch64_reassociation_width): Likewise.
19845 (aarch64_get_condition_code_1): Likewise.
19846 (aarch64_simd_emit_reg_reg_move): Likewise.
19847 (aarch64_simd_attr_length_rglist): Likewise.
19848 (aarch64_reverse_mask): Likewise.
19849 (aarch64_operands_ok_for_ldpstp): Likewise.
19850 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19851 (aarch64_gen_adjusted_ldpstp): Likewise.
19852 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19854 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19855 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19856 (arm_lookup_simd_builtin_type): Likewise.
19857 (arm_simd_builtin_type): Likewise.
19858 (arm_init_simd_builtin_types): Likewise.
19859 (arm_expand_builtin_args): Likewise.
19860 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19861 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19862 (ft32_setup_incoming_varargs): Likewise.
19863 (ft32_function_arg): Likewise.
19864 (ft32_function_arg_advance): Likewise.
19865 (ft32_pass_by_reference): Likewise.
19866 (ft32_arg_partial_bytes): Likewise.
19867 (ft32_valid_pointer_mode): Likewise.
19868 (ft32_addr_space_pointer_mode): Likewise.
19869 (ft32_addr_space_legitimate_address_p): Likewise.
19870 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19872 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19873 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19874 (iamcu_alignment): Likewise.
19875 (canonicalize_vector_int_perm): Likewise.
19876 (ix86_noce_conversion_profitable_p): Likewise.
19877 (ix86_mpx_bound_mode): Likewise.
19878 (ix86_operands_ok_for_move_multiple): Likewise.
19879 * config/microblaze/microblaze-protos.h
19880 (microblaze_expand_conditional_branch_reg): Likewise.
19881 * config/microblaze/microblaze.c
19882 (microblaze_expand_conditional_branch_reg): Likewise.
19883 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19885 (rs6000_reassociation_width): Likewise.
19886 (rs6000_invalid_binary_op): Likewise.
19887 (fusion_p9_p): Likewise.
19888 (emit_fusion_p9_load): Likewise.
19889 (emit_fusion_p9_store): Likewise.
19890 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19892 (riscv_hard_regno_mode_ok_p): Likewise.
19893 (riscv_address_insns): Likewise.
19894 (riscv_split_symbol): Likewise.
19895 (riscv_legitimize_move): Likewise.
19896 (riscv_function_value): Likewise.
19897 (riscv_hard_regno_nregs): Likewise.
19898 (riscv_expand_builtin): Likewise.
19899 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19900 (riscv_build_integer): Likewise.
19901 (riscv_split_integer): Likewise.
19902 (riscv_legitimate_constant_p): Likewise.
19903 (riscv_cannot_force_const_mem): Likewise.
19904 (riscv_regno_mode_ok_for_base_p): Likewise.
19905 (riscv_valid_base_register_p): Likewise.
19906 (riscv_valid_offset_p): Likewise.
19907 (riscv_valid_lo_sum_p): Likewise.
19908 (riscv_classify_address): Likewise.
19909 (riscv_legitimate_address_p): Likewise.
19910 (riscv_address_insns): Likewise.
19911 (riscv_load_store_insns): Likewise.
19912 (riscv_force_binary): Likewise.
19913 (riscv_split_symbol): Likewise.
19914 (riscv_force_address): Likewise.
19915 (riscv_legitimize_address): Likewise.
19916 (riscv_move_integer): Likewise.
19917 (riscv_legitimize_const_move): Likewise.
19918 (riscv_legitimize_move): Likewise.
19919 (riscv_address_cost): Likewise.
19920 (riscv_subword): Likewise.
19921 (riscv_output_move): Likewise.
19922 (riscv_canonicalize_int_order_test): Likewise.
19923 (riscv_emit_int_order_test): Likewise.
19924 (riscv_function_arg_boundary): Likewise.
19925 (riscv_pass_mode_in_fpr_p): Likewise.
19926 (riscv_pass_fpr_single): Likewise.
19927 (riscv_pass_fpr_pair): Likewise.
19928 (riscv_get_arg_info): Likewise.
19929 (riscv_function_arg): Likewise.
19930 (riscv_function_arg_advance): Likewise.
19931 (riscv_arg_partial_bytes): Likewise.
19932 (riscv_function_value): Likewise.
19933 (riscv_pass_by_reference): Likewise.
19934 (riscv_setup_incoming_varargs): Likewise.
19935 (riscv_print_operand): Likewise.
19936 (riscv_elf_select_rtx_section): Likewise.
19937 (riscv_save_restore_reg): Likewise.
19938 (riscv_for_each_saved_reg): Likewise.
19939 (riscv_register_move_cost): Likewise.
19940 (riscv_hard_regno_mode_ok_p): Likewise.
19941 (riscv_hard_regno_nregs): Likewise.
19942 (riscv_class_max_nregs): Likewise.
19943 (riscv_memory_move_cost): Likewise.
19944 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19945 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19946 (rl78_addr_space_address_mode): Likewise.
19947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19949 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19950 (rs6000_reassociation_width): Likewise.
19951 (rs6000_invalid_binary_op): Likewise.
19952 (fusion_p9_p): Likewise.
19953 (emit_fusion_p9_load): Likewise.
19954 (emit_fusion_p9_store): Likewise.
19955 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19956 (ok_for_simple_move_operands): Likewise.
19957 (ok_for_simple_move_strict_operands): Likewise.
19958 (ok_for_simple_arith_logic_operands): Likewise.
19959 (visium_legitimize_reload_address): Likewise.
19960 (visium_select_cc_mode): Likewise.
19961 (output_cbranch): Likewise.
19962 (visium_split_double_move): Likewise.
19963 (visium_expand_copysign): Likewise.
19964 (visium_expand_int_cstore): Likewise.
19965 (visium_expand_fp_cstore): Likewise.
19966 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19967 (visium_function_arg): Likewise.
19968 (visium_function_arg_advance): Likewise.
19969 (visium_libcall_value): Likewise.
19970 (visium_setup_incoming_varargs): Likewise.
19971 (visium_legitimate_constant_p): Likewise.
19972 (visium_legitimate_address_p): Likewise.
19973 (visium_legitimize_address): Likewise.
19974 (visium_secondary_reload): Likewise.
19975 (visium_register_move_cost): Likewise.
19976 (visium_memory_move_cost): Likewise.
19977 (prepare_move_operands): Likewise.
19978 (ok_for_simple_move_operands): Likewise.
19979 (ok_for_simple_move_strict_operands): Likewise.
19980 (ok_for_simple_arith_logic_operands): Likewise.
19981 (visium_function_value_1): Likewise.
19982 (rtx_ok_for_offset_p): Likewise.
19983 (visium_legitimize_reload_address): Likewise.
19984 (visium_split_double_move): Likewise.
19985 (visium_expand_copysign): Likewise.
19986 (visium_expand_int_cstore): Likewise.
19987 (visium_expand_fp_cstore): Likewise.
19988 (visium_split_cstore): Likewise.
19989 (visium_select_cc_mode): Likewise.
19990 (visium_split_cbranch): Likewise.
19991 (output_cbranch): Likewise.
19992 (visium_print_operand_address): Likewise.
19993 * expmed.c (flip_storage_order): Likewise.
19994 * expmed.h (emit_cstore): Likewise.
19995 (flip_storage_order): Likewise.
19996 * genrecog.c (validate_pattern): Likewise.
19997 * hsa-gen.c (gen_hsa_addr): Likewise.
19998 * internal-fn.c (expand_arith_overflow): Likewise.
19999 * ira-color.c (allocno_copy_cost_saving): Likewise.
20000 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20001 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20002 (process_invariant_for_inheritance): Likewise.
20003 * lra-eliminations.c (move_plus_up): Likewise.
20004 * omp-low.c (lower_oacc_reductions): Likewise.
20005 * simplify-rtx.c (simplify_subreg): Likewise.
20006 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20007 (TARGET_CHKP_BOUND_MODE): Likewise..
20008 * targhooks.c (default_chkp_bound_mode): Likewise.
20009 (default_setup_incoming_vararg_bounds): Likewise.
20010 * targhooks.h (default_chkp_bound_mode): Likewise.
20011 (default_setup_incoming_vararg_bounds): Likewise.
20012 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20013 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20014 (have_whole_vector_shift): Likewise.
20015 * tree-vect-stmts.c (vectorizable_load): Likewise.
20016 * doc/tm.texi: Regenerate.
20018 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20020 Graceful degrade if Binutils PR21472 is not available.
20023 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20024 .rodata in flash test fails.
20025 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20026 * confgure: Regenerate.
20027 * config.in: Regenerate.
20028 * config/avr/avr.c (avr_asm_named_section)
20029 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20030 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20031 (avr_asm_init_sections): Same.
20033 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20035 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20036 (fma<VH:mode>4_intrinsic): Likewise.
20037 (*fmsub<VCVTF:mode>4): Likewise.
20038 (*fmsub<VH:mode>4_intrinsic): Likewise.
20040 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20043 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20044 Don't depend on "optimize > 0".
20045 (out_movhi_r_mr, out_movqi_mr_r): Same.
20046 (out_movhi_mr_r, out_movqi_r_mr): Same.
20047 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20048 io_address_operand on "optimize > 0".
20050 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20052 * tree-loop-distribution.c: Add general explanantion on the pass.
20053 (generate_loops_for_partition): Mark distributed loop.
20054 (pg_add_dependence_edges): New parameter. Handle alias data
20055 dependence specially and record it in the parameter if asked.
20056 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20057 (init_partition_graph_vertices, add_partition_graph_edge): New.
20058 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20059 (free_partition_graph_vdata, build_partition_graph): New.
20060 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20061 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20062 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20063 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20064 (version_for_distribution_p, finalize_partitions): New.
20065 (distribute_loop): Handle alias data dependence specially. Factor
20066 out loop fusion code as functions and call these functions.
20068 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20070 * tree-loop-distribution.c (classify_partition): New parameter and
20071 better handle reduction statement.
20072 (rdg_build_partitions): Revise comment.
20073 (distribute_loop): Compute statements in all partitions and pass it
20074 to classify_partition.
20076 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20078 * tree-loop-distribution.c (enum partition_type): New.
20079 (struct partition): New field type.
20080 (partition_merge_into): Add parameter. Update partition type.
20081 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20082 (build_rdg_partition_for_vertex): Compute partition type.
20083 (rdg_build_partitions): Dump partition type.
20084 (distribute_loop): Update calls to partition_merge_into.
20086 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20088 * tree-loop-distribution.c (struct ddr_hasher): New.
20089 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20091 (classify_partition): Call get_data_dependence.
20092 (pg_add_dependence_edges): Ditto.
20093 (distribute_loop): Release data dependence hash table.
20095 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20097 * tree-loop-distribution.c (ref_base_address): Delete.
20098 (similar_memory_accesses): Rename ...
20099 (share_memory_accesses): ... to this. Check if partitions access
20100 the same memory reference.
20101 (distribute_loop): Call share_memory_accesses.
20103 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20105 * tree-loop-distribution.c (struct partition): New field recording
20106 its data reference.
20107 (partition_alloc, partition_free): Init and release data refs.
20108 (partition_merge_into): Merge data refs.
20109 (build_rdg_partition_for_vertex): Collect data refs for partition.
20110 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20112 (distribute_loop): Remve data refs from vertice data of partition
20115 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20117 * tree-loop-distribution.c (params.h): Include header file.
20118 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20119 (datarefs_vec): New global var.
20120 (create_rdg_vertices): Use datarefs_vec directly.
20121 (free_rdg): Don't free data references.
20122 (build_rdg): Update use. Don't free data references.
20123 (distribute_loop): Compute global variable for data references.
20124 Bail out if there are too many data references.
20126 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20128 * tree-loop-distribution.c (loop_nest): New global var.
20129 (build_rdg): Use loop directly, rather than loop nest.
20130 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20132 (distribute_loop): Compute global variable loop nest. Update use.
20134 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20136 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20137 (partition_merge_into): New parameter. Dump reason for fusion.
20138 (distribute_loop): Update use of partition_merge_into.
20140 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20142 * tree-loop-distribution.c (bb_top_order_index): New.
20143 (bb_top_order_index_size, bb_top_order_cmp): New.
20144 (stmts_from_loop): Use topological order.
20145 (pass_loop_distribution::execute): Compute and release topological
20146 order for basic blocks.
20148 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20150 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20153 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20155 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20156 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20157 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20158 * internal-fn.def (LOOP_DIST_ALIAS): New.
20159 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20160 (fold_loop_internal_call): ... this.
20161 (vect_loop_dist_alias_call): New function.
20162 (set_uid_loop_bbs): Call fold_loop_internal_call.
20163 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20166 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20169 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20170 Require dead FLAGS_REG at the beginning of a peephole.
20172 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20175 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20176 arguments in the call to __builtin_ia32_sbb_u32.
20177 (_subborrow_u64): Swap _X and _Y arguments in the call to
20178 __builtin_ia32_sbb_u64.
20180 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20183 * tree-vrp.c (compare_assert_loc): Turn into a function template
20184 with stable template parameter. Only test if a->e is NULL,
20185 !a->e == !b->e has been verified already. Use e == NULL or
20186 e != NULL instead of e or ! e tests. If stable is true, don't use
20187 iterative_hash_expr, on the other side allow a or b or both NULL
20188 and sort the NULLs last.
20189 (process_assert_insertions): Sort using compare_assert_loc<false>
20190 instead of compare_assert_loc, later sort using
20191 compare_assert_loc<true> before calling process_assert_insertions_for
20192 in a loop. Use break instead of continue once seen NULL pointer.
20194 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20196 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20197 Cortex-R7 and Cortex-R8 processors.
20199 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20201 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20202 uninitialized while src is not.
20204 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20206 * common/config/arm/arm-common.c: Adjust include path for
20208 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20209 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20210 (arm-cpu-data.h): Likewise.
20211 (arm-cpu-cdata.h): Likewise.
20212 * config/arm/arm-cpu.h: Delete.
20213 * config/arm/arm-cpu-cdata.h: Delete.
20214 * config/arm/arm-cpu-data.h: Delete.
20216 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20218 * config/arm/arm-cpus.in (cortex-a55): New.
20219 (cortex-a75): Likewise.
20220 (cortex-a75.cortex-a55): Likewise.
20221 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20223 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20224 * config/arm/arm-cpu-cdata.h: Regenerate.
20225 * config/arm/arm-cpu-data.h: Regenerate.
20226 * config/arm/arm-cpu.h: Regenerate.
20227 * config/arm/arm-tables.opt: Regenerate.
20228 * config/arm/arm-tune.md: Regenerate.
20230 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20232 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20234 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20236 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20239 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20241 PR tree-optimization/81292
20242 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20243 full_string_p, also call adjust_related_strinfos if the adjustment
20244 is simple, otherwise invalidate related strinfos.
20246 2017-07-04 Martin Liska <mliska@suse.cz>
20249 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20250 newly created variable as DECL_IGNORED_P.
20252 2017-07-04 Martin Liska <mliska@suse.cz>
20255 * ipa-inline.c (inline_small_functions):
20256 Use xstrdup_for_dump.
20258 2017-07-04 Tom de Vries <tom@codesourcery.com>
20260 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20262 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20265 * config/darwin.c (darwin_function_switched_text_sections):
20268 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20270 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20272 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20274 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20276 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20278 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20279 min_profitable_iters, and th as inclusive lower bounds.
20280 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20281 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20282 for min_profitable_iters and min_profitable_estimate.
20283 (vect_transform_loop): Treat th as an inclusive lower bound.
20284 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20286 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20289 * config/darwin.c (darwin_function_switched_text_sections):
20290 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20291 in two pieces, and suppress the use of buf.
20293 2017-07-03 Nathan Sidwell <nathan@acm.org>
20295 * hash-table.h (hash_table_mod1): Fix indentation.
20297 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20299 PR middle-end/81290
20300 * predict.c (force_edge_cold): Be more careful about propagation
20302 * profile-count.h (profile_probability::guessed,
20303 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20305 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20307 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20309 * doc/invoke.texi (rcpc architecture extension): Document it.
20311 2017-07-03 Richard Biener <rguenther@suse.de>
20313 PR tree-optimization/60510
20314 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20315 the scalar reduction PHI and use it.
20316 (vectorizable_reduction): Properly guard the single_defuse_cycle
20317 path for non-SLP reduction chains where we cannot use it.
20318 Rework reduc_def/index and vector type deduction. Rework
20319 vector operand gathering during reduction op code-gen.
20320 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20321 chains dissolve the chain and leave it to non-SLP reduction
20324 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20326 * tree-data-ref.h (dr_alignment): Declare.
20327 * tree-data-ref.c (dr_alignment): New function.
20328 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20329 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20331 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20333 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20335 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20336 and base_misalignment fields.
20337 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20338 * tree-data-ref.c: Include builtins.h.
20339 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20340 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20341 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20342 * tree-vect-data-refs.c: Include tree-cfg.h.
20343 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20344 fields instead of calculating an alignment here.
20345 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20346 innermost_loop_behavior fields.
20348 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20350 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20352 (DR_STEP_ALIGNMENT): New macro.
20353 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20354 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20355 (create_data_ref): Print it.
20356 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20357 to tell whether the step preserves vector (mis)alignment.
20358 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20359 Move the check for an integer step and generalise to all INTEGER_CST.
20360 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20361 Print the outer step alignment.
20363 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20365 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20366 with offset_alignment.
20367 (DR_ALIGNED_TO): Delete.
20368 (DR_OFFSET_ALIGNMENT): New macro.
20369 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20370 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20371 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20372 (create_data_ref): Likewise.
20373 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20374 (vect_analyze_data_refs): Likewise.
20375 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20376 creating dummy innermost behavior.
20378 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20380 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20381 with a "innermost_loop_behavior *" and refeence tree.
20382 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20383 (create_data_ref): Update call accordingly.
20384 * tree-predcom.c (find_looparound_phi): Likewise.
20386 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20388 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20389 fields with dr_wrt_vec_loop.
20390 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20391 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20392 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20393 (vect_dr_behavior): New function.
20394 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20395 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20396 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20397 track whether the step preserves the misalignment.
20398 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20399 Use vect_dr_behavior.
20400 (vect_setup_realignment): Update call accordingly.
20401 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20402 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20403 call to vect_create_addr_base_for_vector_ref.
20404 (vect_create_cond_for_align_checks): Likewise.
20405 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20406 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20407 (vect_recog_mask_conversion_pattern): Likewise.
20408 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20409 (new_stmt_vec_info): Remove redundant zeroing.
20411 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20413 * common/config/arm/arm-common.c (arm_be8_option): New function.
20414 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20415 (ISA_ARMv6): Add isa_bit_be8.
20416 * config/arm/arm.h (arm_be8_option): Add prototype.
20417 (BE8_SPEC_FUNCTION): New define.
20418 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20419 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20420 (mlittle-endian): Similarly.
20421 (mbe8, mbe32): New options.
20422 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20423 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20425 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20427 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20429 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20431 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20432 (cleanup_tree_cfg_bb): Use it.
20433 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20435 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20437 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20440 * loop-doloop.c (add_test): Update profile.
20442 2017-07-03 Martin Liska <mliska@suse.cz>
20445 * sanopt.c (rewrite_usage_of_param): New function.
20446 (sanitize_rewrite_addressable_params): Likewise.
20447 (pass_sanopt::execute): Call rewrite_usage_of_param.
20449 2017-07-03 Richard Biener <rguenther@suse.de>
20451 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20452 back to using VIEW_CONVERT_EXPR.
20454 2017-07-03 Martin Liska <mliska@suse.cz>
20457 * doc/extend.texi: Document when a resolver function is
20458 generated for target_clones.
20460 2017-07-03 Martin Liska <mliska@suse.cz>
20462 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20463 and shadow memory of auto variables which are subject of
20464 use-after-scope sanitization.
20465 (asan_expand_mark_ifn): Add do set only when is_poison.
20467 2016-07-03 Richard Biener <rguenther@suse.de>
20469 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20471 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20472 (vectorizable_reduction): Perform reduction PHI creation when
20473 visiting a reduction PHI and adjust and simplify code generation
20474 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20475 (vect_transform_loop): Visit reduction PHIs.
20476 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20477 defs into the SLP tree.
20478 (vect_build_slp_tree): Reduction defs terminate the recursion.
20479 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20481 (vect_get_vec_defs_for_stmt_copy): Export.
20482 (vect_get_vec_defs): Likewise.
20483 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20485 (vect_get_vec_defs_for_stmt_copy): Declare.
20486 (vect_get_vec_defs): Likewise.
20488 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20490 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20491 parameter with a "loop" parameter and use it instead of the
20492 loop containing DR_STMT. Don't check simple_iv when doing
20493 BB analysis. Describe the two analysis modes in the comment.
20495 2017-07-03 Tom de Vries <tom@codesourcery.com>
20497 PR tree-optimization/69468
20498 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20499 (find_same_succ_bb): Handle ignore_edge_flags.
20501 2017-07-03 Tom de Vries <tom@codesourcery.com>
20503 PR tree-optimization/81192
20504 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20506 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20508 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20510 2017-07-03 Tom de Vries <tom@codesourcery.com>
20512 PR tree-optimization/81192
20513 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20514 BB_SAME_SUCC (bb) == NULL.
20516 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20518 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20521 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20523 * dumpfile.c: Include profile-count.h
20524 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20526 (insert_cond_bb): Update profile.
20527 * tree-cfg.h (insert_cond_bb): Update prototype.
20528 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20529 * tree-dump.c: Do not include tree-cfg.
20531 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20533 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20535 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20537 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20540 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20542 * tree-complex.c (expand_complex_div_wide): update profile.
20544 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20545 Alan Hayward <alan.hayward@arm.com>
20546 David Sherwood <david.sherwood@arm.com>
20548 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20549 (CORETYPES_H): New define.
20550 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20551 (insn-modes-inline.h, s-modes-inline-h): New rules.
20552 (generated_files): Add insn-modes-inline.h.
20553 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20554 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20555 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20556 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20557 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20558 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20559 (build/gencodes.o, build/genconditions.o): Likewise.
20560 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20561 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20562 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20563 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20564 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20565 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20566 * coretypes.h: Include everything up to real.h for generators.
20567 Include insn-modes.h first. Include wide-int-print.h after
20568 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20569 * machmode.h: Don't include insn-modes.h here.
20570 * function-tests.c: Remove includes of signop.h, machmode.h,
20571 double-int.h and wide-int.h.
20573 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20575 * optc-save-gen.awk: Likewise.
20576 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20577 * godump.c: Remove include of wide-int-print.h.
20578 * pretty-print.h: Likewise.
20579 * wide-int-print.cc: Likewise.
20580 * wide-int.cc: Likewise.
20581 * hash-map-tests.c: Remove include of signop.h.
20582 * hash-set-tests.c: Likewise.
20583 * rtl-tests.c: Likewise.
20584 * mkconfig.sh: Remove include of machmode.h.
20585 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20587 (emit_insn_modes_inline_h): ...this new function. Emit the code
20588 into an insn-modes-inline.h header file, adding appropriate
20589 include guards and end comments.
20590 (emit_insn_modes_c_header): Remove include of machmode.h.
20591 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20593 (main): Handle -i flag and call emit_insn_modes_inline_h when
20596 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20598 * tree-ssa-strlen.c (strinfo): Rename the length field to
20599 nonzero_chars. Add a full_string_p field.
20600 (compare_nonzero_chars, zero_length_string_p): New functions.
20601 (get_addr_stridx): Add an offset_out parameter.
20602 Use compare_nonzero_chars.
20603 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20604 (new_strinfo): Update after above changes to strinfo.
20605 (set_endptr_and_length): Set full_string_p.
20606 (get_string_length): Update after above changes to strinfo.
20607 (unshare_strinfo): Update call to new_strinfo.
20608 (maybe_invalidate): Likewise.
20609 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20610 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20611 as a uhwi instead of an shwi. Update after above changes to
20612 strinfo and new_strinfo.
20613 (zero_length_string): Assert that chainsi contains full strings.
20614 Use zero_length_string_p. Update call to new_strinfo.
20615 (adjust_related_strinfos): Update after above changes to strinfo.
20616 Copy full_string_p from origsi.
20617 (adjust_last_stmt): Use zero_length_string_p.
20618 (handle_builtin_strlen): Update after above changes to strinfo and
20619 new_strinfo. Install the lhs as the string length if the previous
20620 entry didn't describe a full string.
20621 (handle_builtin_strchr): Update after above changes to strinfo
20623 (handle_builtin_strcpy): Likewise.
20624 (handle_builtin_strcat): Likewise.
20625 (handle_builtin_malloc): Likewise.
20626 (handle_pointer_plus): Likewise.
20627 (handle_builtin_memcpy): Likewise. Track nonzero characters
20628 that aren't necessarily followed by a nul terminator.
20629 (handle_char_store): Likewise.
20631 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20633 PR tree-optimization/80769
20634 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20635 for malloc and calloc. Document the new invariant that all related
20636 strinfos have delayed lengths or none do.
20637 (verify_related_strinfos): Move earlier in file.
20638 (set_endptr_and_length): New function, split out from...
20639 (get_string_length): ...here. Also set the lengths of related
20641 (zero_length_string): Assert that chainsi has known (rather than
20643 (adjust_related_strinfos): Likewise.
20645 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20647 PR tree-optimization/81136
20648 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20649 assert that two references with the same misalignment have the same
20650 compile-time misalignment if those compile-time misalignments
20653 2017-07-01 Andi Kleen <ak@linux.intel.com>
20655 * print-tree.c (print_node): Print all attributes.
20657 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20659 * cfg.c (scale_bbs_frequencies): New function.
20660 * cfg.h (scale_bbs_frequencies): Declare it.
20661 * cfgloopanal.c (single_likely_exit): Cleanup.
20662 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20664 (scale_loop_profile): Likewise.
20665 (loop_version): Likewise.
20666 (create_empty_loop_on_edge): Update.
20667 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20668 scale_loop_frequencies, scale_loop_profile, loopify,
20669 loop_version): Update prototypes.
20670 * modulo-sched.c (sms_schedule): Update.
20671 * predict.c (unlikely_executed_edge_p): Also check probability.
20672 (probably_never_executed_edge_p): Fix typo.
20673 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20674 * tree-parloops.c (gen_parallel_loop): Update.
20675 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20676 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20677 * tree-ssa-loop-split.c (split_loop): Update.
20678 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20679 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20680 (vect_loop_versioning): Update.
20681 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20683 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20685 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20687 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20689 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20690 to keep profile consistent.
20692 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20694 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20695 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20696 * profile-count.h (max_safe_multiplier): Make unsigned.
20697 (profile_count::guessed_zero): New.
20699 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20701 * bb-reorder.c (fix_up_crossing_landing_pad,
20702 fix_crossing_conditional_branches): Use make_single_succ_edge
20703 to keep profile consistent.
20705 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20707 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20710 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20713 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20714 the right scopes, make sure cond_jump isn't preserved between multiple
20715 iterations. Search for fallthru edge whenever there are 3+ edges and
20716 use find_fallthru_edge for it.
20718 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20720 Patch by Alexander Monakov <amonakov@ispras.ru>
20721 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20722 probabilities consistently.
20724 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20726 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20727 emit_cmp_and_jump_insns.
20729 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20732 * tree-inline.c (expand_call_inline): Combine profile statuses.
20734 2017-06-30 Andrew Pinski <apinski@cavium.com>
20736 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20737 fold_stmt returned true.
20739 2017-06-30 Nathan Sidwell <nathan@acm.org>
20741 * ggc.h (empty_string): Delete.
20742 * cfgexpand.c (expand_asm_stmt): Use plain "".
20743 * optabs.c (expand_asm_memory_barrier): Likewise.
20744 * stringpool.c (empty_string): Delete.
20745 (digit_vector, digit_string): Delete.
20746 (ggc_alloc_string): Use plain "", don't optimize single digit
20747 strings. Use ggc_alloc_atomic.
20749 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20751 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20752 comparison set and one other set, use the cost of the non-comparison
20755 2017-06-30 Nathan Sidwell <nathan@acm.org>
20757 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20760 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20762 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20763 loops. Remove now unneeded calls to gimple_switch_set_label() that
20764 just set removed labels to NULL_TREE.
20766 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20768 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20769 (set_range_info): ...here. Only call set_range_info_raw if domain
20771 (set_nonzero_bits): Call set_range_info_raw.
20772 * tree-ssanames.h (set_range_info_raw): New.
20774 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20777 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20778 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20779 of nonimmediate_operand and <store_mask_constraint> instead of m
20780 for the input operand. For V8FI iterator, always split if input
20781 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20782 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20783 <store_mask_predicate> instead of register_operand and
20784 <store_mask_constraint> instead of v for the input operand. Make
20785 sure both operands aren't MEMs for if not <mask_applied>.
20787 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20789 * lto-wrapper.c (copy_file) Close both file descriptors before
20792 2017-06-30 Martin Liska <mliska@suse.cz>
20795 * multiple_target.c (create_dispatcher_calls): Make ifunc
20796 also for function that don't have calls or are not referenced.
20798 2017-06-30 Richard Biener <rguenther@suse.de>
20800 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
20801 analyze the first scalar stmt. Move vector type computation
20802 for the BB case here from ...
20803 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
20804 live operation processing in the SLP case properly.
20806 2017-06-30 Richard Biener <rguenther@suse.de>
20808 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
20810 2017-06-30 Martin Liska <mliska@suse.cz>
20813 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
20814 before BUILT_IN_UNWIND_RESUME when ASAN is used.
20816 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
20818 * doc/invoke.texi (AArch64): Add missing options and remove redundant
20821 2017-06-30 Richard Biener <rguenther@suse.de>
20823 PR tree-optimization/81249
20824 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
20825 condition reduction result to original scalar type.
20827 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20829 * profile-count.h (enum profile_quality): Fix typos and whitespace
20832 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20834 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20835 type for branch probabilities.
20837 2017-06-29 Julian Brown <julian@codesourcery.com>
20838 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20840 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20841 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20842 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20843 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20845 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20847 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20848 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20849 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20850 CC usage from generic code to here.
20851 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20852 CC usage into the target macros.
20854 2017-06-29 Maya Rashish <coypu@sdf.org>
20856 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20859 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20861 * arm/arm-builtins.c: Include profile-count.h
20862 * except.c (sjlj_emit_function_enter): Use
20863 profile_probability::unlikely.
20865 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20867 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20868 and tocrel_offset be pointer args rather than implicitly using
20870 (legitimate_constant_pool_address_p, rs6000_emit_move,
20871 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20872 tocrel_offset and use in toc_relative_expr_p call.
20873 (print_operand, print_operand_address): Use static tocrel_base_oac
20874 and tocrel_offset_oac.
20875 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20878 2017-06-29 Maya Rashish <coypu@sdf.org>
20880 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20882 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20884 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20885 objects, take into account only the alignment of 'op0' and 'mode1' if
20888 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20890 * ccmp.c (ccmp_tree_comparison_p): New function.
20891 (ccmp_candidate_p): Update to use above function.
20892 (get_compare_parts): New function.
20893 (expand_ccmp_next): Update to use new functions.
20894 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20896 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20897 take mode as argument.
20898 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20899 * expr.c (expand_expr_real_1): Pass mode as argument.
20901 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20903 * combine.c (combine_instructions): Print insns to dump_file, together
20906 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20908 * asan.c (asan_emit_stack_protection): Update.
20909 (create_cond_insert_point): Update.
20910 * auto-profile.c (afdo_propagate_circuit): Update.
20911 * basic-block.h (struct edge_def): Turn probability to
20912 profile_probability.
20913 (EDGE_FREQUENCY): Update.
20914 * bb-reorder.c (find_traces_1_round): Update.
20915 (better_edge_p): Update.
20916 (sanitize_hot_paths): Update.
20917 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20918 (make_single_succ_edge): Update.
20919 (check_bb_profile): Update.
20920 (dump_edge_info): Update.
20921 (update_bb_profile_for_threading): Update.
20922 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20923 probabilitycount to 0.
20924 * cfgbuild.c (compute_outgoing_frequencies): Update.
20925 * cfgcleanup.c (try_forward_edges): Update.
20926 (outgoing_edges_match): Update.
20927 (try_crossjump_to_edge): Update.
20928 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20929 (expand_gimple_tailcall): Update.
20930 (construct_init_block): Use make_single_succ_edge.
20931 (construct_exit_block): Use make_single_succ_edge.
20932 * cfghooks.c (verify_flow_info): Update.
20933 (redirect_edge_succ_nodup): Update.
20934 (split_edge): Update.
20935 (account_profile_record): Update.
20936 * cfgloopanal.c (single_likely_exit): Update.
20937 * cfgloopmanip.c (scale_loop_profile): Update.
20938 (set_zero_probability): Remove.
20939 (duplicate_loop_to_header_edge): Update.
20940 * cfgloopmanip.h (loop_version): Update prototype.
20941 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20942 (force_nonfallthru_and_redirect): Update.
20943 (update_br_prob_note): Update.
20944 (rtl_verify_edges): Update.
20945 (purge_dead_edges): Update.
20946 (rtl_lv_add_condition_to_bb): Update.
20947 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20948 * cgraphunit.c (init_lowered_empty_function): Update.
20949 (cgraph_node::expand_thunk): Update.
20950 * cilk-common.c: Include profile-count.h
20951 * dojump.c (inv): Remove.
20952 (jumpifnot): Update.
20953 (jumpifnot_1): Update.
20954 (do_jump_1): Update.
20956 (do_jump_by_parts_greater_rtx): Update.
20957 (do_compare_rtx_and_jump): Update.
20958 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20959 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20960 * dwarf2cfi.c: Include profile-count.h
20961 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20962 (sjlj_emit_dispatch_table): Likewise.
20963 * explow.c: Include profile-count.h
20964 * expmed.c (emit_store_flag_force): Update.
20965 (do_cmp_and_jump): Update.
20966 * expr.c (compare_by_pieces_d::generate): Update.
20967 (compare_by_pieces_d::finish_mode): Update.
20968 (emit_block_move_via_loop): Update.
20969 (store_expr_with_bounds): Update.
20970 (store_constructor): Update.
20971 (expand_expr_real_2): Update.
20972 (expand_expr_real_1): Update.
20973 * expr.h (try_casesi, try_tablejump): Update prototypes.
20974 * gimple-pretty-print.c (dump_probability): Update.
20975 (dump_profile): New.
20976 (dump_gimple_label): Update.
20977 (dump_gimple_bb_header): Update.
20978 * graph.c (draw_cfg_node_succ_edges): Update.
20979 * hsa-gen.c (convert_switch_statements): Update.
20980 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20981 (find_if_case_1): Update.
20982 (find_if_case_2): Update.
20983 * internal-fn.c (expand_arith_overflow_result_store): Update.
20984 (expand_addsub_overflow): Update.
20985 (expand_neg_overflow): Update.
20986 (expand_mul_overflow): Update.
20987 (expand_vector_ubsan_overflow): Update.
20988 * ipa-cp.c (good_cloning_opportunity_p): Update.
20989 * ipa-split.c (split_function): Use make_single_succ_edge.
20990 * ipa-utils.c (ipa_merge_profiles): Update.
20991 * loop-doloop.c (add_test): Update.
20992 (doloop_modify): Update.
20993 * loop-unroll.c (compare_and_jump_seq): Update.
20994 (unroll_loop_runtime_iterations): Update.
20995 * lra-constraints.c (lra_inheritance): Update.
20996 * lto-streamer-in.c (input_cfg): Update.
20997 * lto-streamer-out.c (output_cfg): Update.
20998 * mcf.c (adjust_cfg_counts): Update.
20999 * modulo-sched.c (sms_schedule): Update.
21000 * omp-expand.c (expand_omp_for_init_counts): Update.
21001 (extract_omp_for_update_vars): Update.
21002 (expand_omp_ordered_sink): Update.
21003 (expand_omp_for_ordered_loops): Update.
21004 (expand_omp_for_generic): Update.
21005 (expand_omp_for_static_nochunk): Update.
21006 (expand_omp_for_static_chunk): Update.
21007 (expand_cilk_for): Update.
21008 (expand_omp_simd): Update.
21009 (expand_omp_taskloop_for_outer): Update.
21010 (expand_omp_taskloop_for_inner): Update.
21011 * omp-simd-clone.c (simd_clone_adjust): Update.
21012 * optabs.c (expand_doubleword_shift): Update.
21013 (expand_abs): Update.
21014 (emit_cmp_and_jump_insn_1): Update.
21015 (expand_compare_and_swap_loop): Update.
21016 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21017 * predict.c (predictable_edge_p): Update.
21018 (edge_probability_reliable_p): Update.
21019 (set_even_probabilities): Update.
21020 (combine_predictions_for_insn): Update.
21021 (combine_predictions_for_bb): Update.
21022 (propagate_freq): Update.
21023 (estimate_bb_frequencies): Update.
21024 (force_edge_cold): Update.
21025 * profile-count.c (profile_count::dump): Add missing space into dump.
21026 (profile_count::debug): Add newline.
21027 (profile_count::differs_from_p): Explicitly convert to unsigned.
21028 (profile_count::stream_in): Update.
21029 (profile_probability::dump): New member function.
21030 (profile_probability::debug): New member function.
21031 (profile_probability::differs_from_p): New member function.
21032 (profile_probability::differs_lot_from_p): New member function.
21033 (profile_probability::stream_in): New member function.
21034 (profile_probability::stream_out): New member function.
21035 * profile-count.h (profile_count_quality): Rename to ...
21036 (profile_quality): ... this one.
21037 (profile_probability): New.
21038 (profile_count): Update.
21039 * profile.c (compute_branch_probabilities): Update.
21040 * recog.c (peep2_attempt): Update.
21041 * sched-ebb.c (schedule_ebbs): Update.
21042 * sched-rgn.c (find_single_block_region): Update.
21043 (compute_dom_prob_ps): Update.
21044 (schedule_region): Update.
21045 * sel-sched-ir.c (compute_succs_info): Update.
21046 * stmt.c (struct case_node): Update.
21047 (do_jump_if_equal): Update.
21048 (get_outgoing_edge_probs): Update.
21049 (conditional_probability): Update.
21050 (emit_case_dispatch_table): Update.
21051 (expand_case): Update.
21052 (expand_sjlj_dispatch_table): Update.
21053 (emit_case_nodes): Update.
21054 * targhooks.c: Update.
21055 * tracer.c (better_p): Update.
21056 (find_best_successor): Update.
21057 * trans-mem.c (expand_transaction): Update.
21058 * tree-call-cdce.c: Update.
21059 * tree-cfg.c (gimple_split_edge): Upate.
21060 (move_sese_region_to_fn): Upate.
21061 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21062 * tree-eh.c (lower_resx): Upate.
21063 (cleanup_empty_eh_move_lp): Upate.
21064 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21065 * tree-inline.c (copy_edges_for_bb): Update.
21066 (copy_cfg_body): Update.
21067 * tree-parloops.c (gen_parallel_loop): Update.
21068 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21069 (gimple_gen_time_profiler): Update.
21070 * tree-ssa-dce.c (remove_dead_stmt): Update.
21071 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21072 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21073 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21074 (unloop_loops): Update.
21075 (try_peel_loop): Update.
21076 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21077 * tree-ssa-loop-split.c (connect_loops): Update.
21078 (split_loop): Update.
21079 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21080 (hoist_guard): Update.
21081 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21082 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21083 (value_replacement): Update.
21084 * tree-ssa-reassoc.c (branch_fixup): Update.
21085 * tree-ssa-tail-merge.c (replace_block_by): Update.
21086 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21087 (create_edge_and_update_destination_phis): Update.
21088 (compute_path_counts): Update.
21089 (recompute_probabilities): Update.
21090 (update_joiner_offpath_counts): Update.
21091 (freqs_to_counts_path): Update.
21092 (duplicate_thread_path): Update.
21093 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21094 (struct switch_conv_info): Update.
21095 (gen_inbound_check): Update.
21096 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21097 (vect_do_peeling): Update.
21098 (vect_loop_versioning): Update.
21099 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21100 (optimize_mask_stores): Update.
21101 * ubsan.c (ubsan_expand_null_ifn): Update.
21102 * value-prof.c (gimple_divmod_fixed_value): Update.
21103 (gimple_divmod_fixed_value_transform): Update.
21104 (gimple_mod_pow2): Update.
21105 (gimple_mod_pow2_value_transform): Update.
21106 (gimple_mod_subtract): Update.
21107 (gimple_mod_subtract_transform): Update.
21108 (gimple_ic): Update.
21109 (gimple_stringop_fixed_value): Update.
21110 (gimple_stringops_transform): Update.
21111 * value-prof.h: Update.
21113 2017-06-29 Carl Love <cel@us.ibm.com>
21115 * config/rs6000/rs6000-c.c: Add support for built-in functions
21116 vector signed int vec_signed (vector float);
21117 vector signed long long vec_signed (vector double);
21118 vector signed int vec_signed2 (vector double, vector double);
21119 vector signed int vec_signede (vector double);
21120 vector signed int vec_signedo (vector double);
21121 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21122 instruction generator.
21123 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21124 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21125 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21127 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21128 vunsignede_v2df): Add define_expands.
21129 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21130 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21131 VEC_UNSIGNEDO): Add definitions.
21132 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21133 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21134 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21135 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21136 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21137 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21138 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21139 vec_unsignede, vec_unsignedo): Add builtin defines.
21140 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21142 * doc/extend.texi: Update the built-in documentation file for the
21143 new built-in functions.
21145 2017-06-29 Richard Biener <rguenther@suse.de>
21147 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21148 reduction chains to LOOP_VINFO_REDUCTIONS.
21149 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21150 SLP reductions after processing reduction chains.
21152 2017-06-29 Nathan Sidwell <nathan@acm.org>
21154 * builtins.c (fold_builtin_FUNCTION): Use
21155 lang_hooks.decl_printable_name.
21157 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21159 PR middle-end/81194
21160 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21161 with only one label.
21162 * stmt.c (expand_case): Assert NCASES is greater than one.
21164 2017-06-29 Richard Biener <rguenther@suse.de>
21166 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21168 (group_case_labels): Likewise.
21169 (find_taken_edge): Push sanity checking on val to workers...
21170 (find_taken_edge_cond_expr): ... here
21171 (find_taken_edge_switch_expr): ... and here, handle cases
21172 with just a default label.
21173 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21174 (group_case_labels): Likewise.
21175 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21176 group_case_labels does anything cleanup the CFG again.
21178 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21180 PR tree-optimization/81196
21181 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21182 exit condition comparing two IVs.
21184 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21186 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21187 profile to the dummy entry at the end of the list of architectures.
21188 * config/arm/arm-cpu-cdata.h: Regenerated.
21190 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21191 Michael Collison <michael.collison@arm.com>
21194 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21196 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21197 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21198 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21199 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21200 of shift when the shift amount is masked with constant equal to
21201 the size of the mode.
21202 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21205 2017-06-29 Martin Liska <mliska@suse.cz>
21207 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21209 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21212 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21215 (edge_growth_cache_entry::edge_growth_cache_entry): New
21217 (reset_edge_growth_cache): Update to use constructor.
21219 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21221 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21222 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21223 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21225 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21227 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21228 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21230 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21232 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21233 (*-linux-uclibc*): Add t-uclibc tmake_file.
21234 * config/t-musl: New.
21235 * config/t-uclibc: New.
21237 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21239 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21241 (gen_comm_data): Emit architectural setting of arch_prof.
21242 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21244 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21245 (armv8-m.base, armv8-m.main): Likewise.
21246 * arm-protos.h (arm_build_target): Add profile field.
21247 (arch_option): Likewise.
21248 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21250 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21251 arm_active_target.profile.
21253 2017-06-28 Richard Biener <rguenther@suse.de>
21255 PR middle-end/81227
21256 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21257 TYPE_OVERFLOW_WRAPS.
21258 * match.pd (negate_expr_p): Likewise.
21259 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21260 fold_build2, not fold_binary.
21262 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21264 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21265 Convert memory address to Pmode.
21266 (aarch64_print_operand): Assert MEM operands are always Pmode.
21268 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21271 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21272 Remove redundant if.
21273 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21274 * config/arm/aarch-common-protos.h
21275 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21276 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21278 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21281 * multiple_target.c (create_dispatcher_calls): Set the default
21282 clone to be static, not public.
21284 2017-06-28 Richard Biener <rguenther@suse.de>
21286 * tree-vect-loop.c (vectorizable_reduction): Move special
21287 cond reduction IV var creation ...
21288 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21289 parameter. Use STMT_VINFO_VECTYPE.
21290 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21293 2017-06-28 Martin Liska <mliska@suse.cz>
21296 * ipa-visibility.c (non_local_p): Handle visibility.
21298 2017-06-28 Martin Liska <mliska@suse.cz>
21301 * common.opt: Add IntegerRange to various options.
21302 * opt-functions.awk (integer_range_info): New function.
21303 * optc-gen.awk: Add integer_range_info to cl_options struct.
21304 * opts-common.c (decode_cmdline_option): Handle
21305 CL_ERR_INT_RANGE_ARG.
21306 (cmdline_handle_error): Likewise.
21307 * opts.c (print_filtered_help): Show valid interval in
21308 when --help is provided.
21309 * opts.h (struct cl_option): Add range_min and range_max fields.
21310 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21312 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21314 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21315 (x * C EQ/NE y * C): New transformation.
21317 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21319 * genmultilib (combination_space): Accept '+' in option names.
21321 2017-06-28 Martin Liska <mliska@suse.cz>
21324 * asan.c (instrument_derefs): Bail out inner references
21325 that are hard register variables.
21327 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21330 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21331 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21333 2017-06-28 Richard Biener <rguenther@suse.de>
21335 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21336 (vect_get_slp_defs): Adjust.
21337 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21339 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21341 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21342 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21343 (vectorizable_reduction): Adjust.
21344 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21346 (vect_get_slp_defs): Likewise.
21347 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21348 (vectorizable_bswap): Adjust.
21349 (vectorizable_call): Likewise.
21350 (vectorizable_conversion): Likewise.
21351 (vectorizable_assignment): Likewise.
21352 (vectorizable_shift): Likewise.
21353 (vectorizable_operation): Likewise.
21354 (vectorizable_store): Likewise.
21355 (vectorizable_condition): Likewise.
21356 (vectorizable_comparison): Likewise.
21358 2017-06-28 Michael Collison <michael.collison@arm.com>
21361 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21363 (arm_gen_movmemqi): Removed unused variable 'i'.
21364 Convert 'for' loop into 'while' loop.
21365 (arm_expand_prologue): Remove last unnecessary set of insn.
21366 (thumb_pop): Remove unused variable 'pushed_words'.
21367 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21369 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21371 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21372 * config/s390/s390-protos.h: Add prototype of
21373 s390_rel_address_ok_p.
21374 * config/s390/s390.c (s390_got_symbol): New function.
21375 (s390_rel_address_ok_p): New function.
21376 (legitimize_pic_address): Use s390_rel_address_ok_p.
21377 (s390_load_got): Use s390_got_symbol.
21378 (s390_option_override): Issue error if
21379 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21380 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21382 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21384 2017-06-27 Andrew Pinski <apinski@cavium.com>
21386 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21387 (X * copysign (1.0, X)): New pattern.
21388 (X * copysign (1.0, -X)): New pattern.
21389 (copysign (-1.0, CST)): New pattern.
21391 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21393 * genmultilib (combination_space): Remove variable.
21394 Validate reuse rules against regular expression for any sequence
21395 of multilib options in any order.
21397 2017-06-27 Michael Collison <michael.collison@arm.com>
21399 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21400 call aarch64_split_simd_combine.
21401 * (aarch64_combine_internal<mode>): Delete pattern.
21402 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21403 Allow register and subreg operands.
21405 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21407 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21408 specific need, just fallback on defaults.
21409 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21411 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21412 Olivier Hainque <hainque@adacore.com>
21414 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21416 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21417 targets. Pick a default if no particular attempt applied.
21418 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21421 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21423 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21424 (x86_64-wrs-vxworks7): Likewise.
21426 2017-06-27 Marek Polacek <polacek@redhat.com>
21429 * ubsan.c (instrument_null): Check get_base_address's result for null.
21431 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21433 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21435 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21437 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21438 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21439 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21440 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21441 New function types.
21442 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21443 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21444 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21445 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21446 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21447 BUILT_IN_FEUPDATEENV): New builtins.
21448 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21449 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21450 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21451 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21453 (builtin_structptr_types): Adjust size.
21454 * tree.c (builtin_structptr_types): Add four entries.
21456 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21457 Olivier Hainque <hainque@adacore.com>
21459 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21460 (TLS_SYM): New local macro, forcing reference to __tls__ on
21461 link command lines for VxWorks 7 RTPs, triggering initialization
21463 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21464 OS features TLS support, true for RTPs on VxWorks 7.
21465 * config/vxworks.c (vxworks_override_options): Setup emutls
21468 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21470 * predict.c (test_prediction_value_range): Use -1U instead of -1
21471 to avoid narrowing conversion warning.
21472 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21473 to avoid narrowing conversion warning.
21474 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21476 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21478 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21480 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21481 64bit configurations.
21482 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21483 (SIZE_TYPE): Likewise.
21484 * config/vxworks.c (vxworks_emutls_var_fields): Use
21485 long_unsigned_type_node instead of unsigned_type_node as the offset
21486 field type, which is "pointer" mode in emutls.c.
21488 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21491 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21493 PR middle-end/81207
21494 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21495 gimple_vuse copying separately from gimple_vdef copying.
21497 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21499 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21502 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21503 Olivier Hainque <hainque@adacore.com>
21505 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21506 for all vxworks7 targets.
21507 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21508 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21509 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21510 variations for VX6/VX7 and 32/64bits later on in ...
21511 (VXWORKS_LIB_SPEC): Leverage new macros.
21512 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21513 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21515 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21517 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21518 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21520 2017-06-26 Carl Love <cel@us.ibm.com>
21522 * config/rs6000/rs6000-c.c: Add support for built-in functions
21523 vector bool char vec_reve (vector bool char);
21524 vector signed char vec_reve (vector signed char);
21525 vector unsigned char vec_reve (vector unsigned char);
21526 vector bool int vec_reve (vector bool int);
21527 vector signed int vec_reve (vector signed int);
21528 vector unsigned int vec_reve (vector unsigned int);
21529 vector bool long long vec_reve (vector bool long long);
21530 vector signed long long vec_reve (vector signed long long);
21531 vector unsigned long long vec_reve (vector unsigned long long);
21532 vector bool short vec_reve (vector bool short);
21533 vector signed short vec_reve (vector signed short);
21534 vector double vec_reve (vector double);
21535 vector float vec_reve (vector float);
21536 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21537 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21538 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21539 (altivec_vreve): New pattern.
21540 * config/rs6000/altivec.h (vec_reve): New define.
21541 * doc/extend.texi (vec_rev): Update the built-in documentation file
21542 for the new built-in functions.
21544 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21546 PR tree-optimization/71815
21547 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21549 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21551 (slsr_process_phi): Likewise.
21552 (replace_uncond_cands_and_profitable_phis): Don't replace a
21553 multiply candidate with a stride of 1 (copy or cast).
21554 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21556 (lowest_cost_path): Likewise.
21557 (total_savings): Likewise.
21559 2017-06-26 Richard Biener <rguenther@suse.de>
21562 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21563 Use def_builtin_pure for all gather builtins.
21565 2017-06-26 Richard Biener <rguenther@suse.de>
21567 PR tree-optimization/81203
21568 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21569 non-dominating BBs.
21571 2017-06-26 Marek Polacek <polacek@redhat.com>
21574 * doc/invoke.texi: Document -Wmultistatement-macros.
21576 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21578 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21579 arm_neon_ok_no_float_abi effective target.
21581 2017-06-26 Richard Biener <rguenther@suse.de>
21583 PR tree-optimization/80928
21584 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21585 (copy_bbs): Set BB_DUPLICATED flag early.
21586 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21588 (execute_on_shrinking_pred): Likewise.
21589 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21590 BB_DUPLICATED blocks.
21591 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21592 iterate over all PHIs considering removal of *gsi.
21594 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21596 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21599 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21601 * config/rs6000/rs6000-string.c: (expand_block_clear,
21602 do_load_for_compare, select_block_compare_mode,
21603 compute_current_alignment, expand_block_compare,
21604 expand_strncmp_align_check, expand_strn_compare,
21605 expand_block_move, rs6000_output_load_multiple)
21606 Move functions related to string/block move/compare
21607 to a separate file.
21608 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21609 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21610 for this function which is now used in two files.
21611 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21612 * config.gcc: Add rs6000-string.o to extra_objs for
21613 targets powerpc*-*-* and rs6000*-*-*.
21615 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21618 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21619 32-bit, since indexed is not valid for DImode.
21620 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21621 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21622 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21623 (define_peephole2 for Altivec d-form store): Likewise.
21626 * multiple_target.c (create_dispatcher_calls): Only create the
21627 dispatcher call if the function is the default clone of a
21628 versioned function.
21630 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21632 PR middle-end/80902
21633 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21634 a call, force the call to not be a tail call.
21636 2017-06-23 Jeff Law <law@redhat.com>
21638 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21641 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21643 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21644 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21645 handling for early expansion of vector shifts (sl,sr,sra,rl).
21646 (builtin_function_type): Add vector shift right instructions
21647 to the unsigned argument list.
21649 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21651 rtl-optimizatoin/79286
21652 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21653 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21654 trap. PIC register plus a const unspec without offset can never trap.
21656 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21658 * tree.h (builtin_structptr_type): New type.
21659 (builtin_structptr_types): Declare new array.
21660 * tree.c (builtin_structptr_types): New array.
21661 (free_lang_data, build_common_tree_nodes): Use it.
21663 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21666 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21669 2017-06-22 Matt Turner <mattst88@gmail.com>
21671 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21672 Lake models to skylake case. Assume skylake for unknown
21673 models with clflushopt.
21675 2017-06-22 Jeff Law <law@redhat.com>
21677 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21678 frame sizes that do not satisfy aarch64_uimm12_shift.
21680 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21682 * profile-count.h (apply_probability,
21683 apply_scale, probability_in): Fix checks for zero.
21685 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21687 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21688 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21690 2016-06-22 Richard Biener <rguenther@suse.de>
21692 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21693 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21694 REDUC_MAX_EXPR support.
21695 (vectorizable_reduction): Likewise.
21696 (vect_create_epilog_for_reduction): Likewise.
21698 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21700 * match.pd (A / (1 << B) -> A >> B): New.
21701 * generic-match-head.c: Include optabs-tree.h.
21702 * gimple-match-head.c: Likewise.
21703 * optabs-tree.h (target_supports_op_p): New.
21704 * optabs-tree.c (target_supports_op_p): New.
21706 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21708 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21709 $gcc_cv_ld --help output.
21710 (gcc_cv_ld_demangle): Likewise.
21711 (gcc_cv_ld_eh_frame_hdr): Likewise.
21712 (gcc_cv_ld_pie): Likewise.
21713 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21714 (gcc_cv_ld_buildid): Likewise.
21715 (gcc_cv_ld_sysroot): Likewise.
21716 (ld_bndplt_support): Likewise.
21717 (ld_pushpopstate_support): Likewise.
21718 * configure: Regenerate.
21719 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21721 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21724 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21725 operands indexes to avoid gap between operands and match_dups.
21727 2017-06-21 Andrew Pinski <apinski@cavium.com>
21729 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21730 Increment Arith_shift and Arith_shift_reg by 1.
21731 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21733 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21734 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21735 (aarch64_strip_extend): Add new argument and test for it.
21736 (aarch64_cheap_mult_shift_p): New function.
21737 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21738 add a cost if it is true.
21739 Update calls to aarch64_strip_extend.
21740 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21742 2017-06-21 Andrew Pinski <apinski@cavium.com>
21744 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21746 (thunderxt88): Likewise.
21747 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21748 (thunderx_prefetch_tune): New variable.
21749 (thunderx2t99_prefetch_tune): Update for the correct values.
21750 (thunderxt88_tunings): New variable.
21751 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21752 generic_prefetch_tune.
21753 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21755 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21757 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21758 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21759 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21760 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21761 (aarch64_atomic_cas<mode>, GPI): Likewise.
21763 2017-06-21 Martin Liska <mliska@suse.cz>
21765 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21766 statements on cold and hot labels.
21767 * predict.c (tree_estimate_probability_bb): Remove the
21768 prediction from this place.
21770 2017-06-21 Martin Liska <mliska@suse.cz>
21772 PR tree-optimization/79489
21773 * gimplify.c (maybe_add_early_return_predict_stmt): New
21775 (gimplify_return_expr): Call the function.
21776 * predict.c (tree_estimate_probability_bb): Remove handling
21778 * predict.def: Update comment about early return predictor.
21779 * gimple-predict.h (is_gimple_predict): New function.
21780 * predict.def: Change default value of early return to 66.
21781 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21783 * passes.def: Put pass_strip_predict_hints to the beginning of
21786 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21788 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21789 FUNCTION_DECL declarations.
21790 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21792 (dwaf2out_decl): Likewise.
21793 * godump.c (go_early_global_decl): Skip call to the real debug hook
21794 for FUNCTION_DECL declarations.
21795 * passes.c (rest_of_decl_compilation): Skip call to the
21796 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21797 -fdump-go-spec is passed.
21799 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21801 * config/i386/i386.c (struct builtin_isa): New field pure_p.
21802 Reorder for compactness.
21803 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
21804 (def_builtin_pure, def_builtin_pure2): New functions.
21805 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
21807 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21809 * match.pd (nop_convert): New predicate.
21810 ((A +- CST1) +- CST2): Allow some NOP conversions.
21812 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21815 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
21816 with ctors/dtors if GOVD_SHARED is set.
21818 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21820 * config/aarch64/aarch64.md (movti_aarch64):
21821 Emit mov rather than orr.
21822 (movtf_aarch64): Likewise.
21823 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
21824 Emit mov rather than orr.
21826 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21828 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
21829 Swap alternatives, make integer dup more expensive.
21831 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21833 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21834 Return true for non-tls symbols.
21836 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21838 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21839 (cortex-a75): Likewise.
21840 (cortex-a75.cortex-a55): Likewise.
21841 * config/aarch64/aarch64-tune.md: Regenerate.
21842 * doc/invoke.texi (-mtune): Document new values for -mtune.
21844 2017-06-21 Tom de Vries <tom@codesourcery.com>
21846 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21847 stack_size feature.
21848 (Effective-Target Keywords, Other attributes): Suggest using
21849 dg-add-options stack_size feature to get stack limit in stack_size
21850 effective target documentation.
21852 2017-06-21 Julian Brown <julian@codesourcery.com>
21853 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21855 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21856 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21857 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21859 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21860 attribute type list for neon_multiply.
21861 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21862 attribute type list for neon_multiply.
21863 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21864 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21865 attribute type list for neon_multiply.
21866 * config/arm/types.md (crypto_pmull): Add.
21867 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21868 attribute type list.
21870 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21872 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21875 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21877 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21878 to make sure not to dereference a NULL cost_classes_ptr pointer.
21880 2017-06-20 Carl Love <cel@us.ibm.com>
21882 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21883 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21884 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21885 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21886 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21887 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21888 VMULOSW): New enum "unspec" values.
21889 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21890 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21891 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21892 altivec_vmulosw): New patterns.
21893 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21894 VMULOSW): Add definitions.
21896 2017-06-20 Julia Koval <julia.koval@intel.com>
21898 * config/i386/i386.c: Fix rounding expand for new pattern.
21899 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21901 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21903 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21904 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21906 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21908 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21911 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21913 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21914 architecture, then by implementer ID.
21915 * config/aarch64/aarch64-tune.md: Regenerate.
21917 2017-06-20 Richard Biener <rguenther@suse.de>
21919 PR middle-end/81097
21920 * fold-const.c (split_tree): Fold to type before negating.
21922 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21924 * diagnostic-show-locus.c
21925 (selftest::test_fixit_deletion_affecting_newline): New function.
21926 (selftest::diagnostic_show_locus_c_tests): Call it.
21928 2017-06-20 Andreas Schwab <schwab@suse.de>
21931 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21934 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21936 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21937 __ARM_FEATURE_COPROC according to support.
21939 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21941 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21942 Rewritten to avoid overflow for > 32-bit pointers.
21945 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21946 by removing enum keyword.
21947 (ubsan_type_descriptor): Likewise. Formatting fix.
21950 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21951 splitter): Require TARGET_SSE2 in the condition.
21953 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21956 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21957 for doing vector set of SFmode on ISA 3.0.
21958 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21959 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21961 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21962 SFmode value into a V4SF variable that was extracted from another
21963 V4SF variable without converting the element to double precision
21964 and back to single precision vector format.
21965 (vsx_insert_extract_v4sf_p9_2): Likewise.
21967 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21969 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21970 in UWHI to avoid undefined overflow.
21973 * ubsan.h (enum ubsan_encode_value_phase): New.
21974 (ubsan_encode_value): Change second argument to
21975 enum ubsan_encode_value_phase with default value of
21976 UBSAN_ENCODE_VALUE_GENERIC.
21977 * ubsan.c (ubsan_encode_value): Change second argument to
21978 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21979 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21980 create_tmp_var_raw instead of create_tmp_var and use a
21982 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21983 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21984 ubsan_encode_value callers.
21987 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21988 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21989 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21991 2017-06-19 Richard Biener <rguenther@suse.de>
21993 PR middle-end/81118
21994 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21995 estimates if we changed anything.
21997 2017-06-19 Richard Biener <rguenther@suse.de>
21999 PR tree-optimization/80887
22000 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22001 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22002 simplified lookups, then reset mprts_hook.
22003 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22005 (try_to_simplify): Likewise.
22007 2017-06-19 Martin Liska <mliska@suse.cz>
22010 * gimplify.c (gimplify_switch_expr):
22011 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22013 2017-06-19 Martin Liska <mliska@suse.cz>
22015 * doc/install.texi: Document that PGO runs in 4 stages.
22017 2017-06-19 Martin Liska <mliska@suse.cz>
22020 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22021 to dispatcher function name.
22022 * multiple_target.c (replace_function_decl): New function.
22023 (create_dispatcher_calls): Redirect both edges and references.
22025 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22027 * profile-count.c (profile_count::dump): Dump quality.
22028 (profile_count::differs_from_p): Update for unsigned val.
22029 * profile-count.h (profile_count_quality): New enum.
22030 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22032 2017-06-19 Richard Biener <rguenther@suse.de>
22034 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22035 struct function as arg.
22036 (estimate_numbers_of_iterations): Export overload with loop arg.
22037 (free_numbers_of_iterations_estimates_loop): Use an overload of
22038 free_numbers_of_iterations_estimates instead.
22039 * tree-cfg.c (remove_bb): Adjust.
22040 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22041 * tree-parloops.c (gen_parallel_loop): Likewise.
22042 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22044 (tree_unroll_loops_completely): Likewise.
22045 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22046 Use an overload instead and export.
22047 (estimated_loop_iterations): Adjust.
22048 (max_loop_iterations): Likewise.
22049 (likely_max_loop_iterations): Likewise.
22050 (estimate_numbers_of_iterations): Take struct function as arg
22052 (loop_exits_before_overflow): Adjust.
22053 (free_numbers_of_iterations_estimates_loop): Use an overload.
22054 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22055 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22057 2017-06-19 Richard Biener <rguenther@suse.de>
22060 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22061 RANGE_EXPR conservatively.
22063 2017-06-16 Carl Love <cel@us.ibm.com>
22065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22066 definitions for vec_float, vec_float2, vec_floato,
22067 vec_floate built-ins.
22068 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22069 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22071 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22072 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22073 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22074 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22075 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22076 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22077 vec_floato): Add builtin defines.
22078 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22079 Update the built-in documentation file for the new built-in
22082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22084 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22085 (mthumb): Mark as the negative of -marm.
22087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22089 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22091 (ARM Options, -mtune): Document that this accepts the same
22092 extension options as -mcpu.
22093 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22097 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22098 permitted extensions.
22100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22102 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22103 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22104 (armv8-m.main): Add option +nodsp.
22105 * config/arm/arm-cpu-cdata.h: Regenerated.
22107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22109 * config/arm/t-fuchsia: New file.
22110 * config.gcc (arm*-*-fuchsia*): Use it.
22112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22114 * config/arm/t-symbian: Rewrite for new option infrastructure.
22116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22118 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22119 (MULTILIB_REQUIRED): Likewise.
22121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22123 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22124 (MULTILIB_RESUE): Likewise.
22125 (MULTILIB_MATCHES): Likewise.
22126 (MULTLIB_REQUIRED): Likewise.
22128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22130 * config/arm/t-rtems: Rewrite for new option framework.
22132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22134 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22135 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22136 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22137 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22138 * config/arm/t-multilib: ... here.
22139 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22140 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22141 armv7-a and armv8*-a when A-profile libraries have not been built.
22142 * config/arm/t-rmprofile: Rewrite.
22144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22146 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22147 with a backslash. Remove the backslash after substituting unescaped
22149 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22153 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22154 the driver to pass through the default setting of -mfloat-abi.
22155 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22157 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22158 (all_feat_combs): New rule.
22159 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22161 * config/arm/t-aprofile: Rewrite.
22163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22165 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22166 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22167 fpu is not specified by the user/command-line.
22168 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22169 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22170 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22171 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22172 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22173 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22177 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22178 * config/arm/t-arm-elf: Rewritten.
22180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22182 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22183 have some floating-point instructions.
22184 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22185 (TARGET_MAYBE_HARD_FLOAT): New macro.
22186 * config/arm/arm-builtins.c (arm_init_builtins): Use
22187 TARGET_MAYBE_HARD_FLOAT.
22188 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22192 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22193 (configargs.h): Include it.
22194 (arm_print_hint_for_fpu_option): New function.
22195 (arm_parse_fpu_option): New function.
22196 (candidate_extension): New class.
22197 (arm_canon_for_multilib): New function.
22198 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22199 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22200 (ARCH_CANONICAL_SPECS): New macro.
22201 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22203 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22205 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22206 are set after handling multilib fragments. Set target_cpu_default2
22209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22211 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22213 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22214 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22216 (check_arch): Likewise.
22217 * config/arm/arm.c (arm_configure_build_target): Handle
22218 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22219 options in the default.
22221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22223 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22224 when an option is an alias of another.
22225 * config/arm/parsecpu.awk (optalias): New parser token.
22226 (gen_comm_data): Mark non-alias options as such. Emit entries
22227 for extension aliases.
22228 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22229 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22230 (armv6kz, armv6zk, armv6t2): Likewise.
22231 (armv7): Make vfpv3-d16 an alias.
22232 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22234 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22235 Sort in canonical order.
22236 (armv8-a): Sort in canonical order.
22237 (armv8.1-a, armv8.2-a): Likewise.
22238 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22240 (cortex-a9): Sort in canonical order.
22241 * config/arm/arm.c (selftests.h): Include it.
22242 (arm_test_cpu_arch_data): New function.
22243 (arm_run_self_tests): New function.
22244 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22245 (targetm): Move declaration to the end of the file.
22246 * arm-cpu-cdata.h: Regenerated.
22248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22250 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22251 call to target_mode_check describing the type of option passed.
22252 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22253 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22254 arm_parse_cpu_option_name to match parameters against list of
22256 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22257 arm_arch_core_flags data structure.
22258 * config/arm/arm-cpu_cdata.h: Regenerated.
22260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22262 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22264 (arm_print_hint_for_cpu_option): Likewise.
22265 (arm_print_hint_for_arch_option): Likewise.
22266 (arm_parse_cpu_option_name): Likewise.
22267 (arm_parse_arch_option_name): Likewise.
22268 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22269 of entries in the all_fpus list.
22270 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22271 (arm_parse_cpu_option_name): Declare.
22272 (arm_parse_arch_option_name): Declare.
22273 (arm_parse_option_features): Declare.
22274 (arm_intialize_isa): Declare.
22275 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22277 (gen_comm_data): ... here. Make definitions non-static.
22278 * config/arm/arm-cpu-data.h: Regenerated.
22279 * config/arm/arm-cpu-cdata.h: Regenerated.
22281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22283 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22284 (cpu_arch_extension): New structure.
22285 (cpu_arch_option, arch_option, cpu_option): New structures.
22286 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22287 architecture types.
22288 (gen_data): Generate new format data tables.
22289 * config/arm/arm.c (cpu_tune): New structure.
22290 (cpu_option, processors): Delete.
22291 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22292 (arm_print_hint_for_cpu_option): ... this and ...
22293 (arm_print_hint_for_arch_option): ... this.
22294 (arm_parse_arch_cpu_name): Delete. Replace with ...
22295 (arm_parse_cpu_option_name): ... this and ...
22296 (arm_parse_arch_option_name): ... this.
22297 (arm_unrecognized_feature): Change type of target parameter to
22299 (arm_parse_arch_cpu_features): Delete. Replace with ...
22300 (arm_parse_option_features): ... this.
22301 (arm_configure_build_target): Rework to use new configuration data
22303 (arm_print_tune_info): Rework for new configuration data tables.
22304 * config/arm/arm-cpu-data.h: Regenerated.
22305 * config/arm/arm-cpu.h: Regenerated.
22307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22309 * Makefile.in (OBJS): Move sbitmap.o from here ...
22310 (OBJS-libcommon): ... to here.
22312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22314 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22315 (ISA_ALL_CRYPTO): New macro.
22316 (ISA_ALL_SIMD): New macro
22317 (ISA_ALL_FP): New macro.
22318 * config/arm/arm.c (fpu_bitlist): Update initializer.
22319 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22321 (arm9e): Add fpu. Add option for nofp
22322 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22323 (arm926ej-s, arm1026ej-s): Likewise.
22324 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22325 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22326 neon-fp16, neon-vfpv4, nofp and nosimd.
22327 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22328 (cortex-a8): Add fpu. Add option for nofp.
22329 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22330 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22331 (cortex-r4f): Add fpu.
22332 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22333 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22335 (cortex-r8): Likewise.
22336 (cortex-m4): Add fpu. Add option for nofp.
22337 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22338 (cortex-a17.cortex-a7): Likewise.
22339 (cortex-a32): Add fpu. Add options for crypto and nofp.
22340 (cortex-a35, cortex-a53): Likewise.
22341 (cortex-a57): Add fpu. Add option for crypto.
22342 (cortex-a72, cortex-a73): Likewise.
22343 (exynos-m1): Likewise.
22344 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22345 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22346 (cortex-m33): Add fpu. Add option for nofp.
22347 * config/arm/arm-cpu-cdata.h: Regenerated
22348 * config/arm/arm-cpu-data.h: Regenerated.
22350 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22352 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22353 (armv5te, armv5tej): Likewise.
22354 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22355 (armv7): Add options fp and vfpv3-d16.
22356 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22357 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22359 (armv7ve): Likewise.
22360 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22361 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22362 (armv8-a): Add nocrypto option.
22363 (armv8.1-a, armv8.2-a): Likewise.
22364 (armv8-m.main): add options fp, fp.dp and nofp.
22366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22368 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22370 (armv8-a+crc): Delete.
22371 (armv8.1-a): Add options simd, crypto and nofp.
22372 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22373 (armv8.2-a+fp16): Delete.
22374 (armv8-m.main): Add option dsp.
22375 (armv8-m.main+dsp): Delete.
22376 (cortex-a8): Add fpu. Add nofp option.
22377 (cortex-a9): Add fpu. Add nofp and nosimd options.
22378 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22379 link to main cpu and architecture data structures.
22380 (gen_comm_data): Only put isa attributes from the main architecture
22382 (option): New statement for architecture and CPU entries.
22383 * arm.c (struct cpu_option): New structure.
22384 (struct processors): Add entry for options.
22385 (arm_unrecognized_feature): New function.
22386 (arm_parse_arch_cpu_name): Ignore any characters after the first
22388 (arm_parse_arch_cpu_feature): New function.
22389 (arm_configure_build_target): Separate out any CPU and architecture
22390 features and parse separately. Don't error out if -mfpu=auto is
22391 used with only an architecture string.
22392 (arm_print_asm_arch_directives): New function.
22393 (arm_file_start): Call it.
22394 * config/arm/arm-cpu-cdata.h: Regenerated.
22395 * config/arm/arm-cpu-data.h: Likewise.
22396 * config/arm/arm-tables.opt: Likewise.
22398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22400 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22401 assembler when it is not -mfpu=auto.
22403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22405 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22406 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22407 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22408 (ASM_CPU_SPEC): Rewrite.
22409 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22410 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22411 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22412 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22413 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22414 copied string is NUL-terminated. Also strip any characters prefixed
22416 (arm_rewrite_selected_arch): New function.
22417 (arm_rewrite_march): New function.
22419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22421 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22422 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22423 (march, mcpu, mtune): Convert to string-based options.
22424 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22425 (arm_parse_arch_cpu_name): New function.
22426 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22427 identify selected architecture or CPU.
22428 (arm_option_save): New function.
22429 (TARGET_OPTION_SAVE): Redefine.
22430 (arm_option_restore): Restore string options.
22431 (arm_option_print): Print string options.
22433 2017-06-16 Martin Sebor <msebor@redhat.com>
22435 PR tree-optimization/80933
22436 PR tree-optimization/80934
22437 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22438 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22439 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22440 (gimple_fold_builtin): Call them.
22442 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22444 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22445 as unlikely; update profile.
22447 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22449 * predict.c (force_edge_cold): Handle declaring edges impossible
22452 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22454 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22456 (try_unroll_loop_completely): Fix reporting.
22458 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22460 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22462 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22465 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22466 if given a non-constant argument for an intrinsic which requires a
22469 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22471 * profile.c (compare_freqs): New function.
22472 (branch_prob): Sort edge list.
22473 (find_spanning_tree): Assume that the list is priority sorted.
22475 2017-06-16 Richard Biener <rguenther@suse.de>
22477 PR tree-optimization/81090
22478 * passes.def (pass_record_bounds): Remove.
22479 * tree-pass.h (make_pass_record_bounds): Likewise.
22480 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22481 make_pass_record_bounds): Likewise.
22482 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22483 not free niter estimates at the beginning but at the end.
22484 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22486 2017-06-16 Richard Biener <rguenther@suse.de>
22488 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22489 initializer to workaround ICE in host GCC 4.8.
22491 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22493 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22495 (clone_inlined_nodes): Update.
22497 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22499 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22500 prefetch settings, and enable prefetching by default at -O3.
22502 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22504 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22505 Set flag_prefetch_loop_arrays according to tuning data.
22507 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22509 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22510 New tune structure.
22511 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22512 [Unrelated to main purpose of the patch] Place the pointer field last
22513 to enable type checking errors when tune structure are wrongly merged.
22514 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22515 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22516 (thunderx2t99_prefetch_tune): New tune constants.
22517 (tune_params *_tunings): Update all tunings (no functional change).
22518 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22519 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22520 from tunings structures.
22522 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22525 * ubsan.c (instrument_null): Add T argument, use it instead
22526 of computing it based on IS_LHS.
22527 (instrument_object_size): Likewise.
22528 (pass_ubsan::execute): Adjust instrument_null and
22529 instrument_object_size callers to pass gimple_get_lhs or
22530 gimple_assign_rhs1 result to it. Use instrument_null instead of
22531 calling get_base_address and instrument_mem_ref. Handle
22532 aggregate call arguments for object-size sanitization.
22534 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22536 PR tree-optimization/81089
22537 * tree-vrp.c (is_masked_range_test): Validate operands of
22540 2017-06-15 Martin Sebor <msebor@redhat.com>
22543 * dumpfile.c (dump_register): Avoid calling memset to initialize
22544 a class with a default ctor.
22545 * gcc.c (struct compiler): Remove const qualification.
22546 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22547 * hash-table.h: Ditto.
22548 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22550 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22551 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22553 * params.h (struct param_info): Make struct members non-const.
22554 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22555 with default initialization.
22556 * vec.h (vec_copy_construct, vec_default_construct): New helper
22558 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22559 with vec_copy_construct.
22560 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22561 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22562 * doc/invoke.texi (-Wclass-memaccess): Document.
22564 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22566 * emit-rtl.h (is_leaf): Update comment about local
22567 register allocator.
22569 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22572 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22573 for a variable to have a section before checking if the section has a
22575 Set section to.persistent if persistent attribute is set.
22576 Warn if .persistent attribute is used on an automatic variable.
22578 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22580 PR rtl-optimization/80474
22581 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22583 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22585 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22588 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22590 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22591 (casesi): Emit bounds checking as RTL.
22592 (casesi_internal_mips16_<mode>): Remove bounds checking.
22594 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22596 * config/xtensa/xtensa.c (xtensa_option_override): Append
22597 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22598 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22599 xtensa_doloop_hooks): Define unconditionally.
22600 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22602 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22603 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22604 for it in xtensa_option_override.
22605 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22606 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22608 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22610 * doc/cppopts.texi: Document '-' special value to -MF.
22612 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22614 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22615 (cortex_a53_fconst): Likewise.
22616 (cortex_a53_fpmul): Likewise.
22617 (cortex_a53_f_load_64): Likewise.
22618 (cortex_a53_f_load_many): Likewise.
22619 (cortex_a53_advsimd_alu): Likewise.
22620 (cortex_a53_advsimd_alu_q): Likewise.
22621 (cortex_a53_advsimd_mul): Likewise.
22622 (cortex_a53_advsimd_mul_q): Likewise.
22623 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22624 Add missing fmul, r2f_cvt and fconst cases.
22626 2017-06-14 Richard Biener <rguenther@suse.de>
22628 PR middle-end/81088
22629 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22631 (fold_binary_loc): When associating do not treat pre-existing
22632 TREE_OVERFLOW on literal constants as a reason to allow
22633 TREE_OVERFLOW on associated literal constants.
22635 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22637 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22638 (MASK_FEATURES): New macro.
22639 * config/sparc/sparc.c (sparc_option_override): Remove the special
22640 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22642 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22644 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22645 a division of 0 if non-call exceptions are enabled.
22647 2017-06-14 Andrew Pinski <apinski@cavium.com>
22648 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22651 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22652 Improve vector initialization code gen for only variable case.
22654 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22656 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22658 2017-06-14 Richard Biener <rguenther@suse.de>
22660 PR tree-optimization/81083
22661 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22664 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22666 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22667 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22668 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22669 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22670 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22671 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22673 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22675 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22676 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22680 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22682 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22684 * config/rs6000/t-rtems: Don't handle SPE.
22686 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22688 * config/rs6000/t-linux: Don't handle SPE.
22690 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22692 * config/rs6000/eabispe.h: Delete file.
22694 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22696 * config/rs6000/t-spe: Delete file.
22698 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22700 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22701 (rs6000_legitimate_offset_address_p): Return false for anything in
22702 V2SImode or V2SFmode.
22704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22706 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22707 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22708 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22710 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22711 (rs6000_legitimate_offset_address_p): Ditto.
22712 (rs6000_emit_move): Ditto.
22713 (rs6000_init_builtins): Remove V4HI_type_node.
22715 2017-06-13 Martin Liska <mliska@suse.cz>
22718 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22719 (gate_asan): Likewise.
22720 * asan.h (asan_no_sanitize_address_p): Remove the function.
22721 (sanitize_flags_p): New function.
22722 * builtins.def: Fix coding style.
22723 * common.opt: Use renamed enum value.
22724 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22725 * doc/extend.texi: Document no_sanitize attribute.
22726 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22727 to SANITIZE_UNDEFINED_NONDEFAULT.
22728 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22729 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22730 Use sanitize_flags_p.
22731 * gimplify.c (gimplify_function_tree): Likewise.
22732 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22733 * opts.c (parse_no_sanitize_attribute): New function.
22734 (common_handle_option): Use renamed enum value.
22735 * opts.h (parse_no_sanitize_attribute): Declare.
22736 * tree.c (sanitize_flags_p): New function.
22737 * tree.h: Declared here.
22738 * tsan.c: Use sanitize_flags_p.
22739 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22740 (instrument_mem_ref): Likewise.
22741 (instrument_bool_enum_load): Likewise.
22742 (do_ubsan_in_current_function): Remove the function.
22743 (pass_ubsan::execute): Use sanitize_flags_p.
22744 * ubsan.h: Remove do_ubsan_in_current_function
22745 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22746 (dump_function_to_file): Use it here.
22748 2017-06-13 Martin Jambor <mjambor@suse.cz>
22750 PR tree-optimization/80803
22751 PR tree-optimization/81063
22752 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22753 (propagate_subaccesses_across_link): Enqueue subtree whenever
22754 necessary instead of relying on the caller.
22756 2017-06-13 Martin Jambor <mjambor@suse.cz>
22758 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22759 that have a first_link.
22760 (sort_and_splice_var_accesses): Do not check first_link before
22762 (subtree_mark_written_and_enqueue): Likewise.
22763 (propagate_all_subaccesses): Likewise and do not stop at first
22764 parent with a first_link.
22766 2017-06-13 Martin Jambor <mjambor@suse.cz>
22768 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22771 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22773 * match.pd: New pattern.
22775 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22777 * tree-vrp.c (is_masked_range_test): New function.
22778 (register_edge_assert_for): Determine ranges for
22781 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22783 PR tree-optimization/67328
22784 * fold-const.c (maskable_range_p): New function.
22785 (build_range_check): Generate bittests if possible.
22787 2017-06-13 Martin Liska <mliska@suse.cz>
22789 * gimple-pretty-print.c (dump_probability): Add new argument.
22790 (dump_edge_probability): Dump both probability and count.
22791 (dump_gimple_label): Likewise.
22792 (dump_gimple_bb_header): Likewise.
22794 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22797 * config/avr/avr-devices.c: Fix indentation.
22798 * config/avr/gen-avr-mmcu-specs.c: Dito.
22800 2017-06-13 Richard Biener <rguenther@suse.de>
22802 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
22803 instead get vector type from stmt_info.
22804 (vectorizable_reduction): Adjust. Remove dead code.
22806 2017-06-13 Richard Biener <rguenther@suse.de>
22808 PR middle-end/81065
22809 * fold-const.c (extract_muldiv_1): Remove bogus distribution
22810 case of C * (x * C2 + C3).
22811 (fold_addr_of_array_ref_difference): Properly fold index difference.
22813 2017-06-12 David S. Miller <davem@davemloft.net>
22816 * config/sparc/sparc.md (return expander): Emit frame blockage if
22817 function uses alloca.
22819 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
22821 * combine.c (make_field_assignment): Check len rather than the mode
22822 precision when calling force_to_mode.
22824 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
22826 Support multilibs and devices that see flash in RAM address range.
22829 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
22830 (avr_mcu_t) <flash_pm_offset>: New field.
22831 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22832 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22833 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22834 (AVR_TINY_PM_OFFSET): Remove macro.
22835 * config/avr/avr.opt (-mshort-calls): New option.
22836 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22837 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22838 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22839 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22840 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22841 instead of avr_arch->have_jmp_call.
22842 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22843 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22844 avr_arch->flash_pm_offset to define.
22845 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22846 new field flash_pm_offset. Add entry for avrxmega3.
22847 (avr_texinfo): Add entry for avrxmega3.
22848 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22849 attiny212, attiny214,
22850 attiny412, attiny414, attiny416, attiny417,
22851 attiny814, attiny816, attiny817,
22852 attiny1614, attiny1616, attiny1617,
22853 attiny3214, attiny3216, attiny3217.
22854 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22855 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22856 (avr_print_operand_address) [AVR_TINY]: Same.
22857 (avr_asm_init_sections) <readonly_data_section>: Only patch
22858 callback if avr_arch->flash_pm_offset = 0.
22859 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22860 for rodata if avr_arch->flash_pm_offset != 0.
22861 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22862 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22863 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22864 (m_options): Append opt_rcall.
22865 (m_dirnames): Append dir_rcall.
22866 * config/avr/t-multilib: Regenerate.
22868 * configure.ac [target=avr]: Check whether avrxmega3 default
22869 linker description file works as needed.
22870 * configure: Regenerate.
22871 * doc/avr-mmcu.texi: Regenerate.
22872 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22873 <__AVR_ARCH__>: Document avrxmega3 and 103.
22874 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22875 <__AVR_SHORT_CALLS__>: Document it.
22876 <__AVR_PM_BASE_ADDRESS__>: Document it.
22877 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22878 (AVR Variable Attributes) <progmem>: Document this is
22879 not needed for avrxmega3.
22880 (AVR Named Address Spaces) <__flash>: Dito.
22882 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22884 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22886 2017-06-12 Doug Rupp <rupp@adacore.com>
22888 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22889 Append vxworks-stdint.h to the tm_file list.
22890 * config/vxworks-stdint.h: New file.
22892 2017-06-12 Martin Liska <mliska@suse.cz>
22894 PR tree-optimization/81041
22895 * tree-profile.c (gimple_gen_ic_func_profiler):
22896 Create an extra BB in profile-generate
22897 (gimple_gen_time_profiler): Likewise.
22899 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22901 PR tree-optimization/81003
22902 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22903 (update_range_test): Use it instead of force_gimple_operand_gsi.
22905 2017-06-12 Richard Biener <rguenther@suse.de>
22907 PR tree-optimization/81053
22908 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22909 with backedge value not defined in loop. Simplify def stmt
22912 2017-06-11 Tom de Vries <tom@codesourcery.com>
22915 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22917 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22918 nvptx_cannot_force_const_mem.
22920 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22922 * opts.c (finish_options): Move test for flag_split_stack after
22923 it has been initialized.
22925 2017-06-11 Jason Merrill <jason@redhat.com>
22927 * tree.h (id_equal): New.
22928 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22929 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22930 instead of strcmp of IDENTIFIER_POINTER.
22932 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22934 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22935 (mark_all_inlined_calls_cdtor): Fix formating.
22936 (inline_transform): Rescale profile before inlining.
22938 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22940 * cgraph.h (cgraph_edge::clone): Update prototype.
22941 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22942 (cgraph_node::create_clone): Update.
22943 (cgraph_node::create_version_clone): Update.
22944 * tree-inline.c (copy_bb): Update.
22945 (expand_call_inline): Update.
22947 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22949 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22950 factored out from ...
22951 (rs6000_emit_prologue): ... here.
22953 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22955 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22956 factored out from ...
22957 (rs6000_emit_prologue): ... here.
22959 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22961 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22963 (handle_missing_profiles): Fix computation of tp_first_run.
22964 (counts_to_freqs): Do not touch freqs when count is 0.
22966 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22968 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22971 2017-06-10 Tom de Vries <tom@codesourcery.com>
22973 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22974 attributes): Document signal effective target.
22976 2017-06-10 Tom de Vries <tom@codesourcery.com>
22978 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22979 Document effective target stack_size.
22981 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22983 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22984 to the edit_context if they can be auto-applied.
22986 2017-06-9 Ian Lance Taylor <iant@golang.org>
22988 * opts.c (finish_options): If -fsplit-stack, disable implicit
22989 -forder-blocks-and-partition.
22990 * doc/invoke.texi (Optimize Options): Document that when using
22991 -fsplit-stack -forder-blocks-and-partition is not implicitly
22994 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22996 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22997 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22998 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22999 * builtins.def (abort, trap, unreachable): Declare cold.
23000 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23001 * tree-core.h (ECF_COLD): New.
23002 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23003 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23005 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23007 * predict.c (unlikely_executed_stmt_p): Cleanup.
23009 2017-06-09 Richard Biener <rguenther@suse.de>
23011 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23012 model if the ref is always written to.
23014 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23016 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23018 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23020 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23023 2017-06-09 Tom de Vries <tom@codesourcery.com>
23026 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23027 "target cannot support label values" when encountering LABEL_REF.
23029 2017-06-09 Martin Liska <mliska@suse.cz>
23031 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23032 (gimple_gen_ic_func_profiler): Emit direct comparison
23033 of __gcov_indirect_call_callee with NULL.
23034 (gimple_gen_time_profiler): Change probability from
23035 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23037 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23039 * profile.c (edge_gcov_counts): Turn to pointer.
23040 (compute_branch_probabilities, compute_branch_probabilities): Update.
23041 (branch_prob): Do not clear edge_gcov_count.
23042 * profile.h (edge_gcov_counts): Turn to pointer.
23043 (edge_gcov_count): Update.
23045 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23047 * gimple.h (gimple_check_failed): Mark cold.
23049 2017-06-09 Richard Biener <rguenther@suse.de>
23051 PR tree-optimization/66623
23052 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23053 refactor check_reduction into two parts, properly computing
23054 whether we have to check reduction validity for outer loop
23057 2017-06-09 Richard Biener <rguenther@suse.de>
23059 PR tree-optimization/79483
23060 * graphite-scop-detection.c (order): New global.
23061 (get_order): Compute bb to order mapping that satisfies code
23062 generation constraints.
23063 (cmp_pbbs): New helper.
23064 (build_scops): Start domwalk at entry block, sort generated
23067 2017-06-09 Richard Biener <rguenther@suse.de>
23069 PR middle-end/81007
23070 * ipa-polymorphic-call.c
23071 (ipa_polymorphic_call_context::restrict_to_inner_class):
23072 Skip FIELD_DECLs with error_mark_node type.
23073 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23076 2017-06-09 Martin Liska <mliska@suse.cz>
23078 * predict.c (struct branch_predictor): New struct.
23079 (test_prediction_value_range): New test.
23080 (predict_c_tests): New function.
23081 * selftest-run-tests.c (selftest::run_tests): Run the function.
23082 * selftest.h: Declare new tests.
23084 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23087 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23088 gen_add3_insn did not fail.
23089 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23090 r0, construct that number in a temporary reg and add that reg to r0.
23091 If asked to put the result in r0 as well, fail.
23093 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23096 for early expansion of vec_eqv.
23098 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23100 PR middle-end/81005
23101 * ubsan.c (instrument_null): Avoid pointless code temporary.
23102 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23104 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23108 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23111 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23114 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23116 * predict.c (unlikely_executed_edge_p): Move ahead.
23117 (probably_never_executed_edge_p): Use it.
23119 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23121 PR middle-end/79988
23122 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23123 gimple_call_builtin_p call.
23125 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23127 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23128 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23129 rtl_check_failed_type2, rtl_check_failed_code1,
23130 rtl_check_failed_code2, rtl_check_failed_code_mode,
23131 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23132 rtvec_check_failed_bounds, rtl_check_failed_flag,
23133 _fatal_insn_not_found, _fatal_insn): Likewise.
23134 * tree.h (tree_contains_struct_check_failed,
23135 tree_check_failed, tree_not_check_failed,
23136 tree_class_check_failed, tree_range_check_failed,
23137 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23138 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23139 tree_operand_check_failed, omp_clause_check_failed,
23140 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23143 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23145 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23146 flag_branch_probabilities.
23147 * ipa-inline.c (edge_badness): Likewise.
23148 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23149 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23150 * predict.c (maybe_hot_frequency_p): Likewise.
23151 (probably_never_executed): Likewise.
23152 * sched-ebb.c (schedule_ebbs): Likewise.
23153 * sched-rgn.c (find_single_block_region): Likewise.
23154 * tracer.c (tail_duplicate): Likewise.
23156 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23158 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23159 longer requires x_flag_profile_use.
23161 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23163 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23164 instead of flag_reorder_blocks_and_partition.
23165 * dbxout.c (dbxout_function_end): Likewise.
23166 * dwarf2out.c (gen_subprogram_die): Likewise.
23167 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23168 * hw-doloop.c (reorg_loops): Likewise.
23169 * varasm.c (assemble_start_function,
23170 assemble_end_function): Likewise.
23171 (decide_function_section): Do not check for
23172 flag_reorder_blocks_and_partition.
23174 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23176 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23178 (chkp_get_hard_register_fake_addr_expr): Ditto.
23179 (chkp_build_addr_expr): Add check for hard reg case.
23180 (chkp_parse_array_and_component_ref): Ditto.
23181 (chkp_find_bounds_1): Ditto.
23182 (chkp_process_stmt): Don't generate bounds store for
23185 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23187 * predict.c (maybe_hot_bb_p): Do not check profile status.
23188 (maybe_hot_edge_p): Likewise.
23189 (probably_never_executed): Check for zero counts even if profile
23191 (unlikely_executed_edge_p): New function.
23192 (unlikely_executed_stmt_p): New function.
23193 (unlikely_executed_bb_p): New function.
23194 (set_even_probabilities): Use unlikely predicates.
23195 (combine_predictions_for_bb): Likewise.
23196 (predict_paths_for_bb): Likewise.
23197 (predict_paths_leading_to_edge): Likewise.
23198 (determine_unlikely_bbs): New function.
23199 (estimate_bb_frequencies): Use it.
23200 (compute_function_frequency): Use zero counts even if profile is
23202 * profile-count.h: Fix typo.
23204 2017-08-08 Julia Koval <julia.koval@intel.com>
23206 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23207 _mm512_mask_cvtsepi16_storeu_epi8,
23208 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23209 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23210 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23211 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23212 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23213 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23214 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23215 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23216 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23217 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23218 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23219 __builtin_ia32_pmovuswb256mem_mask,
23220 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23221 __builtin_ia32_pmovwb512mem_mask): New builtins.
23223 2017-08-08 Julia Koval <julia.koval@intel.com>
23225 PR target/73350,80862
23226 * config/i386/subst.md (round): Fix round pattern.
23227 * config/i386/i386.c (ix86_erase_embedded_rounding):
23228 Fix erasing rounding for the fixed pattern.
23230 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23232 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23234 2017-06-08 Martin Liska <mliska@suse.cz>
23236 PR gcov-profile/80911
23237 * gcov.c (block_info::block_info): New constructor.
23239 2017-06-07 Carl Love <cel@us.ibm.com>
23241 * config/rs6000/rs6000-c: The return type of the following
23242 built-in functions was implemented as int not long long. Fix sign
23243 of return value for the unsigned version of vec_mulo and vec_mule.
23244 vector unsigned long long vec_bperm (vector unsigned long long,
23245 vector unsigned char)
23246 vector signed long long vec_mule (vector signed int,
23248 vector unsigned long long vec_mule (vector unsigned int,
23249 vector unsigned int)
23250 vector signed long long vec_mulo (vector signed int,
23252 vector unsigned long long vec_mulo (vector unsigned int,
23253 vector unsigned int)
23254 * doc/extend.texi: Fix the documentation for the built-in
23257 2017-06-07 Carl Love <cel@us.ibm.com>
23260 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23263 2017-06-07 Carl Love <cel@us.ibm.com>
23265 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23266 support, Generate doublehv for signed int/float for BE case only.
23268 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23270 * doc/invoke.texi (mcx16): Rewrite.
23272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23274 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23275 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23276 *mov<mode>_softfloat, and an anonymous splitter): Use
23277 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23279 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23281 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23283 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23284 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23285 (rs6000_debug_reg_global): Adjust.
23286 (rs6000_init_hard_regno_mode_ok): Adjust.
23287 (rs6000_dbx_register_number): Adjust.
23288 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23289 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23290 Remove SPE_ACC and SPEFSCR.
23291 (REG_ALLOC_ORDER): Ditto.
23292 (FRAME_POINTER_REGNUM): Change to 111.
23293 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23294 (REG_CLASS_NAMES): Ditto.
23295 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23296 (REGISTER_NAMES): Ditto.
23297 (ADDITIONAL_REG_NAMES): Ditto.
23298 (rs6000_reg_names): Ditto.
23299 * config/rs6000/rs6000.md: Renumber some register number
23302 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23304 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23306 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23307 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23309 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23310 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23311 Delete the SPE high registers.
23312 (REG_ALLOC_ORDER): Ditto.
23313 (enum reg_class): Remove SPE_HIGH_REGS.
23314 (REG_CLASS_NAMES): Ditto.
23315 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23316 (REGISTER_NAMES): Ditto.
23317 (rs6000_reg_names): Ditto.
23318 * doc/tm.texi.in: Remove SPE as example.
23319 * doc/tm.texi: Regenerate.
23321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23323 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23324 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23325 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23326 * config/rs6000/rs6000.md (type): Remove "brinc".
23328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23330 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23331 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23332 * config/rs6000/linuxspe.h: Delete file.
23333 * config/rs6000/rs6000.md: Don't include spe.md.
23334 * config/rs6000/spe.h: Delete file.
23335 * config/rs6000/spe.md: Delete file.
23336 * config/rs6000/t-rs6000: Remove spe.md.
23338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23340 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23341 (reg_or_none500mem_operand): Delete.
23342 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23343 instead of reg_or_none500mem_operand.
23345 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23348 handling of SPE flags.
23349 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23351 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23353 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23355 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23356 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23357 paired_divv2sf3): Similar.
23358 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23359 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23360 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23362 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23363 Rename the paired_* instruction patterns.
23364 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23366 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23367 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23368 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23369 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23370 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23371 PAIRED_VECTOR_MODE.
23372 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23373 (spe_func_has_64bit_regs_p): Delete.
23374 (spe_expand_predicate_builtin): Delete.
23375 (spe_expand_evsel_builtin): Delete.
23376 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23377 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23378 (invalid_e500_subreg): Delete.
23379 (rs6000_legitimize_address): Always force_reg op2 as well, for
23380 paired single memory accesses.
23381 (rs6000_member_type_forces_blk): Delete.
23382 (rs6000_spe_function_arg): Delete.
23383 (rs6000_expand_unop_builtin): Delete SPE handling.
23384 (rs6000_expand_binop_builtin): Ditto.
23385 (spe_expand_stv_builtin): Delete.
23386 (bdesc_2arg_spe): Delete.
23387 (spe_expand_builtin): Delete.
23388 (spe_expand_predicate_builtin): Delete.
23389 (spe_expand_evsel_builtin): Delete.
23390 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23391 (spe_init_builtins): Delete.
23392 (spe_func_has_64bit_regs_p): Delete.
23393 (savres_routine_name): Delete "info" parameter. Adjust callers.
23394 (rs6000_emit_stack_reset): Ditto.
23395 (rs6000_dwarf_register_span): Delete.
23396 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23397 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23398 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23400 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23402 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23403 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23404 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23405 mulv2sf3, divv2sf3): Delete expanders.
23407 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23409 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23413 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23414 * config/rs6000/rs6000.c: Ditto.
23416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23418 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23419 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23420 comparison_operator.
23422 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23424 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23425 * config/rs6000/rs6000.opt: Ditto.
23426 * config/rs6000/t-rtems: Ditto.
23428 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23430 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23431 TARGET_E500_SINGLE by 0, simplify.
23432 * config/rs6000/rs6000.c: Ditto.
23433 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23434 (spe_build_register_parallel): Delete.
23435 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23436 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23437 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23438 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23439 (E500_CONVERT): Delete.
23440 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23442 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23444 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23445 * config/rs6000/dfp.md: Ditto.
23446 (negdd2, *negdd2_fpr): Merge.
23447 (absdd2, *absdd2_fpr): Merge.
23448 (negtd2, *negtd2_fpr): Merge.
23449 (abstd2, *abstd2_fpr): Merge.
23450 * config/rs6000/e500.h: Delete file.
23451 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23452 TARGET_FPRS by 1 and simplify.
23453 * config/rs6000/rs6000-c.c: Ditto.
23454 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23455 TARGET_DF_SPE by 0.
23456 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23458 * config/rs6000/rs6000.md: Ditto.
23459 (floatdidf2, *floatdidf2_fpr): Merge.
23460 (move_from_CR_gt_bit): Delete.
23461 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23462 (E500_CR_IOR_COMPARE): Delete.
23463 (All patterns that require !TARGET_FPRS): Delete.
23464 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23466 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23468 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23470 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23472 * graphds.c (add_edge): Intitialize edge's attached data.
23473 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23474 pointer parameter. Call pointed function on each edge during
23475 graph traversing. Skip traversing the edge when the function
23477 (graphds_dfs, graphds_scc): Ditto.
23478 (for_each_edge): New parameter. Pass the new parameter to callback
23480 * graphds.h (skip_edge_callback): New function pointer type.
23481 (graphds_dfs, graphds_scc): New function pointer parameter.
23482 (graphds_edge_callback, for_each_edge): New parameter.
23484 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23486 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23487 out code checking if runtime alias check is possible to below ...
23488 Call the new function.
23489 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23490 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23492 2017-06-07 Marek Polacek <polacek@redhat.com>
23495 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23496 TYPE_OVERFLOW_WRAPS check.
23498 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23500 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23501 if versioning is required.
23502 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23503 peeling with the check for versioning.
23505 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23507 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23508 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23509 Set true to new parameter if new ssa variable is defined.
23510 (vect_gen_vector_loop_niters): Refactor. Set range information
23511 for the new vector loop bound variable.
23512 (vect_do_peeling): Ditto.
23514 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23516 * tree-affine.c (ssa.h): Include header file.
23517 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23518 has wrapping overflow behavior.
23520 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23522 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23524 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23526 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23527 (tree_to_aff_combination): ... here.
23529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23531 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23532 reg_pressure model function.
23533 (ivopts_global_cost_for_size): Delete.
23534 (determine_set_costs, iv_ca_recount_cost): Call new model function
23535 ivopts_estimate_reg_pressure.
23537 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23539 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23540 expensive than udiv. Remove floating point cases from mod.
23542 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23544 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23545 Increase idiv cost.
23547 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23549 * config/aarch64/aarch64.md
23550 (copysignsf3): Fix mask generation.
23552 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23554 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23556 (class dump_manager): Add register_dumps method.
23557 * dumpfile.c: Include langhooks.h.
23558 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23559 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23560 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23561 (dump_manager::dump_register): Start with 512 entries instead of 32.
23562 (dump_manager::register_dumps): New method.
23563 * toplev.c (general_init): Instead of invoking register_dumps
23564 langhook, invoke register_dumps method on the dump manager.
23565 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23568 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23570 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23571 Say that # requires an associated define_split to exist, and that
23572 the define_split must be suitable for use after register allocation.
23574 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23576 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23577 (compute_outgoing_frequencies): Also initialize zero counts.
23578 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23579 around loops; preserve more of profile when nothing changes.
23581 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23583 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23585 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23586 * config/arm/arm-cpu-cdata.h: Regenerate.
23587 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23588 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23589 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23590 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23591 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23593 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23595 * config/arm/t-rmprofile: Likewise.
23596 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23598 2017-06-06 David S. Miller <davem@davemloft.net>
23601 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23602 blockage if function uses alloca.
23604 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23606 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23607 New "uid" fields to hold pretty-print IDs of group and ref.
23608 Memory references are now identified as <group_id>:<ref_id>
23609 instead of using [random] addresses.
23610 (dump_mem_details): Simplify, no functional change.
23611 (dump_mem_ref): Simplify and make output more concise.
23612 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23613 (find_or_create_group): Initialize group uid.
23614 (record_ref): Initialize ref uid. Improve debug output.
23615 (prune_group_by_reuse, should_issue_prefetch_p,)
23616 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23617 (mark_nontemporal_store, determine_loop_nest_reuse):
23618 Improve debug output.
23620 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23622 * dbgcnt.def (prefetch): New debug counter.
23623 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23624 (schedule_prefetches): Stop issueing prefetches if debug counter
23627 2017-06-06 Tom de Vries <tom@codesourcery.com>
23629 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23630 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23632 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23634 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23635 Use aarch64_reg_or_zero predicate for operand 4.
23636 (aarch64_compare_and_swap<mode> define_insn_and_split):
23637 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23638 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23640 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23642 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23643 (arm_compute_save_core_reg_mask): This.
23644 (thumb1_compute_save_reg_mask): Rename into ...
23645 (thumb1_compute_save_core_reg_mask): This.
23646 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23647 (arm_compute_frame_layout): Likewise.
23649 2017-06-06 Richard Biener <rguenther@suse.de>
23651 PR tree-optimization/80974
23652 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23653 keep or clear leaders SSA info.
23655 2017-06-06 Tom de Vries <tom@codesourcery.com>
23657 * config/nvptx/nvptx.c (split_mode_p): New function.
23658 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23660 2017-06-06 Tom de Vries <tom@codesourcery.com>
23662 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23664 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23667 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23670 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23672 * shrink-wrap.c (handle_simple_exit): Update profile.
23673 (try_shrink_wrapping): Upate profile.
23675 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23677 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23678 (tree_guess_outgoing_edge_probabilities): New.
23679 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23680 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23682 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23684 * ipa-split.c (split_function): Initialize return bb profile.
23686 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23688 * profile.c (compute_branch_probabilities): Also initialize
23689 EXIT_BLOCK profile.
23691 2017-06-06 Richard Biener <rguenther@suse.de>
23693 PR tree-optimization/80928
23694 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23695 (vect_analyze_loop_operations): Properly guard analysis for
23697 (vect_transform_loop): Likewise.
23698 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23699 (vect_model_induction_cost): Do not cost for pure SLP.
23700 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23701 of induction in inner loop vectorization.
23702 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23703 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23704 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23706 (vect_analyze_slp_cost_1): Cost induction.
23707 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23708 (vect_get_slp_vect_defs): Likewise.
23709 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23710 (vect_transform_stmt): Handle SLP reductions.
23711 * tree-vectorizer.h (vectorizable_induction): Adjust.
23713 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23715 * config/rs6000/rs6000.c (make_resolver_func): Update
23716 init_lowered_empty_function call.
23718 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23720 * doc/invoke.texi: Document the -fprofile-abs-path option.
23721 * common.opt (fprofile-abs-path): New option.
23722 * gcov-io.h (gcov_write_filename): Declare.
23723 * gcov-io.c (gcov_write_filename): New function.
23724 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23725 * profile.c (output_location): Likewise.
23727 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23729 * shring-wrap.c: Revert accidental commit.
23731 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23733 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23735 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23737 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23739 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23740 profile in callgraph edge.
23741 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23742 (apply_scale): Likewise.
23743 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23744 Also scale profile when inlining function with zero profile.
23745 (initialize_cfun): Update exit block profile even when it is zero.
23746 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23747 when profile is read.
23749 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23751 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23752 (CLONE_*): New constants to define the processors we can generate
23753 code for with the target_clone attribute.
23754 (rs6000_clone_map): New array to identify which clone processors
23755 the current program is running on.
23756 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23757 target_clone attribute.
23758 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23759 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23760 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23761 (cpu_expand_builtin): Add support for target_clone attribute.
23762 (rs6000_valid_attribute_p): Allow "default" attribute.
23763 (get_decl_name): New debug function to simplify printing the
23764 current function name in debugging statements.
23765 (rs6000_clone_priority): New functions to support the target_clone
23766 attribute, and be able to generate code to switch between ISA 2.05
23767 through ISA 3.0 (power6 through power9).
23768 (rs6000_compare_version_priority): Likewise.
23769 (rs6000_get_function_versions_dispatcher): Likewise.
23770 (make_resolver_func): Likewise.
23771 (add_condition_to_bb): Likewise.
23772 (dispatch_function_versions): Likewise.
23773 (rs6000_generate_version_dispatcher_body): Likewise.
23774 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23775 (fusion_gpr_load_p): Fix a spacing issue.
23776 * doc/extend.texi (Common Function Attributes): Document that the
23777 PowerPC supports the target_clone attribute.
23779 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23781 * config/arm/arm.h: explain F symbol found in description of ARM
23782 register allocation in its legend.
23784 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23786 * config/mips/frame-header-opt.c: Include profile-count.h.
23787 * config/riscv/riscv.c: Include profile-count.h
23789 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23791 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23793 (sm_set_flag_if_changed): Add bbs field.
23794 (execute_sm_if_changed_flag_set): Pass BBS.
23795 (execute_sm): Update.
23797 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23799 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
23802 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23804 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
23805 (peephole2): New peephole2 to emit the above.
23806 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
23808 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23810 * config/aarch64/aarch64.c (define_peephole2 above
23811 *sub_<shift>_<mode>): New peephole.
23813 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23815 * config/i386/i386.c (make_resolver_func): Update.
23816 * Makefile.in: Add profile-count.h and profile-count.o
23817 * auto-profile.c (afdo_indirect_call): Update to new API.
23818 (afdo_set_bb_count): Update.
23819 (afdo_propagate_edge): Update.
23820 (afdo_propagate_circuit): Update.
23821 (afdo_calculate_branch_prob): Update.
23822 (afdo_annotate_cfg): Update.
23823 * basic-block.h: Include profile-count.h
23824 (struct edge_def): Turn count to profile_count.
23825 (struct basic_block_def): Likewie.
23826 (REG_BR_PROB_BASE): Move to profile-count.h
23827 (RDIV): Move to profile-count.h
23828 * bb-reorder.c (max_entry_count): Turn to profile_count.
23829 (find_traces): Update.
23830 (rotate_loop):Update.
23831 (connect_traces):Update.
23832 (sanitize_hot_paths):Update.
23833 * bt-load.c (migrate_btr_defs): Update.
23834 * cfg.c (RDIV): Remove.
23835 (init_flow): Use alloc_block.
23836 (alloc_block): Uninitialize count.
23837 (unchecked_make_edge): Uninitialize count.
23838 (check_bb_profile): Update.
23839 (dump_edge_info): Update.
23840 (dump_bb_info): Update.
23841 (update_bb_profile_for_threading): Update.
23842 (scale_bbs_frequencies_int): Update.
23843 (scale_bbs_frequencies_gcov_type): Update.
23844 (scale_bbs_frequencies_profile_count): New.
23845 * cfg.h (update_bb_profile_for_threading): Update.
23846 (scale_bbs_frequencies_profile_count): Declare.
23847 * cfgbuild.c (compute_outgoing_frequencies): Update.
23848 (find_many_sub_basic_blocks): Update.
23849 * cfgcleanup.c (try_forward_edges): Update.
23850 (try_crossjump_to_edge): Update.
23851 * cfgexpand.c (expand_gimple_tailcall): Update.
23852 (construct_exit_block): Update.
23853 * cfghooks.c (verify_flow_info): Update.
23854 (dump_bb_for_graph): Update.
23855 (split_edge): Update.
23856 (make_forwarder_block): Update.
23857 (duplicate_block): Update.
23858 (account_profile_record): Update.
23859 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23860 (get_estimated_loop_iterations): Update.
23861 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23862 (single_likely_exit): Update.
23863 * cfgloopmanip.c (scale_loop_profile): Update.
23865 (set_zero_probability): Update.
23866 (lv_adjust_loop_entry_edge): Update.
23867 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23868 (purge_dead_edges): Update.
23869 (rtl_account_profile_record): Update.
23870 * cgraph.c (cgraph_node::create): Uninitialize count.
23871 (symbol_table::create_edge): Uninitialize count.
23872 (cgraph_update_edges_for_call_stmt_node): Update.
23873 (cgraph_edge::dump_edge_flags): Update.
23874 (cgraph_node::dump): Update.
23875 (cgraph_edge::maybe_hot_p): Update.
23876 * cgraph.h: Include profile-count.h
23877 (create_clone), create_edge, create_indirect_edge): Update.
23878 (cgraph_node): Turn count to profile_count.
23879 (cgraph_edge0: Likewise.
23880 (make_speculative, clone): Update.
23881 (create_edge): Update.
23882 (init_lowered_empty_function): Update.
23883 * cgraphclones.c (cgraph_edge::clone): Update.
23884 (duplicate_thunk_for_node): Update.
23885 (cgraph_node::create_clone): Update.
23886 * cgraphunit.c (cgraph_node::analyze): Update.
23887 (cgraph_node::expand_thunk): Update.
23888 * final.c (dump_basic_block_info): Update.
23889 * gimple-streamer-in.c (input_bb): Update.
23890 * gimple-streamer-out.c (output_bb): Update.
23891 * graphite.c (print_global_statistics): Update.
23892 (print_graphite_scop_statistics): Update.
23893 * hsa-brig.c: Include basic-block.h.
23894 * hsa-dump.c: Include basic-block.h.
23895 * hsa-gen.c (T sum_slice): Update.
23896 (convert_switch_statements):Update.
23897 * hsa-regalloc.c: Include basic-block.h.
23898 * ipa-chkp.c (chkp_produce_thunks): Update.
23899 * ipa-cp.c (struct caller_statistics): Update.
23900 (init_caller_stats): Update.
23901 (gather_caller_stats): Update.
23902 (ipcp_cloning_candidate_p): Update.
23903 (good_cloning_opportunity_p): Update.
23904 (get_info_about_necessary_edges): Update.
23905 (dump_profile_updates): Update.
23906 (update_profiling_info): Update.
23907 (update_specialized_profile): Update.
23908 (perhaps_add_new_callers): Update.
23909 (decide_about_value): Update.
23910 (ipa_cp_c_finalize): Update.
23911 * ipa-devirt.c (struct odr_type_warn_count): Update.
23912 (struct decl_warn_count): Update.
23913 (struct final_warning_record): Update.
23914 (possible_polymorphic_call_targets): Update.
23915 (ipa_devirt): Update.
23916 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23917 * ipa-icf.c (sem_function::merge): Update.
23918 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23919 * ipa-inline.c (compute_uninlined_call_time): Update.
23920 (compute_inlined_call_time): Update.
23921 (want_inline_small_function_p): Update.
23922 (want_inline_self_recursive_call_p): Update.
23923 (edge_badness): Update.
23924 (lookup_recursive_calls): Update.
23925 (recursive_inlining): Update.
23926 (inline_small_functions): Update.
23927 (dump_overall_stats): Update.
23928 (dump_inline_stats): Update.
23929 * ipa-profile.c (ipa_profile_generate_summary): Update.
23930 (ipa_propagate_frequency): Update.
23931 (ipa_profile): Update.
23932 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23933 * ipa-utils.c (ipa_merge_profiles): Update.
23934 * loop-doloop.c (doloop_modify): Update.
23935 * loop-unroll.c (report_unroll): Update.
23936 (unroll_loop_runtime_iterations): Update.
23937 * lto-cgraph.c (lto_output_edge): Update.
23938 (lto_output_node): Update.
23939 (input_node): Update.
23940 (input_edge): Update.
23941 (merge_profile_summaries): Update.
23942 * lto-streamer-in.c (input_cfg): Update.
23943 * lto-streamer-out.c (output_cfg): Update.
23944 * mcf.c (create_fixup_graph): Update.
23945 (adjust_cfg_counts): Update.
23946 (sum_edge_counts): Update.
23947 * modulo-sched.c (sms_schedule): Update.
23948 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23949 * predict.c (maybe_hot_count_p): Update.
23950 (probably_never_executed): Update.
23951 (dump_prediction): Update.
23952 (combine_predictions_for_bb): Update.
23953 (propagate_freq): Update.
23954 (handle_missing_profiles): Update.
23955 (counts_to_freqs): Update.
23956 (rebuild_frequencies): Update.
23957 (force_edge_cold): Update.
23958 * predict.h: Include profile-count.h
23959 (maybe_hot_count_p, counts_to_freqs): UPdate.
23960 * print-rtl-function.c: Do not include cfg.h
23961 * print-rtl.c: Include basic-block.h
23962 * profile-count.c: New file.
23963 * profile-count.h: New file.
23964 * profile.c (is_edge_inconsistent): Update.
23965 (correct_negative_edge_counts): Update.
23966 (is_inconsistent): Update.
23967 (set_bb_counts): Update.
23968 (read_profile_edge_counts): Update.
23969 (compute_frequency_overlap): Update.
23970 (compute_branch_probabilities): Update; Initialize and deinitialize
23972 (branch_prob): Update.
23973 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23974 (edge_gcov_count): New.
23975 (bb_gcov_count): New.
23976 * shrink-wrap.c (try_shrink_wrapping): Update.
23977 * tracer.c (better_p): Update.
23978 * trans-mem.c (expand_transaction): Update.
23979 (ipa_tm_insert_irr_call): Update.
23980 (ipa_tm_insert_gettmclone_call): Update.
23981 * tree-call-cdce.c: Update.
23982 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23983 (gimple_duplicate_sese_tail): Update.
23984 (gimple_account_profile_record): Update.
23985 (execute_fixup_cfg): Update.
23986 * tree-inline.c (copy_bb): Update.
23987 (copy_edges_for_bb): Update.
23988 (initialize_cfun): Update.
23989 (freqs_to_counts): Update.
23990 (copy_cfg_body): Update.
23991 (expand_call_inline): Update.
23992 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23993 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23994 (try_unroll_loop_completely): Update.
23995 (try_peel_loop): Update.
23996 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23997 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23998 * tree-ssa-loop-split.c (connect_loops): Update.
23999 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24000 * tree-ssa-reassoc.c (branch_fixup): Update.
24001 * tree-ssa-tail-merge.c (replace_block_by): Update.
24002 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24003 (compute_path_counts): Update.
24004 (update_profile): Update.
24005 (recompute_probabilities): Update.
24006 (update_joiner_offpath_counts): Update.
24007 (estimated_freqs_path): Update.
24008 (freqs_to_counts_path): Update.
24009 (clear_counts_path): Update.
24010 (ssa_fix_duplicate_block_edges): Update.
24011 (duplicate_thread_path): Update.
24012 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24013 (struct switch_conv_info): Update.
24014 * tree-tailcall.c (decrease_profile): Update.
24015 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24016 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24017 * value-prof.c (check_counter): Update.
24018 (gimple_divmod_fixed_value): Update.
24019 (gimple_mod_pow2): Update.
24020 (gimple_mod_subtract): Update.
24021 (gimple_ic_transform): Update.
24022 (gimple_stringop_fixed_value): Update.
24023 * value-prof.h (gimple_ic): Update.
24025 2017-06-02 Carl Love <cel@us.ibm.com>
24027 * config/rs6000/rs6000-c: Add support for built-in functions
24028 vector double vec_doublee (vector signed int);
24029 vector double vec_doublee (vector unsigned int);
24030 vector double vec_doublee (vector float);
24031 vector double vec_doubleh (vector signed int);
24032 vector double vec_doubleh (vector unsigned int);
24033 vector double vec_doubleh (vector float);
24034 vector double vec_doublel (vector signed int);
24035 vector double vec_doublel (vector unsigned int);
24036 vector double vec_doublel (vector float);
24037 vector double vec_doubleo (vector signed int);
24038 vector double vec_doubleo (vector unsigned int);
24039 vector double vec_doubleo (vector float);.
24040 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24041 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24043 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24044 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24045 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24047 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24048 vec_doublel, vec_doubleh.
24049 * doc/extend.texi: Update the built-in documentation file for the
24050 new built-in functions.
24052 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24055 * ipa-inline-analysis.c (free_growth_caches): Set
24056 edge_removal_hook_holder to NULL after removing it.
24058 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24060 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24061 comparision with zero.
24063 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24065 for early expansion of vec_min and vec_max builtins.
24066 (builtin_function_type): Add min/max unsigned variants to those
24067 identified as having unsigned arguments.
24069 2017-06-02 Olivier Hainque <hainque@adacore.com>
24071 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24073 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24075 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24076 Use VALL_F16 iterator rather than VALL.
24078 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24080 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24081 Emit CBNZ inside loop when doing a strong exchange and comparing
24082 against zero. Generate the CC flags after the loop.
24084 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24086 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24087 (dl_section_ref): New.
24088 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24089 On AIX, append an expression to subtract the size of the
24090 section length to dl_section_ref.
24092 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24095 for early expansion of vector absolute builtins.
24097 2017-06-02 Richard Biener <rguenther@suse.de>
24099 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24100 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24102 2017-06-02 Richard Biener <rguenther@suse.de>
24104 PR tree-optimization/80948
24105 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24106 stmt order as well.
24108 2017-06-02 Richard Biener <rguenther@suse.de>
24110 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24112 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24113 for inductions as relevant.
24115 2017-06-02 Richard Biener <rguenther@suse.de>
24117 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24118 (vectorizable_induction): ... this. Remove dead code.
24120 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24122 * builtins. (expand_builtin_alloca): Remove second parameter and
24123 infer its value from the first parameter instead.
24124 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24126 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24128 PR rtl-optimization/80903
24129 * loop-doloop.c (add_test): Unshare sequence.
24131 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24133 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24135 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24137 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24139 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24140 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24141 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24142 (xlogue_layout::get_stub_rtx): Make static.
24143 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24144 (xlogue_layout::compute_stub_managed_regs): Rename to...
24145 (xlogue_layout::count_stub_managed_regs): ...this.
24146 (xlogue_layout::is_stub_managed_reg): New function.
24147 (xlogue_layout::m_stub_names): Rename to...
24148 (xlogue_layout::s_stub_names): ...this, make static.
24149 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24150 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24151 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24152 xlogue_layout::s_stub_names): Instantiate statics.
24153 (stub_managed_regs): Remove.
24154 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24155 (disable_call_ms2sysv_xlogues): Rename to...
24156 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24157 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24159 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24160 change after reload_completed.
24161 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24163 (ix86_expand_prologue): Likewise.
24164 (ix86_expand_epilogue): Likewise.
24165 (ix86_expand_split_stack_prologue): Likewise.
24166 (ix86_compute_frame_layout): Remove frame parameter ...
24167 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24168 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24170 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24171 (ix86_frame): Move from here ...
24172 * config/i386/i386.h (ix86_frame): ... to here.
24173 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24174 complete ix86_frame data structure instead. Remove some_ld_name.
24176 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24178 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24179 symbols that hold a DECL_VALUE_EXPR.
24181 2017-06-01 Martin Jambor <mjambor@suse.cz>
24183 PR tree-optimization/80898
24184 * tree-sra.c (process_subtree_disqualification): Removed.
24185 (disqualify_candidate): Do not acll
24186 process_subtree_disqualification.
24187 (subtree_mark_written_and_enqueue): New function.
24188 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24189 RHS has been disqualified and re-queue LHS if necessary. Apart
24190 from that, ignore disqualified RHS.
24192 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24194 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24195 address fetch for z10 or later.
24197 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24199 * config/arc/arc.md (tst_movb): Add guard when splitting.
24201 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24203 * config/arc/arc.c (arc_can_eliminate): Test against
24204 arc_frame_pointer_needed.
24206 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24208 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24209 to prevent store reordering.
24210 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24211 (type): Add block type.
24212 (stack_tie): Define special instruction to be used in
24215 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24217 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24218 constraint. It is not valid for the pattern.
24219 (noncommutative_binary_comparison): Likewise.
24221 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24223 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24226 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24228 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24229 be used by the reg-alloc.
24231 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24233 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24234 reg-alloc when having mul64 or mul32x16 instructions.
24235 (mulsidi3): Likewise.
24236 (umulsidi3): Likewise.
24237 (mulsi32x16): New pattern.
24238 (mulsi64): Likewise.
24239 (mulsidi64): Likewise.
24240 (umulsidi64): Likewise.
24241 (MUL32x16_REG): Define.
24242 (mul64_600): Use MUL32x16_REG.
24243 (mac64_600): Likewise.
24244 (umul64_600): Likewise.
24245 (umac64_600): Likewise.
24247 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24249 * config/arc/arc.md (mulsi3_700): Make it commutative.
24251 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24253 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24255 (*sign_extendsidi2_insn): Likewise for movstosw.
24257 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24259 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24260 the type of the input discriminant value. Convert the
24261 discriminant value of signedness vary.
24263 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24265 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24266 Add to -Wall section.
24268 2017-06-01 Richard Biener <rguenther@suse.de>
24270 PR middle-end/66313
24271 * fold-const.c (fold_plusminus_mult_expr): If the factored
24272 factor may be zero use a wrapping type for the inner operation.
24273 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24274 and handle moved defs.
24275 (process_assignment): Properly guard the unary op case. Return a
24276 tri-state indicating that moving the stmt before the call may allow
24277 to continue. Pass through to_move.
24278 (find_tail_calls): Handle moving unrelated defs before
24281 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24284 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24285 splitter result in the canonical way.
24287 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24289 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24290 also for 32bit target. Update insn attributes.
24291 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24293 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24295 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24296 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24297 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24298 (_mm_maskz_min_ss): New intrinsics.
24300 2017-05-31 Martin Liska <mliska@suse.cz>
24302 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24303 Change comment style to one we normally use.
24304 (vectorizable_reduction): Likewise.
24305 (vectorizable_induction): Likewise.
24306 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24307 (vectorizable_call): Likewise.
24308 (vectorizable_simd_clone_call): Likewise.
24309 (vectorizable_conversion): Likewise.
24310 (vectorizable_assignment): Likewise.
24311 (vectorizable_shift): Likewise.
24312 (vectorizable_operation): Likewise.
24313 (vectorizable_store): Likewise.
24314 (vectorizable_load): Likewise.
24315 * tree-vectorizer.h: Likewise.
24317 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24319 * passes.c (emergency_dump_function): New.
24320 * tree-pass.h (emergency_dump_function): Declare.
24321 * plugin.c (plugins_internal_error_function): Remove.
24322 * plugin.h (plugins_internal_error_function): Remove declaration.
24323 * toplev.c (internal_error_function): New static function. Use it...
24324 (general_init): ...here.
24326 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24328 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24329 (arc_rtx_costs): Add costs for new patterns.
24330 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24331 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24333 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24335 * tree-ssa-strlen.c (get_next_strinfo): New function.
24336 (get_stridx_plus_constant): Use it.
24337 (zero_length_string): Likewise.
24338 (adjust_related_strinfos): Likewise.
24339 (adjust_last_stmt): Likewise.
24341 2017-05-31 Richard Biener <rguenther@suse.de>
24344 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24345 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24347 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24349 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24350 loop_vinfo argument and use of dependence distance vectors.
24351 Check instead whether the two references differ only in their
24352 initial value and assume that they have the same alignment if the
24353 difference is a multiple of the vector alignment.
24354 (vect_analyze_data_refs_alignment): Update call accordingly.
24356 2017-05-31 Martin Liska <mliska@suse.cz>
24359 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24361 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24363 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24364 (create_intersect_range_checks): Move from ...
24365 * tree-data-ref.c (create_intersect_range_checks_index)
24366 (create_intersect_range_checks): ... to here.
24367 (create_runtime_alias_checks): New function factored from ...
24368 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24369 here. Call above function.
24370 * tree-data-ref.h (create_runtime_alias_checks): New function.
24372 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24374 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24375 segment length for dr_b and compute it in wide_int.
24377 2017-05-31 Richard Biener <rguenther@suse.de>
24379 PR tree-optimization/80906
24380 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24381 and pass through iv_map.
24382 (copy_bb_and_scalar_dependences): Adjust.
24383 (translate_pending_phi_nodes): Likewise.
24384 (copy_loop_close_phi_args): Handle code-generating IVs instead
24387 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24389 * diagnostic-color.c (color_dict): Add "type-diff".
24390 (parse_gcc_colors): Update comment.
24391 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24392 -fdiagnostics-show-template-tree and -fno-elide-type.
24393 (GCC_COLORS): Add type-diff to example.
24395 (-fdiagnostics-show-template-tree): New.
24396 (-fno-elide-type): New.
24397 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24398 the pp_format_decoder callback. Call any m_format_postprocessor's
24400 (pretty_printer::pretty_printer): Initialize
24401 m_format_postprocessor.
24402 (pretty_printer::~pretty_printer): Delete any
24403 m_format_postprocessor.
24404 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24405 (class format_postprocessor): New class.
24406 (struct pretty_printer::format_decoder): Document the new parameters.
24407 (struct pretty_printer::m_format_postprocessor): New field.
24408 * tree-diagnostic.c (default_tree_printer): Update for new
24409 bool and const char ** params.
24410 * tree-diagnostic.h (default_tree_printer): Likewise.
24412 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24414 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24415 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24417 rs6000_gen_cell_microcode code.
24418 (rs6000_final_prescan_insn): Delete.
24419 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24420 "warn-cell-microcode" entries.
24421 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24422 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24423 throughout. Change cc_reg_not_micro_cr0_operand to
24424 cc_reg_not_cr0_operand throughout.
24425 (*extendhi<mode>2_noload): Delete.
24426 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24427 (mwarn-cell-microcode): Delete.
24428 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24429 -mgen-cell-microcode and -mwarn-cell-microcode.
24431 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24434 * config/i386/constraints.md (Yd): New constraint.
24436 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24437 and (?Yd, r) alternatives. Update insn attributes.
24438 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24439 and (?*Yd, r) alternatives. Update insn attributes.
24440 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24442 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24444 * gimplify.c (gimplify_modify_expr): Don't create a
24445 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24448 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24450 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24452 2017-05-30 Richard Biener <rguenther@suse.de>
24454 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24455 and reduc_def fields.
24456 (STMT_VINFO_REDUC_TYPE): New define.
24457 (STMT_VINFO_REDUC_DEF): Likewise.
24458 (vect_force_simple_reduction): Adjust prototype.
24459 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24460 (vect_is_simple_reduction): Remove check_reduction argument.
24461 (vect_force_simple_reduction): Adjust and set
24462 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24463 (vectorizable_reduction): Do not re-do reduction analysis
24464 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24465 * tree-parloops.c (gather_scalar_reductions): Adjust.
24467 2017-05-30 Richard Biener <rguenther@suse.de>
24469 PR middle-end/80901
24470 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24473 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24475 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24476 Introduce unknown_misalignment parameter and remove vf.
24477 (vect_peeling_hash_get_lowest_cost):
24478 Pass unknown_misalignment parameter.
24479 (vect_enhance_data_refs_alignment):
24480 Fix unsupportable data ref treatment.
24482 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24484 * tree-vect-data-refs.c (vect_get_data_access_cost):
24485 Workaround for SLP handling.
24486 (vect_enhance_data_refs_alignment):
24487 Compute costs for doing no peeling at all, compare to the best
24488 peeling costs so far and avoid peeling if cheaper.
24490 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24492 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24493 Return peeling info and set costs to zero for unlimited cost
24495 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24496 with unknown misalignment. Compute and costs for unknown
24497 misalignment, compare them to the costs for known misalignment
24498 and choose the cheapest for peeling.
24500 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24502 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24503 (vect_get_peeling_costs_all_drs): Create function.
24504 (vect_peeling_hash_get_lowest_cost):
24505 Use vect_get_peeling_costs_all_drs.
24506 (vect_peeling_supportable): Create function.
24507 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24509 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24511 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24512 DR_HAS_NEGATIVE_STEP.
24513 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24514 (vect_enhance_data_refs_alignment): Use.
24515 (vect_duplicate_ssa_name_ptr_info): Use.
24516 * tree-vectorizer.h (dr_misalignment): Use.
24517 (known_alignment_for_access_p): Use.
24519 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24522 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24524 (has_section_name): New function.
24526 2017-05-30 Martin Liska <mliska@suse.cz>
24529 * auto-profile.c (get_function_decl_from_block): Fix
24532 2017-05-30 Richard Biener <rguenther@suse.de>
24534 PR middle-end/80876
24535 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24537 2017-05-30 Martin Liska <mliska@suse.cz>
24539 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24540 * dumpfile.h (struct dump_file_info): Remove ctors.
24542 2017-05-30 Martin Liska <mliska@suse.cz>
24544 * predict.def: Fix GNU coding style.
24546 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24548 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24549 Mark 'to' argument with ATTRIBUTE_UNUSED.
24551 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24553 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24554 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24555 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24558 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24560 * doc/install.texi (Options specification): Restore entry of
24561 --enable-sjlj-exceptions.
24563 2017-05-27 Michael Eager <eager@eagercon.com>
24566 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24568 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24570 * config/microblaze/microblaze.h
24571 (FIXED_REGISTERS): Update in macro.
24572 (CALL_USED_REGISTERS): Update in macro.
24574 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24576 * doc/install.texi: Add links to macOS binary distributions.
24578 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24582 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24584 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24586 2017-05-26 Martin Liska <mliska@suse.cz>
24588 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24590 2017-05-26 Martin Liska <mliska@suse.cz>
24592 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24593 always leading ';; '.
24594 (dump_bb_info): Likewise.
24595 (brief_dump_cfg): Likewise.
24596 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24597 * dumpfile.c: Remove usage of TDF_VERBOSE.
24598 * dumpfile.h (enum dump_kind): Likewise.
24599 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24600 * print-tree.c (debug_verbose): Remove.
24601 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24602 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24603 * tree-diagnostic.c (default_tree_printer): Replace
24604 TDF_DIAGNOSTIC with TDF_SLIM.
24606 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24608 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24609 in parameter loop, rather than loop_vinfo.
24610 (create_intersect_range_checks): Ditto.
24611 (vect_create_cond_for_alias_checks): Update call to above functions.
24613 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24615 PR tree-optimization/80815
24616 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24617 for merging runtime alias checks. Handle negative DR_STEPs.
24619 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24621 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24623 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24624 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24625 out code pruning runtime alias checks.
24626 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24627 factored out from above.
24628 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24630 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24632 (prune_runtime_alias_test_list): New decalaration.
24634 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24636 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24637 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24638 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24639 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24640 (operator==, comp_dr_with_seg_len_pair): Ditto.
24641 (vect_prune_runtime_alias_test_list): Ditto.
24643 2017-05-26 Martin Liska <mliska@suse.cz>
24646 * params.def: Bound partial-inlining-entry-probability param.
24648 2017-05-26 Marek Polacek <polacek@redhat.com>
24651 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24654 2017-05-26 Richard Biener <rguenther@suse.de>
24656 PR tree-optimization/80842
24657 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24660 2017-05-26 Richard Biener <rguenther@suse.de>
24662 PR tree-optimization/80844
24663 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24665 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24667 * doc/md.texi (Machine Constraints): Update x86 family
24668 machine constraints section to match 'config/i386/constraints.md'.
24670 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24672 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24674 2017-05-25 Nathan Sidwell <nathan@acm.org>
24676 * doc/invoke.texi (--enable-languages): Update documentation.
24678 2017-05-25 Martin Liska <mliska@suse.cz>
24680 * dumpfile.c: Add TDF_FOLDING.
24681 * dumpfile.h (enum dump_kind): Likewise.
24682 * genmatch.c (dt_simplify::gen_1): Use it.
24684 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24686 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24688 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24690 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24691 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24693 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24695 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24696 * match.pd (X == C): Rewrite it here.
24697 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24698 with_certain_nonzero_bits2): New predicates.
24699 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24701 2017-05-24 Nathan Sidwell <nathan@acm.org>
24703 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24706 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24709 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24711 * config/powerpcspe: New port. Files are copied from the rs6000
24712 port, with "rs6000" in filenames replaced by "powerpcspe".
24714 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24716 PR rtl-optimization/80754
24717 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24719 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24721 * config/sparc/sparc.md (length): Return the correct value for -mflat
24722 sibcalls to match output_sibcall.
24724 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24728 * config/rs6000/rs6000.c (struct machine_function): Add new field
24730 (rs6000_get_separate_components): Init that field, use it.
24731 (rs6000_components_for_bb): Use the field.
24733 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24735 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24737 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24739 PR middle-end/80823
24740 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24742 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24745 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24746 address against address_operand predicate.
24747 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24749 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24751 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24752 parameters passed indirectly.
24754 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24756 * config/i386/i386.md (*movdi_internal): Remove SSE4
24757 alternative 18 (?r, *v). Update insn attributes.
24758 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24759 Update insn attributes.
24760 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24761 Update insn attributes.
24762 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24763 alternative 1 (r, v). Remove isa attribute.
24764 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24765 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24766 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24768 2017-05-23 Tom de Vries <tom@codesourcery.com>
24770 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24773 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24775 * cgraphunit.c (symbol_table::process_new_functions): Update.
24776 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24777 (inline_generate_summary): Rename to ...
24778 (ipa_fn_summary_generate): ... this one.
24779 (inline_read_summary): Rename to ...
24780 (ipa_fn_summary_read): ... this one.
24781 (inline_write_summary): Rename to ...
24782 (ipa_fn_summary_write): ... this one.
24783 (inline_free_summary): Rename to ...
24784 (ipa_free_fn_summary): ... this one.
24785 (pass_data_local_fn_summary, pass_local_fn_summary,
24786 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24787 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24788 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24789 make_pass_ipa_fn_summary): New.
24790 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24791 inline_write_summary, inline_free_summary): Remove.
24792 (ipa_free_fn_summary) : New.
24793 * ipa-inline.c (ipa_inline): Update.
24794 (pass_ipa_inline): Do not generate summaries.
24795 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24797 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
24798 and add pass_ipa_fn_summary.
24799 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
24801 (make_pass_inline_parameters): Remove.
24803 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
24805 * omp-low.c (struct omp_context): Remove "default_kind" member.
24808 * omp-offload.c (execute_oacc_device_lower): Remove the
24809 parallelism dimensions function attributes for unparallelized
24810 OpenACC kernels constructs.
24812 2017-05-23 Martin Liska <mliska@suse.cz>
24814 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
24816 (cgraph_edge::make_speculative): Likewise.
24817 (cgraph_edge::resolve_speculation): Likewise.
24818 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24819 (cgraph_node::dump): Likewise.
24820 * cgraph.h: Likewise.
24821 * cgraphunit.c (analyze_functions): Likewise.
24822 (symbol_table::compile): Likewise.
24823 * ipa-cp.c (print_all_lattices): Likewise.
24824 (determine_versionability): Likewise.
24825 (initialize_node_lattices): Likewise.
24826 (ipcp_verify_propagated_values): Likewise.
24827 (estimate_local_effects): Likewise.
24828 (update_profiling_info): Likewise.
24829 (create_specialized_node): Likewise.
24830 (perhaps_add_new_callers): Likewise.
24831 (decide_about_value): Likewise.
24832 (decide_whether_version_node): Likewise.
24833 (identify_dead_nodes): Likewise.
24834 (ipcp_store_bits_results): Likewise.
24835 * ipa-devirt.c (dump_targets): Likewise.
24836 (ipa_devirt): Likewise.
24837 * ipa-icf.c (sem_item::dump): Likewise.
24838 (sem_function::equals): Likewise.
24839 (sem_variable::equals): Likewise.
24840 (sem_item_optimizer::read_section): Likewise.
24841 (sem_item_optimizer::execute): Likewise.
24842 (congruence_class::dump): Likewise.
24843 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24844 (dump_inline_summary): Likewise.
24845 (estimate_node_size_and_time): Likewise.
24846 (inline_analyze_function): Likewise.
24847 * ipa-inline-transform.c (inline_call): Likewise.
24848 * ipa-inline.c (report_inline_failed_reason): Likewise.
24849 (want_early_inline_function_p): Likewise.
24850 (edge_badness): Likewise.
24851 (update_edge_key): Likewise.
24852 (inline_small_functions): Likewise.
24853 * ipa-profile.c (ipa_profile): Likewise.
24854 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24855 (ipa_make_edge_direct_to_target): Likewise.
24856 (remove_described_reference): Likewise.
24857 (ipa_impossible_devirt_target): Likewise.
24858 (propagate_controlled_uses): Likewise.
24859 (ipa_print_node_params): Likewise.
24860 (ipcp_transform_function): Likewise.
24861 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24862 (propagate_pure_const): Likewise.
24863 * ipa-reference.c (generate_summary): Likewise.
24864 (read_write_all_from_decl): Likewise.
24865 (propagate): Likewise.
24866 (ipa_reference_read_optimization_summary): Likewise.
24867 * ipa-utils.c (ipa_merge_profiles): Likewise.
24868 * ipa.c (walk_polymorphic_call_targets): Likewise.
24869 (symbol_table::remove_unreachable_nodes): Likewise.
24870 (ipa_single_use): Likewise.
24871 * passes.c (execute_todo): Likewise.
24872 * predict.c (drop_profile): Likewise.
24873 * symtab.c (symtab_node::get_dump_name): New function.
24874 (symtab_node::dump_name): Likewise.
24875 (symtab_node::dump_asm_name): Likewise.
24876 (symtab_node::dump_references): Likewise.
24877 (symtab_node::dump_referring): Likewise.
24878 (symtab_node::dump_base): Likewise.
24879 (symtab_node::debug_symtab): Likewise.
24880 * tree-sra.c (convert_callers_for_node): Likewise.
24881 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24882 * value-prof.c (init_node_map): Likewise.
24884 2017-05-23 Martin Liska <mliska@suse.cz>
24886 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24887 and symtab_node::debug_symtab to symbol_table::debug.
24888 * cgraphunit.c (analyze_functions): Use the renamed function.
24889 (symbol_table::compile): Likewise.
24890 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24891 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24892 * passes.c (execute_todo): Likewise.
24893 * symtab.c (symbol_table::dump): New function.
24894 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24896 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24898 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24899 that nonconst implies exec.
24901 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24903 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24904 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24905 (inline_edge_summary_vec): Turn into ...
24906 (ipa_call_summaries): ... this one.
24907 (redirect_to_unreachable, edge_set_predicate,
24908 evaluate_properties_for_edge, inline_summary_alloc,
24909 reset_ipa_call_summary, reset_inline_summary,
24910 inline_summary_t::duplicate): Update.
24911 (inline_edge_duplication_hook): Turn to ...
24912 (ipa_call_summary_t::duplicate): ... this one.
24913 (inline_edge_removal_hook): Turn to ...
24914 (ipa_call_summary_t::remove): ... this one.
24915 (dump_inline_edge_summary): Turn to ...
24916 (dump_ipa_call_summary): ... this one.
24917 (estimate_function_body_sizes): Update.
24918 (inline_update_callee_summaries): Update.
24919 (remap_edge_change_prob): Update.
24920 (remap_edge_summaries): Update.
24921 (inline_merge_summary): Update.
24922 (do_estimate_edge_time): Update.
24923 (inline_generate_summary): Update.
24924 (inline_read_section): Update.
24925 (inline_read_summary): Update.
24926 (inline_free_summary): Update.
24927 * ipa-inline.c (can_inline_edge_p): Update.
24928 (compute_inlined_call_time): Update.
24929 (want_inline_small_function_p): Update.
24930 (edge_badness): Update.
24931 (early_inliner): Update.
24932 * ipa-inline.h (inline_edge_summary): Turn to ...
24933 (ipa_call_summary): ... this one.
24934 (ipa_call_summary_t): New class.
24935 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24936 (ipa_call_summaries): New.
24937 (inline_edge_summary): Remove.
24938 (estimate_edge_growth): Update.
24939 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24940 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24941 * ipa-split.c (execute_split_functions): Update.
24942 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24944 2017-05-23 Tom de Vries <tom@codesourcery.com>
24946 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24947 attributes): Document rdrand effective target.
24949 2017-05-23 Tom de Vries <tom@codesourcery.com>
24951 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24952 attributes): Sort alphabetically.
24954 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24956 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24958 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24961 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24962 V2DF/V2DI splat into two separate patterns, one that handles
24963 registers, and the other that only handles memory. Drop support
24964 for splatting from a GPR on ISA 2.07 and then splitting the
24965 splat into direct move and splat.
24966 (vsx_splat_<mode>_reg): Likewise.
24967 (vsx_splat_<mode>_mem): Likewise.
24969 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24971 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24973 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24975 PR middle-end/80809
24976 * omp-low.c (finish_taskreg_remap): New function.
24977 (finish_taskreg_scan): If unit size of ctx->record_type
24978 is non-constant, unshare the size expression and replace
24979 decls in it with possible outer var refs.
24981 PR middle-end/80809
24982 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24983 GOVD_SHARED rather than GOVD_PRIVATE with it.
24984 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24985 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24987 PR middle-end/80853
24988 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24989 as last argument to build_outer_var_ref for pointer bases of array
24990 section reductions.
24992 2017-05-19 Martin Sebor <msebor@redhat.com>
24994 * print-tree.c (print_node): Print DECL_READ_P flag.
24996 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24998 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24999 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25000 * cgraph.c: Likewise.
25001 * cgraphunit.c: Likewise.
25002 * gengtype.c: Likewise.
25003 * ipa-cp.c: Likewise.
25004 * ipa-devirt.c: Likewise.
25005 * ipa-icf.c: Likewise.
25006 * ipa-predicate.c: Likewise.
25007 * ipa-profile.c: Likewise.
25008 * ipa-prop.c: Likewise.
25009 * ipa-split.c: Likewise.
25011 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25012 edge_predicate_pool, dump_inline_hints,
25013 inline_summary::account_size_time, redirect_to_unreachable,
25014 edge_set_predicate, set_hint_predicate,
25015 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25016 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25017 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25018 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25019 ipa_call_summary_t::remove, initialize_growth_caches,
25020 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25021 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25022 mark_modified, unmodified_parm_1, unmodified_parm,
25023 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25024 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25025 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25026 will_be_nonconstant_predicate, record_modified_bb_info,
25027 get_minimal_bb, record_modified, param_change_prob,
25028 phi_result_unknown_predicate, predicate_for_phi_result,
25029 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25030 estimate_function_body_sizes, compute_inline_parameters,
25031 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25032 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25033 inline_update_callee_summaries, remap_edge_change_prob,
25034 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25035 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25036 inline_analyze_function, inline_summary_t::insert,
25037 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25038 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25039 inline_free_summary): Move to ipa-fnsummary.h
25040 (predicate_t): Remove.
25041 * ipa-fnsummary.c: New file.
25042 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25043 (enum inline_hints_vals, inline_hints, agg_position_info,
25044 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25045 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25046 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25047 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25048 inline_read_summary, inline_write_summary, inline_free_summary,
25049 inline_analyze_function, initialize_inline_failed,
25050 inline_merge_summary, inline_update_overall_summary,
25051 compute_inline_parameters): Move to ipa-fnsummary.h
25052 * ipa-fnsummary.h: New file.
25053 * ipa-inline-transform.h: Include ipa-inline.h.
25054 * ipa-inline.c: LIkewise.
25056 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25058 * ipa-inline.c (edge_badness): Use inlined_time instead of
25059 inline_summaries->get.
25061 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25063 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25065 2017-05-22 Nathan Sidwell <nathan@acm.org>
25067 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25068 (fdump-lang): Document 'raw' option.
25069 * dumpfile.h (TDI_tu): Delete.
25070 * dumpfile.c (dump_files): Remove translation-unit.
25071 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25073 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25075 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25076 command option from $(AWK) call.
25077 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25078 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25079 [FORMAT]: Remove handling of variable.
25080 * config/avr/t-multilib: Regenerate.
25082 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25084 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25086 (dump_inline_summary): Do not print self_time.
25087 (estimate_function_body_sizes): Do not set self_time.
25088 (compute_inline_parameters): Likewise.
25089 (inline_read_section, inline_write_summary): Do not stream self_time.
25090 * ipa-inline.h (inline_summary): Drop self_time.
25092 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25094 * ipa-inline-analysis.c (account_size_time): Rename to ...
25095 (inline_summary::account_size_time): ... this one.
25096 (reset_ipa_call_summary): Turn to ...
25097 (ipa_call_summary::reset): ... this one.
25098 (reset_inline_summary): Turn to ...
25099 (inline_summary::reset): ... this one.
25100 (inline_summary_t::remove): Update.
25101 (inline_summary_t::duplicate): Update.
25102 (ipa_call_summary_t::remove): Update.
25103 (dump_inline_summary): Update.
25104 (estimate_function_body_sizes): Update.
25105 (compute_inline_parameters): Update.
25106 (estimate_node_size_and_time): Update.
25107 (inline_merge_summary): Update.
25108 (inline_update_overall_summary): Update.
25109 (inline_read_section): Update.
25110 (inline_write_summary): Update.
25111 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25112 add account_size_time and reset member functions.
25113 (ipa_call_summary): Add reset function.
25114 * ipa-predicate.h (predicate::operator &): Constify.
25116 2017-05-22 Richard Biener <rguenther@suse.de>
25118 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25120 2017-05-19 Jason Merrill <jason@redhat.com>
25122 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25124 2017-05-19 Marek Polacek <polacek@redhat.com>
25127 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25128 TYPE_OVERFLOW_WRAPS checks.
25130 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25132 * tree-core.h (enum omp_clause_default_kind): Add
25133 "OMP_CLAUSE_DEFAULT_PRESENT".
25134 * tree-pretty-print.c (dump_omp_clause): Handle it.
25135 * gimplify.c (enum gimplify_omp_var_data): Add
25136 "GOVD_MAP_FORCE_PRESENT".
25137 (gimplify_adjust_omp_clauses_1): Map it to
25138 "GOMP_MAP_FORCE_PRESENT".
25139 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25141 * gimplify.c (oacc_default_clause): Clarify.
25143 2017-05-19 Nathan Sidwell <nathan@acm.org>
25145 LANG_HOOK_REGISTER_DUMPS
25146 * toplev.c (general_init): Call register dump lang hook.
25147 * doc/invoke.texi: Document -fdump-lang option family.
25148 * dumpfile.c (dump_files): Remove class dump here.
25149 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25150 * dumpfile.h (tree_dump_index): Remove TDI_class.
25151 * langhooks-def.h (lhd_register_dumps): Declare.
25152 (LANG_HOOKS_REGISTER_DUMPS): Define.
25153 (LANG_HOOKS_INITIALIZER): Add it.
25154 * langhooks.c (lhd_register_dumps): Define.
25155 * langhooks.h (struct lang_hooks): Add register_dumps.
25157 2017-05-19 Nathan Sidwell <nathan@acm.org>
25159 * context.h (context::set_passes): New.
25160 * context.c (context::context): Do not create pass manager.
25161 * toplev.c (general_init): Create pass manager here.
25163 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25165 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25166 use this splitter if two add or or instructions would also work for
25167 the constant we want to generate.
25169 2017-05-19 Richard Biener <rguenther@suse.de>
25172 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25173 predicate evaluation.
25175 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25177 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25179 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25182 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25184 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25185 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25186 (inline_edge_summary_vec): Turn into ...
25187 (ipa_call_summaries): ... this one.
25188 (redirect_to_unreachable, edge_set_predicate,
25189 evaluate_properties_for_edge, inline_summary_alloc,
25190 reset_ipa_call_summary, reset_inline_summary,
25191 inline_summary_t::duplicate): Update.
25192 (inline_edge_duplication_hook): Turn to ...
25193 (ipa_call_summary_t::duplicate): ... this one.
25194 (inline_edge_removal_hook): Turn to ...
25195 (ipa_call_summary_t::remove): ... this one.
25196 (dump_inline_edge_summary): Turn to ...
25197 (dump_ipa_call_summary): ... this one.
25198 (estimate_function_body_sizes): Update.
25199 (inline_update_callee_summaries): Update.
25200 (remap_edge_change_prob): Update.
25201 (remap_edge_summaries): Update.
25202 (inline_merge_summary): Update.
25203 (do_estimate_edge_time): Update.
25204 (inline_generate_summary): Update.
25205 (inline_read_section): Update.
25206 (inline_read_summary): Update.
25207 (inline_free_summary): Update.
25208 * ipa-inline.c (can_inline_edge_p): Update.
25209 (compute_inlined_call_time): Update.
25210 (want_inline_small_function_p): Update.
25211 (edge_badness): Update.
25212 (early_inliner): Update.
25213 * ipa-inline.h (inline_edge_summary): Turn to ...
25214 (ipa_call_summary): ... this one.
25215 (ipa_call_summary_t): New class.
25216 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25217 (ipa_call_summaries): New.
25218 (inline_edge_summary): Remove.
25219 (estimate_edge_growth): Update.
25220 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25221 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25222 * ipa-split.c (execute_split_functions): Update.
25223 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25225 2017-05-19 Richard Biener <rguenther@suse.de>
25227 PR middle-end/80764
25228 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25230 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25232 * config/rs6000/rs6000.c (struct machine_function): Add field
25233 fpr_is_wrapped_separately.
25234 (rs6000_get_separate_components): Use 64 components. Handle the
25235 new FPR components.
25236 (rs6000_components_for_bb): Handle the FPR components.
25237 (rs6000_emit_prologue_components): Handle the FPR components.
25238 (rs6000_emit_epilogue_components): Handle the FPR components.
25239 (rs6000_set_handled_components): Handle the FPR components.
25240 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25241 that are already separately shrink-wrapped.
25242 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25243 that are already separately shrink-wrapped.
25245 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25248 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25251 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25252 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25253 cases where the register allocator uses a move and an offsettable
25254 memory operation to/from a FPR register on ISA 2.06/2.07.
25255 (define_peephole2 for Altivec d-form store): Likewise.
25257 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25260 * config/i386/mmx.md (*mov<mode>_internal): Enable
25261 alternatives 11, 12, 13 and 14 also for 32bit targets.
25262 Remove alternatives 15, 16, 17 and 18.
25263 * config/i386/sse.md (vec_concatv2di): Change
25264 alternative (!x, *y) to (x, ?!*Yn).
25266 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25268 * dumpfile.h (enum dump_kind): Remove stray comma.
25270 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25272 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25273 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25274 predicate::num_conditions
25275 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25276 (CHANGED): turn into predicate::changed.
25277 (agg_position_info): Move to ipa-predicate.h
25278 (add_condition, predicate::add_clause, predicate::operator &=,
25279 predicate::or_with, predicate::evaluate, predicate::probability,
25280 dump_condition, dump_clause, predicate::dump,
25281 predicate::remap_after_duplication, predicate::remap_after_inlining,
25282 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25283 (evaluate_conditions_for_known_args): Update.
25284 (set_cond_stmt_execution_predicate): Update.
25285 * ipa-inline.h: Include ipa-predicate.h
25286 (condition, inline_param_summary, conditions, agg_position_info,
25287 predicate): Move to ipa-predicate.h
25288 * ipa-predicate.c: New file.
25289 * ipa-predicate.h: New file.
25291 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25293 * final.c (leaf_function_p): Check we are not in a sequence.
25295 2017-05-18 Martin Liska <mliska@suse.cz>
25297 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25298 * dumpfile.c (dump_register): Use new enum dump_kind.
25299 (get_dump_file_name): Likewise.
25300 (dump_enable_all): Likewise.
25301 (dump_switch_p_1): Likewise.
25302 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25303 * dumpfile.h (enum dump_kind): New enum type.
25304 (struct dump_file_info): Create constructor and
25305 format fields and comments.
25306 * passes.c (pass_manager::register_one_dump_file):
25308 * statistics.c (statistics_early_init): Likewise.
25309 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25310 TDF_TREE with TDF_SLIM.
25311 (gather_memory_references_ref): Likewise.
25313 2017-05-18 Martin Liska <mliska@suse.cz>
25315 * vec.h (struct vnull): Use it.
25317 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25319 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25320 (true_predicate, false_predicate, true_predicate_p,
25321 false_predicate_p): Remove.
25322 (single_cond_predicate, not_inlined_predicate): Turn to member function
25324 (add_condition): Update.
25325 (add_clause): Turn to...
25326 (predicate::add_clause): ... this one; update; allow passing NULL
25328 (and_predicates): Turn to ...
25329 (predicate::operator &=): ... this one.
25330 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25331 (or_predicates): Turn to ...
25332 (predicate::or_with): ... this one.
25333 (evaluate_predicate): Turn to ...
25334 (predicate::evaluate): ... this one.
25335 (predicate_probability): Turn to ...
25336 (predicate::probability): ... this one.
25337 (dump_condition): Update.
25338 (dump_predicate): Turn to ...
25339 (predicate::dump): ... this one.
25340 (account_size_time): Update.
25341 (edge_set_predicate): Update.
25342 (set_hint_predicate): UPdate.
25343 (evaluate_conditions_for_known_args): Update.
25344 (evaluate_properties_for_edge): Update.
25345 (remap_predicate_after_duplication): Turn to...
25346 (predicate::remap_after_duplication): ... this one.
25347 (remap_hint_predicate_after_duplication): Update.
25348 (inline_summary_t::duplicate): UPdate.
25349 (dump_inline_edge_summary): Update.
25350 (dump_inline_summary): Update.
25351 (set_cond_stmt_execution_predicate): Update.
25352 (set_switch_stmt_execution_predicate): Update.
25353 (compute_bb_predicates): Update.
25354 (will_be_nonconstant_expr_predicate): Update.
25355 (will_be_nonconstant_predicate): Update.
25356 (phi_result_unknown_predicate): Update.
25357 (predicate_for_phi_result): Update.
25358 (array_index_predicate): Update.
25359 (estimate_function_body_sizes): Update.
25360 (estimate_node_size_and_time): Update.
25361 (estimate_ipcp_clone_size_and_time): Update.
25362 (remap_predicate): Rename to ...
25363 (predicate::remap_after_inlining): ... this one.
25364 (remap_hint_predicate): Update.
25365 (inline_merge_summary): Update.
25366 (inline_update_overall_summary): Update.
25367 (estimate_size_after_inlining): Update.
25368 (read_predicate): Rename to ...
25369 (predicate::stream_in): ... this one.
25370 (read_inline_edge_summary): Update.
25371 (write_predicate): Rename to ...
25372 (predicate::stream_out): ... this one.
25373 (write_inline_edge_summary): Update.
25374 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25375 (clause_t): Turn to uint32_t
25376 (predicate): Turn to class; implement constructor and operators
25378 (size_time_entry): Update.
25379 (inline_summary): Update.
25380 (inline_edge_summary): Update.
25382 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25384 * fold-const.c (fold_binary_loc): Move transformation...
25385 * match.pd (C - X CMP X): ... here.
25387 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25389 * config/sparc/sparc.c (sparc_option_override): Set function
25390 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25391 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25393 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25395 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25397 2017-05-18 Marek Polacek <polacek@redhat.com>
25400 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25401 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25402 gimple_assign_load_p.
25404 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25406 PR middle-end/80692
25407 * real.c (do_compare): Give decimal_do_compare preference over
25408 comparing just the signs.
25410 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25412 * doc/md.texi (Canonicalization of Instructions): Describe the
25413 canonical form of instructions that inherently set a condition
25416 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25418 PR middle-end/80775
25419 * tree-cfg.c: Move deletion of unreachable case statements to after
25420 the merging of consecutive case labels.
25422 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25424 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25425 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25426 restoring of callee-saved registers.
25428 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25430 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25431 * config/visium/visium.c (single_set_and_flags): Likewise.
25432 * config/visium/visium.md (Substitutions): Likewise.
25434 2017-05-17 Martin Liska <mliska@suse.cz>
25436 * cfg.c: Introduce dump_flags_t type and
25437 use it instead of int type.
25439 * cfghooks.c: Likewise.
25440 * cfghooks.h (struct cfg_hooks): Likewise.
25441 * cfgrtl.c: Likewise.
25442 * cfgrtl.h: Likewise.
25443 * cgraph.c (cgraph_node::get_body): Likewise.
25444 * coretypes.h: Likewise.
25445 * domwalk.c: Likewise.
25446 * domwalk.h: Likewise.
25447 * dumpfile.c (struct dump_option_value_info): Likewise.
25448 (dump_enable_all): Likewise.
25449 (dump_switch_p_1): Likewise.
25450 (opt_info_switch_p): Likewise.
25451 * dumpfile.h (enum tree_dump_index): Likewise.
25452 (struct dump_file_info): Likewise.
25453 * genemit.c: Likewise.
25454 * generic-match-head.c: Likewise.
25455 * gengtype.c (open_base_files): Likewise.
25456 * gimple-pretty-print.c: Likewise.
25457 * gimple-pretty-print.h: Likewise.
25458 * graph.c (print_graph_cfg): Likewise.
25459 * graphite-scop-detection.c (dot_all_sese): Likewise.
25460 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25461 * loop-unroll.c (report_unroll): Likewise.
25462 * passes.c (pass_manager::register_one_dump_file): Likewise.
25463 * print-tree.c: Likewise.
25464 * statistics.c: Likewise.
25465 * tree-cfg.c: Likewise.
25466 * tree-cfg.h: Likewise.
25467 * tree-dfa.c: Likewise.
25468 * tree-dfa.h: Likewise.
25469 * tree-dump.c (dump_function): Likewise.
25470 * tree-dump.h (struct dump_info): Likewise.
25471 * tree-pretty-print.c: Likewise.
25472 * tree-pretty-print.h: Likewise.
25473 * tree-ssa-live.c: Likewise.
25474 * tree-ssa-live.h: Likewise.
25475 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25476 * tree-vect-loop.c: Likewise.
25477 * tree-vect-slp.c: Likewise.
25479 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25480 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25482 PR tree-optimization/80457
25483 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25484 of all arguments to a statement as scalar_to_vec operations.
25485 (vectorizable_call): Adjust call to vect_model_simple_cost for
25487 (vectorizable_conversion): Likewise.
25488 (vectorizable_assignment): Likewise.
25489 (vectorizable_shift): Likewise.
25490 (vectorizable_operation): Likewise.
25491 (vectorizable_comparison): Likewise.
25492 (vect_is_simple_cond): Record the def types for operands.
25493 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25494 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25495 for statement argument count.
25497 2017-05-16 Carl Love <cel@us.ibm.com>
25499 * config/rs6000/rs6000-c: Add support for built-in functions
25500 vector unsigned long long vec_bperm (vector unsigned long long,
25501 vector unsigned char)
25502 vector signed long long vec_mule (vector signed int,
25504 vector unsigned long long vec_mule (vector unsigned int,
25505 vector unsigned int)
25506 vector signed long long vec_mulo (vector signed int,
25508 vector unsigned long long vec_mulo (vector unsigned int,
25509 vector unsigned int)
25510 vector signed char vec_sldw (vector signed char,
25511 vector signed char,
25513 vector unsigned char vec_sldw (vector unsigned char,
25514 vector unsigned char,
25516 vector signed short vec_sldw (vector signed short,
25517 vector signed short,
25519 vector unsigned short vec_sldw (vector unsigned short,
25520 vector unsigned short,
25522 vector signed int vec_sldw (vector signed int,
25525 vector unsigned int vec_sldw (vector unsigned int,
25526 vector unsigned int,
25528 vector signed long long vec_sldw (vector signed long long,
25529 vector signed long long,
25531 vector unsigned long long vec_sldw (vector unsigned long long,
25532 vector unsigned long long,
25534 * config/rs6000/rs6000-c: Add support for built-in functions
25535 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25536 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25537 * doc/extend.texi: Update the built-in documentation for the
25538 new built-in functions.
25540 2017-05-16 Marek Polacek <polacek@redhat.com>
25544 * tree.c (save_expr): Don't fold the expression.
25546 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25548 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25549 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25550 and (?*y,m). Update insn attributes.
25552 2017-05-16 Martin Liska <mliska@suse.cz>
25554 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25555 flags argument of print_gimple_stmt, print_gimple_expr,
25556 print_generic_stmt and print_generic_expr.
25557 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25558 * coretypes.h: Likewise.
25559 * except.c (dump_eh_tree): Likewise.
25560 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25561 * gimple-pretty-print.h: Likewise.
25562 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25563 (backprop::push_to_worklist): Likewise.
25564 (backprop::pop_from_worklist): Likewise.
25565 (backprop::process_use): Likewise.
25566 (backprop::intersect_uses): Likewise.
25567 (note_replacement): Likewise.
25568 * gimple-ssa-store-merging.c
25569 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25570 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25571 (pass_store_merging::execute): Likewise.
25572 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25573 (ssa_base_cand_dump_callback): Likewise.
25574 (dump_incr_vec): Likewise.
25575 (replace_refs): Likewise.
25576 (replace_mult_candidate): Likewise.
25577 (create_add_on_incoming_edge): Likewise.
25578 (create_phi_basis): Likewise.
25579 (insert_initializers): Likewise.
25580 (all_phi_incrs_profitable): Likewise.
25581 (introduce_cast_before_cand): Likewise.
25582 (replace_one_candidate): Likewise.
25583 * gimplify.c (gimplify_expr): Likewise.
25584 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25585 (set_rename): Likewise.
25586 (rename_uses): Likewise.
25587 (copy_loop_phi_nodes): Likewise.
25588 (add_close_phis_to_merge_points): Likewise.
25589 (copy_loop_close_phi_args): Likewise.
25590 (copy_cond_phi_args): Likewise.
25591 (graphite_copy_stmts_from_block): Likewise.
25592 (translate_pending_phi_nodes): Likewise.
25593 * graphite-poly.c (print_pdr): Likewise.
25594 (dump_gbb_cases): Likewise.
25595 (dump_gbb_conditions): Likewise.
25596 (print_scop_params): Likewise.
25597 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25598 (build_cross_bb_scalars_use): Likewise.
25599 (gather_bbs::before_dom_children): Likewise.
25600 * hsa-dump.c (dump_hsa_immed): Likewise.
25601 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25602 (get_replacement_map): Likewise.
25603 * ipa-inline-analysis.c (dump_condition): Likewise.
25604 (estimate_function_body_sizes): Likewise.
25605 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25606 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25607 * ipa-prop.c (ipa_dump_param): Likewise.
25608 (ipa_print_node_jump_functions_for_edge): Likewise.
25609 (ipa_modify_call_arguments): Likewise.
25610 (ipa_modify_expr): Likewise.
25611 (ipa_dump_param_adjustments): Likewise.
25612 (ipa_dump_agg_replacement_values): Likewise.
25613 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25614 * ipa-pure-const.c (check_stmt): Likewise.
25615 (pass_nothrow::execute): Likewise.
25616 * ipa-split.c (execute_split_functions): Likewise.
25617 * omp-offload.c (dump_oacc_loop_part): Likewise.
25618 (dump_oacc_loop): Likewise.
25619 * trans-mem.c (tm_log_emit): Likewise.
25620 (tm_memopt_accumulate_memops): Likewise.
25621 (dump_tm_memopt_set): Likewise.
25622 (dump_tm_memopt_transform): Likewise.
25623 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25624 (print_loop): Likewise.
25625 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25626 (chkp_gather_checks_info): Likewise.
25627 (chkp_get_check_result): Likewise.
25628 (chkp_remove_check_if_pass): Likewise.
25629 (chkp_use_outer_bounds_if_possible): Likewise.
25630 (chkp_reduce_bounds_lifetime): Likewise.
25631 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25632 (chkp_mark_completed_bounds): Likewise.
25633 (chkp_register_incomplete_bounds): Likewise.
25634 (chkp_mark_invalid_bounds): Likewise.
25635 (chkp_maybe_copy_and_register_bounds): Likewise.
25636 (chkp_build_returned_bound): Likewise.
25637 (chkp_get_bound_for_parm): Likewise.
25638 (chkp_build_bndldx): Likewise.
25639 (chkp_get_bounds_by_definition): Likewise.
25640 (chkp_generate_extern_var_bounds): Likewise.
25641 (chkp_get_bounds_for_decl_addr): Likewise.
25642 * tree-chrec.c (chrec_apply): Likewise.
25643 * tree-data-ref.c (dump_data_reference): Likewise.
25644 (dump_subscript): Likewise.
25645 (dump_data_dependence_relation): Likewise.
25646 (analyze_overlapping_iterations): Likewise.
25647 * tree-inline.c (expand_call_inline): Likewise.
25648 (tree_function_versioning): Likewise.
25649 * tree-into-ssa.c (dump_defs_stack): Likewise.
25650 (dump_currdefs): Likewise.
25651 (dump_names_replaced_by): Likewise.
25652 (dump_update_ssa): Likewise.
25653 (update_ssa): Likewise.
25654 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25655 * tree-parloops.c (build_new_reduction): Likewise.
25656 (try_create_reduction_list): Likewise.
25657 (ref_conflicts_with_region): Likewise.
25658 (oacc_entry_exit_ok_1): Likewise.
25659 (oacc_entry_exit_single_gang): Likewise.
25660 * tree-pretty-print.h: Likewise.
25661 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25662 (get_scalar_evolution): Likewise.
25663 (add_to_evolution): Likewise.
25664 (get_loop_exit_condition): Likewise.
25665 (analyze_evolution_in_loop): Likewise.
25666 (analyze_initial_condition): Likewise.
25667 (analyze_scalar_evolution): Likewise.
25668 (instantiate_scev): Likewise.
25669 (number_of_latch_executions): Likewise.
25670 (gather_chrec_stats): Likewise.
25671 (final_value_replacement_loop): Likewise.
25672 (scev_const_prop): Likewise.
25673 * tree-sra.c (dump_access): Likewise.
25674 (disqualify_candidate): Likewise.
25675 (create_access): Likewise.
25676 (reject): Likewise.
25677 (maybe_add_sra_candidate): Likewise.
25678 (create_access_replacement): Likewise.
25679 (analyze_access_subtree): Likewise.
25680 (analyze_all_variable_accesses): Likewise.
25681 (sra_modify_assign): Likewise.
25682 (initialize_constant_pool_replacements): Likewise.
25683 (find_param_candidates): Likewise.
25684 (decide_one_param_reduction): Likewise.
25685 (replace_removed_params_ssa_names): Likewise.
25686 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25687 * tree-ssa-copy.c (dump_copy_of): Likewise.
25688 (copy_prop_visit_cond_stmt): Likewise.
25689 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25690 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25691 (record_equivalences_from_stmt): Likewise.
25692 * tree-ssa-dse.c (compute_trims): Likewise.
25693 (delete_dead_call): Likewise.
25694 (delete_dead_assignment): Likewise.
25695 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25696 (forward_propagate_into_cond): Likewise.
25697 (pass_forwprop::execute): Likewise.
25698 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25699 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25701 (move_computations_worker): Likewise.
25702 (execute_sm): Likewise.
25703 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25704 (remove_exits_and_undefined_stmts): Likewise.
25705 (remove_redundant_iv_tests): Likewise.
25706 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25707 (adjust_iv_update_pos): Likewise.
25708 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25709 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25710 (value_replacement): Likewise.
25711 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25712 * tree-ssa-pre.c (print_pre_expr): Likewise.
25713 (get_representative_for): Likewise.
25714 (create_expression_by_pieces): Likewise.
25715 (insert_into_preds_of_block): Likewise.
25716 (eliminate_insert): Likewise.
25717 (eliminate_dom_walker::before_dom_children): Likewise.
25718 (eliminate): Likewise.
25719 (remove_dead_inserted_code): Likewise.
25720 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25721 * tree-ssa-reassoc.c (get_rank): Likewise.
25722 (eliminate_duplicate_pair): Likewise.
25723 (eliminate_plus_minus_pair): Likewise.
25724 (eliminate_not_pairs): Likewise.
25725 (undistribute_ops_list): Likewise.
25726 (eliminate_redundant_comparison): Likewise.
25727 (update_range_test): Likewise.
25728 (optimize_range_tests_var_bound): Likewise.
25729 (optimize_vec_cond_expr): Likewise.
25730 (rewrite_expr_tree): Likewise.
25731 (rewrite_expr_tree_parallel): Likewise.
25732 (linearize_expr): Likewise.
25733 (break_up_subtract): Likewise.
25734 (linearize_expr_tree): Likewise.
25735 (attempt_builtin_powi): Likewise.
25736 (attempt_builtin_copysign): Likewise.
25737 (transform_stmt_to_copy): Likewise.
25738 (transform_stmt_to_multiply): Likewise.
25739 (dump_ops_vector): Likewise.
25740 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25741 (print_scc): Likewise.
25742 (set_ssa_val_to): Likewise.
25743 (visit_reference_op_store): Likewise.
25744 (visit_use): Likewise.
25745 (sccvn_dom_walker::before_dom_children): Likewise.
25746 (run_scc_vn): Likewise.
25747 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25749 (expr_hash_elt::print): Likewise.
25750 (const_and_copies::pop_to_marker): Likewise.
25751 (const_and_copies::record_const_or_copy_raw): Likewise.
25752 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25753 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25754 (dump_predicates): Likewise.
25755 (find_uninit_use): Likewise.
25756 (warn_uninitialized_phi): Likewise.
25757 (pass_late_warn_uninitialized::execute): Likewise.
25758 * tree-ssa.c (verify_vssa): Likewise.
25759 (verify_ssa): Likewise.
25760 (maybe_optimize_var): Likewise.
25761 * tree-vrp.c (dump_value_range): Likewise.
25762 (dump_all_value_ranges): Likewise.
25763 (dump_asserts_for): Likewise.
25764 (register_edge_assert_for_2): Likewise.
25765 (vrp_visit_cond_stmt): Likewise.
25766 (vrp_visit_switch_stmt): Likewise.
25767 (vrp_visit_stmt): Likewise.
25768 (vrp_visit_phi_node): Likewise.
25769 (simplify_cond_using_ranges_1): Likewise.
25770 (fold_predicate_in): Likewise.
25771 (evrp_dom_walker::before_dom_children): Likewise.
25772 (evrp_dom_walker::push_value_range): Likewise.
25773 (evrp_dom_walker::pop_value_range): Likewise.
25774 (execute_early_vrp): Likewise.
25776 2017-05-16 Richard Biener <rguenther@suse.de>
25778 * dwarf2out.c (loc_list_from_tree_1): Do not create
25779 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25781 2017-05-16 Richard Biener <rguenther@suse.de>
25783 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25785 (note_variable_value_in_expr): If we resolved the decl ref
25786 do not push to the stack.
25788 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25790 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25791 operations in fast-math mode.
25792 (vaddq_f16): Likewise.
25793 (vmul_f16): Likewise.
25794 (vmulq_f16): Likewise.
25795 (vsub_f16): Likewise.
25796 (vsubq_f16): Likewise.
25797 * config/arm/neon.md (add<mode>3): New.
25799 (fma:<VH:mode>3): New. Also remove outdated comment.
25802 2017-05-16 Martin Liska <mliska@suse.cz>
25806 * ipa-devirt.c (warn_types_mismatch): Fix typo.
25807 (odr_types_equivalent_p): Likewise.
25809 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
25811 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
25813 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
25816 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
25817 non-interunit SSE move alternatives with '?'.
25818 (zero-extendsidi peephole2): New peephole to skip intermediate
25819 general register in SSE zero-extend sequence.
25821 2017-05-15 Jeff Law <law@redhat.com>
25823 * reorg.c (relax_delay_slots): Create a new variable to hold
25824 the temporary target rather than clobbering TARGET_LABEL.
25826 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
25827 missing argument to extract_bit_field call.
25828 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
25830 2017-05-15 Martin Liska <mliska@suse.cz>
25833 * gcc.c (process_command): Do not allow empty argument of -o option.
25835 2017-05-15 Renlin Li <renlin.li@arm.com>
25837 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25838 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25839 * config/aarch64/constraints.md (Usf): Add long call check.
25840 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25841 (call_value): Likewise.
25842 (sibcall): Likewise.
25843 (sibcall_value): Likewise.
25845 (call_value_insn): New.
25846 (sibcall_insn): Update rtx pattern.
25847 (sibcall_value_insn): Likewise.
25848 (call_internal): Remove.
25849 (call_value_internal): Likewise.
25850 (sibcall_internal): Likewise.
25851 (sibcall_value_internal): Likewise.
25852 (call_reg): Likewise.
25853 (call_symbol): Likewise.
25854 (call_value_reg): Likewise.
25855 (call_value_symbol): Likewise.
25857 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25860 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25862 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25864 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25865 compatible with CCGOCmode and with CCZmode.
25867 2017-05-14 Martin Sebor <msebor@redhat.com>
25869 PR middle-end/77671
25870 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25871 (gimple_fold_builtin_snprintf): Same.
25872 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25873 (gimple_fold_builtin_snprintf): Same.
25874 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25875 of character types.
25876 (is_call_safe): New function.
25877 (try_substitute_return_value): Call it.
25878 (try_simplify_call): New function.
25879 (pass_sprintf_length::handle_gimple_call): Call it.
25881 2017-05-14 Martin Sebor <msebor@redhat.com>
25883 PR middle-end/80669
25884 * builtins.c (expand_builtin_stpncpy): Simplify.
25886 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25888 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25889 * config/i386/i386.h
25890 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25891 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25892 (struct machine_function): Add new members call_ms2sysv,
25893 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25894 (struct machine_frame_state): New fields sp_realigned and
25895 sp_realigned_offset.
25896 * config/i386/i386.c
25897 (enum xlogue_stub): New enum.
25898 (enum xlogue_stub_sets): New enum.
25899 (class xlogue_layout): New class.
25900 (struct ix86_frame): New fields stack_realign_allocate_offset,
25901 stack_realign_offset and outlined_save_offset. Modify comments to
25902 detail stack layout when using out-of-line stubs.
25903 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25904 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25905 -mcall-ms2sysv-xlogues.
25906 (stub_managed_regs): New static variable.
25907 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25908 registers managed by out-of-line stub.
25909 (disable_call_ms2sysv_xlogues): New function.
25910 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25911 m->call_ms2sysv when appropriate and compute frame layout for
25913 (sp_valid_at, fp_valid_at): New inline functions.
25914 (choose_basereg): New function.
25915 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25917 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25918 Use align parameter of choose_baseaddr to generated aligned SSE movs
25920 (pro_epilogue_adjust_stack): Modify to track
25921 machine_frame_state::sp_realigned.
25922 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25923 (ix86_nsaved_sseregs): Likewise.
25924 (ix86_emit_save_regs): Likewise.
25925 (ix86_emit_save_regs_using_mov): Likewise.
25926 (ix86_emit_save_sse_regs_using_mov): Likewise.
25927 (get_scratch_register_on_entry): Likewise.
25928 (gen_frame_set): New function.
25929 (gen_frame_load): Likewise.
25930 (gen_frame_store): Likewise.
25931 (emit_outlined_ms2sysv_save): Likewise.
25932 (emit_outlined_ms2sysv_restore): Likewise.
25933 (ix86_expand_prologue): Modify stack re-alignment code and call
25934 emit_outlined_ms2sysv_save when appropriate.
25935 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25936 parameter rtx_insn *insn, which allows the function to be used to only
25937 generate the notes.
25938 (ix86_expand_epilogue): Modify validity checks of frame and stack
25939 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25940 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25941 * config/i386/predicates.md
25942 (save_multiple): New predicate.
25943 (restore_multiple): Likewise.
25944 * config/i386/sse.md
25945 (save_multiple<mode>): New pattern.
25946 (save_multiple_realign<mode>): Likewise.
25947 (restore_multiple<mode>): Likewise.
25948 (restore_multiple_and_return<mode>): Likewise.
25949 (restore_multiple_leave_return<mode>): Likewise.
25950 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25952 2017-05-14 Julia Koval <julia.koval@intel.com>
25954 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25955 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25956 (__builtin_ia32_xsetbv): New builtins.
25957 * config/i386/i386.c (ix86_expand_special_args_builtin):
25959 (ix86_expand_builtin): Special expand for new intrinsics.
25960 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25961 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25962 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25964 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25966 * cfganal.c (inverted_post_order_compute): Change argument type
25968 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25969 * df-core.c (rest_of_handle_df_initialize): Adjust.
25970 (rest_of_handle_df_finish): Likewise.
25971 (df_analyze_1): Likewise.
25972 (df_analyze): Likewise.
25973 (loop_inverted_post_order_compute): Change argument to be a vec *.
25974 (df_analyze_loop): Adjust.
25975 (df_get_n_blocks): Likewise.
25976 (df_get_postorder): Likewise.
25977 * df.h (struct df_d): Change field to be a vec.
25978 * lcm.c (compute_laterin): Adjust.
25979 (compute_available): Likewise.
25980 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25981 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25982 * tree-ssa-pre.c (compute_antic): Likewise.
25984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25986 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25987 (depth_first_search::depth_first_search): Change structure init
25988 function to this constructor.
25989 (depth_first_search::add_bb): Rename function to this member.
25990 (depth_first_search::execute): Likewise.
25991 (flow_dfs_compute_reverse_finish): Adjust.
25993 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25995 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25996 (longest_simple_path): Likewise.
25997 * shrink-wrap.c (spread_components): Likewise.
25998 (disqualify_problematic_components): Likewise.
25999 (emit_common_heads_for_components): Likewise.
26000 (emit_common_tails_for_components): Likewise.
26001 (insert_prologue_epilogue_for_components): Likewise.
26003 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26005 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26008 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26010 * df-core.c (df_set_blocks): Start using auto_bitmap.
26011 (df_compact_blocks): Likewise.
26012 * df-problems.c (df_rd_confluence_n): Likewise.
26013 * df-scan.c (df_insn_rescan_all): Likewise.
26014 (df_process_deferred_rescans): Likewise.
26015 (df_update_entry_block_defs): Likewise.
26016 (df_update_exit_block_uses): Likewise.
26017 (df_entry_block_bitmap_verify): Likewise.
26018 (df_exit_block_bitmap_verify): Likewise.
26019 (df_scan_verify): Likewise.
26020 * lra-constraints.c (lra_constraints): Likewise.
26021 (undo_optional_reloads): Likewise.
26022 (lra_undo_inheritance): Likewise.
26023 * lra-remat.c (calculate_gen_cands): Likewise.
26024 (do_remat): Likewise.
26025 * lra-spills.c (assign_spill_hard_regs): Likewise.
26026 (spill_pseudos): Likewise.
26027 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26028 (bitmap_set_subtract_values): Likewise.
26030 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26032 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26033 management with auto_bitmap.
26034 (fix_inter_tick): Likewise.
26035 (fix_recovery_deps): Likewise.
26036 * ira.c (add_store_equivs): Likewise.
26037 (find_moveable_pseudos): Likewise.
26038 (split_live_ranges_for_shrink_wrap): Likewise.
26039 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26040 (rtx_reuse_manager::seen_def_p): Likewise.
26041 (rtx_reuse_manager::set_seen_def): Likewise.
26042 * print-rtl.h (class rtx_reuse_manager): Likewise.
26044 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26046 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26048 (migrate_btr_def): Likewise.
26049 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26050 * df-core.c (loop_post_order_compute): Likewise.
26051 (loop_inverted_post_order_compute): Likewise.
26052 * hsa-common.h: Likewise.
26053 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26054 * init-regs.c (initialize_uninitialized_regs): Likewise.
26055 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26056 (inline_small_functions): Likewise.
26057 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26058 * ira.c (combine_and_move_insns): Likewise.
26059 (build_insn_chain): Likewise.
26060 * loop-invariant.c (find_invariants): Likewise.
26061 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26062 * predict.c (tree_predict_by_opcode): Likewise.
26063 (predict_paths_leading_to): Likewise.
26064 (predict_paths_leading_to_edge): Likewise.
26065 (estimate_loops_at_level): Likewise.
26066 (estimate_loops): Likewise.
26067 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26068 (spread_components): Likewise.
26069 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26070 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26071 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26072 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26073 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26074 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26075 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26076 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26077 (mark_threaded_blocks): Likewise.
26078 (thread_through_all_blocks): Likewise.
26079 * tree-ssa.c (verify_ssa): Likewise.
26080 (execute_update_addresses_taken): Likewise.
26081 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26083 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26085 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26087 (post_order_compute): Likewise.
26088 (inverted_post_order_compute): Likewise.
26089 (pre_and_rev_post_order_compute_fn): Likewise.
26091 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26093 * genrecog.c (int_set::int_set): Explicitly construct our
26094 auto_vec base class.
26095 * vec.h (auto_vec::auto_vec): New constructor.
26097 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26099 * bitmap.h (class auto_bitmap): New constructor taking
26100 bitmap_obstack * argument.
26102 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26104 * bitmap.h (class auto_bitmap): Change type of m_bits to
26105 bitmap_head, and adjust ctor / dtor and member operators.
26107 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26109 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26110 when returned register mode doesn't match original mode.
26112 2017-05-12 Jeff Law <law@redhat.com>
26113 Jakub Jelinek <jakub@redhat.com>
26115 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26116 we look for cc setter after the compare-elim changes.
26117 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26118 within the vector to match what compare-elim now expects.
26119 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26120 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26122 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26123 after the compare-elim changes.
26124 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26125 the vector to match what compare-elim now expects.
26126 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26127 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26128 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26129 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26130 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26132 * config/visium/visium.c (single_set_and_flags): Fix where
26133 we look for cc setter after the compare-elim changes.
26134 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26135 with the vector to match what compare-elim now expects.
26136 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26137 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26138 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26139 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26140 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26141 (neg<mode>2_insn_set_overflow): Likewise.
26143 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26145 PR middle-end/79794
26146 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26147 maybe_expand_insn call, set ops[0].target. If still set after call,
26148 set alt_rtl. Add extra arg to recursive calls.
26149 (extract_bit_field): Add alt_rtl argument. Pass to
26151 * expmed.h (extract_bit_field): Fix prototype.
26152 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26153 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26154 to extract_bit_field_calls.
26155 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26156 Pass alt_rtl to extract_bit_field calls.
26157 * calls.c (store_unaligned_arguments_into_psuedos)
26158 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26159 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26161 * optabs.h (struct expand_operand): Add target bitfield.
26163 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26165 * compare-elim.c (try_eliminate_compare): Canonicalize
26166 operation with embedded compare to
26167 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26168 (set (reg) (operation)].
26170 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26172 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26175 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26176 cost of adding a carry flag for ADC instruction.
26177 [case MINUS]: Ignore the cost of subtracting a carry flag
26178 for SBB instruction.
26180 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26182 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26184 * config/rs6000/bmiintrin.h: New file.
26185 * config/rs6000/bmi2intrin.h: New file.
26186 * config/rs6000/x86intrin.h: New file.
26188 2017-05-12 Jeff Law <law@redhat.com>
26190 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26193 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26195 PR middle-end/80707
26196 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26198 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26200 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26201 early expansion of vector divide builtins.
26202 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26203 builtins identified as having unsigned arguments.
26205 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26207 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26208 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26209 expansion of vector logical operations (and, andc, or, xor,
26212 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26214 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26215 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26217 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26219 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26220 early GIMPLE expansion of vector multiplies.
26222 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26224 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26225 TARGET_HAVE_MOVT conditional.
26226 (movt splitter): Likewise.
26228 2017-05-12 Richard Biener <rguenther@suse.de>
26230 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26231 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26232 Fold all stmts not inplace.
26234 2017-05-12 Richard Biener <rguenther@suse.de>
26236 PR tree-optimization/80713
26237 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26238 inserted_exprs bit for not removed stmts.
26240 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26242 PR middle-end/69921
26243 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26244 parallelized" attribute for parallelized OpenACC kernels.
26245 * omp-offload.c (execute_oacc_device_lower): Use it.
26247 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26248 Set "oacc kernels" attribute.
26249 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26250 parameter. Adjust all users.
26251 (oacc_fn_attrib_kernels_p): Remove function.
26252 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26253 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26254 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26255 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26256 assert "oacc kernels" attribute is set.
26258 2017-05-11 Carl Love <cel@us.ibm.com>
26260 * config/rs6000/rs6000-c: Add support for built-in functions
26261 vector unsigned char vec_popcnt (vector signed char)
26262 vector unsigned char vec_popcnt (vector unsigned char)
26263 vector unsigned short vec_popcnt (vector signed short)
26264 vector unsigned short vec_popcnt (vector unsigned short)
26265 vector unsigned int vec_popcnt (vector signed int)
26266 vector unsigned int vec_popcnt (vector unsigned int)
26267 vector unsigned long long vec_popcnt (vector signed long long)
26268 vector unsigned long long vec_popcnt (vector unsigned long long)
26269 vector signed long long vec_slo (vector signed long long,
26270 vector signed char)
26271 vector signed long long vec_slo (vector signed long long,
26272 vector unsigned char)
26273 vector unsigned long long vec_slo (vector unsigned long long,
26274 vector signed char)
26275 vector unsigned long long vec_slo (vector unsigned long long,
26276 vector unsigned char)
26277 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26278 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26279 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26280 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26281 * doc/extend.texi: Update the built-in documentation file for the
26282 new built-in functions.
26284 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26286 * attribs.h (sorted_attr_string): Move machine independent
26287 functions for target clone support from the i386 port to common
26288 code. Rename ix86_function_versions to common_function_versions.
26289 Rename make_name to make_unique_name.
26290 (common_function_versions): Likewise.
26291 (make_unique_name): Likewise.
26292 (make_dispatcher_decl): Likewise.
26293 (is_function_default_version): Likewise.
26294 * attribs.c (attr_strcmp): Likewise.
26295 (sorted_attr_string): Likewise.
26296 (common_function_versions): Likewise.
26297 (make_unique_name): Likewise.
26298 (make_dispatcher_decl): Likewise.
26299 (is_function_default_version): Likewise.
26300 * config/i386/i386.c (attr_strcmp): Likewise.
26301 (sorted_attr_string): Likewise.
26302 (ix86_function_versions): Likewise.
26303 (make_name): Likewise.
26304 (make_dispatcher_decl): Likewise.
26305 (is_function_default_version): Likewise.
26306 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26308 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26311 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26312 Account for direct move costs for vec_construct of integer
26315 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26318 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26319 (UNSPEC_STX_ATOMIC): Ditto.
26320 (loaddi_via_sse): New insn.
26321 (storedi_via_sse): Ditto.
26322 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26323 Update corresponding peephole2 patterns.
26324 (atomic_storedi_fpu): Ditto.
26326 2017-05-11 Julia Koval <julia.koval@intel.com>
26328 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26329 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26331 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26332 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26333 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26335 2017-05-11 Nathan Sidwell <nathan@acm.org>
26337 * graphite-poly.c: Include dumpfile.h.
26339 * dumpfle.h (dump_function): Declare here ...
26340 * tree-dump.h (dump_function): ... not here.
26341 * dumpfile.c: #include tree-cfg.h.
26342 (dump_function): Move here from ...
26343 * tree-dump.c (dump_function): ... here.
26344 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26345 * graphite-poly.c: Don't include tree-dump.h.
26346 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26347 * print-tree.c: Likewise.
26348 * stor-layout.c: Likewise.
26349 * tree-nested.c: Likewise.
26351 * dumpfile.c (dump_start): Use TDF_FLAGS.
26352 (dump_enable_all): Fix TDF_KIND check thinko.
26354 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26356 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26357 array entries to represent two legal parameterizations of the
26358 overloaded __builtin_cmpb function, as represented by the
26359 P6_OV_BUILTIN_CMPB constant.
26360 (altivec_resolve_overloaded_builtin): Add special case handling
26361 for the __builtin_cmpb function, as represented by the
26362 P6_OV_BUILTIN_CMPB constant.
26363 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26364 (BU_P6_64BIT_2): New macro.
26365 (BU_P6_OVERLOAD_2): New macro
26366 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26367 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26368 (CMPB): Add overload support to represent both 32-bit and 64-bit
26369 compare-bytes function.
26370 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26371 support for TARGET_CMPB.
26372 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26373 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26374 documentation of the __builtin_cmpb overloaded built-in function.
26376 2017-05-11 Richard Biener <rguenther@suse.de>
26378 PR tree-optimization/80705
26379 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26380 bases are not vectorizable.
26382 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26384 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26385 when counting register pressure.
26387 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26389 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26390 (iv_ca_more_deps): Renamed to ...
26391 (iv_ca_compare_deps): ... this.
26392 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26394 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26396 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26398 (determine_group_iv_costs): ... here.
26399 (find_inv_vars_cb): Record inv var if it's not recorded before.
26401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26403 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26404 (get_shiftadd_cost): Ditto.
26406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26408 * tree-ssa-address.c: Include header file.
26409 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26411 (add_to_parts): Refactor.
26412 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26413 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26416 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26418 PR tree-optimization/53090
26419 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26421 (extract_cond_operands): Detect condition with IV on both sides
26422 and return COMP_IV_EXPR_2.
26423 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26424 (rewrite_use_compare): Simplify by removing call to function
26425 extract_cond_operands.
26427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26429 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26430 (extract_cond_operands): Detect condition comparing against non-
26431 invariant bound and return appropriate enum value.
26432 (find_interesting_uses_cond): Update use of extract_cond_operands.
26433 Handle its return value accordingly.
26434 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26436 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26438 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26439 nonlinear iv_use computation in loop invariant sensitive way.
26441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26443 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26444 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26448 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26449 (dump_cand): Support iv_cand.inv_exprs.
26450 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26452 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26457 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26459 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26460 as local function. Include necessary header files.
26461 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26463 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26465 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26469 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26470 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26471 RSHIFT_EXPR and BIT_NOT_EXPR.
26473 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26475 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26476 (adjust_setup_cost): New parameter supporting round up adjustment.
26477 (struct address_cost_data): Delete.
26478 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26479 (split_address_cost, ptr_difference_cost): Delete.
26480 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26481 (struct ainc_cost_data): New struct.
26482 (get_address_cost_ainc): New function.
26483 (get_address_cost, get_computation_cost): Reimplement.
26484 (determine_group_iv_cost_address): Record inv_expr for all uses of
26486 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26487 (iv_ca_has_deps): Reimplemented to ...
26488 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26490 (iv_ca_extend): Call iv_ca_more_deps.
26492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26494 * tree-ssa-address.c (struct mem_address): Move to header file.
26495 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26496 * tree-ssa-address.h (struct mem_address): Move from C file.
26497 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26499 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26501 * tree-affine.h (aff_combination_type): New interface.
26502 (aff_combination_zero_p): Remove static.
26503 (aff_combination_const_p): New interface.
26504 (aff_combination_singleton_var_p): New interfaces.
26506 2017-05-11 Richard Biener <rguenther@suse.de>
26508 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26509 Skip unreachable blocks and destinations.
26510 (eliminate): Move stmt removal and fixup ...
26511 (fini_eliminate): ... here. Skip inserted exprs.
26512 (pass_pre::execute): Move fini_pre after fini_eliminate.
26513 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26514 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26515 PRE to get rid of dead code that has invalid SSA form and
26516 split critical edges again.
26518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26520 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26522 2017-05-11 Richard Biener <rguenther@suse.de>
26524 * passes.c (execute_function_todo): Verify loops if they are
26525 said to be up-to-date.
26526 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26527 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26529 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26532 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26533 handle calling assemble_external ourself.
26536 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26537 modes with zero size. Enhance comment.
26539 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26541 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26542 built-ins for vec_xl and vec_xst with short and char pointer
26545 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26547 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26548 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26549 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26550 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26551 (_mm_maskz_min_round_ss): New intrinsics.
26552 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26553 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26554 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26555 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26556 (__builtin_ia32_minss_mask_round): New builtins.
26557 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26558 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26559 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26561 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26562 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26564 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26567 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26569 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26570 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26571 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26572 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26573 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26574 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26575 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26576 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26577 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26578 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26579 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26580 (__builtin_ia32_mulss_mask_round): New builtins.
26581 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26582 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26583 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26585 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26586 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26588 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26591 2017-05-10 Julia Koval <julia.koval@intel.com>
26593 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26594 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26595 (_mm256_setr_m128i): New intrinsics.
26597 2017-05-10 Julia Koval <julia.koval@intel.com>
26599 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26600 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26601 (_mm_maskz_rcp14_ss): New intrinsics.
26602 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26603 (__builtin_ia32_rcp14ss_mask): New builtins.
26604 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26606 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26608 PR tree-optimization/51513
26609 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26610 (assert_unreachable_fallthru_edge_p): Use it.
26611 (group_case_labels_stmt): Likewise.
26612 * tree-cfg.h: Prototype it.
26613 * stmt.c: Include cfghooks.h and tree-cfg.h.
26614 (emit_case_dispatch_table) <gap_label>: New local variable.
26615 Use it to fill dispatch table gaps.
26616 Test for default_label before updating probabilities.
26617 (expand_case) <default_label>: Remove unneeded initialization.
26618 Test for unreachable default case statement and remove its edge.
26619 Set default_label accordingly.
26620 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26622 2017-05-10 Carl Love <cel@us.ibm.com>
26624 * config/rs6000/rs6000-c: Add support for built-in functions
26625 vector signed char vec_neg (vector signed char)
26626 vector signed short int vec_neg (vector short int)
26627 vector signed int vec_neg (vector signed int)
26628 vector signed long long vec_neg (vector signed long long)
26629 vector float vec_neg (vector float)
26630 vector double vec_neg (vector double)
26631 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26633 * config/rs6000/altivec.h: Add define for vec_neg
26634 * doc/extend.texi: Update the built-in documentation for the
26635 new built-in functions.
26637 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26639 PR tree-optimization/77644
26640 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26642 2017-05-10 Nathan Sidwell <nathan@acm.org>
26644 * dumpfile.h (TDI_lang_all): New.
26645 (TDF_KIND): New. Renumber others
26646 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26648 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26650 (get_dump_file_name): Adjust suffix generation.
26651 (dump_enable_all): Use TDF_KIND.
26652 * doc/invoke.texi (-fdump-lang-all): Document.
26654 * dumpfile.h: Tabify.
26656 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26659 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26660 Move member access before delete.
26662 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26664 * tree-inline.c (expand_call_inline): Split block at stmt
26667 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26670 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26671 are now unused after splitting mov{sf,sd}_hardfloat.
26672 (f32_lr2): Likewise.
26673 (f32_lm): Likewise.
26674 (f32_lm2): Likewise.
26675 (f32_li): Likewise.
26676 (f32_li2): Likewise.
26677 (f32_lv): Likewise.
26678 (f32_sr): Likewise.
26679 (f32_sr2): Likewise.
26680 (f32_sm): Likewise.
26681 (f32_sm2): Likewise.
26682 (f32_si): Likewise.
26683 (f32_si2): Likewise.
26684 (f32_sv): Likewise.
26685 (f32_dm): Likewise.
26686 (f32_vsx): Likewise.
26687 (f32_av): Likewise.
26688 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26689 For movsf, order stores so the VSX stores occur before the GPR
26690 store which encourages the register allocator to use a traditional
26691 FPR instead of a GPR. For movsd, order the stores so that the GPR
26692 store comes before the VSX stores to allow the power6 to work.
26693 This is due to the power6 not having a 32-bit integer store
26694 instruction from a FPR.
26695 (movsf_hardfloat): Likewise.
26696 (movsd_hardfloat): Likewise.
26698 2017-05-09 Martin Sebor <msebor@redhat.com>
26700 PR translation/80280
26701 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26704 PR translation/80280
26705 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26706 data member added in r247778.
26707 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26709 2017-05-09 Nathan Sidwell <nathan@acm.org>
26711 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26713 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26716 2017-05-09 Marek Polacek <polacek@redhat.com>
26718 * doc/invoke.texi: Fix typo.
26720 2017-05-09 Richard Biener <rguenther@suse.de>
26722 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26723 (vrp_val_is_min): Likewise.
26724 (set_value_range_to_value): Likewise.
26725 (set_value_range_to_nonnegative): Likewise.
26726 (gimple_assign_nonzero_p): Likewise.
26727 (gimple_stmt_nonzero_p): Likewise.
26728 (vrp_int_const_binop): Likewise. Remove unreachable case.
26729 (adjust_range_with_scev): Adjust comments.
26730 (compare_range_with_value): Likewise.
26731 (extract_range_from_phi_node): Likewise.
26732 (test_for_singularity): Likewise.
26734 2017-05-09 Richard Biener <rguenther@suse.de>
26736 * tree-vrp.c (get_single_symbol): Add assert that we don't
26737 get overflowed constants as invariant part.
26738 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26739 checks. Use wi::cmp instead of recursing for integer constants.
26740 (compare_values): Just ignore whether we assumed undefined
26741 overflow instead of failing the compare.
26742 (extract_range_for_var_from_comparison_expr): Add comment before the
26743 TREE_NO_WARNING sets.
26744 (test_for_singularity): Likewise.
26745 (extract_range_from_comparison): Do not disable optimization
26746 when we assumed undefined overflow.
26747 (extract_range_basic): Remove init of unused var.
26749 2017-05-09 Richard Biener <rguenther@suse.de>
26751 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26752 (extract_range_from_multiplicative_op_1): Adjust.
26753 (extract_range_from_binary_expr_1): Use int_const_binop.
26755 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26758 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26759 rs6000_store_data_bypass_p in seven define_bypass directives and
26760 in several comments.
26761 * config/rs6000/rs6000-protos.h: Add prototype for
26762 rs6000_store_data_bypass_p function.
26763 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26764 function implements slightly different (rs6000-specific) semantics
26765 than store_data_bypass_p, returning false rather than aborting
26766 with assertion error when arguments do not satisfy the
26767 requirements of store data bypass.
26768 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26769 rs6000_store_data_bypass_p.
26771 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26773 * config/xtensa/xtensa-protos.h
26774 (xtensa_initial_elimination_offset): New declaration.
26775 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26776 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26777 macro definition, add case for FRAME_POINTER_REGNUM when
26778 FRAME_GROWS_DOWNWARD.
26779 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26780 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26781 xtensa_initial_elimination_offset.
26783 2017-05-08 Nathan Sidwell <nathan@acm.org>
26785 * doc/invoke.texi: Alphabetize -fdump options.
26787 2017-05-08 Martin Sebor <msebor@redhat.com>
26789 PR translation/80280
26790 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26792 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26794 * target.def (compute_frame_layout): New optional target hook.
26795 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26796 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26797 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
26799 * reload1.c (verify_initial_elim_offsets): Likewise.
26800 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
26801 (use_simple_return_p): Call arm_compute_frame_layout if needed.
26802 (arm_get_frame_offsets): Split up into this ...
26803 (arm_compute_frame_layout): ... and this function.
26805 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
26807 * config/aarch64/constraints.md (Usa): New constraint.
26808 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
26810 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
26812 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
26813 with_multilib_list after it has been checked.
26815 2017-05-08 Richard Biener <rguenther@suse.de>
26817 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
26818 (bitmap_set_subtract_values): Likewise.
26820 2017-05-08 Richard Biener <rguenther@suse.de>
26822 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
26823 (gimple_assign_nonzero): ... this and remove strict_overflow_p
26825 (gimple_stmt_nonzero_warnv_p): Rename to ...
26826 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
26828 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
26829 (extract_range_basic): Adjust, do not disable propagation on
26830 strict overflow sensitive simplification.
26831 (vrp_visit_cond_stmt): Likewise.
26833 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26835 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26836 body size unconditionally.
26838 2017-05-07 Jeff Law <law@redhat.com>
26841 2017-05-06 Jeff Law <law@redhat.com>
26842 PR tree-optimization/78496
26843 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26846 PR tree-optimization/78496
26847 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26848 (simplify_stmt_using_ranges): Call it.
26849 (vrp_dom_walker::before_dom_children): Extract equivalences
26850 from an ASSERT_EXPR with an equality comparison against a
26853 2017-05-06 Jeff Law <law@redhat.com>
26855 PR tree-optimization/78496
26856 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26859 PR tree-optimization/78496
26860 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26861 (simplify_stmt_using_ranges): Call it.
26862 (vrp_dom_walker::before_dom_children): Extract equivalences
26863 from an ASSERT_EXPR with an equality comparison against a
26866 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26868 * lra-constraints.c (lra_copy_reg_equiv): New function.
26869 (split_reg): Use it to copy equivalence information from the
26870 original register to the spill register.
26872 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26874 PR rtl-optimization/75964
26875 * simplify-rtx.c (simplify_const_relational_operation): Remove
26876 invalid handling of comparisons of integer ABS.
26878 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26880 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26881 initialize to zero.
26882 (init_regs): Remove declaration.
26883 (function_arg_advance_32): Initialize error_p as boolean variable.
26885 2017-05-05 Nathan Sidwell <nathan@acm.org>
26887 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26888 lines. Use for (;;).
26890 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26892 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26893 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26894 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26895 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26896 VF=2 that require versioning.
26898 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26900 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26903 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26905 * diagnostic.h (diagnostic_override_option_index): Convert from
26906 macro to inline function.
26908 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26910 * diagnostic.c (last_module_changed_p): New function.
26911 (set_last_module): New function.
26912 (diagnostic_report_current_module): Convert macro usage to
26913 the above functions.
26914 * diagnostic.h (diagnostic_context::last_module): Strengthen
26915 from const line_map * to const line_map_ordinary *.
26916 (diagnostic_last_module_changed): Delete macro.
26917 (diagnostic_set_last_module): Delete macro.
26919 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26921 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26922 with diagnostic_report_diagnostic.
26923 (diagnostic_n_impl_richloc): Likewise.
26924 * diagnostic.h (report_diagnostic): Delete macro.
26925 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26926 with diagnostic_report_diagnostic.
26927 * substring-locations.c (format_warning_va): Likewise.
26929 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26931 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26932 save/restor of format_spec. Move option-printing code to...
26933 (print_option_information): ...this new function, and
26934 reimplement by simply printing to the pretty_printer,
26935 rather than appending to the format string.
26937 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26939 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26940 handling logic into...
26941 (update_effective_level_from_pragmas): ...this new function.
26943 2017-05-04 Andrew Waterman <andrew@sifive.com>
26945 * config/riscv/riscv.opt (mstrict-align): New option.
26946 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26947 (SLOW_UNALIGNED_ACCESS): Define.
26948 (riscv_slow_unaligned_access): Declare.
26949 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26951 (riscv_slow_unaligned_access): New variable.
26952 (rocket_tune_info): Set slow_unaligned_access to true.
26953 (optimize_size_tune_info): Set slow_unaligned_access to false.
26954 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26955 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26956 (riscv_option_override): Set riscv_slow_unaligned_access.
26957 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26959 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26961 * config/riscv/riscv.md: Unify indentation.
26963 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26968 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26970 (extendsi<mode>2): Add support for doing sign extension via
26971 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26972 don't have ISA 3.0 instructions.
26973 (extendsi<mode>2 splitter): Likewise.
26974 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26975 generate the normal insns since SImode can now go in vector
26976 registers. Disallow the special UNSPECs needed for previous
26977 machines to hide SImode being used. Add new insns
26978 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26979 (fix_trunc<mode>si2_stfiwx): Likewise.
26980 (fix_trunc<mode>si2_internal): Likewise.
26981 (fixuns_trunc<mode>si2): Likewise.
26982 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26983 (fctiw<u>z_<mode>_smallint): Likewise.
26984 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26985 of floating point to 32-bit integer from doing a direct move to
26986 the GPR registers to do a store.
26987 (fctiwz_<mode>): Break long line.
26989 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26991 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26992 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26993 (addr_list, addr_offset_valid_p): New.
26994 (split_address_groups): Check offset validity with above function.
26995 (gt-tree-ssa-loop-ivopts.h): Include header file.
26997 2017-05-05 Nathan Sidwell <nathan@acm.org>
26999 * config.gcc (arm*-*-*): Add missing 'fi'.
27001 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27003 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27004 included in -fopt-info does not matter.
27005 * doc/optinfo.texi (-fopt-info): Fix description of default
27006 behavour. Explicitly say order of options included in -fopt-info
27009 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27011 * config.gcc: Allow combinations of aprofile and rmprofile values for
27012 --with-multilib-list.
27013 * config/arm/t-multilib: New file.
27014 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27015 variables. Remove setting of ISA and floating-point ABI in
27016 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27017 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27018 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27019 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27021 * config/arm/t-rmprofile: Likewise except for the matches changes.
27022 * doc/install.texi (--with-multilib-list): Document the combination of
27023 aprofile and rmprofile values and warn about pitfalls in doing that.
27025 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27027 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27028 (movdi_aarch64): Likewise.
27030 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27032 PR tree-optimization/80632
27033 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27035 (build_arrays): Initialize it for virtual phis.
27036 (fix_phi_nodes): Use it for virtual phis.
27038 PR tree-optimization/80558
27039 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27040 [x, y] op z into [x op, y op z] for op & or | if conditions
27043 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27044 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27047 * config/arm/arm.md (use_literal_pool): Remove.
27048 (64-bit immediate split): No longer takes cost into consideration
27049 if arm_disable_literal_pool is enabled.
27050 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27051 used when arm_disable_literal_pool is enabled.
27052 (arm_max_const_double_inline_cost): Remove use of
27053 arm_disable_literal_pool.
27054 (push_minipool_fix): Add assert.
27055 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27056 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27057 (no_literal_pool_sf_immediate): New.
27059 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27061 PR tree-optimization/80613
27062 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27063 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27065 2017-05-05 Richard Biener <rguenther@suse.de>
27067 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27069 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27071 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27072 of this flag from insn conditions due to removal from r247495.
27074 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27076 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27078 (arm_early_store_addr_dep_ptr): Likewise.
27079 * config/arm/aarch-common-protos.h
27080 (arm_early_load_addr_dep_ptr): Add prototype.
27081 (arm_early_store_addr_dep_ptr): Likewise.
27082 * config/arm/cortex-a53.md: Add new bypasses.
27084 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27086 * tree.c (next_type_uid): Change type to unsigned.
27087 (type_hash_canon): Decrement back next_type_uid if
27088 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27089 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27092 2017-05-04 Martin Sebor <msebor@redhat.com>
27094 * builtins.c: Fix a trivial typo in a comment.
27096 PR middle-end/79234
27097 * builtins.c (check_sizes): Adjust to handle reading past the end.
27098 Avoid printing excessive upper bound of ranges. Use %E to print
27099 tree nodes instead of converting them to %wu.
27100 (expand_builtin_memchr): New function.
27101 (compute_dest_size): Rename...
27102 (compute_objsize): ...to this.
27103 (expand_builtin_memcpy): Adjust.
27104 (expand_builtin_mempcpy): Adjust.
27105 (expand_builtin_strcat): Adjust.
27106 (expand_builtin_strcpy): Adjust.
27107 (check_strncat_sizes): Adjust.
27108 (expand_builtin_strncat): Adjust.
27109 (expand_builtin_strncpy): Adjust and simplify.
27110 (expand_builtin_memset): Adjust.
27111 (expand_builtin_bzero): Adjust.
27112 (expand_builtin_memcmp): Adjust.
27113 (expand_builtin): Handle memcmp.
27114 (maybe_emit_chk_warning): Check strncat just once.
27116 2017-05-04 Martin Sebor <msebor@redhat.com>
27118 PR preprocessor/79214
27119 PR middle-end/79222
27120 PR middle-end/79223
27121 * builtins.c (check_sizes): Add inlining context and issue
27122 warnings even when -Wno-system-headers is set.
27123 (check_strncat_sizes): Same.
27124 (expand_builtin_strncat): Same.
27125 (expand_builtin_memmove): New function.
27126 (expand_builtin_stpncpy): Same.
27127 (expand_builtin): Handle memmove and stpncpy.
27129 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27131 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27132 which is not used any more.
27134 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27136 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27138 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27140 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27141 (cortexa53_tunings): Likewise.
27142 (cortexa57_tunings): Likewise.
27143 (cortexa72_tunings): Likewise.
27144 (cortexa73_tunings): Likewise.
27146 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27148 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27149 Set loop alignment to 8.
27151 2017-05-04 Martin Sebor <msebor@redhat.com>
27153 PR translation/80280
27154 * builtins.c (expand_builtin_object_size): Add missing quoting to
27155 %D and like directives.
27156 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27157 (hsa_type_for_tree_type): Same.
27158 (verify_function_arguments): Same.
27159 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27160 * varasm.c (get_section): Same.
27163 2017-05-04 Martin Sebor <msebor@redhat.com>
27165 PR translation/80280
27166 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27168 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27170 * config/aarch64/aarch64.c (generic_addrcost_table):
27171 Change HI/TI mode setting.
27173 2017-05-04 Martin Jambor <mjambor@suse.cz>
27175 PR tree-optimization/80622
27176 * tree-sra.c (comes_initialized_p): New function.
27177 (build_accesses_from_assign): Only set write lazily when
27178 comes_initialized_p is false.
27179 (analyze_access_subtree): Use comes_initialized_p.
27180 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27181 instead of testing for PARM_DECL.
27183 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27185 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27186 constraint on operand 0 to allow more general addressing modes.
27187 Adjust output template.
27188 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27190 * config/aarch64/aarch64-protos.h
27191 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27192 * config/aarch64/constraints.md (Dp): New address constraint.
27193 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27196 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27198 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27199 update use of estimate_ipcp_clone_size_and_time.
27200 (estimate_local_effects): Update use of
27201 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27202 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27203 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27204 Return nonspecialized time.
27206 2017-05-04 Richard Biener <rguenther@suse.de>
27208 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27209 for the last VUSE which def dominates the PHI. Directly call
27211 (get_continuation_for_phi_1): Remove.
27213 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27215 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27216 to explain the use of truncating division. Cap the number of
27217 iterations to the maximum given by nb_iterations_upper_bound,
27220 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27222 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27223 * configure: Regenerate.
27224 * config.in: Regenerate.
27225 * config/i386/driver-mingw32.c: new file.
27226 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27227 * config.host: Link driver-mingw32.o on MinGW host.
27228 * doc/install.texi: Document new --enable-mingw-wildcard configure
27231 2017-05-04 Marek Polacek <polacek@redhat.com>
27233 PR tree-optimization/80612
27234 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27236 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27237 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27239 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27240 (movt splitter): Likewise.
27241 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27242 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27243 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27244 block for Thumb-1 with MOVT.
27245 (thumb2_legitimate_address_p): Move code block ...
27246 (can_avoid_literal_pool_for_label_p): ... into this new function.
27247 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27249 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27250 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27251 "M-profile targets with the MOVT instruction".
27253 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27255 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27256 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27257 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27259 2017-05-04 Martin Liska <mliska@suse.cz>
27261 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27262 variable cond_code.
27264 2017-05-04 Richard Biener <rguenther@suse.de>
27266 * tree.c (array_at_struct_end_p): Handle arrays at struct
27267 end with flexarrays more conservatively. Refactor and treat
27268 arrays of arrays or aggregates more strict. Fix
27269 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27270 * tree.c (array_at_struct_end_p): Adjust prototype.
27271 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27272 * gimple-fold.c (get_range_strlen): Likewise.
27273 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27275 2017-05-04 Richard Biener <rguenther@suse.de>
27277 PR tree-optimization/31130
27278 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27280 (supports_overflow_infinity): Likewise.
27281 (is_negative_overflow_infinity): Likewise.
27282 (is_positive_overflow_infinity): Likewise.
27283 (is_overflow_infinity): Likewise.
27284 (stmt_overflow_infinity): Likewise.
27285 (overflow_infinity_range_p): Likewise.
27286 (usable_range_p): Remove as always returning true.
27287 (make_overflow_infinity): Remove.
27288 (negative_overflow_infinity): Likewise.
27289 (positive_overflow_infinity): Likewise.
27290 (avoid_overflow_infinity): Likewise.
27291 (set_value_range): Adjust accordingly.
27292 (set_value_range_to_nonnegative): Likewise, remove now unused
27293 overflow_infinity arg.
27294 (vrp_operand_equal_p): Adjust.
27295 (update_value_range): Likewise.
27296 (range_int_cst_singleton_p): Likewise.
27297 (operand_less_p): Likewise.
27298 (compare_values_warnv): Likewise.
27299 (extract_range_for_var_from_comparison_expr): Likewise.
27300 (vrp_int_const_binop): Likewise.
27301 (zero_nonzero_bits_from_vr): Likewise.
27302 (extract_range_from_multiplicative_op_1): Likewise.
27303 (extract_range_from_binary_expr_1): Likewise.
27304 (extract_range_from_unary_expr): Likewise.
27305 (extract_range_from_comparison): Likewise.
27306 (extract_range_basic): Likewise.
27307 (adjust_range_with_scev): Likewise.
27308 (compare_ranges): Likewise.
27309 (compare_range_with_value): Likewise.
27310 (dump_value_range): Likewise.
27311 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27313 (simplify_cond_using_ranges): Adjust.
27315 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27317 * brig-builtins.def: Added a builtin for class_f64.
27318 * builtin-types.def: Added a builtin type needed by class_f64.
27320 2017-05-03 Jason Merrill <jason@redhat.com>
27322 * timevar.def: Add TV_CONSTEXPR.
27324 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27326 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27328 2017-05-03 Martin Jambor <mjambor@suse.cz>
27330 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27331 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27332 * ipa-cp.c (ipcp_propagate_stage): Do not call
27333 ipa_update_after_lto_read.
27334 * ipa-inline.c (ipa_inline): Likewise.
27336 2017-05-03 Martin Jambor <mjambor@suse.cz>
27338 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27339 tag. Added a default constructor and a destructor.
27340 (ipa_edge_args_sum_t): New class;
27341 (ipa_edge_args_sum): Declare.
27342 (ipa_edge_args_vector): Remove declaration.
27343 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27344 (ipa_free_edge_args_substructures): Remove declaration.
27345 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27346 (ipa_edge_args_info_available_for_edge_p): Likewise.
27347 * ipa-prop.c (ipa_edge_args_vector): Removed.
27348 (edge_removal_hook_holder): Likewise.
27349 (edge_duplication_hook_holder): Likewise.
27350 (ipa_edge_args_sum): New variable.
27351 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27352 ipa_edge_args_vector.
27353 (ipa_free_edge_args_substructures): Likewise.
27354 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27355 ipa_edge_args_vector.
27356 (ipa_edge_removal_hook): Turned into method
27357 ipa_edge_args_sum_t::remove.
27358 (ipa_edge_duplication_hook): Turned into method
27359 ipa_edge_args_sum_t::duplicate.
27360 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27361 registering edge hooks.
27362 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27363 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27364 ipa_edge_args_sum instead of ipa_edge_args_vector.
27365 * ipa-profile.c (ipa_profile): Likewise.
27367 2017-05-03 Martin Jambor <mjambor@suse.cz>
27369 * symbol-summary.h (function_summary): New method exists.
27370 (function_summary::symtab_removal): Deallocate through release.
27371 (call_summary): New class.
27372 (gt_ggc_mx): New overload.
27373 (gt_pch_nx): Likewise.
27374 (gt_pch_nx): Likewise.
27376 2017-05-03 Jeff Law <law@redhat.com>
27378 PR tree-optimization/78496
27379 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27380 from simplify_cond_using_ranges. Split off code to walk
27381 backwards through casts into ...
27382 (simplify_cond_using_ranges_2): New function.
27383 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27384 (execute_vrp): After identifying jump threads, call
27385 simplify_cond_using_ranges_2.
27387 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27390 * ipa-inline.h (inline_summary): Add ctor.
27391 (create_ggc): Do not use ggc_cleared_alloc.
27393 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27394 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27396 * gcc.c (handle_braces): Support escaping in switch matching
27398 * doc/invoke.texi (Spec Files): Document it.
27399 Remove superfluous @code markup in items.
27401 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27403 * diagnostic-show-locus.c (struct column_range): New struct.
27404 (get_affected_columns): New function.
27405 (get_printed_columns): New function.
27406 (struct correction): New struct.
27407 (correction::ensure_capacity): New function.
27408 (correction::ensure_terminated): New function.
27409 (struct line_corrections): New struct.
27410 (line_corrections::~line_corrections): New dtor.
27411 (line_corrections::add_hint): New function.
27412 (layout::print_trailing_fixits): Reimplement in terms of the new
27414 (selftest::test_overlapped_fixit_printing): New function.
27415 (selftest::diagnostic_show_locus_c_tests): Call it.
27417 2017-05-03 Nathan Sidwell <nathan@acm.org>
27419 Canonicalize canonical type hashing
27420 * tree.h (type_hash_canon_hash): Declare.
27421 * tree.c (type_hash_list, attribute_hash_list): Move into
27422 type_hash_canon_hash.
27423 (build_type_attribute_qual_variant): Break out hash code calc into
27424 type_hash_canon_hash.
27425 (type_hash_canon_hash): New. Generic type hash computation.
27426 (build_range_type_1, build_array_type_1, build_function_type,
27427 build_method_type_directly, build_offset_type, build_complex_type,
27428 make_vector_type): Call it.
27430 2017-05-03 Richard Biener <rguenther@suse.de>
27432 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27433 When all DRs have unknown misaligned do not always peel
27434 when there is a store but apply the same costing model as if
27435 there were only loads.
27437 2017-05-03 Richard Biener <rguenther@suse.de>
27440 PR tree-optimization/80492
27441 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27442 compare_base_decls returning dont-know properly.
27444 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27446 * config/arm/iterators.md (CCSI): New mode iterator.
27447 (arch): New mode attribute.
27448 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27449 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27450 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27451 code iterator for success result mode.
27452 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27453 the corresponding new insn generators.
27455 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27458 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27459 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27461 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27463 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27464 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27465 (DDR_A): Wrap DDR argument in brackets.
27466 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27467 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27468 (DDR_REVERSED_P): Likewise.
27470 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27472 PR tree-optimization/79472
27473 * tree-switch-conversion.c (struct switch_conv_info): Add
27474 contiguous_range and default_case_nonstandard fields.
27475 (collect_switch_conv_info): Compute contiguous_range and
27476 default_case_nonstandard fields, don't clear final_bb if
27477 contiguous_range and only the default case doesn't have the required
27479 (check_all_empty_except_final): Set default_case_nonstandard instead
27480 of failing if contiguous_range and the default case doesn't have empty
27482 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27483 and only the default case doesn't have the required constants. Skip
27485 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27486 if default_case_nonstandard.
27487 (build_constructors): Build constant 1 just once. Assert that default
27488 values aren't inserted in between cases if contiguous_range. Skip
27490 (build_arrays): Skip virtual phis.
27491 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27492 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27493 Handle virtual phis.
27494 (gen_inbound_check): Handle default_case_nonstandard case.
27495 (process_switch): Adjust check_final_bb caller. Call
27496 gather_default_values with the first non-default case instead of
27497 default case if default_case_nonstandard.
27499 2017-05-02 Nathan Sidwell <nathan@acm.org>
27501 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27502 check. Fix formatting.
27504 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27506 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27507 errors when comparing specialized and unspecialized times.
27509 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27511 * diagnostic-show-locus.c
27512 (layout::should_print_annotation_line_p): Make private.
27513 (layout::print_annotation_line): Make private.
27514 (layout::annotation_line_showed_range_p): Make private.
27515 (layout::show_ruler): Make private.
27516 (layout::print_source_line): Make private. Pass in line and
27517 line_width, rather than calling location_get_source_line. Drop
27519 (layout::print_leading_fixits): New method.
27520 (layout::print_any_fixits): Rename to...
27521 (layout::print_trailing_fixits): ...this, and make private.
27522 Don't print newline fixits.
27523 (diagnostic_show_locus): Move logic for printing one row into...
27524 (layout::print_line): ...this new function. Move the
27525 location_get_source_line call and error-handling from
27526 print_source_line to here. Call print_leading_fixits, and rename
27527 print_any_fixits to print_trailing_fixits.
27528 (selftest::test_fixit_insert_containing_newline): Update now that
27529 newlines are partially supported.
27530 (selftest::test_fixit_insert_containing_newline_2): New test.
27531 (selftest::test_fixit_replace_containing_newline): Update comments.
27532 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27533 * edit-context.c (class added_line): New class.
27534 (class edited_line): Describe newline handling in comment.
27535 (edited_line::actually_edited_p): New method.
27536 (edited_line::print_content): Delete redundant decl.
27537 (edited_line::m_predecessors): New field.
27538 (edited_file::print_content): Call edited_line::print_content.
27539 (edited_file::print_diff): Update to support newlines.
27540 (edited_file::print_diff_hunk): Likewise.
27541 (edited_file::print_run_of_changed_lines): New function.
27542 (edited_file::print_diff_line): Convert to...
27543 (print_diff_line): ...this.
27544 (edited_file::get_effective_line_count): New function.
27545 (edited_line::edited_line): Initialize new field m_predecessors.
27546 (edited_line::~edited_line): Clean up m_predecessors.
27547 (edited_line::apply_fixit): Handle newlines.
27548 (edited_line::get_effective_line_count): New function.
27549 (edited_line::print_content): New function.
27550 (edited_line::print_diff_lines): New function.
27551 (selftest::test_applying_fixits_insert_containing_newline): New
27553 (selftest::test_applying_fixits_replace_containing_newline): New
27555 (selftest::insert_line): New function.
27556 (selftest::test_applying_fixits_multiple_lines): Add example of
27558 (selftest::edit_context_c_tests): Call the new tests.
27560 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27562 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27563 parameter cand. Update dump information.
27564 (get_computation_cost): Update uses.
27566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27568 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27569 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27570 (get_computation_at, rewrite_use_address): Update use of
27571 get_computation_aff.
27573 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27575 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27576 (get_computation): Delete.
27577 (get_computation_cost): Implement like get_computation_cost_at.
27578 Use get_computation_at.
27579 (get_computation_cost_at): Delete.
27580 (rewrite_use_nonlinear_expr): Use get_computation_at.
27581 (rewrite_use_compare, remove_unused_ivs): Ditto.
27583 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27585 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27587 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27589 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27590 (ivopts_global_cost_for_size): Rename parameter and update uses.
27591 (iv_ca_recount_cost): Update uses.
27592 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27593 candidates seperately in n_invs and n_cands.
27594 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27596 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27598 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27599 (find_inv_vars_cb): New.
27600 (find_depends): Renamed to ...
27601 (find_inv_vars): ... this.
27602 (add_candidate_1, force_var_cost): Call find_inv_vars.
27603 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27605 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27607 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27608 inv_vars. Add inv_exprs.
27609 (struct iv_cand): Rename depends_on to inv_vars.
27610 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27611 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27612 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27613 (dump_cand): Dump inv_vars.
27614 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27615 (record_invariant, find_depends, add_candidate_1): Ditto.
27616 (set_group_iv_cost, force_var_cost): Ditto.
27617 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27618 (get_computation_cost_at, get_computation_cost): Ditto.
27619 (determine_group_iv_cost_generic): Ditto.
27620 (determine_group_iv_cost_address): Ditto.
27621 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27622 (determine_group_iv_costs): Ditto.
27623 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27624 (iv_ca_set_remove_invariants): Renamed to ...
27625 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27626 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27627 (iv_ca_set_add_invariants): Renamed to ...
27628 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27629 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27630 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27631 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27632 (create_new_ivs): Remove useless dump.
27634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27636 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27638 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27639 (iv_ca_set_no_cp, create_new_iv): Ditto.
27641 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27643 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27645 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27647 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27648 function tree_check2.
27650 2017-05-02 Martin Liska <mliska@suse.cz>
27652 * doc/gcov.texi: Add missing preposition.
27653 * gcov.c (function_info::function_info): Properly fill up
27654 all member variables.
27656 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27658 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27660 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27662 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27664 2017-05-02 Martin Liska <mliska@suse.cz>
27667 * lto-streamer-in.c (lto_read_tree_1): Remove
27668 LTO_STREAMER_DEBUG.
27669 * lto-streamer.c (struct tree_hash_entry): Likewise.
27670 (struct tree_entry_hasher): Likewise.
27671 (tree_entry_hasher::hash): Likewise.
27672 (tree_entry_hasher::equal): Likewise.
27673 (lto_streamer_init): Likewise.
27674 (lto_orig_address_map): Likewise.
27675 (lto_orig_address_get): Likewise.
27676 (lto_orig_address_remove): Likewise.
27677 * lto-streamer.h: Likewise.
27678 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27679 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27681 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27683 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27684 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27685 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27686 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27687 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27688 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27689 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27690 (mm_maskz_sub_ss): New intrinsics.
27691 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27692 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27693 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27694 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27695 (__builtin_ia32_subss_mask_round): New builtins.
27696 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27697 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27698 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27700 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27701 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27703 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27706 2017-05-02 Martin Jambor <mjambor@suse.cz>
27708 PR tree-optimization/78687
27709 * tree-sra.c (access): New field parent.
27710 (process_subtree_disqualification): New function.
27711 (disqualify_candidate): Call it.
27712 (build_accesses_from_assign): Reset write flag if creating an
27714 (build_access_subtree): Fill in parent field and also prpagate
27715 down grp_write flag.
27716 (create_artificial_child_access): New parameter set_grp_write, set
27717 grp_write to its value.
27718 (propagate_subaccesses_across_link): Also propagate grp_write flag
27720 (propagate_all_subaccesses): Push the closest parent back to work
27721 queue if add_access_to_work_queue returned true.
27723 2017-05-02 Richard Biener <rguenther@suse.de>
27725 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27726 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27727 -fstrict-overflow documentation.
27728 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27729 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27730 flag_strict_overflow.
27731 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27732 * lto-opts.c (lto_write_options): Do not stream it.
27733 * lto-wrapper.c (merge_and_complain): Do not handle it.
27734 * opts.c (default_options_table): Do not set -fstrict-overflow.
27735 (finish_options): Likewise do not clear it when sanitizing.
27736 * simplify-rtx.c (simplify_const_relational_operation): Do not
27737 test flag_strict_overflow.
27739 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27741 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27742 using enabled attribute.
27743 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27744 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27745 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27746 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27747 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27748 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27749 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27750 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27751 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27752 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27754 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27756 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27758 2017-05-02 Richard Biener <rguenther@suse.de>
27760 PR tree-optimization/80591
27762 2017-04-10 Richard Biener <rguenther@suse.de>
27764 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27767 2017-05-02 Richard Biener <rguenther@suse.de>
27769 PR tree-optimization/80549
27770 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27771 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27772 headers if they do not have a preheader.
27774 2017-05-02 Martin Liska <mliska@suse.cz>
27777 * common.opt: Fix typo.
27778 * doc/invoke.texi: Likewise.
27780 2017-05-01 Jan Beulich <jbeulich@suse.com>
27782 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27783 swapping, add (x,x,m,x,n) alternative.
27785 2017-05-01 Nathan Sidwell <nathan@acm.org>
27787 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27788 unnecessary unadjusted_alignment check.
27790 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27793 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27795 * gimplify.c (gimplify_cilk_detach): New function.
27796 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27797 * tree-core.h: Document EXPR_CILK_SPAWN.
27798 * tree.h (EXPR_CILK_SPAWN): Define.
27800 2017-05-01 David Malcolm <dmalcolm@redhat.com>
27802 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
27803 to use new fixit_hint representation, using the "replace" logic.
27804 (get_line_span_for_fixit_hint): Likewise.
27805 (layout::print_any_fixits): Likewise.
27806 (selftest::test_one_liner_many_fixits): Rename to...
27807 (selftest::test_one_liner_many_fixits_1): ...this, and update
27808 comment and expected output to reflect that the multiple fix-it
27809 hints are now consolidated into one insertion.
27810 (selftest::test_one_liner_many_fixits_2): New test.
27811 (selftest::test_diagnostic_show_locus_one_liner): Update for
27813 (selftest::test_fixit_consolidation): Update for fix-it API
27815 * diagnostic.c (print_parseable_fixits): Likewise.
27816 * edit-context.c (edited_line::m_line_events): Convert from
27817 auto_vec <line_event *> to auto_vec <line_event>.
27818 (class line_event): Convert from abstract base class to a concrete
27819 class, taking over the role of replace_event.
27820 (class insert_event): Delete.
27821 (class replace_event): Rename to class line_event. Convert to
27823 (edit_context::add_fixits): Reimplement.
27824 (edit_context::apply_insert): Delete.
27825 (edit_context::apply_replace): Rename to...
27826 (edit_context::apply_fixit): ...this. Convert to half-open range.
27827 (edited_file::apply_insert): Delete.
27828 (edited_file::apply_replace): Rename to...
27829 (edited_file::apply_fixit): ...this.
27830 (edited_line::~edited_line): Drop deletion of events.
27831 (edited_line::apply_insert): Delete.
27832 (edited_line::apply_replace): Rename to...
27833 (edited_line::apply_fixit): ...this. Convert to half-open range.
27834 Update for change to type of m_line_events.
27835 * edit-context.h (edit_context::apply_insert): Delete.
27836 (edit_context::apply_replace): Rename to...
27837 (edit_context::apply_fixit): ...this.
27839 2017-05-01 Martin Sebor <msebor@redhat.com>
27841 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27844 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27847 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27848 __get_cpuid_max returns 0.
27849 (__get_cpuid_count): Ditto.
27851 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27853 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27854 replacement expression is another instance of one of its arguments.
27856 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27859 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27860 check for stack push/pop autoinc.
27861 * config/i386/i386.c (ix86_agi_dependent): Return false
27862 if the only reason why modified_in_p returned true is that
27863 addr is SP based and set_insn is a push or pop.
27865 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27867 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27870 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27873 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27874 (account_size_time): Use two predicates - exec_pred and
27876 (evaluate_conditions_for_known_args): Compute both clause and
27878 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27879 (inline_summary_t::duplicate): Update.
27880 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27882 (compute_inline_parameters): Likewise.
27883 (estimate_edge_size_and_time): Update caluclation of time.
27884 (estimate_node_size_and_time): Compute both time and nonspecialized
27886 (estimate_ipcp_clone_size_and_time): Update.
27887 (inline_merge_summary): Update.
27888 (do_estimate_edge_time): Update.
27889 (do_estimate_edge_size): Update.
27890 (do_estimate_edge_hints): Update.
27891 (inline_read_section, inline_write_summary): Stream both new predicates.
27892 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27894 (compute_inlined_call_time): Cleanup.
27895 (big_speedup_p): Update.
27896 (edge_badness): Update.
27897 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27898 (size_time_entry): Replace predicate by exec_predicate and
27899 nonconst_predicate.
27900 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27901 (estimate_edge_time): Return also nonspec_time.
27902 (reset_edge_growth_cache): Update.
27904 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27906 PR rtl-optimization/80491
27907 * ifcvt.c (noce_process_if_block): When looking for x setter
27908 with missing else_bb, don't check only the insn right before
27909 cond_earliest, but look for the last insn that x is modified in
27910 within the same bb.
27912 PR rtl-optimization/80491
27913 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27915 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27917 PR tree-optimization/80487
27918 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27920 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27922 PR tree-optimization/79697
27923 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27924 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27925 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27927 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27928 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27930 2017-04-28 Martin Sebor <msebor@redhat.com>
27932 PR tree-optimization/80523
27933 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27934 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27936 (maybe_warn, format_directive, parse_directive): Use new functions.
27937 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27939 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27941 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27943 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27945 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27946 target_header_dir): Set correctly.
27947 * configure: Regenerated.
27948 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27949 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27950 instead of SYSTEM_HEADER_DIR.
27952 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27954 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27955 (estimate_local_effects): Likewise.
27956 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27957 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27958 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27959 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27960 do_estimate_edge_time, estimate_edge_time): Likewise.
27961 * ipa-inline-analysis.c (estimate_node_size_and_time,
27962 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27963 (estimate_time_after_inlining): Remove.
27965 2017-04-28 Martin Liska <mliska@suse.cz>
27967 * doc/gcov.texi: Enhance documentation of gcov.
27969 2017-04-28 Martin Liska <mliska@suse.cz>
27971 * doc/gcov.texi: Sort options in alphabetic order.
27972 * doc/gcov-dump.texi: Likewise.
27973 * doc/gcov-tool.texi: Likewise.
27974 * gcov.c (print_usage): Likewise.
27975 * gcov-dump.c (print_usage): Likewise.
27976 * gcov-tool.c (print_merge_usage_message): Likewise.
27977 (print_rewrite_usage_message): Likewise.
27978 (print_overlap_usage_message): Likewise.
27980 2017-04-28 Martin Liska <mliska@suse.cz>
27982 PR gcov-profile/53915
27983 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27985 2017-04-28 Martin Liska <mliska@suse.cz>
27987 PR gcov-profile/79891
27988 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27989 is marked by compiler as living on a line.
27990 (get_cycles_count): Remove usage of the union.
27991 (output_intermediate_file): Likewise.
27992 (find_source): Fix GNU coding style.
27993 (accumulate_line_counts): Remove old non-all block mode.
27994 (output_lines): Remove usage of the union.
27995 * profile.c (output_location): Include all BBs, even if
27996 belonging to a same line (and file) as a previous BB.
27998 2017-04-28 Martin Liska <mliska@suse.cz>
28000 * gcov.c (process_args): Handle new argument 'w'.
28001 (read_graph_file): Assign ID to BBs.
28002 (output_branch_count): Display BB # if verbose flag is set.
28003 (output_lines): Likewise for arcs.
28004 (print_usage): Add '--verbose' option help.
28005 * doc/gcov.texi: Document --verbose (-w) option.
28007 2017-04-28 Martin Liska <mliska@suse.cz>
28009 * gcov.c (struct block_location_info): New struct.
28010 (process_file): Fill up the new structure.
28011 (read_graph_file): Replace usage of encoding by the newly added
28013 (add_line_counts): Likewise.
28014 (accumulate_line_counts): Remove usage of the union.
28015 (function_info::function_info): New function.
28016 (function_info::~function_info): Likewise.
28017 (process_file): Call delete instead of release_function.
28018 (release_function): Release the function.
28019 (release_structures): Call delete instead of release_function.
28020 (solve_flow_graph): Replace usage of num_blocks.
28021 (find_exception_blocks): Likewise.
28022 (output_lines): Fix GNU coding style.
28024 2017-04-28 Martin Liska <mliska@suse.cz>
28027 * coverage.c (coverage_remove_note_file): New function.
28028 * coverage.h: Declare the function.
28029 * toplev.c (finalize): Clean if an error has been seen.
28031 2017-04-28 Martin Liska <mliska@suse.cz>
28033 PR gcov-profile/80031
28034 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28035 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28036 * gcov.c (read_graph_file): Read just number of blocks.
28037 * profile.c (branch_prob): Do not stream 0 flags per a basic
28040 2017-04-28 Martin Liska <mliska@suse.cz>
28042 * gcov-dump.c (tag_*): Add new argument to declarations.
28043 (dump_gcov_file): Likewise.
28044 (tag_blocks): Add and use new argument depth.
28045 (tag_arcs): Likewise.
28046 (tag_lines): Likewise.
28047 (tag_counters): Likewise.
28048 (tag_summary): Likewise.
28049 (dump_working_sets): Use depth to do a proper indentation.
28051 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28054 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28055 * symtab.c (symtab_node::debug_symtab): Move definition here.
28057 2017-04-28 Richard Biener <rguenther@suse.de>
28059 * lto-streamer.h (LTO_major_version): Bump to 7.
28061 2017-04-28 Richard Biener <rguenther@suse.de>
28063 * tree-vrp.c (assert_info): New struct.
28064 (add_assert_info): New helper.
28065 (register_edge_assert_for_2): Refactor to add asserts to a vector
28067 (register_edge_assert_for_1): Likewise.
28068 (register_edge_assert_for): Likewise.
28069 (finish_register_edge_assert_for): New helper actually registering
28070 asserts where live on edge.
28071 (find_conditional_asserts): Adjust.
28072 (find_switch_asserts): Likewise.
28073 (evrp_dom_walker::try_find_new_range): Generalize.
28074 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28076 2017-04-27 Marek Polacek <polacek@redhat.com>
28079 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28080 arg10 and arg11 to itype.
28082 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28084 * doc/extend.texi (Object Size Checking): Improve grammar.
28086 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28089 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28090 that the logic for permitting reciprocal estimates matches that
28093 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28096 * tree.c (type_cache_hasher::equal): Only compare
28097 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28098 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28099 non-aggregate element types.
28100 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28101 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28103 2017-04-27 Richard Biener <rguenther@suse.de>
28105 PR middle-end/80533
28106 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28107 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28108 keeping a reference to a trailing array.
28110 2017-04-27 Richard Biener <rguenther@suse.de>
28112 PR middle-end/80539
28113 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28114 being in loop-closed SSA form conservatively.
28115 (chrec_fold_multiply_poly_poly): Likewise.
28117 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28119 PR middle-end/79665
28120 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28121 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28123 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28126 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28127 (aarch64_function_arg_alignment): Return unsigned int again, but still
28128 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28129 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28130 Don't emit -Wpsabi note.
28131 (aarch64_function_arg_boundary): Likewise.
28132 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28135 2017-04-26 Nathan Sidwell <nathan@acm.org>
28137 * tree.h (crc32_unsigned_n): Declare.
28138 (crc32_unsigned, crc32_unsigned): Make inline.
28139 * tree.c (crc32_unsigned_bits): Replace with ...
28140 (crc32_unsigned_n): ... this.
28141 (crc32_unsigned, crc32_byte): Remove.
28142 (crc32_string): Remove unnecessary braces.
28144 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28146 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28147 * ipa-inline-analysis.c (MAX_TIME): Remove.
28148 (account_size_time): Use sreal for time.
28149 (dump_inline_summary): Update.
28150 (estimate_function_body_sizes): Update.
28151 (estimate_edge_size_and_time): Update.
28152 (estimate_calls_size_and_time): Update.
28153 (estimate_node_size_and_time): Update.
28154 (inline_merge_summary): Update.
28155 (inline_update_overall_summary): Update.
28156 (estimate_time_after_inlining): Update.
28157 (inline_read_section): Update.
28158 (inline_write_summary): Update.
28159 * ipa-inline.c (compute_uninlined_call_time): Update.
28160 (compute_inlined_call_time): Update.
28161 (recursive_inlining): Update.
28162 (inline_small_functions): Update.
28163 (dump_overall_stats): Update.
28164 * ipa-inline.h: Include sreal.h.
28165 (size_time_entry): Turn time to sreal.
28166 (inline_summary): Turn self_time nad time to sreal.
28168 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28170 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28172 (sreal::stream_out, sreal::stream_in): New.
28173 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28175 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28177 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28180 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28183 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28185 (dimode_scalar_chain::compute_convert_gain): Ditto.
28186 (dimode_scalar_chain::make_vector_copies): Ditto.
28187 (dimode_scalar_chain::convert_reg): Ditto.
28188 (dimode_scalar_chain::convert_insn): Ditto.
28189 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28190 (VI248_AVX512BW_1): New mode iterator.
28191 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28192 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28195 2017-04-25 Martin Sebor <msebor@redhat.com>
28197 PR tree-optimization/80497
28198 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28199 constants are representable in HOST_WIDE_INT.
28200 (parse_directive): Ditto.
28202 2017-04-25 Martin Sebor <msebor@redhat.com>
28205 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28206 (new_zero_array): Adjust signature.
28207 (dom_info::dom_init): Used unsigned rather that size_t.
28208 (dom_info::dom_info): Same.
28210 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28211 Jakub Jelinek <jakub@redhat.com>
28214 * config/arm/arm.c: Include gimple.h.
28215 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28216 returns negative, increment ncrn only if it returned positive.
28217 (arm_needs_doubleword_align): Return int instead of bool,
28218 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28219 members, but if there is any such non-FIELD_DECL
28220 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28221 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28222 returns negative, increment nregs only if it returned positive.
28223 (arm_setup_incoming_varargs): Likewise.
28224 (arm_function_arg_boundary): Emit -Wpsabi note if
28225 arm_needs_doubleword_align returns negative, return
28226 DOUBLEWORD_ALIGNMENT only if it returned positive.
28228 2017-04-25 Marek Polacek <polacek@redhat.com>
28231 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28232 first argument to type.
28234 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28238 type checks to test for compatibility instead of equality.
28240 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28241 Jakub Jelinek <jakub@redhat.com>
28244 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28246 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28247 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28248 the alignment computation, but return their maximum in warn_alignment.
28249 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28250 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28252 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28253 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28256 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28258 * config/arc/simdext.md (dmpyh): Fix typo.
28260 2017-04-25 Richard Biener <rguenther@suse.de>
28262 PR tree-optimization/80492
28263 * alias.c (compare_base_decls): Handle registers with asm
28264 specification conservatively.
28265 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28266 compare_base_decls returning dont-know properly.
28268 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28270 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28271 (legitimate_offset_address_p): New function.
28272 (arc_legitimate_address_p): Use above function.
28274 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28276 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28278 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28280 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28281 ACCH registers whenever they are available.
28283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28285 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28286 double regs fix when not used.
28288 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28290 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28292 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28293 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28295 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28297 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28298 when emitting short ADD instructions.
28300 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28302 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28304 (cmpsi_cc_c_insn): Likewise.
28305 (cbranchsi4_scratch): Compute proper instruction length using
28306 compact_hreg_operand.
28307 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28309 2017-04-25 Richard Biener <rguenther@suse.de>
28311 PR middle-end/80509
28312 * passes.c (pass_manager::pass_manager): Initialize
28313 m_name_to_pass_map.
28315 2017-04-25 Richard Biener <rguenther@suse.de>
28317 PR tree-optimization/79201
28318 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28320 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28323 * config/s390/vector.md: Split MEM->GPR vector moves for
28324 non-s_operand addresses.
28326 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28329 * config/s390/predicates.md (reload_const_wide_int_operand): New
28331 * config/s390/s390.md ("movti"): Remove d/P alternative.
28332 ("movti_bigconst"): New pattern definition.
28334 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28337 * s390-protos.h (s390_expand_cs_hqi): Removed.
28338 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28339 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28340 modes as well as CCZ1mode and CCZmode.
28341 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28342 signature of s390_emit_compare_and_swap.
28343 (s390_expand_cs_hqi): Likewise, make static.
28344 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28345 compare-and-swap, in some cases.
28346 (s390_expand_cs): Wrapper function.
28347 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28349 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28350 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28351 patterns for small and large integers. Forbid symref memory operands.
28352 Move expander to s390.c. Require cc register.
28353 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28354 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28355 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28356 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28357 symref memory operands. Remove CC mode and call s390_match_ccmode
28359 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28361 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28363 * config/s390/s390.md (define_peephole2): New peephole to help
28364 combining the load-and-test pattern with volatile memory.
28366 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28368 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28369 with CCZmode for TARGET_Z196.
28371 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28373 PR rtl-optimization/80501
28374 * combine.c (make_compound_operation_int): Set subreg_code to SET
28375 even for AND with mask of the sign bit of mode.
28377 PR rtl-optimization/80500
28378 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28379 sum's initial value.
28381 2017-04-25 Julian Brown <julian@codesourcery.com>
28382 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28384 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28386 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28388 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28390 2017-04-25 Julian Brown <julian@codesourcery.com>
28391 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28393 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28394 (thunderx2t99_sha): New Reservation.
28396 2017-04-25 Julian Brown <julian@codesourcery.com>
28397 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28399 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28400 type for 1-element load.
28402 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28404 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28406 2017-04-24 Martin Jambor <mjambor@suse.cz>
28408 PR tree-optimization/80293
28409 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28410 char arrays not totally scalarizable if it is false.
28411 (analyze_all_variable_accesses): Pass correct value in the new
28412 parameter. Add a statistics counter.
28414 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28416 PR middle-end/79931
28417 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28419 2017-04-24 Richard Biener <rguenther@suse.de>
28421 PR tree-optimization/80494
28422 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28423 out for complex types.
28425 2017-04-24 Richard Biener <rguenther@suse.de>
28427 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28428 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28429 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28430 (DFS): Adjust and never fail.
28431 (sccvn_dom_walker::fail): Remove.
28432 (sccvn_dom_walker::before_dom_children): Adjust.
28433 (run_scc_vn): Likewise and never fail.
28434 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28435 (pass_fre::execute): Likewise.
28437 2017-04-24 Richard Biener <rguenther@suse.de>
28439 PR tree-optimization/79725
28440 * tree-ssa-sink.c (statement_sink_location): Return whether
28441 failure reason was zero uses. Move that check later.
28442 (sink_code_in_bb): Deal with zero uses by removing the stmt
28445 2017-04-24 Richard Biener <rguenther@suse.de>
28448 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28449 pointer-based references.
28451 2017-04-24 Richard Biener <rguenther@suse.de>
28454 * pass_manager.h (pass_manager::operator new): Remove.
28455 (pass_manager::operator delete): Likewise.
28456 * passes.c (pass_manager::operator new): Remove.
28457 (pass_manager::operator delete): Likewise.
28458 (pass_manager::pass_manager): Zero individual pass members.
28460 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28463 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28464 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28465 Check "XEXP (src, 1)" operand here.
28466 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28467 Check "XEXP (src, 1)" operand here.
28468 (dimode_scalar_chain::make_vector_copies): Detect count register
28469 of a shift instruction. Zero extend count register from QImode
28470 to DImode to satisfy vector shift pattern count operand predicate.
28471 Substitute vector shift count operand with a DImode copy.
28472 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28475 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28477 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28478 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28479 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28480 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28481 (UNSPEC_NOREX_MEM): Remove definition.
28483 2017-04-21 Richard Biener <rguenther@suse.de>
28485 PR tree-optimization/79547
28486 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28487 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28488 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28489 without any constraints.
28491 2017-04-21 Richard Biener <rguenther@suse.de>
28493 PR tree-optimization/78847
28494 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28496 2017-04-21 Richard Biener <rguenther@suse.de>
28498 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28499 (build_distinct_type_copy): Likewise.
28500 (build_variant_type_copy): Likewise.
28501 * tree.c (build_qualified_type): Pass down mem-stat info.
28502 (build_distinct_type_copy): Likewise.
28503 (build_variant_type_copy): Likewise.
28505 2017-04-21 Richard Biener <rguenther@suse.de>
28507 PR tree-optimization/80237
28508 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28510 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28511 for a simplified result.
28513 2016-04-21 Richard Biener <rguenther@suse.de>
28515 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28516 sth as strict as a simple_iv but a chrec without symbols and an
28517 operand defined in the loop we are peeling (and not some subloop).
28518 (propagate_constants_for_unrolling): Propagate all constants.
28520 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28523 * config/i386/i386.c (print_reg): Remove assert for disalowed
28524 regno values, call output_operand_lossage instead.
28526 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28529 * config/i386/constraints.md (Yc): New register constraint.
28530 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28531 Use Yc constraint for alternative 2 of operand 0. Remove
28532 preferred_for_speed attribute.
28534 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28536 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28537 lastprivate clauses in SIMT case.
28539 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28541 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28543 2017-04-20 Richard Biener <rguenther@suse.de>
28545 PR tree-optimization/57796
28546 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28547 as N scalar stores.
28548 (vect_model_load_cost): Cost gathers as N scalar loads.
28550 2017-04-20 Richard Biener <rguenther@suse.de>
28552 * ggc-page.c (ggc_allocated_p): Rename to ...
28553 (safe_lookup_page_table_entry): ... this and return the lookup
28555 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28557 2017-04-20 Richard Biener <rguenther@suse.de>
28559 PR tree-optimization/80453
28560 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28561 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28562 from the conditions.
28563 (vn_phi_eq): Pass them down.
28564 (vn_phi_lookup): Record them.
28565 (vn_phi_insert): Likewise.
28567 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28569 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28570 uninitialized variable warning to avoid buffer overrun.
28572 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28575 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28576 is suppressed for '{ 0 }' in C.
28578 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28580 * BASE-VER: Set to 8.0.0.
28582 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28584 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28585 priority .init_array and .fini_array section with SECTION_NOTYPE
28588 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28590 PR middle-end/80423
28591 * tree.h (build_array_type): Add typeless_storage default argument.
28592 * tree.c (type_cache_hasher::equal): Also compare
28593 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28594 (build_array_type): Add typeless_storage argument, set
28595 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28597 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28598 (build_array_type): Likewise. Add typeless_storage argument.
28600 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28601 Jakub Jelinek <jakub@redhat.com>
28603 PR tree-optimization/80426
28604 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28605 operation on symbolic operands, also compute the overflow for the
28606 invariant part when the operation degenerates into a negation.
28608 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28611 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28612 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28615 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28617 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28620 * config/avr/avr.c (tree.h): Include it.
28621 (cgraph.h): Include it.
28622 (avr_encode_section_info): Don't warn for uninitialized progmem
28623 variable if it's just an alias.
28625 2017-04-19 Richard Biener <rguenther@suse.de>
28628 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28629 when needed by AutoPGO.
28631 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28634 * doc/lto.texi: Remove an extra 'that'.
28636 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28638 PR rtl-optimization/80429
28639 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28640 are only used in debug insns.
28642 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28643 Vladimir Makarov <vmakarov@redhat.com>
28645 * config/sparc/predicates.md (input_operand): Add comment. Return
28646 true for any memory operand when LRA is in progress.
28647 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28649 2017-04-18 Jeff Law <law@redhat.com>
28652 * mips.md ({return,simple_return}_internal): Do not overwrite
28655 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28657 PR tree-optimization/80443
28658 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28659 instead of adding 1, subtract -1 and similarly instead of subtracting
28662 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28664 PR rtl-optimization/80357
28665 * haifa-sched.c (tmp_bitmap): New variable.
28666 (model_recompute): Handle duplicate use records.
28667 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28668 (free_global_sched_pressure_data): Free it.
28670 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28673 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28674 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28675 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28676 instead of SYSTEM_HEADER_DIR.
28678 2017-04-18 Jeff Law <law@redhat.com>
28680 PR middle-end/80422
28681 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28682 predecessors after walking up the insn chain.
28684 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28687 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28688 sizetype type into debug info.
28690 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28693 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28694 unneeded test for TARGET_UPPER_REGS_SF.
28695 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28697 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28700 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28701 instead of gsi_after_labels.
28703 2017-04-18 Jeff Law <law@redhat.com>
28705 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28709 2017-04-13 Jeff Law <law@redhat.com>
28710 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28711 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28713 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28716 * config/avr/avr.c (intl.h): Include it.
28717 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28719 2017-04-18 Martin Liska <mliska@suse.cz>
28721 PR gcov-profile/78783
28722 * gcov-tool.c (gcov_output_files): Validate that destination
28723 file is either removed by the tool or by a user.
28725 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28726 Guy Benyei <guybe@mellanox.com>
28728 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28729 block, and do not negate it, the stored id is already negative.
28731 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28733 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28735 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28738 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28739 masks of options that should be turned off if the VSX vector
28740 options are turned off.
28741 (OTHER_P8_VECTOR_MASKS): Likewise.
28742 (OTHER_VSX_VECTOR_MASKS): Likewise.
28743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28744 rs6000_disable_incompatible_switches to validate no type switches
28746 (rs6000_incompatible_switch): New function to disallow turning on
28747 other vector options if -mno-vsx, -mno-power8-vector, or
28748 -mno-power9-vector are specified.
28750 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28752 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28754 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28756 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28757 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28758 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28759 (ARG_POINTER_CFA_OFFSET): Likewise.
28761 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28763 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28764 conditions to take advantage of various optimizations.
28766 2017-04-13 Jeff Law <law@redhat.com>
28768 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28769 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28770 (zero_extendsidi2_dext): Likewise.
28772 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28775 * fold-const.c (fold_ternary_loc): Revert
28776 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28779 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28781 PR rtl-optimization/80343
28782 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28783 new scratch pseudo.
28785 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28788 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28789 to ubsan_encode_value.
28791 2017-04-13 Jeff Law <law@redhat.com>
28793 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28794 appearing in DEBUG_INSNs.
28796 2017-04-13 Martin Liska <mliska@suse.cz>
28798 PR gcov-profile/80413
28799 * gcov-io.c (gcov_write_string): Copy to buffer just when
28800 allocated size is greater than zero.
28802 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28805 * dwarf2out.c (decls_for_scope): Ignore declarations of
28806 current_function_decl in BLOCK_NONLOCALIZED_VARS.
28808 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
28811 * ipa-visibility.c (non_local_p): Fix typos.
28812 (localize_node): When localizing symbol in same comdat group,
28813 dissolve the group only when we know external symbols are going
28815 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
28817 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28819 PR tree-optimization/79390
28820 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
28821 order does not result in usable sequence, retry with reversed operand
28827 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
28828 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
28829 op0 instead of fold_convert_loc (loc, type, arg0).
28831 2017-04-12 Jeff Law <law@redhat.com>
28833 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28834 has a delay slot in the generated code.
28836 * config/cris/cris.md (cris_preferred_reload_class): Return
28837 GENNONACR_REGS rather than GENERAL_REGS.
28839 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28842 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28843 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28844 signedness of the result type.
28846 2017-04-12 Richard Biener <rguenther@suse.de>
28847 Jeff Law <law@redhat.com>
28849 PR tree-optimization/80359
28850 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28851 trim stores to TARGET_MEM_REFs.
28853 2017-04-12 Richard Biener <rguenther@suse.de>
28855 PR tree-optimization/79390
28856 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28857 threading case even more.
28859 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28862 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28863 for quad_address_p for TImode, instead of just not indexed_address.
28865 2017-04-12 Richard Biener <rguenther@suse.de>
28866 Bernd Edlinger <bernd.edlinger@hotmail.de>
28868 PR middle-end/79671
28869 * alias.c (component_uses_parent_alias_set_from): Handle
28870 TYPE_TYPELESS_STORAGE.
28871 (get_alias_set): Likewise.
28872 * tree-core.h (tree_type_common): Add typeless_storage flag.
28873 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28874 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28875 for types containing members with TYPE_TYPELESS_STORAGE.
28876 (place_field): Likewise.
28877 (layout_type): Likewise for ARRAY_TYPE.
28878 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28879 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28880 TYPE_TYPELESS_STORAGE.
28881 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28883 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28886 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28887 first argument to type.
28889 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28893 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28894 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28895 (rs6000_expand_binop_builtin): Likewise.
28896 (rs6000_expand_ternop_builtin): Likewise; also add missing
28897 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28898 vshasigma built-ins.
28899 * doc/extend.texi: Document that vec_xxpermdi's third argument
28900 must be a constant.
28902 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28904 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28905 Use shift_const cost parameter when calculating gain of STV shifts.
28907 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28909 PR rtl-optimization/70478
28910 * lra-constraints.c (process_alt_operands): Check memory for
28911 disfavoring memory insn operand.
28913 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28915 PR middle-end/80100
28916 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28917 left shift in unsigned HOST_WIDE_INT type.
28919 PR rtl-optimization/80385
28920 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28921 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28924 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28925 if they have any depend clauses.
28927 2017-04-11 Martin Liska <mliska@suse.cz>
28930 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28931 * ipa-split.c (split_function): Create a local comdat symbol
28932 if caller is in a comdat group.
28934 2017-04-11 Martin Liska <mliska@suse.cz>
28937 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28940 2017-04-11 Martin Sebor <msebor@redhat.com>
28942 PR middle-end/80364
28943 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28944 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28946 (directive::set_width, directive::set_precision, format_character):
28948 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28951 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28954 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28955 conflict, set target->arch_name instead of target->cpu_name.
28957 2017-04-11 Richard Biener <rguenther@suse.de>
28959 PR tree-optimization/80374
28960 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28961 build_zero_cst, remove fold_convertible_p check again.
28963 2017-04-11 Martin Liska <mliska@suse.cz>
28966 * ubsan.c (instrument_object_size): Do not instrument register
28969 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28972 * config/i386/i386-builtin-types.def
28973 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28974 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28975 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28976 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28977 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28978 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28979 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28980 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28981 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28982 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28983 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28984 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28985 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28986 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28987 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28988 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28989 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28990 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28991 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28992 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28993 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28994 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28995 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28996 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28997 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28998 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28999 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29000 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29001 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29002 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29003 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29004 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29005 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29006 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29007 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29008 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29009 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29010 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29011 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29012 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29013 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29014 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29015 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29016 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29017 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29018 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29020 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29021 flag to second_arg_count, handle 4 argument function type _COUNT
29022 aliases, handle second_arg_count on second argument rather than last.
29024 2017-04-10 Jeff Law <law@redhat.com>
29026 PR tree-optimization/80374
29027 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29028 record anything if we can not convert integer_zero_node to the
29031 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29034 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29035 Enhance special handling given to the TARGET_P9_MINMAX option in
29036 relation to certain other options.
29038 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29040 PR tree-optimization/80153
29041 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29042 remove POINTER_PLUS_EXPR's base part directly, rather than through
29045 2017-04-10 Richard Biener <rguenther@suse.de>
29046 Bin Cheng <bin.cheng@arm.com>
29048 PR tree-optimization/80153
29049 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29050 the first element of pointer type aff_tree. Build result expr in
29052 (add_elt_to_tree): Convert to type unconditionally. Remove other
29053 fold_convert calls.
29054 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29055 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29057 2017-04-10 Richard Biener <rguenther@suse.de>
29059 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29062 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29064 PR rtl-optimization/70478
29065 * lra-constraints.c (curr_small_class_check): New.
29066 (update_and_check_small_class_inputs): New.
29067 (process_alt_operands): Update curr_small_class_check. Disfavor
29068 alternative insn memory operands. Check available regs for small
29071 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29074 * config/mips/mips.opt (-mvirt): Update description.
29075 * doc/invoke.texi (-mvirt): Likewise.
29077 2017-04-10 Richard Biener <rguenther@suse.de>
29079 PR middle-end/80362
29080 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29081 looking for NEGATE_EXPR in -A / -B to A / B folding.
29083 2017-04-10 Martin Liska <mliska@suse.cz>
29085 PR gcov-profile/80224
29086 * gcov.c (print_usage): Fix usage string.
29087 (get_gcov_intermediate_filename): Remove.
29088 (output_gcov_file): Use both for normal and intermediate format.
29089 (generate_results): Do not initialize special file for
29090 intermediate format.
29092 2017-04-10 Richard Biener <rguenther@suse.de>
29094 PR tree-optimization/80304
29095 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29098 2017-04-10 Nathan Sidwell <nathan@acm.org>
29101 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29102 (rs6000_init_builtins): Use it.
29104 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29106 * config/arm/arm.md (<mrc>): Add mode to SET source.
29107 (<mrrc>): Likewise.
29109 2017-04-10 Richard Biener <rguenther@suse.de>
29111 PR middle-end/80344
29112 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29114 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29117 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29118 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29119 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29120 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29121 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29122 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29123 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29124 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29125 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29126 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29127 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29128 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29129 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29130 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29131 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29132 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29133 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29134 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29135 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29136 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29137 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29138 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29139 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29141 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29143 PR rtl-optimization/70478
29144 * lra-constraints.c: Reverse the last patch.
29146 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29148 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29149 Add comment for WCHAR_T.
29151 2017-04-08 Martin Liska <mliska@suse.cz>
29154 2017-04-07 Martin Liska <mliska@suse.cz>
29157 * ipa-split.c (split_function): Add function part to a same comdat
29160 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29163 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29165 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29167 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29168 (vec_pairing): ...to this.
29169 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29170 (rs6000_sched_init): Adjust for name change.
29171 (struct rs6000_sched_context): Likewise.
29172 (rs6000_init_sched_context): Likewise.
29173 (rs6000_set_sched_context): Likewise.
29175 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29181 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29183 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29184 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29185 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29187 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29189 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29191 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29193 PR rtl-optimization/70703
29194 * ira-color.c (update_conflict_hard_regno_costs): Use
29195 int64_t instead of HOST_WIDE_INT.
29197 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29199 PR rtl-optimization/70478
29200 * lra-constraints.c (process_alt_operands): Disfavor alternative
29201 insn memory operands.
29203 2017-04-07 Jeff Law <law@redhat.com>
29205 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29206 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29208 2017-04-07 Martin Liska <mliska@suse.cz>
29211 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29212 Show error message instead of an ICE.
29214 2017-04-07 Martin Liska <mliska@suse.cz>
29217 * ipa-split.c (split_function): Add function part to a same comdat
29220 2017-04-07 Richard Biener <rguenther@suse.de>
29222 PR middle-end/80341
29223 * tree.c (get_unwidened): Also handle ! for_type case for
29225 * convert.c (do_narrow): Split out from ...
29226 (convert_to_integer_1): ... here. Do not pass final truncation
29227 type to get_unwidened for TRUNC_DIV_EXPR.
29229 2017-04-07 Richard Biener <rguenther@suse.de>
29231 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29233 (aff_combination_const): Adjust.
29234 (aff_combination_scale): Likewise.
29235 (aff_combination_add_elt): Likewise.
29236 (aff_combination_add_cst): Likewise.
29237 (aff_combination_convert): Likewise.
29238 (add_elt_to_tree): Likewise. Remove unused argument.
29239 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29241 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29243 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29245 * config/arm/arm.c (arm_default_short_enums): Use
29246 ARM_DEFAULT_SHORT_ENUMS.
29247 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29249 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29252 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29253 members with redundant out-of-class redeclaration.
29255 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29258 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29259 * config/i386/i386.md (*zero_extendsidi2):
29260 Add (?*x,*x) and (?*v,*v) alternatives.
29262 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29265 * config/i386/i386.c (ix86_expand_builtin)
29266 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29267 mode from insn data. Convert operands to insn operand mode.
29268 Copy operands that don't satisfy insn predicate to a register.
29270 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29272 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29275 2017-04-06 Richard Biener <rguenther@suse.de>
29277 PR tree-optimization/80334
29278 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29279 preserve alignment of accesses.
29281 2017-04-06 Richard Biener <rguenther@suse.de>
29283 PR tree-optimization/80262
29284 * tree-sra.c (build_ref_for_offset): Preserve address-space
29286 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29287 Drop useless address-space information on MEM_REF offsets.
29289 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29291 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29293 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29295 PR rtl-optimization/70703
29296 * ira-color.c (update_conflict_hard_regno_costs): Use
29297 HOST_WIDE_INT instead of long.
29299 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29302 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29303 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29305 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29306 for x86_64 target. Handle -m3dnowa option.
29308 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29310 PR rtl-optimization/70703
29311 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29312 (update_conflict_hard_regno_costs): Use long instead of unsigned
29313 arithmetic for cost calculation.
29315 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29316 Bernd Edlinger <bernd.edlinger@hotmail.de>
29319 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29322 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29325 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29326 ptr_mode with Pmode throughout.
29327 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29328 into probe_stack_range and use DImode.
29330 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29333 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29334 call_eh_return is true.
29336 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29338 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29339 Initialize last_match_fntype_index.
29341 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29344 * tree-nvr.c: Include internal-fn.h.
29345 (pass_return_slot::execute): Ignore internal calls without
29348 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29349 Richard Biener <rguenther@suse.de>
29352 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29353 captures used multiple times, except for the last use.
29354 * generic-match-head.c: Include gimplify.h.
29356 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29358 PR tree-optimization/79390
29359 * target.h (struct noce_if_info): Declare.
29360 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29361 * target.def (noce_conversion_profitable_p): New target hook.
29362 * ifcvt.h (struct noce_if_info): New type, moved from ...
29363 * ifcvt.c (struct noce_if_info): ... here.
29364 (noce_conversion_profitable_p): Renamed to ...
29365 (default_noce_conversion_profitable_p): ... this. No longer
29367 (noce_try_store_flag_constants, noce_try_addcc,
29368 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29369 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29370 instead of noce_conversion_profitable_p.
29371 * config/i386/i386.c: Include ifcvt.h.
29372 (ix86_option_override_internal): Don't override
29373 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29374 (ix86_noce_conversion_profitable_p): New function.
29375 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29376 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29377 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29378 * doc/tm.texi: Regenerated.
29380 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29382 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29385 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29388 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29389 instructions for small multiply cores.
29391 2017-04-04 Jeff Law <law@redhat.com>
29393 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29395 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29396 that are not set by the loop over the elements.
29398 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29401 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29402 int mode, convert_modes it to mode as unsigned, otherwise use
29403 lowpart_subreg to mode rather than SImode.
29404 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29405 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29406 Use DImode instead of SImode for the shift count operand.
29407 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29410 2017-04-04 Richard Biener <rguenther@suse.de>
29412 PR middle-end/80281
29413 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29414 arithmetic done for the negate or the plus. Simplify.
29415 (A - (-B) -> A + B): Likewise.
29416 * fold-const.c (split_tree): Make sure to not negate pointers.
29418 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29420 PR rtl-optimization/60818
29421 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29422 a compare of comparisons with the thing compared if this results
29423 in a different machine mode.
29425 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29427 * alias.c (base_alias_check): Fix typo in comment.
29428 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29429 * cgraphunit.c (symbol_table::compile): Likewise.
29430 * collect2.c (maybe_run_lto_and_relink): Likewise.
29431 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29432 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29433 * config/avr/avr.c (avr_map_op_t): Likewise.
29434 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29435 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29436 * config/epiphany/epiphany.md (movcc): Likewise.
29437 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29438 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29440 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29441 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29442 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29443 * config/sh/sh.c (sh_rtx_costs): Likewise.
29444 * fold-const.c (fold_truth_andor): Likewise.
29445 * genautomata.c (collapse_flag): Likewise.
29446 * gengtype.h (struct type::u::s): Likewise.
29447 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29448 * input.c (FORMAT_AMOUNT): Likewise.
29449 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29450 (known_aggs_to_agg_replacement_list): Likewise.
29451 * ipa-inline-analysis.c: Likewise.
29452 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29453 * ipa-polymorphic-call.c
29454 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29455 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29456 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29458 * modulo-sched.c (apply_reg_moves): Likewise.
29459 * omp-expand.c (build_omp_regions_1): Likewise.
29460 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29461 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29462 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29463 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29464 * value-prof.c: Likewise.
29465 * var-tracking.c (val_reset): Likewise.
29467 2017-04-03 Richard Biener <rguenther@suse.de>
29469 PR tree-optimization/80275
29470 * fold-const.c (split_address_to_core_and_offset): Handle
29473 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29475 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29476 descriptors is at least equal to that of functions.
29478 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29480 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29482 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29485 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29486 (mov<IMOD4:mode>): New expander.
29487 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29489 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29491 PR rtl-optimization/79405
29492 * fwprop.c (propagations_left): New variable.
29493 (forward_propagate_into): Decrement it.
29494 (fwprop_init): Initialize it.
29495 (fw_prop): If the variable has reached zero, stop propagating.
29496 (fwprop_addr): Ditto.
29498 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29501 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29502 a FUNCTION_DECL, pass it as decl instead of origin to
29505 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29507 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29510 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29513 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29514 TARGET_VSX_SMALL_INTEGER.
29516 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29518 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29519 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29521 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29523 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29524 extraction from odd-numbered MSA register.
29526 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29528 PR middle-end/80173
29529 * expmed.c (store_bit_field_1): Don't attempt to create
29530 a word subreg out of hard registers wider than word if they
29531 have HARD_REGNO_NREGS of 1 for their mode.
29533 PR middle-end/80163
29534 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29535 conversions to integer types wider than word and pointer.
29538 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29539 (rtx_equal_for_cselib_p): Pass 0 to it.
29540 * cselib.c (cselib_hasher::equal): Likewise.
29541 (rtx_equal_for_cselib_1): Add depth argument. If depth
29542 is 128, don't look up VALUE locs and punt. Increment
29543 depth in recursive calls when walking VALUE locs.
29545 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29547 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29548 (make_gcov_file_name): Use the canonical path name for generating
29550 (read_line): Fix handling of files with ascii null bytes.
29552 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29554 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29555 to initialise a vector register instead
29556 of using a const_int.
29558 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29560 PR translation/80189
29561 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29562 diagnostic messages.
29564 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29567 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29568 (dfp_diex_<mode>): Update mode of operand 1.
29569 * doc/extend.texi (dxex, dxexq): Document change to return type.
29570 (diex, diexq): Document change to argument type.
29572 2017-03-30 Martin Jambor <mjambor@suse.cz>
29575 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29576 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29577 it reflects the signature changes performed at the callee side.
29578 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29579 to cgraph_build_function_type_skip_args.
29580 (build_function_decl_skip_args): Adjust call to the above function.
29582 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29585 * config/i386/sse.md
29586 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29587 register as dest whenever it is a MEM not rtx_equal_p to the
29588 corresponding dup operand, and when forcing into reg move the
29589 reg into the memory afterwards.
29590 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29591 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29592 for the force_reg mode.
29593 (avx512vl_vextractf128<mode>): Use register as dest either
29594 always when a MEM, or when it is a MEM not rtx_equal_p to the
29595 corresponding dup operand, or even not when it is a CONST_VECTOR
29596 depending on the mode and lo vs. hi.
29597 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29599 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29600 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29601 Likewise. Require that operands[2] is even.
29602 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29603 Remove extraneous parens. Require that operands[2] is a multiple
29605 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29606 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29607 disallow memory then.
29609 2017-03-30 Richard Biener <rguenther@suse.de>
29611 PR tree-optimization/77498
29612 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29613 to non-constants over backedges.
29615 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29617 PR rtl-optimization/80233
29618 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29619 as last_combined_insn. Do not test for BARRIER_P separately.
29621 2017-03-29 Andreas Schwab <schwab@suse.de>
29624 * calls.c (prepare_call_address): Convert funexp to Pmode before
29625 copying to temp reg.
29627 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29629 PR tree-optimization/80158
29630 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29631 Handle possible future case of more than one alternate
29633 (replace_rhs_if_not_dup): Likewise.
29634 (replace_one_candidate): Likewise.
29636 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29638 PR rtl-optimization/80193
29639 * ira.c (ira): Do not check allocation for LRA.
29641 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29643 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29644 (nvptx_output_simt_exit): Declare.
29645 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29646 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29647 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29648 (nvptx_declare_function_name): ...here. Emit declaration of local
29649 memory space buffer for omp_simt_enter insn.
29650 (nvptx_output_unisimt_switch): New.
29651 (nvptx_output_softstack_switch): New.
29652 (nvptx_output_simt_enter): New.
29653 (nvptx_output_simt_exit): New.
29654 * config/nvptx/nvptx.h (struct machine_function): New fields
29655 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29656 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29657 (UNSPECV_SIMT_EXIT): Ditto.
29658 (omp_simt_enter_insn): New insn.
29659 (omp_simt_enter): New expansion.
29660 (omp_simt_exit): New insn.
29661 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29663 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29664 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29665 (expand_GOMP_SIMT_EXIT): New.
29666 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29667 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29668 (GOMP_SIMT_EXIT): Ditto.
29669 * target-insns.def (omp_simt_enter): New insn.
29670 (omp_simt_exit): Ditto.
29671 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29673 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29674 (lower_rec_input_clauses): Likewise.
29675 (lower_lastprivate_clauses): Handle SIMT privatization.
29677 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29678 (ompdevlow_adjust_simt_enter): New.
29679 (find_simtpriv_var_op): New.
29680 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29681 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29683 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29684 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29685 (copy_decl_for_dup_finish): Ditto.
29687 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29689 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29692 * config/i386/i386.c (ix86_option_override_internal): Always
29693 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29695 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29697 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29699 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29701 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29702 mark new edge's irreducible flag accordign to it.
29703 (vect_do_peeling): Check loop preheader edge's irreducible flag
29704 and pass it to function slpeel_add_loop_guard.
29706 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29708 PR tree-optimization/80218
29709 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29710 Update block frequencies and counts.
29712 2017-03-28 Richard Biener <rguenther@suse.de>
29714 PR tree-optimization/78644
29715 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29716 of a simplification result we may not use it at all.
29718 2017-03-28 Richard Biener <rguenther@suse.de>
29721 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29722 without arguments, generate default definition of a SSA name.
29724 2017-03-28 Richard Biener <rguenther@suse.de>
29726 PR middle-end/80222
29727 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29728 TYPE_REF_CAN_ALIAS_ALL references.
29729 * fold-const.c (fold_indirect_ref_1): Likewise.
29731 2017-03-28 Martin Liska <mliska@suse.cz>
29734 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29735 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29737 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29738 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29740 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29741 (EXTRA_SPECS): Define.
29742 (SUBTARGET_EXTRA_SPECS): Likewise.
29743 (SUBTARGET_CPP_SPEC): Likewise.
29744 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29745 SUBTARGET_EXTRA_SPECS.
29746 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29748 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29750 * config/arc/simdext.md (vst64_insn): Update pattern.
29751 (vld32wh_insn): Likewise.
29752 (vld32wl_insn): Likewise.
29753 (vld64_insn): Likewise.
29754 (vld32_insn): Likewise.
29756 2017-03-28 Marek Polacek <polacek@redhat.com>
29759 * fold-const.c (fold_comparison): Use protected_set_expr_location
29760 instead of SET_EXPR_LOCATION.
29762 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29764 * tree.c (add_expr): Avoid name lookup warning.
29766 2017-03-27 Jeff Law <law@redhat.com>
29768 PR tree-optimization/80216
29769 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29770 function name. Limit recursion depth.
29771 (record_temporary_equivalences): Corresponding changes.
29773 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29775 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29778 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29781 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29783 * cfgcleanup.c (reg_note_cfa_p): New array.
29784 (insns_have_identical_cfa_notes): New function.
29785 (old_insns_match_p): Don't cross-jump in between /f
29786 and non-/f instructions. If both i1 and i2 are frame related,
29787 verify all CFA notes, their order and content.
29789 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29792 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29793 HImode and SImode with zero extend to DImode to one insn.
29794 (bswap<mode>2_extenddi): Likewise.
29795 (bswapsi2_extenddi): Likewise.
29796 (bswaphi2_extendsi): Likewise.
29797 (bswaphi2): Combine bswap HImode and SImode into one insn.
29798 Separate memory insns from swapping register.
29799 (bswapsi2): Likewise.
29800 (bswap<mode>2): Likewise.
29801 (bswaphi2_internal): Delete, no longer used.
29802 (bswapsi2_internal): Likewise.
29803 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
29804 store, and gpr<-gpr swap insns.
29805 (bswap<mode>2_store): Likewise.
29806 (bswaphi2_reg): Register only splitter, combine with the splitter.
29807 (bswaphi2 splitter): Likewise.
29808 (bswapsi2_reg): Likewise.
29809 (bswapsi2 splitter): Likewise.
29810 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
29811 the insns into load, store, and register/register insns.
29812 (bswapdi2_ldbrx): Likewise.
29813 (bswapdi2_load): Likewise.
29814 (bswapdi2_store): Likewise.
29815 (bswapdi2_reg): Likewise.
29817 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
29819 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
29820 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
29822 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
29825 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
29827 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29828 special handling for target option conflicts between dform
29829 options (-mpower9-dform, -mpower9-dform-vector,
29830 -mpower9-dform-scalar) and -mno-direct-move.
29832 2017-03-27 Richard Biener <rguenther@suse.de>
29834 PR tree-optimization/80181
29835 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29837 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29839 * config/arc/predicates.md (move_double_src_operand): Replace the
29840 call to move_double_src_operand with a call to address_operand.
29842 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29844 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29845 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29846 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29848 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29850 * config/arc/predicates.md (long_immediate_loadstore_operand):
29851 Consider scaled addresses cases.
29853 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29855 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29856 restored when in interrupt.
29857 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29858 doesn't have delay slot.
29860 2017-03-27 Richard Biener <rguenther@suse.de>
29863 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29864 inlined thunk clones.
29866 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29869 * asan.c (instrument_derefs): Copy over last operand from
29870 original COMPONENT_REF to the new COMPONENT_REF with
29871 DECL_BIT_FIELD_REPRESENTATIVE.
29872 * ubsan.c (instrument_object_size): Likewise.
29874 2017-03-27 Richard Biener <rguenther@suse.de>
29876 PR tree-optimization/80170
29877 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29878 sure DR/SCEV didnt fold in constants we do not see when looking
29879 at the reference base alignment.
29881 2017-03-27 Richard Biener <rguenther@suse.de>
29883 PR middle-end/80171
29884 * gimple-fold.c (fold_ctor_reference): Properly guard against
29885 NULL return value from canonicalize_constructor_val.
29887 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29890 * config/i386/i386.c (ix86_expand_builtin)
29891 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29892 flags reg setting and flags reg using instructions.
29893 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29894 clobbering instructions to zero extend op2.
29896 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29898 * doc/install.texi (Configuration) <--with-aix-soname>:
29899 Update link to AIX ld.
29901 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29903 PR rtl-optimization/80160
29904 PR rtl-optimization/80159
29905 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29906 reg_alternate_class into account.
29908 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29911 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29912 to consider in curr_insn_transform.
29914 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29916 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29917 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29918 and emit_mode_inner.
29920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29922 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29923 argument to the overloaded builtin variants. Use the new flag to
29924 deprecate certain builtin variants.
29925 * config/s390/s390-builtin-types.def: Add new builtin types.
29926 * config/s390/s390-builtins.h: Support new flags field for
29927 overloaded builtins.
29928 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29929 (s390_macro_to_expand): Enable vector float data type.
29930 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29931 builtins by incrementing the __VEC__ version number.
29932 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29934 (s390_resolve_overloaded_builtin): Emit error messages depending
29935 on the builtin flags.
29936 * config/s390/s390.c (s390_expand_builtin): Support additional
29937 flags argument. Change error message to match the messages
29938 emitted in s390-c.c.
29939 * config/s390/s390.md: New UNSPEC_* constants.
29940 (op_type): Add new instruction types.
29941 * config/s390/vecintrin.h: Add new builtins and test data class
29943 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29944 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29945 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29946 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29947 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29948 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29950 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29951 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29952 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29953 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29955 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29956 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29957 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29958 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29960 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29961 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29962 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29963 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29964 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29965 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29966 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29968 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29969 ("vec_scatter_element<V_HW_4:mode>_DI")
29970 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29971 ("vec_fpint<mode>", "vflls")
29972 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29973 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29974 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29975 ("*vec_cmphe<mode>_cc"): ... these.
29977 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29978 mode constant instead of magic value.
29980 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29982 * config/s390/s390.c (s390_expand_vec_compare): Support other
29983 vector floating point modes than just V2DF.
29984 (s390_expand_vcond): Likewise.
29985 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29986 (s390_cannot_change_mode_class): Prevent mode changes between TF
29987 and V1TF in vector registers.
29988 * config/s390/s390.md (DF, SF): New mode attributes.
29989 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29990 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29991 SFmode support for VRs.
29992 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29994 (VFT, VF_HW): New mode iterators.
29995 (vw, sdx): New mode attributes.
29996 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29997 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29998 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29999 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30000 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30001 also the new vector floating point modes. Renaming to ...
30003 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30004 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30005 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30006 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30007 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30008 ("vec_unordered<mode>"): ... these.
30010 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30011 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30012 ("*vec_extendv2df"): New insn definitions.
30014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30016 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30017 ("mulditi3_2", "*muldi3_sign"): New patterns.
30018 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30019 rename the pattern definition.
30021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30023 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30025 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30027 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30029 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30030 instruction if possible.
30031 * config/s390/vector.md (vec_halfnumelts): New mode
30033 ("*vec_vllezlf<mode>"): New pattern.
30035 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30037 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30038 ("popcountv4si2", "popcountv2di2"): Rename to ...
30039 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30040 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30042 ("popcount<mode>2_vxe"): New pattern.
30044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30046 * common/config/s390/s390-common.c (processor_flags_table): Add
30048 * config.gcc: Add arch12.
30049 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30050 Default to arch12 for unknown CPU model numbers.
30051 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30052 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30053 PROCESSOR_max sanity check.
30054 * config/s390/s390-opts.h (enum processor_type): Add
30056 * config/s390/s390.c (processor_table): Add arch12.
30057 (s390_expand_builtin): Add check for B_VXE flag.
30058 (s390_issue_rate): Add PROCESSOR_ARCH12.
30059 (s390_get_sched_attrmask): Likewise.
30060 (s390_get_unit_mask): Likewise.
30061 (s390_sched_score): Enable z13 scheduling for arch12.
30062 (s390_sched_reorder): Likewise.
30063 (s390_sched_variable_issue): Likewise.
30064 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30066 (s390_tune_attr): Use z13 scheduling also for arch12.
30067 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30068 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30069 (TARGET_VXE_P): New macros.
30070 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30071 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30072 * config/s390/s390.opt: Add arch12 as processor_type.
30074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30076 * config/s390/s390.md
30077 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30078 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30079 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30081 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30082 Rename expanders to ...
30084 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30085 ("fixuns_truncdddi2_emu"): ... these.
30087 ("fixuns_trunc<mode>si2_emu"): New expander.
30089 ("*fixuns_truncdfdi2_z13"): Rename to ...
30090 ("*fixuns_truncdfdi2_vx"): ... this.
30092 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30094 * config/s390/2964.md: Remove the single element vector compare
30095 instructions which are no longer used.
30096 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30098 (s390_canonicalize_comparison): Remove handling of DFmode
30100 (s390_expand_vec_compare_scalar): Remove function.
30101 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30102 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30104 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30106 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30108 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30110 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30111 will anyway by matched by mov<mode>_64dfp.
30113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30115 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30116 vlef/vstef. Add missing operand to vleif.
30118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30120 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30121 pair for all vector types with 64 bit elements.
30122 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30123 * config/s390/vector.md (V_HW_64): ... here.
30124 (V_128_NOSINGLE): New mode iterator.
30125 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30126 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30127 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30128 ("*vec_load_pairv2di"): Change to ...
30129 ("*vec_load_pair<mode>"): ... this one.
30131 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30133 * config/s390/constraints.md: Add comments.
30134 (jKK): Reject element sizes > 8 bytes.
30135 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30137 * config/s390/s390.md: Add the s_operand checks formerly in
30138 s390_split_ok_p to various splitters where they are still
30140 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30141 for 128 bit vectors. Plus two splitters.
30143 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30145 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30148 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30151 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30152 error if the boundary argument is not constant.
30154 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30156 PR rtl-optimization/80112
30157 * loop-doloop.c (doloop_condition_get): Don't check condition
30158 if cmp isn't SET with IF_THEN_ELSE src.
30160 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30162 PR tree-optimization/80158
30163 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30164 replacing a candidate statement, also replace it for the
30165 candidate's alternate interpretation.
30166 (replace_rhs_if_not_dup): Likewise.
30167 (replace_one_candidate): Likewise.
30169 2017-03-24 Richard Biener <rguenther@suse.de>
30171 PR tree-optimization/80167
30172 * graphite-isl-ast-to-gimple.c
30173 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30175 (translate_isl_ast_to_gimple::get_rename): Likewise.
30177 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30180 handling of certain combinations of target options, including the
30181 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30182 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30184 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30187 * config/arm/arm.md (*load_multiple): Add reload_completed to
30188 matching condition.
30190 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30191 Richard Biener <rguenth@suse.de>
30193 PR tree-optimization/79908
30194 PR tree-optimization/80136
30195 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30196 been cast away, gimplify_and_add suffices.
30198 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30200 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30202 2017-03-23 Richard Biener <rguenther@suse.de>
30204 PR tree-optimization/80032
30205 * gimplify.c (gimple_push_cleanup): Forced unconditional
30206 cleanups still have to go to the conditional_cleanups
30209 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30211 PR tree-optimization/80072
30212 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30214 (next_operand_entry_id): Change type to unsigned int.
30215 (sort_by_operand_rank): Make sure to return the right return value
30216 even if unsigned fields are bigger than INT_MAX.
30217 (struct oecount): Change cnt and id type to unsigned int.
30218 (oecount_hasher::equal): Formatting fix.
30219 (oecount_cmp): Make sure to return the right return value
30220 even if unsigned fields are bigger than INT_MAX.
30221 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30224 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30225 TREE_READONLY on result if writing it more than once.
30228 * doc/invoke.texi (-fsanitize=thread): Document that with
30229 -fnon-call-exceptions atomics are not able to throw
30233 * tsan.c: Include tree-eh.h.
30234 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30235 maybe_clean_or_replace_eh_stmt where needed.
30236 (instrument_memory_accesses): Add cfg_changed argument.
30237 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30238 if it returned true.
30239 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30241 PR rtl-optimization/63191
30242 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30243 wrapper function, moved the whole old content into ...
30244 (ix86_delegitimize_address_1): ... this. New inline function.
30245 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30246 true as last argument instead of ix86_delegitimize_address.
30248 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30250 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30251 cortexa57_branch_cost.
30253 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30255 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30257 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30260 * doc/md.texi (Constraints): Document wA constraint.
30261 * config/rs6000/constraints.md (wA): New.
30262 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30263 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30264 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30265 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30267 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30270 * gimplify.c (is_oacc_declared): New function.
30271 (oacc_default_clause): Use it to set default flags for acc declared
30272 variables inside parallel regions.
30273 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30274 declared variables.
30275 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30276 declare attribute to any decl as necessary.
30278 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30281 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30282 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30283 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30284 (arm_arch_lpae): This.
30285 * config/arm/arm.c (arm_arch7ve): Rename into ...
30286 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30287 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30290 2017-03-22 Martin Liska <mliska@suse.cz>
30293 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30294 error message instead of an ICE.
30296 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30298 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30300 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30302 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30307 2017-03-21 Martin Sebor <msebor@redhat.com>
30309 * doc/extend.texi: Use "cannot" instead of "can't."
30310 * doc/hostconfig.texi: Same.
30311 * doc/install.texi: Same.
30312 * doc/invoke.texi: Same.
30313 * doc/loop.texi: Same.
30314 * doc/md.texi: Same.
30315 * doc/objc.texi: Same.
30316 * doc/rtl.texi: Same.
30317 * doc/tm.texi: Same.
30318 * doc/tm.texi.in: Same.
30319 * doc/trouble.texi: Same.
30321 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30324 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30325 (collect_checksum_attributes): Set it.
30326 (die_checksum_ordered): Use it.
30328 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30330 PR tree-optimization/79908
30331 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30332 change: For a VA_ARG whose LHS has been cast away, use
30333 force_gimple_operand to construct the side effects.
30335 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30337 PR translation/80001
30338 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30339 more amenable to translation.
30340 (oacc_loop_auto_partitions): Likewise.
30342 2017-03-21 Marek Polacek <polacek@redhat.com>
30343 Martin Sebor <msebor@redhat.com>
30345 PR tree-optimization/80109
30346 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30347 on INTEGRAL_TYPE_P.
30349 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30350 Segher Boessenkool <segher@kernel.crashing.org>
30353 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30354 check reg_used_between_p between insn and one of succ or succ2
30355 depending on if succ is artificial insn not inserted into insn
30358 2017-03-21 Martin Liska <mliska@suse.cz>
30360 PR gcov-profile/80081
30361 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30362 * doc/gcc.texi: Include gcov-dump stuff.
30363 * doc/gcov-dump.texi: New file.
30365 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30367 PR rtl-optimization/79150
30368 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30369 conditional jump, if the jump is the last insn of the loop.
30371 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30372 Richard Biener <rguenth@suse.de>
30374 PR tree-optimization/79908
30375 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30376 been cast away, use force_gimple_operand to construct the side
30379 2017-03-21 Martin Liska <mliska@suse.cz>
30381 PR libfortran/79956
30382 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30385 2017-03-21 Brad Spengler <spender@grsecurity.net>
30388 * plugin.c (htab_hash_plugin): New function.
30389 (add_new_plugin): Use it and adjust.
30390 (parse_plugin_arg_opt): Adjust.
30391 (init_one_plugin): Likewise.
30393 2017-03-21 Richard Biener <rguenther@suse.de>
30395 PR tree-optimization/80032
30396 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30397 if set force the cleanup to happen unconditionally.
30398 (gimplify_target_expr): Push inserted clobbers with force_uncond
30399 to avoid them being removed by control-dependent DCE.
30401 2017-03-21 Richard Biener <rguenther@suse.de>
30403 PR tree-optimization/80122
30404 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30405 va_arg_pack_len when the inlined call stmt requires pack
30407 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30409 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30412 * tsan.c (instrument_builtin_call): If the memory model argument
30413 is not a constant, assume it is valid.
30416 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30419 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30421 PR rtl-optimization/79910
30422 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30423 if its dest is used by an insn before I2 (other than the combined
30424 insns themselves, which are properly handled already).
30426 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30429 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30431 * combine.c (record_used_regs): New static function.
30432 (try_combine): Handle situations where there is an additional
30433 instruction between I2 and I3 which needs to have a LOG_LINK
30437 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30439 * combine.c (try_combine): Delete redundant i1 test. Call
30440 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30442 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30445 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30446 alternatives 13/14.
30448 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30450 PR tree-optimization/80054
30451 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30452 the optimization if a PHI or any of its arguments is not dominated
30453 by the candidate's basis. Use gphi* rather than gimple* as
30455 (replace_profitable_candidates): Clean up a gimple* variable that
30456 should be a gphi* variable.
30458 2017-03-20 Martin Sebor <msebor@redhat.com>
30461 * doc/extend.texi (attribute constructor): Document present limitation.
30463 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30466 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30467 __POWER9_VECTOR__ #ifdef control, change template definition to
30468 use Power9-specific built-in function.
30469 (vec_any_eq): Likewise.
30470 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30471 to control outcomes from this test.
30472 (vector_ae_<mode>p): For VEC_F modes, likewise.
30474 2017-03-20 Ian Lance Taylor <iant@google.com>
30476 * config/i386/i386.c (ix86_function_regparm): Save an extra
30477 register for -fsplit-stack with DECL_STATIC_CHAIN.
30479 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30482 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30483 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30485 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30487 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30489 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30492 2017-03-20 Marek Polacek <polacek@redhat.com>
30495 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30497 2017-03-20 Richard Biener <rguenther@suse.de>
30499 PR tree-optimization/80113
30500 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30501 allocate extra SSA name for PHI def.
30502 (add_close_phis_to_outer_loops): Likewise.
30503 (add_close_phis_to_merge_points): Likewise.
30504 (copy_loop_close_phi_args): Likewise.
30505 (copy_cond_phi_nodes): Likewise.
30507 2017-03-20 Martin Liska <mliska@suse.cz>
30509 PR middle-end/79753
30510 * tree-chkp.c (chkp_build_returned_bound): Do not build
30511 returned bounds for a LHS that's not a BOUNDED_P type.
30513 2017-03-20 Martin Liska <mliska@suse.cz>
30517 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30518 COMPLEX_CST and VECTOR_CST.
30520 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30523 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30524 target operand. A new splitter adds the clobber statement in case
30525 the target operand is dead anyway.
30527 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30529 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30530 to age-old versions of binutils and glibc.
30532 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30534 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30536 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30538 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30540 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30542 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30543 requirement for binutils 2.13.
30545 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30547 * combine.c (try_combine): Delete redundant i1 test. Call
30548 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30550 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30552 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30553 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30555 <riscv64-*-elf>: Re-arrange section
30556 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30557 <riscv32-*-linux>: Likewise.
30558 <riscv64-*-elf>: Likewise
30559 <riscv64-*-linux>: Likewise.
30561 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30564 * aarch64.opt(verbose-cost-dump): Fix typo.
30566 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30569 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30570 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30572 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30574 * reload.c (find_reloads): When reloading a nonoffsettable address,
30575 use RELOAD_OTHER for it and its address reloads.
30577 PR rtl-optimization/79910
30578 * combine.c (record_used_regs): New static function.
30579 (try_combine): Handle situations where there is an additional
30580 instruction between I2 and I3 which needs to have a LOG_LINK
30583 2017-03-17 Jeff Law <law@redhat.com>
30585 PR tree-optimization/71437
30586 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30587 conditional in the hash table first.
30588 (vrp_dom_walker::before_dom_children): Extract condition from
30589 ASSERT_EXPR. Record condition, its inverion and any implied
30590 conditions as well.
30592 2017-03-17 Marek Polacek <polacek@redhat.com>
30593 Markus Trippelsdorf <markus@trippelsdorf.de>
30595 PR tree-optimization/80079
30596 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30599 2017-03-17 Richard Biener <rguenther@suse.de>
30601 PR middle-end/80075
30602 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30603 Properly verify the LHS before the RHS possibly claims to be
30605 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30608 2017-03-17 Martin Jambor <mjambor@suse.cz>
30610 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30611 (List of -O2 options): Likewise.
30612 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30615 2017-03-17 Tom de Vries <tom@codesourcery.com>
30617 * gcov-dump.c (print_usage): Print bug_report_url.
30619 2017-03-17 Richard Biener <rguenther@suse.de>
30621 PR middle-end/80050
30622 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30623 (parser::peek): Likewise.
30625 2017-03-17 Richard Biener <rguenther@suse.de>
30627 PR tree-optimization/80048
30628 * sese.c (free_sese_info): Properly release rename_map and
30629 copied_bb_map elements.
30631 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30633 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30634 Add linked-list forward and backlinks. Insert on
30635 construction, remove on destruction.
30636 (class pass_store_merging): Add m_stores_head field.
30637 (pass_store_merging::terminate_and_process_all_chains):
30638 Iterate over m_stores_head list.
30639 (pass_store_merging::terminate_all_aliasing_chains):
30641 (pass_store_merging::execute): Check for debug stmts first.
30642 Push new chains onto the m_stores_head stack.
30644 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30647 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30648 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30649 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30651 2017-03-16 Jeff Law <law@redhat.com>
30653 PR tree-optimization/71437
30654 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30655 member function. Implementation moved into after_dom_children
30656 member function and into the threader's thread_outgoing_edges
30658 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30659 some code into new thread_outgoing_edges.
30660 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30661 definition. Simplify marker handling (do it here). Assume we always
30662 have the available expression and the const/copies tables.
30663 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30665 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30666 * tree-vrp.c (equiv_stack): No longer file scoped.
30667 (vrp_dom_walker): New class.
30668 (vrp_dom_walker::before_dom_children): New member function.
30669 (vrp_dom_walker::after_dom_children): Likewise.
30670 (identify_jump_threads): Setup domwalker. Use it rather than
30671 walking edges in a random order by hand. Simplify setup/finalization.
30672 (finalize_jump_threads): Remove.
30673 (vrp_finalize): Do not call identify_jump_threads here.
30674 (execute_vrp): Do it here instead and call thread_through_all_blocks
30677 PR tree-optimization/71437
30678 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30680 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30682 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30683 (dom_opt_dom_walker::thread_across_edge): Remove
30684 handle_dominating_asserts argument. All callers changed.
30685 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30686 changes. Remove calls to lhs_of_dominating_assert. Other
30687 uses of handle_dominating_asserts turn into unconditional code
30688 (simplify_control_stmt_condition_1): Likewise.
30689 (simplify_control_stmt_condition): Likewise.
30690 (thread_through_normal_block, thread_across_edge): Likewise.
30691 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30692 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30693 object if it is not an SSA_NAME.
30694 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30695 before calling into the VRP specific simplifiers.
30696 (identify_jump_threads): Remove handle_dominating_asserts
30699 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30702 * tree-diagnostic.c (default_tree_printer): No longer static.
30703 * tree-diagnostic.h (default_tree_printer): New prototype.
30705 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30707 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30708 Change ins into fmov.
30710 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30712 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30713 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30714 Use h_con constraint for operand 1.
30715 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30716 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30718 2017-03-15 Jeff Law <law@redhat.com>
30720 PR tree-optimization/71437
30721 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30722 (record_temporary_equivalences): Use it.
30724 PR tree-optimization/71437
30725 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30726 tree-ssa-scopedtables.
30727 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30728 (record_conditions, record_cond, vuse_eq): Likewise.
30729 (record_edge_info): Adjust to API tweak of record_conditions.
30730 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30731 (record_temporary_equivalences, optimize_stmt): Likewise.
30732 (eliminate_redundant_computations): Likewise.
30733 (record_equivalences_from_stmt): Likewise.
30734 * tree-ssa-scopedtables.c: Include options.h and params.h.
30735 (vuse_eq): New function, moved from tree-ssa-dom.c
30736 (build_and_record_new_cond): Likewise.
30737 (record_conditions): Likewise. Accept vector of conditions rather
30738 than edge_equivalence structure for first argument.
30739 for the first argument.
30740 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30741 from tree-ssa-dom.c.
30742 (avail_exprs_stack::record_cond): Likewise.
30743 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30744 from tree-ssa-dom.c.
30745 (avail_exprs_stack): Add new member functions lookup_avail_expr
30747 (record_conditions): Declare.
30749 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30752 * lra-constraints.c (process_alt_operands): Increase reject for
30753 reloading an input/output operand.
30755 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30758 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30759 insns to convert from signed/unsigned char/short to IEEE 128-bit
30761 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30763 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30766 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30767 subreg of inner mode for values already in registers.
30769 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30771 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30772 iteration reg is used after the loop.
30774 2017-03-14 Martin Sebor <msebor@redhat.com>
30776 PR tree-optimization/79800
30777 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30778 precision in negative-positive range.
30779 (format_floating): Call non-const overload with adjusted precision.
30781 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30784 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30787 2017-03-14 Martin Sebor <msebor@redhat.com>
30789 PR middle-end/80020
30790 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30791 * builtins.def (aligned_alloc): Use it.
30794 * Makefile.in (GTFILES): Add calls.c.
30795 * calls.c: Include "gt-calls.h".
30797 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
30799 PR rtl-optimization/79728
30800 * regs.h (struct target_regs): New field
30801 x_contains_allocatable_regs_of_mode.
30802 (contains_allocatable_regs_of_mode): New macro.
30803 * reginfo.c (init_reg_sets_1): Initialize it, and change
30804 contains_reg_of_mode so it includes global regs as well.
30805 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
30806 rather than contains_regs_of_mode.
30808 2017-03-14 Martin Liska <mliska@suse.cz>
30810 * doc/invoke.texi: Document options that can't be combined with
30811 -fcheck-pointer-bounds.
30813 2017-03-14 Martin Liska <mliska@suse.cz>
30815 PR middle-end/79831
30816 * doc/invoke.texi (-Wchkp): Document the option.
30818 2017-03-14 Martin Liska <mliska@suse.cz>
30820 * Makefile.in: Install gcov-dump.
30822 2017-03-14 Martin Liska <mliska@suse.cz>
30824 * multiple_target.c (expand_target_clones): Bail out for
30825 an invalid attribute.
30827 2017-03-14 Richard Biener <rguenther@suse.de>
30829 * alias.c (struct alias_set_entry): Pack properly.
30830 * cfgloop.h (struct loop): Likewise.
30831 * cse.c (struct set): Likewise.
30832 * ipa-utils.c (struct searchc_env): Likewise.
30833 * loop-invariant.c (struct invariant): Likewise.
30834 * lra-remat.c (struct cand): Likewise.
30835 * recog.c (struct change_t): Likewise.
30836 * rtl.h (struct address_info): Likewise.
30837 * symbol-summary.h (function_summary): Likewise.
30838 * tree-loop-distribution.c (struct partition): Likewise.
30839 * tree-object-size.c (struct object_size_info): Likewise.
30840 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30841 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30842 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30843 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30844 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30845 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30846 (struct _stmt_vec_info): Likewise.
30848 2017-03-14 Martin Liska <mliska@suse.cz>
30851 * multiple_target.c (create_dispatcher_calls): Check that
30852 a target can create a function dispatcher.
30854 2017-03-14 Martin Liska <mliska@suse.cz>
30857 * multiple_target.c (expand_target_clones): Drop local.local
30858 flag for default implementation.
30860 2017-03-14 Richard Biener <rguenther@suse.de>
30862 PR tree-optimization/80030
30863 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30865 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30867 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30868 gcc_fallthrough() instead of __attribute__((fallthrough));
30870 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30872 * doc/gcc.texi: Remove "up" link to (DIR).
30873 * doc/gccint.texi: Ditto.
30875 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30877 * doc/install.texi (Specific) <avr>: Remove reference to
30880 2017-03-13 Jeff Law <law@redhat.com>
30882 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30883 attribute rather than comments.
30885 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30886 match_scratch operand is highest.
30888 2017-03-13 Martin Liska <mliska@suse.cz>
30890 PR middle-end/78339
30891 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30892 is a CHKP clone, use original declaration.
30894 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30896 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30897 (arc_conditional_register_usage): Use a different allocation order
30898 when optimizing for size.
30899 * common/config/arc/arc-common.c (arc_option_optimization_table):
30900 Section anchors default on when optimizing for size.
30902 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30904 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30906 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30908 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30909 * config/arc/arc.md (cpu_facility): Add cd variant.
30910 (*movqi_insn): Add code density variant.
30911 (*movhi_insn): Likewise.
30912 (*movqi_insn): Likewise.
30913 (*addsi3_mixed): Likewise.
30914 (subsi3_insn): Likewise.
30916 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30918 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30920 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30922 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30923 expressions with MINUS and UNARY ops.
30925 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30928 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30930 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30931 between vec_select and vector argument.
30932 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30933 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30934 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30935 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30936 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30937 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30939 2017-03-13 Richard Biener <rguenther@suse.de>
30942 * params.def (vect-max-peeling-for-alignment): Fix typo.
30944 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30946 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30947 issue that only occurred with binutils below 2.18.
30949 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30951 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30952 refer to binutils 2.11/2.12 minimum.
30954 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30956 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30957 ftp.kernel.org and simplify binutils requirement.
30959 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30961 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30963 (Optimize Options): Ditto. Also remove redundancy.
30965 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30967 PR translation/79848
30968 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30970 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30971 to G_ to avoid double translation.
30973 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30975 PR translation/79923
30976 * auto-profile.c (get_combined_location): Convert leading
30977 character of diagnostics to lower case and remove trailing period.
30978 (read_profile): Likewise for various diagnostics.
30979 * config/arm/arm.c (arm_option_override): Remove trailing period
30980 from various diagnostics.
30981 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30982 (msp430_expand_delay_cycles): Likewise.
30984 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30987 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30988 full command-line argument, rather than just "str".
30989 (aarch64_validate_march): Likewise.
30990 (aarch64_validate_mtune): Likewise.
30992 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30994 PR rtl-optimization/78911
30995 * lra-assigns.c (must_not_spill_p): New function.
30996 (spill_for): Use it.
30998 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31000 PR tree-optimization/79981
31001 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31002 ATOMIC_COMPARE_EXCHANGE ifn result.
31003 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31004 IFN_ATOMIC_COMPARE_EXCHANGE.
31006 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31009 * opts.c (parse_sanitizer_options): Add missing question mark to
31010 "did you mean" message.
31012 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31014 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31016 (VMULEUH_UNS): Likewise.
31017 (VMULOUB_UNS): Likewise.
31018 (VMULOUH_UNS): Likewise.
31019 * config/rs6000/rs6000.c (builtin_function_type): Remove
31020 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31022 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31025 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31026 x with result of extra_parsing_for_operand_code_0.
31027 (function_reader::extra_parsing_for_operand_code_0): Convert
31028 return type from void to rtx, returning x. When reading
31029 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31030 larger size containing struct block_symbol.
31032 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31035 -mfloat128-hardware without -m64.
31037 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31040 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31041 entries to the case statement that marks unsigned arguments to
31042 overloaded functions.
31044 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31046 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31047 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31049 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31052 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31053 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31055 2017-03-10 Martin Liska <mliska@suse.cz>
31059 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31060 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31063 2017-03-10 Olivier Hainque <hainque@adacore.com>
31065 * tree-switch-conversion (array_value_type): Start by resetting
31066 candidate type to it's main variant.
31068 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31070 PR rtl-optimization/79909
31071 * combine.c (try_combine): Use simplify_replace_rtx on individual
31072 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31073 of the whole CALL_INSN_FUNCTION_USAGE.
31075 PR tree-optimization/79972
31076 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31077 get_range_info on SSA_NAMEs. Formatting fixes.
31079 2017-03-10 Richard Biener <rguenther@suse.de>
31080 Jakub Jelinek <jakub@redhat.com>
31082 PR tree-optimization/77975
31083 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31084 edge to be constant.
31085 (get_val_for): For constant x return it. Formatting fix.
31086 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31087 would use the same bases as the current one.
31089 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31091 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31092 instead of vec_select for V1TImode.
31093 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31095 (VSX_LE_128): Add V1TI to this mode iterator.
31096 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31097 (*vsx_le_perm_store_<mode>): Likewise.
31098 (pre-reload splitter for VSX stores): Likewise.
31099 (post-reload splitter for VSX stores): Likewise.
31100 (*vsx_xxpermdi2_le_<mode>): Likewise.
31101 (*vsx_lxvd2x2_le_<mode>): Likewise.
31102 (*vsx_stxvd2x2_le_<mode>): Likewise.
31104 2017-03-09 Michael Eager <eager@eagercon.com>
31106 Correct failures with --enable-checking=yes,rtl.
31108 * config/microblaze/microblaze.c (microblaze_expand_shift):
31109 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31110 test for const0_rtx.
31111 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31112 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31114 2017-03-09 Richard Biener <rguenther@suse.de>
31116 PR tree-optimization/79977
31117 * graphite-scop-detection.c (scop_detection::merge_sese):
31118 Handle the case of extra exits to blocks dominating the entry.
31120 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31122 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31125 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31127 PR rtl-optimization/79949
31128 * lra-constraints.c (process_alt_operands): Check memory when
31129 trying to predict a cycle. Print about the overall increase.
31131 2017-03-09 Richard Biener <rguenther@suse.de>
31133 PR middle-end/79971
31134 * gimple-expr.c (useless_type_conversion_p): Preserve
31135 TYPE_SATURATING for fixed-point types.
31137 2017-03-09 Richard Biener <rguenther@suse.de>
31140 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31141 alignment of BLKmode params.
31143 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31146 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31147 (VALL_NO_V2Q): Likewise.
31149 * config/aarch64/aarch64-simd.md
31150 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31152 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31153 VALL_NO_V2Q mode iterator.
31154 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31156 2017-03-09 Martin Liska <mliska@suse.cz>
31158 PR tree-optimization/79631
31159 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31160 tree_int_cst_sign_bit just for INTEGER constants.
31162 2017-03-09 Martin Liska <mliska@suse.cz>
31166 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31169 2017-03-09 Marek Polacek <polacek@redhat.com>
31172 * tree.c (inchash::add_expr): Handle TREE_VEC.
31174 2017-03-09 Martin Liska <mliska@suse.cz>
31177 (chkp_narrow_size_and_offset): New function.
31178 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31179 (void chkp_parse_bit_field_ref): New function.
31180 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31181 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31183 2017-03-09 Martin Liska <mliska@suse.cz>
31186 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31187 (chkp_find_bounds_1): Remove gcc_unreachable.
31189 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31192 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31193 BUILT_IN_SYNC*, determine the access type from the size suffix and
31194 always build a MEM_REF with that type. Handle forgotten
31195 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31198 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31199 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31200 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31201 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31202 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31203 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31204 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31205 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31206 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31207 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31208 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31209 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31210 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31211 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31212 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31213 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31214 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31215 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31216 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31217 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31218 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31219 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31220 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31221 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31222 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31223 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31224 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31225 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31226 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31227 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31228 definitions outside of __OPTIMIZE__ guarded section.
31231 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31232 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31233 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31234 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31237 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31239 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31240 ("vfenez<mode>"): Add missing constraints.
31242 2017-03-08 Martin Sebor <msebor@redhat.com>
31245 * config/nds32/nds32.c (nds32_option_override):
31246 Fix misspelled diagnostic.
31248 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31251 * gimplify.c (gimplify_omp_for): Replace index var in outer
31252 taskloop statement with an artificial variable and add
31253 OMP_CLAUSE_PRIVATE clause for it.
31255 2017-03-08 Richard Biener <rguenther@suse.de>
31257 PR tree-optimization/79955
31258 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31259 for accesses that are completely outside of the variable.
31261 2017-03-08 Andrew Haley <aph@redhat.com>
31263 PR tree-optimization/79943
31264 * tree-ssa-loop-split.c (compute_new_first_bound): When
31265 calculating the new upper bound, (END-BEG) should be added, not
31268 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31270 * config/avr/avr.md (setmemhi): Make sure match_dup
31271 operand number comes before match_scratch.
31273 2017-03-08 Richard Biener <rguenther@suse.de>
31275 PR tree-optimization/79920
31276 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31277 with ncopies == 1 to ...
31278 (vect_transform_slp_perm_load): ... here. Properly compute
31279 all element loads by iterating VF times over the group. Do
31280 not handle ncopies (computed in a broken way) in
31281 vect_create_mask_and_perm.
31283 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31286 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31287 is a uniform vector, use uniform_vector_p return value instead of
31288 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31290 2017-03-07 Marek Polacek <polacek@redhat.com>
31292 PR middle-end/79809
31293 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31294 (alloca_call_type): Likewise.
31296 2017-03-07 Martin Liska <mliska@suse.cz>
31298 * gcov.c (process_args): Put comment to correct location.
31300 2017-03-07 Martin Liska <mliska@suse.cz>
31302 PR middle-end/68270
31303 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31304 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31305 (chkp_narrow_bounds_for_field): Likewise.
31306 (chkp_parse_array_and_component_ref): Pass one more argument to
31309 2017-03-07 Richard Biener <rguenther@suse.de>
31311 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31314 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31316 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31317 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31319 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31322 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31323 to end of description.
31324 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31326 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31328 PR rtl-optimization/79901
31329 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31331 (*avx512f_<code><mode>3<mask_name>): ... this.
31332 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31333 iterator instead of VI8_AVX2_AVX512BW.
31335 PR rtl-optimization/79901
31336 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31337 min/max expander, expand it using expand_vec_cond_expr.
31340 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31343 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31346 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31347 to void * for PCH reasons.
31348 * dwarf2out.c (output_loc_operands, output_die): Cast
31349 v.val_vec.array to unsigned char *.
31351 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31354 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31357 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31359 PR rtl-optimization/79571
31360 * lra-constraints.c (process_alt_operands): Calculate static
31361 reject and subtract it from overall when only addresses will be
31364 2017-03-06 Julia Koval <julia.koval@intel.com>
31367 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31368 incoming stack boundary to 128 for 64-bit targets.
31370 2017-03-06 Richard Biener <rguenther@suse.de>
31372 PR tree-optimization/79894
31373 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31374 to NULL after folding it.
31376 2017-03-06 Richard Biener <rguenther@suse.de>
31378 PR tree-optimization/79824
31379 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31380 check disabling peeling for gaps.
31382 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31384 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31385 attributes): Document gettimeofday.
31387 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31389 * config/s390/s390.c (s390_option_override_internal): Set
31390 PARAM_MIN_VECT_LOOP_BOUND
31392 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31394 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31395 * config/s390/s390.md: Likewise.
31397 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31400 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31401 (<avx2_avx512>_perm<mode>): Rename to ...
31402 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31404 (<avx512>_perm<mode>_mask): Rename to ...
31405 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31407 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31408 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31409 instead of VI8F_256_512.
31410 (avx512f_perm<mode>): New define_expand.
31411 (avx512f_perm<mode>_mask): Likewise.
31412 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31413 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31415 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31417 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31418 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31420 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31422 2017-03-06 Martin Liska <mliska@suse.cz>
31425 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31426 when having a SSA NAME w/o VAR_DECL assigned to it.
31428 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31430 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31431 msa_dpsub_<su>_d): Fix MODE for vec_select.
31433 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31435 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31437 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31439 2017-03-06 Richard Biener <rguenther@suse.de>
31441 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31442 * plugin.c (register_plugin_info): Likewise.
31443 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31445 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31447 * config/i386/sse.md (sse_storehps, sse_storelps,
31448 avx_<castmode><avxsizesuffix>_<castmode>,
31449 avx512f_<castmode><avxsizesuffix>_<castmode>,
31450 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31451 in condition that at least one operand is not a MEM.
31453 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31455 PR middle-end/79805
31456 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31457 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31459 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31460 gimple_call_nothrow_p flag based on whether original builtin can throw.
31461 If it can, emit following stmts on the fallthrough edge.
31462 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31463 don't create new bb if inserting just debug stmts on the edge, try to
31464 insert them on the fallthru bb or just reset debug stmts.
31466 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31469 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31470 restore recog_data (including the operand rtxes inside it) around
31471 the call to get_insn_template.
31473 2017-03-03 Martin Sebor <msebor@redhat.com>
31475 PR tree-optimization/79699
31476 * context.c (context::~context): Free MPFR caches to avoid
31477 a memory leak on program exit.
31479 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31481 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31482 Use wide_int::ulow () instead of .elt (0).
31484 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31486 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31487 (*pushxf): Limit oF constraint to 32bit targets and add oC
31488 constraint for 64bit targets.
31489 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31490 (*pushdf): Change rmF constraint to rmC.
31492 2017-03-03 Martin Liska <mliska@suse.cz>
31494 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31495 Remove unused variable.
31497 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31500 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31501 is a memory operand, increase num_memory.
31502 (ix86_expand_args_builtin): Likewise.
31504 2017-03-03 Jan Hubicka <jh@suse.cz>
31507 * ipa-devirt.c (maybe_record_node): Properly handle
31508 __cxa_pure_virtual visibility.
31510 2017-03-03 Martin Liska <mliska@suse.cz>
31512 PR tree-optimization/79803
31513 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31515 (pass_loop_prefetch::execute): Disabled optimization if an
31516 assumption about L1 cache size is not met.
31518 2017-03-03 Martin Liska <mliska@suse.cz>
31520 PR rtl-optimization/79574
31521 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31522 (hash_scan_set): Likewise.
31523 (dump_hash_table): Likewise.
31524 (hoist_code): Likewise.
31526 2017-03-03 Richard Biener <rguenther@suse.de>
31528 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31529 in place of uhigh ().
31530 (fixed_convert_from_real): Likewise.
31532 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31535 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31537 2017-03-03 Richard Biener <rguenther@suse.de>
31539 PR middle-end/79818
31540 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31541 TYPE_OVERFLOW_UNDEFINED check.
31543 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31545 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31547 (vector_ae_<mode>_p): Likewise.
31548 (vector_nez_<mode>_p): Likewise.
31549 (vector_ne_v2di_p): Likewise.
31550 (vector_ae_v2di_p): Likewise.
31551 (vector_ne_<mode>_p): Likewise.
31552 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31554 (vsx_tsqrt<mode>2_fe): Likewise.
31556 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31559 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31561 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31563 PR rtl-optimization/79780
31564 * cprop.c (one_cprop_pass): When second and further conditional trap
31565 in a single basic block is turned into an unconditional trap, turn it
31566 into a deleted note to avoid RTL verification failures.
31568 2017-03-02 Richard Biener <rguenther@suse.de>
31570 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31572 2017-03-02 Richard Biener <rguenther@suse.de>
31574 PR tree-optimization/79345
31576 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31577 param and abort the walk, returning -1 if it is hit.
31578 (walk_aliased_vdefs): Take a limit param and pass it on.
31579 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31580 defaulting to 0 and return a signed int.
31581 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31582 (check_defs): New helper.
31583 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31584 about uninitialized memory.
31585 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31586 bogus uninitialized warning.
31587 (fixed_convert_from_real): Likewise.
31589 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31591 PR tree-optimization/66768
31592 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31593 iv_use if base object can't be determined.
31595 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31597 PR tree-optimization/79345
31598 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31599 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31600 (get_pattern_stats): Initialize it.
31601 * genemit.c (gen_expand): Verify match_scratch numbers come after
31602 match_operand/match_dup numbers.
31603 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31604 match_scratch numbers.
31605 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31607 * config/s390/s390.md (trunctdsd2): Likewise.
31609 2017-03-02 Richard Biener <rguenther@suse.de>
31611 * wide-int.h (wide_int_storage::operator=): Implement in terms
31614 2017-03-02 Richard Biener <rguenther@suse.de>
31616 PR tree-optimization/79777
31617 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31618 the to insert expression to sth existing.
31620 2017-03-01 Martin Sebor <msebor@redhat.com>
31622 PR middle-end/79692
31623 * gimple-ssa-sprintf.c
31624 (directive::known_width_and_precision): New function.
31625 (format_integer): Use it.
31626 (get_mpfr_format_length): Consider the full range of precision
31627 when computing %g output with the # flag. Set the likely byte
31628 count to 3 rather than 1 when precision is indeterminate.
31629 (format_floating): Correct the lower bound of precision.
31631 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31633 * doc/invoke.texi: Document default code model for 64-bit Linux.
31635 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31638 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31639 udiv rather than div since input pattern is unsigned.
31641 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31643 * config/i386/i386.c (print_reg): Warn for values of
31644 unsupported size in integer register.
31646 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31649 * config/rs6000/predicates.md (current_file_function_operand): Do
31650 not allow self calls to be local if the function is replaceable.
31652 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31655 * config/rs6000/altivec.h (vec_ctz and others): Change the
31656 preprocessor macro that controls conditional compilation from
31657 _ARCH_PWR9 to __POWER9_VECTOR__.
31658 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31659 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31660 control (instead of _ARCH_PWR9 control) so that template
31661 definition uses power9-specific function.
31662 (vec_any_eq): Likewise.
31663 (vec_all_ne): Change macro definition to use a power9-specific
31664 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31665 _ARCH_PWR9 control).
31666 (vec_any_eq) Likewise.
31667 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31668 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31669 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31670 support for xvcmpnedp instruction.
31671 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31672 macro expansion so that Power9 implementation of vec_all_ne does
31673 not use the AltiVec predicate framework.
31674 (VCMPNEH_P): Likewise.
31675 (VCMPNEW_P): Likewise.
31676 (VCMPNED_P): Likewise.
31677 (VCMPNEFP_P): Likewise.
31678 (VCMPNEDP_P): Likewise.
31679 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31680 implementation of vec_any_eq to not use AltiVec predicate
31682 (VCMPAEH_P): Likewise.
31683 (VCMPAEW_P): Likewise.
31684 (VCMPAED_P): Likewise.
31685 (VCMPAEFP_P): Likewise.
31686 (VCMPAEDP_P): Likewise.
31687 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31688 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31689 not use the AltiVec predicate framework.
31690 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31691 of vec_any_eq to not use AltiVec predicate framework.
31692 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31693 support for predefined __POWER9_VECTOR__ macro to indicate that
31694 Power9 instruction selection is enabled.
31695 (altivec_overloaded_builtins): Remove extraneous
31696 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31697 function argument types RS6000_BTI_bool_V16QI and
31698 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31699 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31700 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31701 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31702 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31703 Power9 for implementations of vec_cmpne. Change the signature for
31704 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31705 (representing vec_all_ne) to remove the previously described first
31706 argument of type RS6000_BTI_INTSI, as this was an artifact of
31707 reliance on the AltiVec predicate framework, which is no longer
31708 used in the implementation of these functions. Add
31709 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31710 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31711 since, unlike the AltiVec predicate framework implementation, we
31712 do not share function descriptors between vec_alle and vec_anyeq.
31713 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31714 set of modes that receive special treatment even when
31715 TARGET_P9_VECTOR is true. The special treatment emits code that
31716 does not depend on Power9 instructions.
31717 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31718 define_expand to not rely on AltiVec predicate framework.
31719 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31721 (vector_ne_v2di_p): Change this define_expand to not rely on
31722 AltiVec predicate framework.
31723 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31725 (vector_ne_<mode>_p): Change this define_expand to not rely on
31726 AltiVec predicate framework.
31727 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31729 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31730 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31731 define_insn pattern.
31732 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31733 define_insn pattern because the xvcmpne<VSs>. instruction is not
31735 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31736 instruction is not supported.
31738 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31740 * config/nvptx/nvptx.c: Include intl.h.
31742 2017-03-01 Martin Jambor <mjambor@suse.cz>
31745 * ipa-prop.h (ipa_bits): Removed field known.
31746 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31747 to pointers. Adjusted their comments to warn about their sharing.
31748 (ipcp_transformation_summary): Change bits to a vector of pointers.
31749 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31750 (ipa_get_ipa_bits_for_value): Declare.
31751 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31752 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31753 (ipa_bits_hash_table): Likewise.
31754 (ipa_vr_ggc_hash_traits): Likewise.
31755 (ipa_vr_hash_table): Likewise.
31756 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31757 being pointers and vr_known being removed.
31758 (ipa_set_jf_unknown): Likewise.
31759 (ipa_get_ipa_bits_for_value): New function.
31760 (ipa_set_jfunc_bits): Likewise.
31761 (ipa_get_value_range): New overloaded functions.
31762 (ipa_set_jfunc_vr): Likewise.
31763 (ipa_compute_jump_functions_for_edge): Use the above functions to
31764 construct bits and vr parts of jump functions.
31765 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31766 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31768 (ipcp_grow_transformations_if_necessary): Also allocate
31769 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31771 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31772 them. Fix too long lines.
31773 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31774 vr_known being removed.
31775 (ipa_read_jump_function): Use new setter functions to construct bits
31776 and vr parts of jump functions or set them to NULL.
31777 (write_ipcp_transformation_info): Adjust for bits being pointers.
31778 (read_ipcp_transformation_info): Likewise.
31779 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31781 Include gt-ipa-prop.h.
31782 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31784 (ipcp_store_bits_results): Likewise.
31785 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31786 Do not write to existing jump functions but use a temporary instead.
31788 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31791 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31792 attempt to use its first operand as BIT_FIELD_REF base.
31794 2017-03-01 Richard Biener <rguenther@suse.de>
31796 PR middle-end/79721
31797 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
31798 interpolating formula in wrapping arithmetic.
31799 (chrec_apply): Convert chrec_evaluate return value to wanted type.
31801 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31803 PR tree-optimization/79734
31804 * tree-vect-generic.c (expand_vector_condition): Optimize
31805 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
31806 Handle VEC_COND_EXPR where comparison has different inner width from
31807 type's inner width.
31809 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
31811 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
31812 markup, and similar issues. Remove @opindex entries for things
31813 that aren't options. Add missing -mmpy-option entries.
31815 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31817 PR tree-optimization/79737
31818 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
31819 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
31820 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
31821 instead of byte_size. Formatting fix.
31822 (shift_bytes_in_array_right): Formatting fix.
31824 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
31827 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
31828 condition on optimize for the leaf function test.
31830 2017-02-28 Martin Liska <mliska@suse.cz>
31833 * read-rtl-function.c (function_reader::handle_unknown_directive):
31834 Bail out when one uses -flto.
31836 2017-02-28 Martin Liska <mliska@suse.cz>
31838 * common.opt: Replace space with tabular for options of <number>
31840 * config/i386/i386.opt: Show <number> value for
31841 -mlarge-data-threshold.
31842 * opts.c (print_filtered_help): Do not display number in hexadecimal
31845 2017-02-28 Martin Liska <mliska@suse.cz>
31847 * common.opt: Fix --help=option -Q for options which are of
31850 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31852 * config/i386/i386.c (print_reg): Error out for values
31853 of 8-bit size in invalid integer register.
31855 2017-02-28 Martin Sebor <msebor@redhat.com>
31857 PR tree-optimization/79691
31858 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31860 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31863 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31864 gcc_unreachable with output_operand_lossage.
31866 2017-02-28 Richard Biener <rguenther@suse.de>
31868 PR tree-optimization/79740
31869 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31871 (visit_nary_op): Insert the nary into the hashtable if we
31872 pattern-matched sth.
31873 * tree-ssa-pre.c (eliminate_insert): Robustify.
31875 2017-02-28 Richard Biener <rguenther@suse.de>
31877 PR middle-end/79731
31878 * fold-const.c (decode_field_reference): Reject out-of-bound
31881 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31883 * config/i386/i386.c: Include intl.h.
31884 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31885 instead of just cond ? "..." : "...".
31886 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31887 * coverage.c (read_counts_file): Likewise.
31888 * omp-offload.c: Include intl.h.
31889 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31890 of just cond ? "..." : "...".
31891 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31892 of just cond ? "..." : "...".
31894 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31897 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31899 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31900 'tune for' CPU name.
31901 * config/arm/arm-cpu-data.h: Regenerated.
31903 2017-02-28 Richard Biener <rguenther@suse.de>
31905 PR tree-optimization/79732
31906 * tree-inline.c (expand_call_inline): Do not shadow var.
31908 2017-02-28 Richard Biener <rguenther@suse.de>
31910 PR tree-optimization/79723
31911 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31912 address-space properly.
31914 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31916 * doc/optinfo.texi (Optimization groups): Fix option used for
31918 * doc/invoke.texi (-fopt-info): Document "omp".
31919 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31920 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31921 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31922 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31923 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31925 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31927 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31928 OPTGROUP_OMP with "omp".
31930 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31933 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31934 for arithmetic shift of unsigned V2DI.
31936 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31938 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31939 arc/linux.h headers.
31940 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31941 (LINK_SPEC): Likewise.
31942 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31943 (EXTRA_SPECS): Likewise.
31944 (STARTFILE_SPEC): Likewise.
31945 (ENDFILE_SPEC): Likewise.
31946 (LIB_SPEC): Likewise.
31947 (TARGET_SDATA_DEFAULT): Likewise.
31948 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31949 (MULTILIB_DEFAULTS): Likewise.
31950 (DWARF2_UNWIND_INFO): Likewise.
31951 * config/arc/big.h: New file.
31952 * config/arc/elf.h: Likewise.
31953 * config/arc/linux.h: Likewise.
31954 * config/arc/t-uClibc: Remove.
31956 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31958 PR tree-optimization/77536
31959 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31960 (tree_transform_and_unroll_loop): Use above function to compute the
31961 estimated niter of unrolled loop and use it when scaling profile.
31962 Also use count info rather than frequency if it's non-zero.
31963 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31964 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31965 (vect_transform_loop): Call above function.
31967 2017-02-27 Richard Biener <rguenther@suse.de>
31969 PR tree-optimization/45397
31970 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31971 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31972 (visit_nary_op): Add pattern matching for CSEing sign-changed
31973 or truncated operations with wider ones.
31975 2017-02-27 Richard Biener <rguenther@suse.de>
31977 PR tree-optimization/79690
31978 * tree-vect-stmts.c (vectorizable_store): Use vector type
31979 built from the DR with address-space.
31981 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31983 * doc/invoke.texi (Optimize Options): Refine the description
31984 of asan-use-after-return.
31986 2017-02-25 Alan Modra <amodra@gmail.com>
31988 PR rtl-optimization/79584
31989 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31990 base, not ad->base_term, the reg within base. Remove assertion
31991 that ad->base == ad->base_term. Replace gen_int_mode using
31992 bogus mode with const0_rtx.
31994 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31996 PR middle-end/79396
31997 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31998 FMA_EXPR like tcc_binary or tcc_unary.
32000 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32003 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32005 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32006 (output_loc_operands): Handle DW_OP_call_ref and
32007 DW_OP_GNU_variable_value.
32008 (struct variable_value_struct): New type.
32009 (struct variable_value_hasher): Likewise.
32010 (variable_value_hash): New variable.
32011 (string_types): Remove.
32012 (copy_loc_descr): New function.
32013 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32014 (prepend_loc_descr_to_each): New function.
32015 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32016 instead of add_loc_descr_to_each if the first argument is single
32017 location list and the second has multiple.
32018 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32019 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32020 when looking for variable value which doesn't have other location info.
32021 (loc_list_from_tree): Formatting fix.
32022 (gen_array_type_die): Simplify DW_AT_string_length handling.
32023 (adjust_string_types): Remove.
32024 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32025 string_types. Call resolve_variable_values.
32026 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32027 (resolve_addr_in_expr): Likewise. Add A argument.
32028 (copy_deref_exprloc): Remove deref argument. Adjust for the
32029 original expression being DW_OP_GNU_variable_value with optionally
32030 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32031 optionally after it.
32032 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32033 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32034 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32035 (variable_value_hasher::hash, variable_value_hasher::equal): New
32037 (resolve_variable_value_in_expr, resolve_variable_value,
32038 resolve_variable_values, note_variable_value_in_expr,
32039 note_variable_value): New functions.
32040 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32043 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32046 * opts.h (handle_generated_option): Add GENERATED_P argument.
32047 * opts-common.c (handle_option): Adjust function comment.
32048 (handle_generated_option): Add GENERATED_P argument, pass it to
32050 (control_warning_option): Pass false to handle_generated_option
32052 * opts.c (maybe_default_option): Pass true to handle_generated_option
32054 * optc-gen.awk: Likewise.
32056 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32058 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32059 a REG, look at the REG it is a SUBREG of.
32060 (splitter for cmpeqsi_t): Ditto.
32062 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32064 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32065 the special USEs with the pattern of the insn, not the insn itself.
32067 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32070 * doc/invoke.texi: Document -mload-store-pairs.
32072 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32073 Sandra Loosemore <sandra@codesourcery.com>
32075 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32076 argument isn't a CONST_INT.
32077 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32078 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32079 (nios2_validate_compare): Bypass alternate compare logic if *op2
32080 is not a CONST_INT.
32081 (ldstwm_operation_p): Return false if first_base is not a REG or
32082 if first_offset is not a CONST_INT.
32084 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32086 * config/cris/cris.md: Use correct operand in a define_peephole2.
32088 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32090 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32092 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32094 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32095 this_insn if it is an INSN or JUMP_INSN.
32096 (force_offsettable): Look at base, not at addr.
32097 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32098 on things that aren't necessarily CONST_INTs.
32100 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32102 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32103 -mfpmath=sse is the default also for x86-32 targets with SSE2
32104 instruction set when @option{-ffast-math} is enabled
32106 2017-02-24 Jeff Law <law@redhat.com>
32108 PR rtl-optimizatoin/79286
32109 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32112 2017-02-24 Richard Biener <rguenther@suse.de>
32114 PR tree-optimization/79389
32115 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32118 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32120 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32121 function comment to reflect reality.
32122 (loop_exits_before_overflow): Fix typo in function description.
32124 2017-02-24 Richard Biener <rguenther@suse.de>
32126 PR tree-optimization/79389
32127 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32128 properly that a threading opportunity exists. Detect conditional
32129 copy/constant propagation opportunities.
32131 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32133 * config/visium/visium.md (type): Add trap.
32134 (b): New mode attribute.
32135 (*btst): Rename into...
32136 (*btst<mode>): ...this and adjust.
32137 (*cbranchsi4_btst_insn): Rename into...
32138 (*cbranch<mode>4_btst_insn): ...this and adjust.
32139 (trap): New define_insn.
32141 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32143 PR tree-optimization/79389
32144 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32145 (noce_reversed_cond_code): New function.
32146 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32147 reversed_comparison_code. Formatting fix.
32148 (noce_try_store_flag): Test rev_cond != NULL in addition to
32149 reversed_comparison_code.
32150 (noce_try_store_flag_constants): Likewise.
32151 (noce_try_store_flag_mask): Likewise.
32152 (noce_try_addcc): Use rev_cond if non-NULL instead of
32153 reversed_comparison_code.
32154 (noce_try_cmove_arith): Likewise. Formatting fixes.
32155 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32156 (noce_find_if_block): Initialize rev_cond.
32157 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32158 instead of false as last argument never attempt to reverse it
32161 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32163 PR tree-optimization/79663
32164 * tree-predcom.c (combine_chains): Process refs in reverse order
32165 only for ZERO length chains, and add explaining comment.
32167 2017-02-23 Jeff Law <law@redhat.com>
32169 PR tree-optimization/79578
32170 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32171 in call to operand_equal_p.
32173 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32176 * config/i386/cpuid.h: Fix another undefined behavior.
32178 2017-02-23 Richard Biener <rguenther@suse.de>
32180 PR tree-optimization/79683
32181 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32182 vector types for data-refs.
32184 2017-02-23 Martin Liska <mliska@suse.cz>
32186 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32188 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32190 PR middle-end/79665
32191 * internal-fn.c (get_range_pos_neg): Moved to ...
32192 * tree.c (get_range_pos_neg): ... here. No longer static.
32193 * tree.h (get_range_pos_neg): New prototype.
32194 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32195 are known to be in between 0 and signed maximum inclusive, try to
32196 expand both unsigned and signed divmod and use the cheaper one from
32199 2017-02-22 Jeff Law <law@redhat.com>
32201 PR tree-optimization/79578
32202 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32203 to compare base operands.
32205 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32208 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32209 gpc_reg_operand instead of fpr_reg_operand.
32211 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32213 * config/mips/mips.c (mips_return_in_memory): Force FP
32214 vector types to be returned in memory for o32 ABI.
32216 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32218 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32219 instead of DW_TAG_member for static data member declarations and don't
32220 set no_linkage_name for static inline data members.
32221 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32224 2017-02-22 Martin Liska <mliska@suse.cz>
32226 * doc/invoke.texi: Replace inequality signs with square brackets
32229 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32232 * lra-constraints.c (simplify_operand_subreg): Handle
32233 WORD_REGISTER_OPERATIONS targets.
32235 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32238 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32239 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32240 elimination by swapping fld*.
32242 2017-02-22 Richard Biener <rguenther@suse.de>
32244 PR tree-optimization/79673
32245 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32246 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32247 irrelevant address-space qualifiers and avoiding a
32248 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32250 2017-02-22 Richard Biener <rguenther@suse.de>
32252 PR tree-optimization/79666
32253 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32254 to not symbolically negate if that may introduce undefined
32257 2017-02-22 Martin Liska <mliska@suse.cz>
32260 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32261 * data-streamer-out.c (streamer_write_gcov_count_stream):
32263 * value-prof.c (stream_out_histogram_value): Make assert more
32264 precise based on type of counter.
32266 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32269 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32270 Use nonimmediate_operand instead of memory_operand for operand 1.
32271 (float-extend standard_x87sse_constant_load splitter): Ditto.
32273 2017-02-21 Jeff Law <law@redhat.com>
32275 PR tree-optimization/79621
32276 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32277 blocks with edges to themselves.
32279 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32282 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32283 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32284 Use gimple_call_builtin_p.
32287 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32288 on temporarily removed DEBUG_INSNs.
32290 PR tree-optimization/79649
32291 * tree-loop-distribution.c (classify_partition): Give up on
32292 non-generic address space loads/stores.
32294 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32296 * doc/loop.texi (Loop manipulation): Remove nonexistent
32297 tree_ssa_loop_version from the documentation.
32298 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32300 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32303 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32304 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32305 * config/rs6000/rs6000.c: Include except.h.
32306 (rs6000_expand_split_stack_prologue): Call
32307 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32309 2017-02-21 Martin Jambor <mjambor@suse.cz>
32312 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32313 have been analyzed.
32315 2017-02-21 Martin Jambor <mjambor@suse.cz>
32317 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32318 for backward compatibility only.
32319 * doc/invoke.texi (Option Summary): Remove all references to
32320 -fipa-cp-alignment.
32322 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32326 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32328 * lra-constraints.c (curr_insn_transform): Handle
32329 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32331 2017-02-21 Martin Liska <mliska@suse.cz>
32333 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32335 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32337 PR translation/79638
32338 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32340 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32343 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32344 (arm_function_ok_for_sibcall): Return false for an indirect call by
32345 descriptor if all the argument registers are used.
32346 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32347 alignment of the function.
32349 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32351 PR tree-optimization/61441
32352 * simplify-rtx.c (simplify_const_unary_operation): For
32353 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32354 the sNaN unmodified.
32356 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32358 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32359 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32360 instead of SYSTEM_HEADER_DIR.
32362 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32363 Martin Liška <mliska@suse.cz>
32365 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32366 Fix typos and grammar, use active voice, and clarify.
32368 2017-02-20 Marek Polacek <polacek@redhat.com>
32370 PR middle-end/79537
32371 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32374 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32376 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32379 * config/i386/i386.c (ix86_expand_builtin): Handle
32380 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32381 ix86_builtins_isa[fcode].isa as a requirement of those
32382 flags and any other flag in the bitmask.
32383 (ix86_init_mmx_sse_builtins): Use 0 instead of
32384 ~OPTION_MASK_ISA_64BIT as mask.
32385 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32386 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32387 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32388 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32390 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32393 * lra-constraints.c (split_reg): Check requested split mode
32394 is supported by the register.
32396 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32398 * lra-constraints.c (simplify_operand_subreg): Remove early
32401 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32404 * lra-constraints.c (curr_insn_transform): Tighten condition
32405 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32407 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32410 * lra-constraints.c (curr_insn_transform): Handle
32411 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32413 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32416 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32418 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32420 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32423 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32426 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32428 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32431 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32433 * final.c (last_columnnum, override_columnnum): New variables.
32434 (final_start_function): Set last_columnnum, pass it to begin_prologue
32435 hook and pass 0 to dwarf2out_begin_prologue.
32436 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32437 to source_line debug hook.
32438 (notice_source_line): Compute last_columnnum and for debug_column_info
32439 return true on column changes.
32440 * debug.h (struct gcc_debug_hooks): Add column argument to
32441 source_line and begin_prologue hooks.
32442 (debug_nothing_int_charstar_int_bool): Remove prototype.
32443 (debug_nothing_int_int_charstar,
32444 debug_nothing_int_int_charstar_int_bool): New prototypes.
32445 (dwarf2out_begin_prologue): Add column argument.
32446 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32447 begin_prologue hooks.
32448 (debug_nothing_int_charstar_int_bool): Remove.
32449 (debug_nothing_int_int_charstar,
32450 debug_nothing_int_int_charstar_int_bool): New functions.
32451 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32452 through to dwarf2out_source_line.
32453 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32454 (dwarf2out_source_line): Add column argument, emit it if requested.
32455 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32457 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32458 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32459 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32460 through to dwarf2out_begin_prologue.
32461 (vmsdbgout_source_line): Add column argument, pass it through to
32462 dwarf2out_source_line.
32463 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32464 dbxout_source_line caller.
32465 (dbxout_source_line): Add column argument.
32467 * common.opt (gno-column-info, gcolumn-info): New options.
32468 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32469 (check_die): Also test for multiple DW_AT_decl_column attributes.
32470 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32471 DW_AT_decl_column if requested.
32472 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32474 (gen_variable_die): Likewise.
32475 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32476 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32479 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32480 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32481 (ix86_handle_option): Handle OPT_m3dnowa.
32482 * doc/invoke.texi (-m3dnowa): Document.
32483 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32484 -m3dnowa instead of -m3dnow -march=athlon.
32487 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32488 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32490 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32493 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32494 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32495 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32496 generator for vsx_xxpermdi_<mode>_be.
32497 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32498 force big-endian semantics.
32499 (vsx_xxpermdi_<mode>_be): New define_expand with same
32500 implementation as previous version of vsx_xxpermdi_<mode>.
32502 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32504 PR tree-optimization/79327
32505 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32506 variable, its initialization and use.
32508 2017-02-17 Julia Koval <julia.koval@intel.com>
32510 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32511 (OPTION_MASK_ISA_PKU_UNSET): New.
32512 (ix86_handle_option): Handle -mrdpid.
32513 * config/i386/cpuid.h (bit_RDPID): New.
32514 * config/i386/driver-i386.c (host_detect_local_cpu):
32515 Detect RDPID feature.
32516 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32517 * config/i386/i386-c.c (ix86_target_macros_internal):
32519 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32520 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32521 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32522 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32523 * config/i386/i386.md (define_insn "rdpid"): New.
32524 * config/i386/i386.opt Add -mrdpid.
32525 * config/i386/immintrin.h (_rdpid_u32): New.
32527 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32529 PR rtl-optimization/79541
32530 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32531 instead of transforming it into USE.
32533 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32535 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32536 If HONOR_SNANS (SFmode) force the input to a register.
32537 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32538 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32539 an frsp or similar insn.
32541 2017-02-17 Martin Liska <mliska@suse.cz>
32543 PR rtl-optimization/79577
32544 * params.def (selsched-max-sched-times): Increase minimum to 1.
32546 2017-02-17 Martin Liska <mliska@suse.cz>
32548 PR rtl-optimization/79574
32549 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32551 2017-02-17 Martin Liska <mliska@suse.cz>
32553 PR tree-optimization/79529
32554 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32555 ssa_defined_default_def_p to handle cases which are implicitly
32557 * tree-ssa.c (ssa_defined_default_def_p): New function.
32558 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32559 which are implicitly defined.
32560 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32562 2017-02-17 Richard Biener <rguenther@suse.de>
32564 PR middle-end/79576
32565 * params.def (max-ssa-name-query-depth): Limit to 10.
32567 2017-02-17 Richard Biener <rguenther@suse.de>
32569 PR tree-optimization/79552
32570 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32573 2017-02-17 Richard Biener <rguenther@suse.de>
32576 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32578 2017-02-17 Marek Polacek <polacek@redhat.com>
32580 PR middle-end/79536
32581 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32582 (fold_negate_expr): New wrapper.
32584 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32586 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32587 Correct terminology and de-emphasize pre-standard behavior.
32589 2017-02-16 Alan Modra <amodra@gmail.com>
32591 PR rtl-optimization/79286
32592 * ira.c (def_dominates_uses): New function.
32593 (update_equiv_regs): Don't create an equivalence for insns that
32594 may trap where the register def does not dominate the use.
32596 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32598 PR rtl-optimization/78127
32599 * lra.c (lra): Call lra_eliminate before finish the loop after
32602 2017-02-16 Richard Biener <rguenther@suse.de>
32604 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32606 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32607 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32608 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32609 (isl_val_int_from_wi): New function.
32610 (extract_affine_gmp): Rename to ...
32611 (extract_affine_wi): ... this, take a widest_int.
32612 (extract_affine_int): Just wrap extract_affine_wi.
32613 (add_param_constraints): Use isl_val_int_from_wi.
32614 (add_loop_constraints): Likewise, and extract_affine_wi.
32616 2017-02-15 Jeff Law <law@redhat.com>
32618 PR middle-end/79521
32619 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32620 ira_init_register_move_cost_if_necessary.
32622 2017-02-15 Martin Sebor <msebor@redhat.com>
32624 PR middle-end/32003
32625 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32626 removed in a prior commit.
32628 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32630 PR tree-optimization/79347
32631 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32632 counters during peeling.
32634 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32636 * Makefile.in (site.exp): Remove "set ISLVER".
32638 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32641 * real.c (real_from_integer): Call real_convert even for decimal.
32643 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32646 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32648 2017-02-14 Andrew Pinski <apinski@cavium.com>
32650 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32651 cores and change the partno/implementer to be correct.
32652 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32653 the 'B" as the implementer.
32654 * config/aarch64/aarch64-tune.md: Regenerate.
32656 2017-02-14 Carl Love <cel@us.ibm.com>
32658 * config/rs6000/rs6000.c: Add case statement entry to make the
32659 xvcvuxdsp built-in argument unsigned.
32660 * config/rs6000/vsx.md: Fix the source and return operand types so they
32661 match the instruction definitions from the ISA document. Fix typo
32662 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32665 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32668 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32669 member early_clobber_alts.
32670 * lra-lives.c (reg_early_clobber_p): New.
32671 (process_bb_lives): Use it.
32672 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32673 (debug_operand_data): Initialize early_clobber_alts.
32674 (setup_operand_alternative): Set up early_clobber_alts.
32675 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32676 alternatives to new_insn_reg.
32677 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32679 (lra_update_insn_regno_info): Pass the new arg.
32681 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32683 PR middle-end/79505
32684 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32685 (new_oacc_loop_raw): Don't clear already cleared fields.
32688 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32689 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32690 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32692 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32695 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32697 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32700 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32701 the extra instruction to the right place to store 128-bit constant
32704 2017-02-14 Martin Sebor <msebor@redhat.com>
32706 PR middle-end/79448
32707 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32708 warning for strings of unknown length.
32710 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32712 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32714 2017-02-14 Jeff Law <law@redhat.com>
32717 * ira-costs.c (scan_one_insn): Initialize register move costs
32718 for pseudos seen in USE/CLOBBER insns.
32720 PR tree-optimization/79095
32721 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32722 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32723 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32724 if the operands are known to be not equal, then the resulting range
32726 (intersect_ranges): If the new range is ~[0,0] and the old range is
32727 wide, then prefer ~[0,0].
32728 * tree-vrp.c (overflow_comparison_p_1): New function.
32729 (overflow_comparison_p): New function.
32730 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32731 if NAME is used in an overflow test.
32732 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32733 overflow check that can be expressed as an equality test, then adjust
32734 ops to be that equality test.
32736 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32738 * config/s390/s390-builtin-types.def: Remove flags argument.
32739 * config/s390/s390.c (s390_init_builtins): Likewise.
32741 2017-02-14 Martin Liska <mliska@suse.cz>
32743 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32744 vector. Fix trailing white spaces.
32746 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32748 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32751 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32753 PR rtl-optimization/68664
32754 * config/arm/arm.c (arm_sched_can_speculate_insn):
32755 New function. Declare prototype.
32756 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32758 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32760 PR rtl-optimization/68664
32761 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32763 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32765 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32767 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32768 max skip bytes for function, loop and jump.
32770 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32772 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32773 ABS_EXPR for gimple dump.
32775 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32778 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32780 PR tree-optimization/79408
32781 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32782 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32783 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32784 also if rhs1 is INTEGER_CST.
32786 2017-02-14 Richard Biener <rguenther@suse.de>
32788 PR middle-end/79432
32789 * tree-into-ssa.c (insert_phi_nodes): When the function can
32790 have abnormal edges rewrite SSA names with broken use-def
32791 dominance out of SSA and register them for PHI insertion.
32793 2017-02-13 Martin Sebor <msebor@redhat.com>
32795 PR middle-end/79496
32796 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32797 clearing info.nowrite flag when snprintf size argument is a range.
32799 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32801 * cprop.c (cprop_jump): Add missing space in string literal.
32802 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
32803 (get_constraint_for_component_ref): Likewise.
32804 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
32805 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
32806 * lra-constraints.c (process_alt_operands): Likewise.
32807 * ipa-inline.c (inline_small_functions): Likewise.
32808 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
32809 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
32810 * trans-mem.c (diagnose_tm_1_op): Likewise.
32811 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
32812 (grid_parallel_clauses_gridifiable): Likewise.
32814 * config/nvptx/mkoffload.c (process): Add space in between
32817 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
32818 "MOD4_SSE_REGS" and "ALL_REGS".
32820 * spellcheck.c (test_data): Add , in between "foo" and "food".
32822 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32825 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
32826 boundary crossing check and subsequent code generation agree.
32828 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32830 * config/aarch64/aarch64.c (has_memory_op): Delete.
32831 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32834 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32836 PR rtl-optimization/79388
32837 PR rtl-optimization/79450
32838 * combine.c (distribute_notes): When removing TEM_INSN for which
32839 corresponding dest has last value recorded, invalidate that last
32842 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32844 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32845 of explicit '@'. Add missing assembly comment marker on branch costs
32848 2017-02-13 Nathan Sidwell <nathan@acm.org>
32850 * gengtype-lex.l (<in_struct>): Add '/'.
32852 2017-02-13 Martin Liska <mliska@suse.cz>
32855 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32857 2017-02-13 Richard Biener <rguenther@suse.de>
32859 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32861 * configure: Re-generate.
32862 * config.in: Likewise.
32863 * graphite-dependences.c: Simplify as if
32864 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32865 * graphite-isl-ast-to-gimple.c: Likewise.
32866 * graphite-optimize-isl.c: Likewise.
32867 * graphite-poly.c: Likewise.
32868 * graphite-sese-to-poly.c: Likewise.
32869 * graphite.h: Likewise.
32870 * toplev.c: Include isl/version.h and use isl_version () for
32871 printing the ISL version.
32872 * doc/install.texi: Update ISL requirement.
32874 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32876 * doc/standards.texi (Standards): Update reference to
32879 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32881 * doc/extend.texi (Named Address Spaces): sourceware.org now
32883 * doc/install.texi (Binaries): Ditto.
32886 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32888 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32889 standard terminology "stringize"/"stringizing" throughout.
32890 * doc/cppinternals.texi: Likewise.
32892 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32894 * doc/extend.texi: Fix some spelling mistakes and typos.
32895 * doc/invoke.texi: Likewise.
32897 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32900 * params.def (inline-min-speedup) Change from 10 to 8.
32902 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32904 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32907 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32910 * ipa-inline-analysis.c (get_minimal_bb): New function.
32911 (record_modified): Use it.
32912 (remap_edge_change_prob): Handle also ancestor functions.
32914 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32916 * doc/contrib.texi (Contributors): Remove broken link into
32917 the Mauve CVS repository.
32919 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32921 PR middle-end/79454
32922 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32923 result computation whenever lhs doesn't have vector mode, not
32924 just when it has BLKmode.
32926 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32928 * doc/makefile.texi (profiledbootstrap): Refer to the
32929 installation instructions only in textual form.
32931 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32934 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32936 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32938 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32939 (Specific): Update mingw-w64 reference.
32941 (Specific): Remove broken link to Renesas RX processor.
32943 2017-02-10 Richard Biener <rguenther@suse.de>
32945 * toplev.c (process_options): Do not mention obsolete graphite
32946 options when printing sorry message about missing graphite support.
32947 Mention -floop-nest-optimize.
32949 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32951 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32952 (vtst_p16): Likewise.
32953 (vtstq_p8): Likewise.
32954 (vtstq_p16): Likewise.
32956 (vtstq_p64): Likewise.
32957 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32958 (vset_lane_p64): New.
32959 (vsetq_lane_p64): New.
32961 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32963 PR tree-optimization/79411
32964 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32965 stmt operands are SSA_NAMEs used in abnormal phis.
32966 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32969 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32972 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32975 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32977 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32979 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32981 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32985 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32986 not arbitrary TREE_CONSTANT.
32989 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32990 "omp declare target link" attribute unless is_global_var.
32991 * omp-offload.c (find_link_var_op): Likewise.
32993 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32994 Chung-Lin Tang <cltang@codesourcery.com>
32996 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32998 (gimplify_adjust_omp_clauses): Don't delete TILE.
32999 (gimplify_omp_for): Deal with TILE.
33000 * internal-fn.c (expand_GOACC_TILE): New function.
33001 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33003 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33004 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33006 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33007 avoid DIV for outermost collapse var.
33008 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33009 Remove out of date comments, fix whitespace.
33010 * omp-general.c (omp_extract_for_data): Deal with tiling.
33011 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33012 adjust OLF_DIM_BASE value.
33013 (struct omp_for_data): Add tiling field.
33014 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33015 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33016 for auto loops. Remove default auto determining, moved to
33017 oacc_loop_fixed_partitions.
33018 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33019 stmts, add e_mask field.
33020 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33021 (oacc_thread_numbers): Use oacc_dim_call.
33022 (oacc_xform_tile): New.
33023 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33024 (finish_oacc_loop): Adjust for ifns vector.
33025 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33026 add case for GOACC_TILE fns.
33027 (oacc_loop_xform_loop): Delete.
33028 (oacc_loop_process): Iterate over call list directly, and add
33029 handling for GOACC_TILE fns.
33030 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33032 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33033 vector partitioning to outer loops. Assign 2 partitions to loops
33034 when available. Add TILE handling.
33035 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33036 (execite_oacc_device_lower): Process GOACC_TILE fns,
33037 ignore unknown specs.
33038 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33039 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33040 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33042 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33044 * configure.ac (ACX_BUGURL): Update.
33045 * configure: Regenerate.
33047 2017-02-09 Richard Biener <rguenther@suse.de>
33049 PR tree-optimization/69823
33050 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33051 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33053 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33055 * config/arc/arc-c.def: Add __NPS400__ definition.
33056 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33057 (TARGET_NPS400): Define.
33059 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33061 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33063 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33064 pointer, arch_info.
33065 (arc_cpu_types): Fill the arch_info field with a pointer into the
33066 arc_arch_types table.
33067 (arc_selected_cpu): Declare.
33068 * config/arc/arc.c (arc_selected_cpu): Make global.
33069 (arc_selected_arch): Delete.
33070 (arc_base_cpu): Delete.
33071 (arc_override_options): Remove references to deleted variables,
33072 update access to arch information.
33073 (ARC_OPT): Update access to arch information.
33074 (ARC_OPTX): Likewise.
33075 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33076 (TARGET_ARC600): Update access to arch information.
33077 (TARGET_ARC601): Likewise.
33078 (TARGET_ARC700): Likewise.
33079 (TARGET_EM): Likewise.
33080 (TARGET_HS): Likewise.
33081 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33084 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33087 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33088 condition/operands for integer GE/LE/GEU/LEU operations.
33090 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33092 PR translation/79397
33093 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33096 2017-02-08 Martin Jambor <mjambor@suse.cz>
33099 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33100 whether allocation happened.
33101 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33102 nothing was allocated.
33104 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33106 PR tree-optimization/79408
33107 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33108 constant, but SSA_NAME with a known integer range, use the minimum
33109 of that range instead of op1 to determine if modulo can be replaced
33110 with its first operand.
33112 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33114 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33116 2017-02-08 Richard Biener <rguenther@suse.de>
33118 PR tree-optimization/71824
33119 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33120 Check all loops contained in the merged region.
33122 2017-02-07 Andrew Pinski <apinski@cavium.com>
33124 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33126 2017-02-07 Andrew Pinski <apinski@cavium.com>
33128 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33129 (thunderxt88): Likewise.
33130 (thunderxt81): Disable LSE and change v8.1 to v8.
33131 (thunderxt83): Likewise.
33133 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33134 Richard Biener <rguenther@suse.de>
33136 PR middle-end/79399
33137 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33138 type from int to size_t.
33139 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33141 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33143 PR rtl-optimization/79386
33144 * cprop.c (bypass_conditional_jumps): Initialize
33145 bypass_last_basic_block already before splitting bbs after
33146 unconditional traps...
33147 (bypass_conditional_jumps): ... rather than here.
33150 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33151 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33152 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33153 fix -masm=intel patterns.
33155 2017-02-07 Richard Biener <rguenther@suse.de>
33157 PR tree-optimization/79256
33158 PR middle-end/79278
33159 * builtins.c (get_object_alignment_2): Use min_align_of_type
33160 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33161 and ADJUST_FIELD_ALIGN.
33163 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33165 * doc/tm.texi: Regenerate.
33166 * stor-layout.c (layout_decl): Adjust.
33167 (update_alignment_for_field): Likewise.
33168 (place_field): Likewise.
33169 (min_align_of_type): Likewise.
33170 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33171 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33172 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33173 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33174 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33175 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33176 * config/i386/i386.c (x86_field_alignment): Likewise.
33177 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33178 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33179 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33180 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33181 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33182 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33186 2017-01-30 Richard Biener <rguenther@suse.de>
33188 PR tree-optimization/79256
33189 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33190 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33193 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33195 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33196 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33197 builtins to SImode and emit a zero-extend, if necessary.
33199 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33201 * docs/invoke.texi (RISC-V Options): Alphabetize.
33203 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33205 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33208 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33210 * config/riscv/riscv.c: New file.
33211 * common/config/riscv/riscv-common.c: Likewise.
33212 * config.gcc: Likewise.
33213 * config/riscv/constraints.md: Likewise.
33214 * config/riscv/elf.h: Likewise.
33215 * config/riscv/generic.md: Likewise.
33216 * config/riscv/linux.h: Likewise.
33217 * config/riscv/multilib-generator: Likewise.
33218 * config/riscv/peephole.md: Likewise.
33219 * config/riscv/pic.md: Likewise.
33220 * config/riscv/predicates.md: Likewise.
33221 * config/riscv/riscv-builtins.c: Likewise.
33222 * config/riscv/riscv-c.c: Likewise.
33223 * config/riscv/riscv-ftypes.def: Likewise.
33224 * config/riscv/riscv-modes.def: Likewise.
33225 * config/riscv/riscv-opts.h: Likewise.
33226 * config/riscv/riscv-protos.h: Likewise.
33227 * config/riscv/riscv.h: Likewise.
33228 * config/riscv/riscv.md: Likewise.
33229 * config/riscv/riscv.opt: Likewise.
33230 * config/riscv/sync.md: Likewise.
33231 * config/riscv/t-elf-multilib: Likewise.
33232 * config/riscv/t-linux: Likewise.
33233 * config/riscv/t-linux-multilib: Likewise.
33234 * config/riscv/t-riscv: Likewise.
33235 * configure.ac: Likewise.
33236 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33237 Waterman as RISC-V maintainers.
33238 * doc/install.texi: Add RISC-V entries.
33239 * doc/invoke.texi: Add RISC-V options section.
33240 * doc/md.texi: Add RISC-V constraints section.
33241 * configure: Regenerated.
33243 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33246 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33247 false values to be constant vectors with all 0 or all 1 bits set.
33248 (vcondu<mode><mode>): Likewise.
33249 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33251 (fpmask_comparison_operator): Update comment.
33252 (vecint_comparison_operator): New predicate.
33253 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33254 vector conditionals when the true and false values are constant
33255 vectors with all 0 bits or all 1 bits set.
33257 2017-02-06 Martin Sebor <msebor@redhat.com>
33259 PR tree-optimization/79376
33260 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33262 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33264 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33265 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33266 to simplify split condition.
33268 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33270 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33271 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33274 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33276 PR rtl-optimization/68664
33277 * target.def (can_speculate_insn): New hook.
33278 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33279 * doc/tm.texi: Regenerate.
33280 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33281 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33282 (rs6000_sched_can_speculate_insn): New function.
33284 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33286 PR tree-optimization/79284
33287 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33288 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33289 vectorizable_mask_load_store, vectorizable_operation,
33290 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33291 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33292 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33293 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33294 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33295 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33296 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33297 is_gimple_assign (stmt). Replace another such test with
33298 is_gimple_assign (stmt).
33300 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33303 * config/avr/avr.c (rtl-iter.h): Include it.
33304 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33305 (avr_legitimate_combined_insn): ...and implementation.
33307 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33309 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33310 * config/s390/s390.c (s390_const_operand_ok)
33311 (s390_canonicalize_comparison, s390_extract_part)
33312 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33313 (s390_contiguous_bitmask_p, s390_rtx_costs)
33314 (legitimize_pic_address): Likewise.
33315 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33316 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33317 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33318 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33319 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33321 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33323 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33324 REGNO($0) == REGNO($1).
33326 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33328 * config/s390/linux.h(SIZE_TYPE): Add comment.
33330 2017-02-06 Julian Brown <julian@codesourcery.com>
33331 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33332 Virendra Pathak <virendra.pathak@broadcom.com>
33334 * config/aarch64/aarch64-cores.def: Change the scheduler
33336 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33337 * config/aarch64/thunderx2t99.md: New file.
33339 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33341 * doc/standards.texi (Go Language): Update link to language
33344 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33346 * tree-eh.c (lower_resx): Sanitize profile.
33347 (cleanup_empty_eh_move_lp): Likewise.
33349 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33352 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33354 * cfgloopmanip.h (loop_version): Update prototype.
33355 * modulo-sched.c (sms_schedule): Update call of loop_version.
33356 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33357 * tree-parloops.c (gen_parallel_loop): Likewise.
33358 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33359 * tree-ssa-loop-split.c (split_loop): Likewise.
33360 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33361 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33363 2017-02-05 Martin Liska <mliska@suse.cz>
33366 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33368 (print_operand_address): Initialize a struct to zero.
33370 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33372 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33373 garbage collector only in textual form.
33375 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33377 * doc/extend.texi (x86 specific memory model extensions for
33378 transactional memory): Simplify a phrase.
33380 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33383 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33384 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33385 (atomic_storedi_1): Likewise.
33387 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33389 PR tree-optimization/79338
33390 * tree-parloops.c (gather_scalar_reductions): Don't call
33391 vect_analyze_loop_form for loop->inner before destroying loop's
33394 2017-02-03 Martin Sebor <msebor@redhat.com>
33396 PR tree-optimization/79327
33397 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33398 when precision has resulted in leading zeros.
33399 (format_integer): Adjust the likely counter to assume an unknown
33400 argument that may be zero is non-zero.
33402 2017-02-03 Jason Merrill <jason@redhat.com>
33405 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33406 avoid copying non-taken branch.
33408 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33410 PR tree-optimization/79340
33411 * tree-vect-loop.c (vectorizable_reduction): Release
33412 vec_defs elements after safe_splicing them into other vectors.
33415 PR tree-optimization/79327
33416 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33417 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33419 (format_integer): Use wide_int_to_tree instead of build_int_cst
33420 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33421 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33422 of shortest and longest sequence.
33424 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33426 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33427 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33429 2017-02-03 Walter Lee <walt@tilera.com>
33432 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33433 after initial stackframe link reg save.
33434 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33436 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33439 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33440 wu for stxssp alternative.
33442 2017-02-03 Martin Sebor <msebor@redhat.com>
33444 PR tree-optimization/79352
33445 * gimple-fold.c (get_range_strlen): Add argument.
33446 (get_range_strlen): Change return type to bool.
33447 (get_maxval_strlen): Pass in a dummy argument.
33448 * gimple-fold.h (get_range_strlen): Change return type to bool.
33449 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33450 * tree.h (array_at_struct_end_p): Add argument.
33451 * tree.c (array_at_struct_end_p): Handle it.
33453 2017-02-03 Martin Liska <mliska@suse.cz>
33456 * multiple_target.c (create_dispatcher_calls): Redirect edge
33457 from a caller of a dispatcher.
33458 (expand_target_clones): Make the clones local.
33459 (ipa_target_clone): Do both target clones and resolvers.
33460 (ipa_dispatcher_calls): Remove the pass.
33461 (pass_dispatcher_calls::gate): Likewise.
33462 (make_pass_dispatcher_calls): Likewise.
33463 * passes.def (pass_target_clone): Put as very first IPA early
33466 2017-02-03 Martin Liska <mliska@suse.cz>
33468 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33469 in case of a function with ifunc attribute.
33471 2017-02-03 Martin Liska <mliska@suse.cz>
33473 * cgraph.c (cgraph_node::dump): Dump function version info.
33474 * symtab.c (symtab_node::dump_base): Add missing new line.
33476 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33478 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33479 (ifcombine_ifandif): Use it.
33481 2017-02-03 Martin Liska <mliska@suse.cz>
33483 * doc/invoke.texi: Document default value for
33484 use-after-scope-direct-emission-threshold.
33486 2017-02-03 Martin Liska <mliska@suse.cz>
33488 PR tree-optimization/79339
33489 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33490 (format_floating): Likewise.
33492 2017-02-03 Martin Liska <mliska@suse.cz>
33495 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33497 (ipa_node_params_t::remove): Likewise.
33498 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33499 initialization from removed ipa_node_params_t::insert.
33500 (ipa_node_params::~ipa_node_params): Move from removed
33501 ipa_node_params_t::release.
33502 * symbol-summary.h (symbol_summary::m_released): New member.
33503 Do not release a summary twice. Do not allow to call finalizer
33504 for types of a summary that live in GGC memory.
33506 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33508 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33511 2017-02-02 Martin Sebor <msebor@redhat.com>
33513 PR middle-end/79275
33514 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33515 (format_string): Tighten up the range of output for non-constant
33516 strings and correct the expected range for wide non-constant strings.
33518 2017-02-02 Martin Sebor <msebor@redhat.com>
33520 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33522 PR middle-end/32003
33523 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33525 (-fdump-tree-@var): Add to index and document how to come up
33526 with pass-specific option and dump file names.
33527 (-fdump-passes): Clarify where to look for output.
33529 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33531 PR middle-end/77445
33532 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33533 statistics of the analyzed path; allow threading for speed when
33534 any of BBs along the path are optimized for speed.
33536 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33538 PR middle-end/78468
33539 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33540 settings of the virtual registers.
33543 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33545 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33546 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33549 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33551 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33552 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33554 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33556 * config/s390/s390.md: Add missing comments with the expanded
33558 * config/s390/vector.md: Likewise.
33559 * config/s390/vx-builtins.md: Likewise.
33561 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33564 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33565 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33566 conditions on a single line.
33568 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33570 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33571 __S390_VX__ to __VX__.
33573 2017-02-01 Andrew Pinski <apinski@cavium.com>
33575 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33576 stmt_info to record_stmt_cost.
33577 (vect_get_known_peeling_cost): Pass stmt_info if known to
33579 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33580 cpu_vector_cost field into
33581 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33582 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33583 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33584 splitting of scalar_stmt_cost and vec_stmt_cost.
33585 (thunderx_vector_cost): Likewise.
33586 (cortexa57_vector_cost): LIkewise.
33587 (exynosm1_vector_cost): Likewise.
33588 (xgene1_vector_cost): Likewise.
33589 (thunderx2t99_vector_cost): Improve after the splitting of the two
33591 (aarch64_builtin_vectorization_cost): Update for the splitting of
33592 scalar_stmt_cost and vec_stmt_cost.
33594 2017-02-01 Torvald Riegel <triegel@redhat.com>
33595 Richard Henderson <rth@redhat.com>
33597 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33598 conditional on existance of a fast atomic load.
33599 * optabs-query.c (can_atomic_load_p): New function.
33600 * optabs-query.h (can_atomic_load_p): Declare it.
33601 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33602 no fast atomic load is available for the particular size of access.
33603 (expand_atomic_compare_and_swap): Likewise.
33604 (expand_atomic_load): Likewise.
33605 (expand_atomic_store): Likewise.
33606 (expand_atomic_fetch_op): Likewise.
33607 * testsuite/lib/target-supports.exp
33608 (check_effective_target_sync_int_128): Remove x86 because it provides
33609 no fast atomic load.
33610 (check_effective_target_sync_int_128_runtime): Likewise.
33612 2017-02-01 Richard Biener <rguenther@suse.de>
33614 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33615 (graphite_transform_loops): Provide opt-info for optimized nests.
33616 * tree-parloop.c (parallelize_loops): Provide opt-info for
33617 parallelized loops.
33619 2017-02-01 Richard Biener <rguenther@suse.de>
33621 PR middle-end/79315
33622 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33623 was not set before.
33625 2017-02-01 Richard Biener <rguenther@suse.de>
33627 PR tree-optimization/71824
33628 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33629 Verify the loops are valid in the merged SESE region.
33630 (scop_detection::can_represent_loop_1): Check analyzing the
33631 evolution of the number of iterations in the region succeeds.
33633 2017-01-31 Ian Lance Taylor <iant@golang.org>
33635 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33636 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33638 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33640 PR preprocessor/79210
33641 * input.c (get_substring_ranges_for_loc): Replace line_width
33642 assertion with error-handling.
33644 2017-01-31 Richard Biener <rguenther@suse.de>
33646 PR tree-optimization/77318
33647 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33648 (create_pw_aff_from_tree): Take loop parameter.
33649 (add_condition_to_pbb): Pass loop of the condition to
33650 create_pw_aff_from_tree.
33652 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33654 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33655 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33657 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33661 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33663 (convert_int_to_float128): Likewise.
33664 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33665 (convert_int_to_float128): Likewise.
33666 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33667 (UNSPEC_IEEE128_CONVERT): Likewise.
33668 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33669 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33670 Use local variables for IBM extended format.
33671 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33672 (fix_trunc<mode>si2_fprs): Likewise.
33673 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33674 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33675 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33676 to know that we can now have integers of all sizes in vector
33678 (fix<uns>_<mode>di2_hw): Likewise.
33679 (float<uns>_<mode>si2_hw): Likewise.
33680 (fix_<mode>si2_hw): Likewise.
33681 (fixuns_<mode>si2_hw): Likewise.
33682 (float<uns>_<mode>di2_hw): Likewise.
33683 (float_<mode>di2_hw): Likewise.
33684 (float_<mode>si2_hw): Likewise.
33685 (floatuns_<mode>di2_hw): Likewise.
33686 (floatuns_<mode>si2_hw): Likewise.
33687 (xscvqp<su>wz_<mode>): Delete, no longer used.
33688 (xscvqp<su>dz_<mode>): Likewise.
33689 (xscv<su>dqp_<mode>): Likewise.
33690 (ieee128_mfvsrd_64bit): Likewise.
33691 (ieee128_mfvsrd_32bit): Likewise.
33692 (ieee128_mfvsrwz): Likewise.
33693 (ieee128_mtvsrw): Likewise.
33694 (ieee128_mtvsrd_64bit): Likewise.
33695 (ieee128_mtvsrd_32bit): Likewise.
33697 2017-01-31 Martin Liska <mliska@suse.cz>
33700 * ipa-prop.c (ipa_free_all_node_params): Call release method
33701 instead of ~sumbol_summary to not to trigger double times
33704 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33706 PR tree-optimization/71691
33707 * bitmap.h (class auto_bitmap): New.
33708 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33709 is_maybe_undefined instead of ssa_undefined_value_p.
33711 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33713 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33714 __S390_ARCH_LEVEL__ to __ARCH__.
33716 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33718 PR tree-optimization/79267
33719 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33720 if should_remove_lhs_p is true.
33722 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33725 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33726 (add_alignment_attribute): New.
33727 (base_type_die): Add alignment attribute.
33728 (subrange_type_die): Likewise.
33729 (modified_type_die): Likewise.
33730 (gen_array_type_die): Likewise.
33731 (gen_descr_array_type_die: Likewise.
33732 (gen_enumeration_type_die): Likewise.
33733 (gen_subprogram_die): Likewise.
33734 (gen_variable_die): Likewise.
33735 (gen_field_die): Likewise.
33736 (gen_ptr_to_mbr_type_die): Likewise.
33737 (gen_struct_or_union_type_die): Likewise.
33738 (gen_subroutine_type_die): Likewise.
33739 (gen_typedef_die): Likewise.
33740 (base_type_cmp): Compare alignment attribute.
33742 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33745 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33746 (setb_unsigned) New pattern for setb with CCUNS.
33747 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33748 subfc./subfe sequence to avoid overflow problems. Generate a
33749 shorter sequence with cmpld/setb for power9.
33750 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33751 for generating subfc. instruction.
33752 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33753 now uses this instruction.
33755 2017-01-30 Ian Lance Taylor <iant@google.com>
33758 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33759 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33761 2017-01-30 Martin Sebor <msebor@redhat.com>
33763 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33764 Move constant to the right of a relational operator.
33765 (get_mpfr_format_length, format_character, format_string): Ditto.
33766 (should_warn_p, maybe_warn): Same.
33768 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33770 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33773 * asan.c (get_translation_unit_decl): Remove function.
33774 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33776 2017-01-30 Martin Liska <mliska@suse.cz>
33778 PR gcov-profile/79259
33779 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33780 -fprofile-generate.
33782 2017-01-30 Martin Liska <mliska@suse.cz>
33785 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33786 Initialize variables with NULL value.
33788 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33791 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33793 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33795 2017-01-30 Richard Biener <rguenther@suse.de>
33797 PR tree-optimization/79276
33798 * tree-vrp.c (process_assert_insertions): Properly adjust common
33799 when removing a duplicate.
33801 2017-01-30 Richard Biener <rguenther@suse.de>
33803 PR tree-optimization/79256
33804 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33805 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33807 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
33809 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
33812 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
33813 ("*r<noxa>sbg_<mode>_sll_bitmask")
33814 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
33815 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
33816 Use contiguous_bitmask_nowrap_operand.
33818 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33821 * config/rs6000/altivec.h (vec_xl): Revise #define.
33822 (vec_xst): Likewise.
33824 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
33826 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
33828 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
33830 PR rtl-optimization/79194
33831 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33832 traps before call to bypass_conditional_jumps.
33834 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33836 PR tree-optimization/71374
33837 * lra-constraints.c (check_conflict_input_operands): New.
33838 (match_reload): Use it.
33840 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33843 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33844 account to calculate conflict_set.
33846 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33848 PR rtl-optimization/78559
33849 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33850 other_insn in combine.
33852 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33854 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33855 uint16_type_node for BT_UINT16.
33857 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33859 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33860 "RTL Tests" to menu.
33861 (GIMPLE Tests): New node.
33862 (RTL Tests): New node.
33864 2017-01-27 Richard Biener <rguenther@suse.de>
33866 PR tree-optimization/79245
33867 * tree-loop-distribution.c (distribute_loop): Apply cost
33868 modeling also to detected patterns.
33870 2017-01-27 Richard Biener <rguenther@suse.de>
33872 PR tree-optimization/71433
33873 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33874 (compare_assert_loc): New function.
33875 (process_assert_insertions): Sort and optimize assert locations
33876 to remove duplicates and push down identical assertions on
33877 edges to their destination block.
33879 2017-01-27 Richard Biener <rguenther@suse.de>
33881 PR tree-optimization/79244
33882 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33883 out SSA names even if abnormal.
33885 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33887 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33888 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33889 instead of MPFR_RNDN.
33891 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33894 * arm.c (arm_option_override): Don't call build_target_option_node
33895 until after doing all option overrides.
33896 (arm_valid_target_attribute_tree): Likewise.
33898 2017-01-27 Martin Liska <mliska@suse.cz>
33900 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33901 for {cd}tors and C++ {cd}tors.
33903 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33905 * config/s390/s390.md ("*setmem_long_and")
33906 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33908 2017-01-26 Martin Sebor <msebor@redhat.com>
33910 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33913 2017-01-26 Martin Sebor <msebor@redhat.com>
33915 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33916 HAVE_DFmode before using XFmode or DFmode.
33917 (parse_directive): Avoid using the z length modifier to avoid
33918 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33920 PR middle-end/78703
33921 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33922 to accept adjustment as an array.
33923 (get_int_range): New function.
33924 (struct directive): Make width and prec arrays.
33925 (directive::set_width, directive::set_precision): Call get_int_range.
33926 (format_integer, format_floating): Handle width and precision ranges.
33927 (format_string, parse_directive): Same.
33929 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33932 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33933 comdat_type_p set, just clone them, but keep the children in the
33937 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33938 which have direct callers with -fvar-tracking-assignments enabled
33940 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33941 inside of type units.
33943 2017-01-26 Martin Sebor <msebor@redhat.com>
33945 PR middle-end/78703
33946 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33948 (struct format_result): Replace number_chars, number_chars_min,
33949 and number_chars_max with a single member of struct result_range.
33951 (format_result::operator+=): Adjust.
33952 (struct fmtresult): Remove bounded. Handle likely and unlikely
33954 (fmtresult::adjust_for_width_or_precision): New function.
33955 (fmtresult:type_max_digits): New function.
33956 (bytes_remaining): Handle likely and unlikely counters.
33957 (min_bytes_remaining): Remove.
33958 (format_percent): Simplify.
33959 (format_integer, format_floating): Set likely and unlikely counters.
33960 (get_string_length, format_character, format_string): Same.
33961 (format_plain, should_warn_p): New function.
33962 (maybe_warn): Call should_warn_p. Update diagnostic messages
33963 and handle those for all directives, including plain strings.
33964 (format_directive): Handle likely and unlikely counters.
33965 Remove unnecessary quoting from diagnostics. Add an informational
33967 (add_bytes): Remove.
33968 (pass_sprintf_length::compute_format_length): Simplify.
33969 (try_substitute_return_value): Handle likely and unlikely counters.
33971 2017-01-26 Carl Love <cel@us.ibm.com>
33973 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33974 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33976 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33979 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33980 endianess for subregs into account.
33981 * lra-constraints.c (lra_constraints): Do risky transformations
33982 always on the first iteration.
33983 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33984 last_call_used_reg_set.
33985 (process_bb_lives): Define and use last_call_used_reg_set.
33986 * lra.c (lra): Always continue after lra_constraints on the first
33989 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33991 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33993 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33995 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33997 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33998 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33999 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34000 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34001 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34002 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34003 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34004 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34005 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34007 2017-01-26 Marek Polacek <polacek@redhat.com>
34010 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34011 for the third operand.
34013 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34015 PR middle-end/79236
34016 * omp-low.c (struct omp_context): Add simt_stmt field.
34017 (scan_omp_for): Return omp_context *.
34018 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34019 context to the _simt_ SIMD stmt.
34020 (lower_omp_for): For combined SIMD with sibling _simt_
34021 SIMD, make sure to use the same decls in _looptemp_
34022 clauses as in the sibling.
34024 2017-01-26 David Sherwood <david.sherwood@arm.com>
34026 PR middle-end/79212
34027 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34030 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34033 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34034 emit fld b; fld a; if possible.
34036 * brig-builtins.def: Update copyright years.
34037 * config/arm/arm_acle_builtins.def: Update copyright years.
34039 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34042 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34043 constraint instead of o for the stxsd instruction.
34045 2017-01-25 Carl Love <cel@us.ibm.com>
34047 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34048 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34050 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34052 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34054 2017-01-25 Richard Biener <rguenther@suse.de>
34056 PR tree-optimization/69264
34057 * target.def (vector_alignment_reachable): Improve documentation.
34058 * doc/tm.texi: Regenerate.
34059 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34061 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34062 earlier changes with respect to TYPE_USER_ALIGN.
34063 (vector_alignment_reachable_p): Likewise. Improve dumping.
34065 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34068 * config/arm/arm.md (xordi3): Force constant operand into a register
34071 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34073 * doc/invoke.texi (-fstore-merging): Correct default optimization
34074 levels at which it is enabled.
34075 (-O): Move -fstore-merging from list to...
34078 2017-01-25 Richard Biener <rguenther@suse.de>
34081 * omp-expand.c: Include debug.h.
34082 (expand_omp_taskreg): Make sure to generate early debug before
34083 outlining anything from a function.
34084 (expand_omp_target): Likewise.
34085 (grid_expand_target_grid_body): Likewise.
34087 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34090 * asan.c (get_translation_unit_decl): New function.
34091 (asan_add_global): Extract modules file name from globals
34092 TRANSLATION_UNIT_DECL name.
34094 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34097 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34098 for long calls with APCS frame and VFP.
34100 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34102 * cfg.c (original_copy_tables_initialized_p): New function.
34103 * cfg.h (original_copy_tables_initialized_p): New decl.
34104 * cfgrtl.c (relink_block_chain): Guard the call to
34105 free_original_copy_tables with a call to
34106 original_copy_tables_initialized_p.
34107 * cgraph.h (symtab_node::native_rtl_p): New decl.
34108 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34109 (symtab_node::needed_p): Don't assert for early assembly output
34110 for __RTL functions.
34111 (cgraph_node::finalize_function): Set "force_output" for __RTL
34113 (cgraph_node::analyze): Bail out early for __RTL functions.
34114 (analyze_functions): Update assertion to support __RTL functions.
34115 (cgraph_node::expand): Bail out early for __RTL functions.
34116 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34118 * function.h (struct function): Update comment for field
34120 * gimple-expr.c: Include "tree-pass.h".
34121 (gimple_has_body_p): Return false for __RTL functions.
34122 * Makefile.in (OBJS): Add run-rtl-passes.o.
34123 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34125 (gcc::pass_manager::get_clean_slate): New accessor.
34126 * passes.c: Include "insn-addr.h".
34127 (should_skip_pass_p): Add logging. Update logic for running
34128 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34129 property-provider override so it is only done for gimple passes.
34131 (skip_pass): New function.
34132 (execute_one_pass): Call skip_pass when skipping passes.
34133 * read-md.c (md_reader::read_char): Support filtering
34134 the input to a subset of line numbers.
34135 (md_reader::md_reader): Initialize fields
34136 m_first_line and m_last_line.
34137 (md_reader::read_file_fragment): New function.
34138 * read-md.h (md_reader::read_file_fragment): New decl.
34139 (md_reader::m_first_line): New field.
34140 (md_reader::m_last_line): New field.
34141 * read-rtl-function.c (function_reader::create_function): Only
34142 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34143 curr_properties. Set DECL_INITIAL to a dummy block.
34144 (read_rtl_function_body_from_file_range): New function.
34145 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34147 * run-rtl-passes.c: New file.
34148 * run-rtl-passes.h: New file.
34150 2017-01-24 Jeff Law <law@redhat.com>
34152 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34155 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34157 PR tree-optimization/79159
34158 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34159 (record_nonwrapping_iv): Improve boundary using above function if no
34160 value range information.
34162 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34163 Martin Jambor <mjambor@suse.cz>
34165 * brig-builtins.def: New file.
34166 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34167 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34168 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34169 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34170 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34171 * builtin-types.def (BT_INT8): New.
34172 (BT_INT16): Likewise.
34173 (BT_UINT8): Likewise.
34174 (BT_UINT16): Likewise.
34175 (BT_FN_ULONG): Likewise.
34176 (BT_FN_UINT_INT): Likewise.
34177 (BT_FN_UINT_ULONG): Likewise.
34178 (BT_FN_UINT_LONG): Likewise.
34179 (BT_FN_UINT_PTR): Likewise.
34180 (BT_FN_ULONG_PTR): Likewise.
34181 (BT_FN_INT8_FLOAT): Likewise.
34182 (BT_FN_INT16_FLOAT): Likewise.
34183 (BT_FN_UINT32_FLOAT): Likewise.
34184 (BT_FN_UINT16_FLOAT): Likewise.
34185 (BT_FN_UINT8_FLOAT): Likewise.
34186 (BT_FN_UINT64_FLOAT): Likewise.
34187 (BT_FN_UINT16_UINT32): Likewise.
34188 (BT_FN_UINT32_UINT16): Likewise.
34189 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34190 (BT_FN_INT_PTR_INT): Likewise.
34191 (BT_FN_UINT_PTR_UINT): Likewise.
34192 (BT_FN_LONG_PTR_LONG): Likewise.
34193 (BT_FN_ULONG_PTR_ULONG): Likewise.
34194 (BT_FN_VOID_UINT64_UINT64): Likewise.
34195 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34196 (BT_FN_INT8_INT8_INT8): Likewise.
34197 (BT_FN_INT16_INT16_INT16): Likewise.
34198 (BT_FN_INT_INT_INT): Likewise.
34199 (BT_FN_UINT_FLOAT_UINT): Likewise.
34200 (BT_FN_FLOAT_UINT_UINT): Likewise.
34201 (BT_FN_ULONG_UINT_UINT): Likewise.
34202 (BT_FN_ULONG_UINT_PTR): Likewise.
34203 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34204 (BT_FN_UINT_UINT_UINT): Likewise.
34205 (BT_FN_VOID_UINT_PTR): Likewise.
34206 (BT_FN_UINT_UINT_PTR: Likewise.
34207 (BT_FN_UINT32_UINT64_PTR): Likewise.
34208 (BT_FN_INT_INT_UINT_UINT): Likewise.
34209 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34210 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34211 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34212 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34213 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34214 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34215 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34216 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34217 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34218 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34219 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34220 * doc/frontends.texi: List BRIG FE.
34221 * doc/install.texi (Testing): Add BRIG tesring requirements.
34222 * doc/invoke.texi (Overall Options): Mention BRIG.
34223 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34225 2017-01-24 Richard Biener <rguenther@suse.de>
34227 PR translation/79208
34228 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34230 2017-01-24 Martin Jambor <mjambor@suse.cz>
34233 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34234 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34235 and known_contexts.
34237 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34239 PR middle-end/79123
34240 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34241 casts from signed to unsigned really don't have a range.
34243 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34245 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34246 GMP_RNDx for compatiblity.
34248 2017-01-24 Martin Liska <mliska@suse.cz>
34251 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34252 that would prevent us to call alloca with -1 as argument.
34254 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34256 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34257 Avoid -Wformat-security warning.
34259 2017-01-23 Andrew Pinski <apinski@cavium.com>
34261 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34264 2017-01-23 Martin Sebor <msebor@redhat.com>
34266 PR middle-end/78703
34267 * gimple-ssa-sprintf.c (warn_level): New global.
34268 (format_integer): Use it here and throughout the rest of the file.
34269 Use the same switch to compute sign as base.
34270 (maybe_warn): New function.
34271 (format_directive): Factor out warnings into maybe_warn.
34272 Add debugging output. Use warn_level.
34273 (add_bytes): Use warn_level.
34274 (pass_sprintf_length::compute_format_length): Add debugging output.
34275 (try_substitute_return_value): Same.
34276 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34278 PR middle-end/78703
34279 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34280 (struct fmtresult, format_integer, format_floating): Adjust.
34281 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34282 (get_string_length, format_string,format_directive): Same.
34283 (pass_sprintf_length::compute_format_length): Same.
34284 (try_substitute_return_value): Simplify slightly.
34286 PR middle-end/78703
34287 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34288 (fmtresult::operator+=): Outlined.
34289 (struct fmtresult): Add ctors.
34290 (struct conversion_spec): Rename...
34291 (struct directive): ...to this. Add and remove data members.
34292 (directive::set_width, directive::set_precision): New functions.
34293 (format_percent): Use fmtresult ctor.
34294 (get_width_and_precision): Remove.
34295 (format_integer): Make naming changes. Avoid computing width and
34297 (format_floating): Same. Adjust indentation.
34298 (format_character, format_none): New functions.
34299 (format_string): Moved character handling to format_character.
34300 (format_directive): Remove arguments, change return type.
34301 (parse_directive): New function.
34302 (pass_sprintf_length::compute_format_length): Move directive
34303 parsing to parse_directive.
34305 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34307 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34308 (assign_assembler_name_if_needed): ... this.
34309 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34310 (assign_assembler_name_if_needed): ... this.
34311 (free_lang_data_in_cgraph): Adjust callers.
34312 * cgraphunit.c (cgraph_node::analyze): Likewise.
34313 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34316 2017-01-23 Richard Biener <rguenther@suse.de>
34318 PR tree-optimization/79088
34319 PR tree-optimization/79188
34320 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34321 resetting loop bounds after last path deletion. Reset loop
34322 bounds of the target loop, make code match the comments.
34323 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34324 Make sure loops need no fixups.
34326 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34328 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34329 exponent support with double type for first argument.
34330 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34331 type returned by __builtin_vec_extract_sig,
34332 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34333 functions from "vector int" to "vector unsigned int" or from
34334 "vector long long int" to "vector unsigned long long int".
34335 Changed type returned by __builtin_vec_extract_exp,
34336 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34337 functions from "vector int" to "vector unsigned int" or from
34338 "vector long long int" to "vector unsigned long long int".
34339 Changed return type of __builtin_vec_test_data_class,
34340 __builtin_vec_test_data_class_sp, and
34341 __builtin_vec_test_data_class_dp from "vector int" to
34342 "vector bool int" or from "vector long long int" to "vector bool
34343 long long int" and changed second argument type from "unsigned
34344 int" to "int". Added new overloaded function forms "vector float
34345 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34346 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34347 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34348 double, vector unsigned long long int)" and "vector double
34349 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34350 long int)". Changed return type of
34351 __builtin_scalar_test_data_class and
34352 __builtin_scalar_test_data_class_sp and
34353 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34354 int" and changed second argument from "unsigned int" to "int".
34355 Changed type returned by __builtin_scalar_test_neg,
34356 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34357 from "int" to "bool int". Added new overloaded function form
34358 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34359 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34360 exponent double-precision with floating point first argument.
34361 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34362 documentation of scalar_test_data_class, scalar_test_neg,
34363 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34364 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34365 vec_test_data_class built-in functions to reflect refinements in
34366 their type signatures.
34368 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34370 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34372 (aarch64_elf_asm_destructor): Likewise.
34374 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34376 PR rtl-optimization/78634
34377 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34378 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34379 * ifcvt.c (noce_try_cmove): Add missing cost check.
34381 PR rtl-optimization/71724
34382 * combine.c (if_then_else_cond): Look for situations where it is
34383 beneficial to undo the work of one of the recursive calls.
34385 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34387 PR tree-optimization/70754
34388 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34389 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34390 combined stmt before it if not NULL.
34391 (combine_chains): Process refs reversely and compute dominance point
34394 2017-01-23 Martin Liska <mliska@suse.cz>
34396 PR tree-optimization/79196
34397 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34398 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34400 (strlen_optimize_stmt): Call the renamed function.
34402 2017-01-23 Michael Matz <matz@suse.de>
34404 PR tree-optimization/78384
34405 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34407 2017-01-23 Richard Biener <rguenther@suse.de>
34409 PR tree-optimization/79186
34410 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34411 both incoming edges before moving an assert.
34413 2017-01-23 Martin Jambor <mjambor@suse.cz>
34415 * ipa-prop.c (load_from_param_1): Removed.
34416 (load_from_unmodified_param): Bits from load_from_param_1 put back
34418 (load_from_param): Removed.
34419 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34420 with stmt. Reverted back to use of load_from_unmodified_param.
34422 2017-01-23 Martin Jambor <mjambor@suse.cz>
34425 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34426 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34427 field a pointer to garbage collected vector, mark lattices and
34428 ipcp_orig_node with GTY((skip)).
34429 (ipa_get_param_count): Adjust to descriptors being a pointer.
34430 (ipa_get_param): Likewise.
34431 (ipa_get_type): Likewise.
34432 (ipa_get_param_move_cost): Likewise.
34433 (ipa_set_param_used): Likewise.
34434 (ipa_get_controlled_uses): Likewise.
34435 (ipa_set_controlled_uses): Likewise.
34436 (ipa_is_param_used): Likewise.
34437 (ipa_node_params_t): Move into garbage collector. New methods insert
34439 (ipa_node_params_sum): Annotate wth GTY(()).
34440 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34442 (ipa_load_from_parm_agg): Adjust declaration.
34443 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34444 * ipa-profile.c (ipa_profile): Likewise.
34445 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34446 (ipa_populate_param_decls): Make descriptors parameter garbage
34448 (ipa_dump_param): Adjust to descriptors being a pointer.
34449 (ipa_alloc_node_params): Likewise.
34450 (ipa_initialize_node_params): Likewise.
34451 (load_from_param_1): Make descriptors parameter garbage collected.
34452 (load_from_unmodified_param): Likewise.
34453 (load_from_param): Likewise.
34454 (ipa_load_from_parm_agg): Likewise.
34455 (ipa_node_params::~ipa_node_params): Removed.
34456 (ipa_free_all_node_params): Remove call to delete operator.
34457 (ipa_node_params_t::insert): New.
34458 (ipa_node_params_t::remove): Likewise.
34459 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34460 copy known_csts and known_contexts vectors.
34461 (ipa_read_node_info): Adjust to descriptors being a pointer.
34462 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34464 (ipcp_transform_function): Make descriptors variable garbage
34467 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34469 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34470 * config/i386/avx512dqintrin.h: Ditto.
34471 * config/i386/avx512fintrin.h: Ditto.
34472 * config/i386/i386.c: Handle new builtins.
34473 * config/i386/i386-builtin.def: Add new builtins.
34474 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34475 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34477 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34478 Martin Liska <mliska@suse.cz>
34480 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34481 * asan.c (asan_expand_poison_ifn): Support stores and use
34482 appropriate ASAN report function.
34483 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34484 * internal-fn.def (ASAN_POISON_USE): Declare.
34485 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34486 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34487 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34488 ASAN_POISON calls w/o LHS.
34489 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34490 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34491 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34492 * gimplify.c (asan_poison_variables): Add attribute
34493 use_after_scope_memory to variables that really needs to live
34495 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34496 having the attribute.
34498 2017-01-23 Martin Liska <mliska@suse.cz>
34500 * asan.c (create_asan_shadow_var): New function.
34501 (asan_expand_poison_ifn): Likewise.
34502 * asan.h (asan_expand_poison_ifn): New declaration.
34503 * internal-fn.c (expand_ASAN_POISON): Likewise.
34504 * internal-fn.def (ASAN_POISON): New builtin.
34505 * sanopt.c (pass_sanopt::execute): Expand
34506 asan_expand_poison_ifn.
34507 * tree-inline.c (copy_decl_for_dup_finish): Make function
34509 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34510 * tree-ssa.c (is_asan_mark_p): New function.
34511 (execute_update_addresses_taken): Rewrite local variables
34512 (identified just by use-after-scope as addressable) into SSA.
34514 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34516 * doc/install.texi (Specific): opensource.apple.com uses https
34517 now. Remove trailing slash.
34519 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34521 * README.Portability: Remove note on an Irix compatibility issue.
34523 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34525 * gcov.c (INCLUDE_ALGORITHM): Define.
34526 (INCLUDE_VECTOR): Define.
34527 No longer include <vector> and <algorithm> directly.
34529 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34531 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34533 * doc/invoke.texi (Code Gen Options): Ditto.
34535 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34538 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34540 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34542 rtl-optimization/79125
34543 * cprop.c (local_cprop_pass): Handle cases where we make an
34544 unconditional trap.
34546 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34550 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34551 read from, for big endian.
34553 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34555 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34556 register pauth builtins for LP64 only.
34558 2017-01-20 Marek Polacek <polacek@redhat.com>
34561 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34564 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34566 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34568 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34569 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34570 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34572 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34575 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34576 in big-endian mode when they are not a single duplicated value.
34578 2017-01-20 Richard Biener <rguenther@suse.de>
34580 * BASE-VER: Bump to 7.0.1.
34582 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34584 * omp-low.c (omplow_simd_context): New struct. Use it...
34585 (lower_rec_simd_input_clauses): ...here and...
34586 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34587 references to idx, lane, max_vf, is_simt.
34589 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34591 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34594 2017-01-20 Martin Jambor <mjambor@suse.cz>
34596 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34597 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34599 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34600 (GTFILES): Rename hsa.c to hsa-common.c.
34601 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34602 * hsa-dump.c: Likewise.
34603 * hsa-gen.c: Likewise.
34604 * hsa-regalloc.c: Likewise.
34605 * ipa-hsa.c: Likewise.
34606 * omp-expand.c: Likewise.
34607 * omp-low.c: Likewise.
34608 * toplev.c: Likewise.
34610 2017-01-20 Marek Polacek <polacek@redhat.com>
34613 * doc/invoke.texi: Document -Wduplicated-branches.
34614 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34615 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34616 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34617 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34618 return 0 only when not OEP_LEXICOGRAPHIC.
34619 (fold_build_cleanup_point_expr): Use the expression
34620 location when building CLEANUP_POINT_EXPR.
34621 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34622 * tree.c (add_expr): Handle error_mark_node.
34624 2017-01-20 Martin Liska <mliska@suse.cz>
34627 * tree-profile.c (init_ic_make_global_vars): Do not call
34629 (gimple_init_gcov_profiler): Likewise.
34631 2017-01-20 Martin Liska <mliska@suse.cz>
34634 * cgraph.h (maybe_create_reference): Remove argument and
34636 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34638 * ipa-cp.c (create_specialized_node): Likewise.
34639 * symtab.c (symtab_node::maybe_create_reference): Handle
34640 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34642 2017-01-20 Martin Liska <mliska@suse.cz>
34644 * read-rtl-function.c (function_reader::create_function): Use
34645 build_decl instread of build_decl_stat.
34647 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34649 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34650 * config/i386/avx512dqintrin.h: Ditto.
34651 * config/i386/avx512fintrin.h: Ditto.
34652 * config/i386/i386-builtin-types.def: Add new types.
34653 * config/i386/i386.c: Handle new types.
34654 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34655 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34656 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34657 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34658 (__builtin_ia32_kshiftridi): New.
34659 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34661 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34665 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34666 define to rs6000_init_stack_protect_guard.
34667 (rs6000_init_stack_protect_guard): New function.
34669 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34670 Yunqiang Su <yunqiang.su@imgtec.com>
34672 * config.gcc (supported_defaults): Add madd4.
34673 (with_madd4): Add validation.
34674 (all_defaults): Add madd4.
34675 * config/mips/mips.opt (mmadd4): New option.
34676 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34678 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34680 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34681 (ISA_HAS_FUSED_MADD4): Likewise.
34682 * doc/invoke.texi (-mmadd4): Document the new option.
34683 * doc/install.texi (--with-madd4): Document the new option.
34685 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34687 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34688 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34689 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34690 (aarch64_init_pauth_hint_builtins): New.
34691 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34692 (aarch64_expand_builtin): Expand new builtins.
34694 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34696 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34697 * combine-stack-adj.c (no_unhandled_cfa): Handle
34698 REG_CFA_TOGGLE_RA_MANGLE.
34699 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34700 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34701 info for return address signing.
34702 (aarch64_expand_epilogue): Likewise.
34704 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34706 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34707 * config/aarch64/aarch64-protos.h
34708 (aarch64_return_address_signing_enabled): New declaration.
34709 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34711 (aarch64_expand_prologue): Sign return address before it's pushed onto
34713 (aarch64_expand_epilogue): Authenticate return address fetched from
34715 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34716 (aarch64_attributes): New function attributes for "sign-return-address".
34717 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34718 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34719 ("*do_return"): Generate combined instructions according to key index.
34720 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34721 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34723 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34724 * config/aarch64/aarch64.opt (msign-return-address=): New.
34725 * doc/extend.texi (AArch64 Function Attributes): Documents
34726 "sign-return-address=".
34727 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34729 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34731 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34732 overall option summary.
34734 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34736 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34737 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34738 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34739 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34741 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34743 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34744 -mpower9-minmax by default for -mcpu=power9.
34745 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34746 128-bit floating point.
34748 2017-01-20 Alan Modra <amodra@gmail.com>
34750 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34751 optimizing for size.
34753 2017-01-20 Alan Modra <amodra@gmail.com>
34756 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34757 for strcmp and strncmp from corresponding builtin decl.
34759 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34761 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34762 instead of i386/rtems-64.h.
34763 * config/i386/rtems-64.h: Remove.
34765 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34769 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34771 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34773 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34775 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34776 Change int to HOST_WIDE_INT.
34777 * config/aarch64/aarch64-protos.h
34778 (aarch64_simd_gen_const_vector_dup): Likewise.
34779 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34781 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34783 * langhooks-def.h (lhd_type_for_size): New decl.
34784 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34785 * langhooks.c (lhd_type_for_size): New function, taken from
34788 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34790 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34791 define_bypass for CR latency.
34792 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34793 (power9-alu2): Add define_bypass for CR latency.
34795 (power9-mul): Update insn latency.
34796 (power9-mul-compare): Update insn latency, bypass latency and remove
34799 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34801 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
34803 * config/aarch64/aarch64.md
34804 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
34805 aarch64_nopcrelative_literal_loads.
34806 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
34808 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
34810 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
34811 TARGET_LOONGSON_3A.
34812 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
34814 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
34817 * config.gcc (supported_defaults): Add lxc1-sxc1.
34818 (with_lxc1_sxc1): Add validation.
34819 (all_defaults): Add lxc1-sxc1.
34820 * config/mips/mips.opt (mlxc1-sxc1): New option.
34821 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34823 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34824 __mips_no_lxc1_sxc1.
34825 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
34826 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
34827 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
34829 2017-01-19 Richard Biener <rguenther@suse.de>
34831 PR tree-optimization/72488
34832 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34833 sure to restore SSA info.
34834 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34836 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34838 PR rtl-optimization/79121
34839 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34840 of the inner type when shifting an extended value.
34842 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34845 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34846 interposable aliases.
34848 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34851 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34852 Use the evmergelohi instruction.
34853 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34854 (mov_sitf_e500_subreg8_2_be): Likewise.
34855 (mov_sitf_e500_subreg12_2_le): Likewise.
34856 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34857 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34858 (mov_sitf_e500_subreg8_2_le): Likewise.
34859 (mov_sitf_e500_subreg12_2_be): Likewise.
34861 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34863 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34864 attribute from vecsimple to vecperm.
34865 (altivec_vbpermq2): Likewise.
34867 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34870 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34872 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34873 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34874 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34875 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34876 case where N arg is SIZE_MAX.
34877 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34878 (cmpstrsi): Add pattern.
34880 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34882 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34883 __builtin_vec_revb builtins.
34884 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34885 built-in functions to support generation of the ISA 3.0 XXBR<x>
34886 vector byte reverse instructions.
34887 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34888 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34889 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34890 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34891 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34892 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34893 (P9V_BUILTIN_VEC_REVB): Likewise.
34894 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34895 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34896 (p9_xxbrq_v16qi): Likewise.
34897 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34898 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34899 (p9_xxbrh_v8hi): Likewise.
34900 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34901 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34902 vec_revb built-in functions.
34904 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34906 PR rtl-optimization/78952
34907 * config/i386/i386.md (any_extract): New code iterator.
34908 (*insvqi_2): Use any_extract for source operand.
34909 (*insvqi_3): Use any_shiftrt for source operand.
34911 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34913 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34915 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34917 2017-01-18 Matthias Klose <doko@ubuntu.com>
34919 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34921 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34923 * config/rs6000/altivec.h (vec_bperm): Change #define.
34924 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34925 (altivec_vbpermq2): New define_insn.
34926 (altivec_vbpermd): Likewise.
34927 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34928 function interface.
34929 (VBPERMD): Likewise.
34930 (VBPERM): New polymorphic function interface.
34931 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34932 Add entries for P9V_BUILTIN_VEC_VBPERM.
34933 * doc/extend.texi: Add interfaces for vec_bperm.
34935 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34937 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34938 first letter of error messages.
34939 (s390_resolve_overloaded_builtin): Likewise.
34940 * config/s390/s390.c (s390_expand_builtin): Likewise.
34941 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34942 (s390_valid_target_attribute_inner_p): Likewise.
34943 * config/s390/s390.md ("tabort"): Likewise.
34945 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34947 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34948 (ISA_AVOID_DIV_HILO): New macro.
34949 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34950 (ISA_HAS_DDIV): Likewise.
34952 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34954 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34956 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34958 * doc/invoke.texi (fabi-version): Spelling fix.
34960 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34963 * doc/invoke.texi (fabi-version): Mention mangling fix for
34966 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34969 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34971 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34974 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34977 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34979 (stack_protect_test): Ditto.
34980 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34981 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34983 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34984 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34985 -mstack-protector-guard-offset=.
34986 (RS/6000 and PowerPC Options): Ditto.
34988 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34990 * config/i386/i386.h (MASK_CLASS_P): New define.
34991 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34992 there are no registers from different register sets also when
34993 mask registers are used. Update function comment.
34994 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34995 to (*k/*r) and (*k/*km) alternatives.
34997 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34999 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35000 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35001 (EH_RETURN_HANDLER_RTX): New define.
35002 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35003 Force frame pointer in EH return functions.
35004 (aarch64_expand_epilogue): Add barrier for eh_return.
35005 (aarch64_final_eh_return_addr): Remove.
35006 (aarch64_eh_return_handler_rtx): New function.
35007 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35009 (aarch64_eh_return_handler_rtx): New prototype.
35011 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35013 * config/rs6000/altivec.h (vec_rlmi): New #define.
35014 (vec_vrlnm): Likewise.
35015 (vec_rlnm): Likewise.
35016 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35017 (UNSPEC_VRLNM): Likewise.
35018 (VIlong): New mode iterator.
35019 (altivec_vrl<VI_char>mi): New define_insn.
35020 (altivec_vrl<VI_char>nm): Likewise.
35021 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35023 (VRLDNM): Likewise.
35024 (RLNM): New polymorphic function entry.
35025 (VRLWMI): New monomorphic function entry.
35026 (VRLDMI): Likewise.
35027 (RLMI): New polymorphic function entry.
35028 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35029 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35030 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35033 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35036 * dwarf2out.c (field_byte_offset): Restore the
35037 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35038 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35039 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35042 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35045 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35047 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35050 * dwarf2out.c (add_data_member_location_attribute): For constant
35051 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35052 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35053 DW_AT_byte_size attributes.
35055 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35057 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35058 after forcing to constant memory when the code model is medium.
35060 2017-01-17 Julia Koval <julia.koval@intel.com>
35063 * config/i386/avx512fintrin.h
35064 (_mm512_i32gather_ps): Change __addr type to void const*.
35065 (_mm512_mask_i32gather_ps): Ditto.
35066 (_mm512_i32gather_pd): Ditto.
35067 (_mm512_mask_i32gather_pd): Ditto.
35068 (_mm512_i64gather_ps): Ditto.
35069 (_mm512_mask_i64gather_ps): Ditto.
35070 (_mm512_i64gather_pd): Ditto.
35071 (_mm512_mask_i64gather_pd): Ditto.
35072 (_mm512_i32gather_epi32): Ditto.
35073 (_mm512_mask_i32gather_epi32): Ditto.
35074 (_mm512_i32gather_epi64): Ditto.
35075 (_mm512_mask_i32gather_epi64): Ditto.
35076 (_mm512_i64gather_epi32): Ditto.
35077 (_mm512_mask_i64gather_epi32): Ditto.
35078 (_mm512_i64gather_epi64): Ditto.
35079 (_mm512_mask_i64gather_epi64): Ditto.
35080 (_mm512_i32scatter_ps): Change __addr type to void*.
35081 (_mm512_mask_i32scatter_ps): Ditto.
35082 (_mm512_i32scatter_pd): Ditto.
35083 (_mm512_mask_i32scatter_pd): Ditto.
35084 (_mm512_i64scatter_ps): Ditto.
35085 (_mm512_mask_i64scatter_ps): Ditto.
35086 (_mm512_i64scatter_pd): Ditto.
35087 (_mm512_mask_i64scatter_pd): Ditto.
35088 (_mm512_i32scatter_epi32): Ditto.
35089 (_mm512_mask_i32scatter_epi32): Ditto.
35090 (_mm512_i32scatter_epi64): Ditto.
35091 (_mm512_mask_i32scatter_epi64): Ditto.
35092 (_mm512_i64scatter_epi32): Ditto.
35093 (_mm512_mask_i64scatter_epi32): Ditto.
35094 (_mm512_i64scatter_epi64): Ditto.
35095 (_mm512_mask_i64scatter_epi64): Ditto.
35096 * config/i386/avx512pfintrin.h
35097 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35098 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35099 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35100 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35101 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35102 (_mm512_prefetch_i32scatter_ps): Ditto.
35103 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35104 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35105 (_mm512_prefetch_i64scatter_pd): Ditto.
35106 (_mm512_prefetch_i64scatter_ps): Ditto.
35107 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35108 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35109 * config/i386/avx512vlintrin.h
35110 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35111 (_mm_mmask_i32gather_ps): Ditto.
35112 (_mm256_mmask_i32gather_pd): Ditto.
35113 (_mm_mmask_i32gather_pd): Ditto.
35114 (_mm256_mmask_i64gather_ps): Ditto.
35115 (_mm_mmask_i64gather_ps): Ditto.
35116 (_mm256_mmask_i64gather_pd): Ditto.
35117 (_mm_mmask_i64gather_pd): Ditto.
35118 (_mm256_mmask_i32gather_epi32): Ditto.
35119 (_mm_mmask_i32gather_epi32): Ditto.
35120 (_mm256_mmask_i32gather_epi64): Ditto.
35121 (_mm_mmask_i32gather_epi64): Ditto.
35122 (_mm256_mmask_i64gather_epi32): Ditto.
35123 (_mm_mmask_i64gather_epi32): Ditto.
35124 (_mm256_mmask_i64gather_epi64): Ditto.
35125 (_mm_mmask_i64gather_epi64): Ditto.
35126 (_mm256_i32scatter_ps): Change __addr type to void*.
35127 (_mm256_mask_i32scatter_ps): Ditto.
35128 (_mm_i32scatter_ps): Ditto.
35129 (_mm_mask_i32scatter_ps): Ditto.
35130 (_mm256_i32scatter_pd): Ditto.
35131 (_mm256_mask_i32scatter_pd): Ditto.
35132 (_mm_i32scatter_pd): Ditto.
35133 (_mm_mask_i32scatter_pd): Ditto.
35134 (_mm256_i64scatter_ps): Ditto.
35135 (_mm256_mask_i64scatter_ps): Ditto.
35136 (_mm_i64scatter_ps): Ditto.
35137 (_mm_mask_i64scatter_ps): Ditto.
35138 (_mm256_i64scatter_pd): Ditto.
35139 (_mm256_mask_i64scatter_pd): Ditto.
35140 (_mm_i64scatter_pd): Ditto.
35141 (_mm_mask_i64scatter_pd): Ditto.
35142 (_mm256_i32scatter_epi32): Ditto.
35143 (_mm256_mask_i32scatter_epi32): Ditto.
35144 (_mm_i32scatter_epi32): Ditto.
35145 (_mm_mask_i32scatter_epi32): Ditto.
35146 (_mm256_i32scatter_epi64): Ditto.
35147 (_mm256_mask_i32scatter_epi64): Ditto.
35148 (_mm_i32scatter_epi64): Ditto.
35149 (_mm_mask_i32scatter_epi64): Ditto.
35150 (_mm256_i64scatter_epi32): Ditto.
35151 (_mm256_mask_i64scatter_epi32): Ditto.
35152 (_mm_i64scatter_epi32): Ditto.
35153 (_mm_mask_i64scatter_epi32): Ditto.
35154 (_mm256_i64scatter_epi64): Ditto.
35155 (_mm256_mask_i64scatter_epi64): Ditto.
35156 (_mm_i64scatter_epi64): Ditto.
35157 (_mm_mask_i64scatter_epi64): Ditto.
35158 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35159 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35160 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35161 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35162 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35163 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35164 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35165 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35166 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35167 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35168 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35169 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35170 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35171 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35172 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35173 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35174 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35175 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35176 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35177 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35178 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35179 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35180 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35181 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35182 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35183 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35184 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35185 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35186 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35187 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35188 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35189 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35190 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35191 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35192 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35193 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35194 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35195 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35196 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35197 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35198 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35199 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35200 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35201 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35202 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35203 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35204 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35205 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35206 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35207 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35208 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35209 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35210 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35211 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35212 definitions accordingly.
35214 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35215 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35218 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35221 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35224 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35225 hard regs for inner regno.
35227 2017-01-17 Martin Liska <mliska@suse.cz>
35230 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35231 assumption and add comment.
35233 2017-01-17 Nathan Sidwell <nathan@acm.org>
35235 * ipa-visibility.c (localize_node): New function, broken out of ...
35236 (function_and_variable_visibility): ... here. Call it.
35238 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35240 PR middle-end/77445
35241 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35242 correctly set frequency of oudgoing edge.
35243 (duplicate_thread_path): Fix profile updating.
35245 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35248 * configure.ac: Add GCC_BASE_VER.
35249 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35250 version from BASE-VER file.
35251 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35252 (gcc.o): Depend on $(BASEVER).
35253 * common.opt (dumpfullversion): New option.
35254 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35255 * doc/invoke.texi: Document -dumpfullversion.
35256 * doc/install.texi: Document --with-gcc-major-version-only.
35257 * configure: Regenerated.
35259 2017-01-17 Richard Biener <rguenther@suse.de>
35261 PR tree-optimization/71433
35262 * tree-vrp.c (register_new_assert_for): Merge same asserts
35263 on all incoming edges.
35264 (process_assert_insertions_for): Handle insertions at the
35267 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35269 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35270 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35272 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35275 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35278 2017-01-17 Alan Modra <amodra@gmail.com>
35281 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35282 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35283 symbolic stack limit when pic.
35285 2017-01-16 Martin Sebor <msebor@redhat.com>
35287 PR tree-optimization/78608
35288 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35290 2017-01-16 Jeff Law <law@redhat.com>
35293 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35294 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35295 for several include directories that may be relative to sysroot.
35296 * config/i386/x-mingw32 (gplus_includedir): Define.
35297 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35298 (native_system_includedir): Likewise.
35299 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35300 override if TARGET_SYSTEM_ROOT is defined.
35301 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35303 PR tree-optimization/79090
35304 PR tree-optimization/33562
35305 PR tree-optimization/61912
35306 PR tree-optimization/77485
35307 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35308 and computed trims into the dump file.
35310 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35312 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35314 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35317 * gimplify.c (gimplify_init_constructor): If want_value and
35318 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35322 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35323 sequence. Formatting fixes.
35324 (doloop_optimize): Formatting fixes.
35327 * gcc.c (debug_level_greater_than_spec_func): New function.
35328 (static_spec_functions): Add debug-level-gt spec function.
35329 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35331 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35332 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35333 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35334 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35335 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35336 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35338 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35340 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35341 QImode fixups to general and mask registers only.
35343 2017-01-16 Carl Love <cel@us.ibm.com>
35345 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35346 for built-in functions
35347 vector signed char vec_nabs (vector signed char)
35348 vector signed short vec_nabs (vector signed short)
35349 vector signed int vec_nabs (vector signed int)
35350 vector signed long long vec_nabs (vector signed long long)
35351 vector float vec_nabs (vector float)
35352 vector double vec_nabs (vector double)
35353 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35355 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35356 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35357 * doc/extend.texi: Update the documentation file for the new built-in
35360 2017-01-16 Martin Sebor <msebor@redhat.com>
35362 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35365 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35367 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35368 UNSPEC_VSX__XXSPLTD to require special splat handling.
35370 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35373 * system.h: Poison strndup.
35375 2017-01-16 Alan Modra <amodra@gmail.com>
35378 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35381 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35383 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35385 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35387 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35388 call recog here. Assert that INSN_CODE (insn) is non-negative.
35390 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35393 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35395 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35396 in the currently scheduled RTL fragment.
35398 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35400 PR rtl-optimization/78751
35401 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35404 2017-01-14 Jeff Law <law@redhat.com>
35406 PR tree-optimization/79090
35407 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35408 variable length stores.
35409 (compute_trims): Delete dead assignment to *trim_tail.
35410 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35413 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35415 PR rtl-optimization/78626
35416 PR rtl-optimization/78727
35417 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35418 of a block, and split such blocks after everything else is finished.
35420 2017-01-14 Alan Modra <amodra@gmail.com>
35423 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35424 target legitimate_combined_insn.
35425 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35426 (rs6000_legitimate_combined_insn): New function.
35427 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35429 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35430 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35431 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35433 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35435 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35437 2017-01-13 Jeff Law <law@redhat.com>
35439 PR tree-optimization/33562
35440 PR tree-optimization/61912
35441 PR tree-optimization/77485
35442 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35444 (delete_dead_assignment): Likewise.
35445 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35446 statement to delete_dead_call and delete_dead_assignment.
35448 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35451 * substring-locations.c (format_warning_va): Strengthen case 1 so
35452 that both endpoints of the substring must be within the format
35453 range for just the substring to be printed.
35455 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35457 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35458 * config/i386/i386.c (ix86_target_string): Add missing options
35459 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35460 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35461 flags_other and ix86_target_other to flags2_other. Display unknown
35463 (ix86_valid_target_attribute_inner_p): Add missing options and
35464 reorder options by implied ISAs, as in ix86_target_string.
35466 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35468 * hash-table.h (hash_table::too_empty_p): New function.
35469 (hash_table::expand): Use it.
35470 (hash_table::traverse): Likewise.
35471 (hash_table::empty_slot): Use sizeof (value_type) instead of
35472 sizeof (PTR) to convert bytes to elements. Shrink the table
35473 if the current size is excessive for the current number of
35476 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35478 * ira-costs.c (record_reg_classes): Break from the inner loop
35479 early once alt_fail is known to be true. Update outer loop
35480 handling accordingly.
35482 2017-01-13 Jeff Law <law@redhat.com>
35484 * tree-ssa-dse.c (decrement_count): New function.
35485 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35486 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35487 when we know the partially dead statement is a mem* function.
35489 PR tree-optimization/61912
35490 PR tree-optimization/77485
35491 * tree-ssa-dse.c: Include expr.h.
35492 (maybe_trim_constructor_store): New function.
35493 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35495 PR tree-optimization/33562
35496 PR tree-optimization/61912
35497 PR tree-optimization/77485
35498 * doc/invoke.texi: Document new dse-max-object-size param.
35499 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35500 * tree-ssa-dse.c: Include params.h.
35501 (dse_store_status): New enum.
35502 (initialize_ao_ref_for_dse): New, partially extracted from
35504 (valid_ao_ref_for_dse, normalize_ref): New.
35505 (setup_live_bytes_from_ref, compute_trims): Likewise.
35506 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35507 (maybe_trim_partially_dead_store): Likewise.
35508 (maybe_trim_complex_store): Likewise.
35509 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35510 Track what bytes live from the original store. Return tri-state
35511 for dead, partially dead or live.
35512 (dse_dom_walker): Add constructor, destructor and new private members.
35513 (delete_dead_call, delete_dead_assignment): New extracted from
35515 (dse_optimize_stmt): Make a member of dse_dom_walker.
35516 Use initialize_ao_ref_for_dse.
35518 PR tree-optimization/33562
35519 PR tree-optimization/61912
35520 PR tree-optimization/77485
35521 * sbitmap.h (bitmap_count_bits): Prototype.
35522 (bitmap_clear_range, bitmap_set_range): Likewise.
35523 * sbitmap.c (bitmap_clear_range): New function.
35524 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35526 2017-01-13 Martin Liska <mliska@suse.cz>
35529 * function.c (set_cfun): Add new argument force.
35530 * function.h (set_cfun): Likewise.
35531 * ipa-inline-transform.c (inline_call): Use the function when
35532 strict alising from is dropped for function we inline to.
35534 2017-01-13 Richard Biener <rguenther@suse.de>
35536 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35537 for dumping GIMPLE INTEGER_CSTs.
35539 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35541 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35542 to 201112L since C++17.
35544 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35547 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35548 if -fsanitize=kernel-address is present.
35550 2017-01-13 Richard Biener <rguenther@suse.de>
35552 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35553 as _Literal ( type ) number in case usual suffixes do not
35554 preserve all information.
35556 2017-01-13 Richard Biener <rguenther@suse.de>
35558 PR tree-optimization/77283
35559 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35560 and ssa-iterators.h.
35561 (is_feasible_trace): Implement a cost model based on joiner
35564 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35567 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35568 char or short to __float128/_Float128 directly.
35570 2017-01-12 Martin Sebor <msebor@redhat.com>
35572 to -Wformat-overflow.
35573 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35574 (min_bytes_remaining): Same.
35575 (get_string_length): Same.
35576 (format_string): Same.
35577 (format_directive): Same.
35579 (pass_sprintf_length::handle_gimple_call): Same.
35581 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35583 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35584 info.nowrite calls with no lhs that can't throw. Return bool
35585 whether gsi_remove has been called or not.
35586 (pass_sprintf_length::handle_gimple_call): Return bool whether
35587 try_substitute_return_value called gsi_remove. Formatting fix.
35588 (pass_sprintf_length::execute): Don't use gsi_remove if
35589 handle_gimple_call returned true.
35592 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35593 be removed due to side-effects, don't remove following barrier nor
35594 turn the successor edge into fallthru edge.
35596 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35599 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35600 element-reversing loads and stores as not swappable.
35602 2017-01-12 Nathan Sidwell <nathan@acm.org>
35603 Nicolai Stange <nicstange@gmail.com>
35605 * combine.c (try_combine): Don't ignore result of overlap checking
35606 loop. Combine overlap & asm check into single loop.
35608 2017-01-12 Richard Biener <rguenther@suse.de>
35610 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35611 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35613 2017-01-12 Richard Biener <rguenther@suse.de>
35615 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35616 and TS_TARGET_OPTION directly derive from TS_BASE.
35617 * tree-core.h (tree_optimization_option): Derive from tree_base.
35618 (tree_target_option): Likewise.
35620 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35622 * config/i386/i386.c (memory_address_length): Increase len
35623 only when rip_relative_addr_p returns false.
35625 2017-01-11 Julia Koval <julia.koval@intel.com>
35627 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35628 (OPTION_MASK_ISA_SGX_SET): New.
35629 (ix86_handle_option): Handle OPT_msgx.
35630 * config.gcc: Added sgxintrin.h.
35631 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35632 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35633 * config/i386/i386.c (ix86_target_string): Add -msgx.
35635 (ix86_option_override_internal): Handle new options.
35636 (ix86_valid_target_attribute_inner_p): Add sgx.
35637 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35638 * config/i386/i386.opt: Add msgx.
35639 * config/i386/sgxintrin.h: New file.
35640 * config/i386/x86intrin.h: Add sgxintrin.h.
35642 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35645 * fold-const.c (maybe_nonzero_address): Return 1 for function
35647 (tree_single_nonzero_warnv_p): Don't handle function local objects
35651 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35654 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35657 * opts.c: Include "spellcheck.h"
35658 (struct string_fragment): New struct.
35659 (struct edit_distance_traits<const string_fragment &>): New
35661 (get_closest_sanitizer_option): New function.
35662 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35664 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35666 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35668 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35669 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35670 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35671 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35672 for initial die_offset if dwarf_split_debug_info.
35673 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35674 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35676 (output_skeleton_debug_sections): Formatting fix. Use
35677 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35678 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35680 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35682 * config/arm/cortex-a53.md: Add bypasses for
35683 cortex_a53_r2f_cvt.
35684 (cortex_a53_r2f): Only use for transfers.
35685 (cortex_a53_f2r): Likewise.
35686 (cortex_a53_r2f_cvt): Add reservation for conversions.
35687 (cortex_a53_f2r_cvt): Likewise.
35689 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35691 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35692 to all inlined functions, change static to extern.
35694 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35697 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35699 (arm_assemble_integer): Likewise.
35701 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35703 * config.gcc: Use new awk script to check CPU, FPU and architecture
35704 parameters for --with-... options.
35705 * config/arm/parsecpu.awk: New file
35706 * config/arm/arm-cpus.in: New file.
35707 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35709 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35711 * config/arm/t-arm: Update dependency rules.
35712 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35713 of processing .def files.
35714 * config/arm/genopt.sh: Deleted.
35715 * config/arm/gentune.sh: Deleted.
35716 * config/arm/arm-cores.def: Deleted.
35717 * config/arm/arm-arches.def: Deleted.
35718 * config/arm/arm-fpus.def: Deleted.
35719 * config/arm/arm-tune.md: Regenerated.
35720 * config/arm/arm-tables.opt: Regenerated.
35721 * config/arm/arm-cpu.h: New generated file.
35722 * config/arm/arm-cpu-data.h: New generated file.
35723 * config/arm/arm-cpu-cdata.h: New generated file.
35725 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35728 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35730 (input_varpool_node): Unpack dynamically_initialized bit.
35732 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35734 PR rtl-optimization/79032
35735 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35736 the alignment of the adjusted memory reference against that of MODE,
35737 instead of the alignment of the original memory reference.
35739 2017-01-11 Martin Jambor <mjambor@suse.cz>
35741 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35743 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35744 decorated functions.
35746 2017-01-11 Richard Biener <rguenther@suse.de>
35748 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35749 set range/nonnull info for PHI results. Do not set it on
35750 stmts marked for removal.
35752 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35754 * expr.c (store_field): In the bitfield case, fetch the return value
35755 from the registers before applying a single big-endian adjustment.
35756 Always do a final load for a BLKmode value not larger than a word.
35758 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35761 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35762 that we correctly handle column numbers greater than
35763 LINE_MAP_MAX_COLUMN_NUMBER.
35765 2017-01-10 Martin Sebor <msebor@redhat.com>
35767 PR middle-end/78245
35768 * gimple-ssa-sprintf.c (get_destination_size): Call
35769 {init,fini}object_sizes.
35770 * tree-object-size.c (addr_object_size): Adjust.
35771 (pass_through_call): Adjust.
35772 (pass_object_sizes::execute): Adjust.
35773 * tree-object-size.h (fini_object_sizes): Declare.
35775 2017-01-10 Martin Sebor <msebor@redhat.com>
35777 PR tree-optimization/78775
35778 * builtins.c (get_size_range): Move...
35779 * calls.c: ...to here.
35780 (alloc_max_size): Accept zero argument.
35781 (operand_signed_p): Remove.
35782 (maybe_warn_alloc_args_overflow): Call get_size_range.
35783 * calls.h (get_size_range): Declare.
35785 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35787 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35788 from TI's devices.csv file as of September 2016.
35789 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35791 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35793 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35794 * doc/invoke.texi: Likewise.
35795 * doc/md.texi: Likewise.
35796 * doc/objc.texi: Likewise.
35798 2017-01-10 Joshua Conner <joshconner@google.com>
35800 * config/arm/fuchsia-elf.h: New file.
35801 * config/fuchsia.h: New file.
35802 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
35803 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
35805 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
35807 2016-01-10 Richard Biener <rguenther@suse.de>
35809 PR tree-optimization/79034
35810 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35811 Propagate out degenerate PHIs in the joiner.
35813 2017-01-10 Martin Liska <mliska@suse.cz>
35815 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
35816 (sort_congruence_classes_by_decl_uid): Likewise.
35817 (sort_congruence_class_groups_by_decl_uid): Likewise.
35818 (sem_item_optimizer::merge_classes): Sort class, groups in these
35819 classes and members in the groups by DECL_UID of declarations.
35820 This would make merge operations stable.
35822 2017-01-10 Martin Liska <mliska@suse.cz>
35824 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
35825 usage of m_classes_vec.
35826 (sem_item_optimizer::~sem_item_optimizer): Likewise.
35827 (sem_item_optimizer::get_group_by_hash): Likewise.
35828 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
35829 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
35830 (sem_item_optimizer::verify_classes): Likewise.
35831 (sem_item_optimizer::process_cong_reduction): Likewise.
35832 (sem_item_optimizer::dump_cong_classes): Likewise.
35833 (sem_item_optimizer::merge_classes): Likewise.
35834 * ipa-icf.h (congruence_class_hash): Rename from
35835 congruence_class_group_hash. Remove declaration of m_classes_vec.
35837 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35839 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35840 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35841 * config.gcc: Add avx512vpopcntdqintrin.h.
35842 * config/i386/avx512vpopcntdqintrin.h: New.
35843 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35844 * config/i386/i386-builtin-types.def: Add new types.
35845 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35846 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35847 __builtin_ia32_vpopcountq_v8di_mask): New.
35848 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35849 __AVX512VPOPCNTDQ__.
35850 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35851 (PTA_AVX512VPOPCNTDQ): Define.
35852 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35853 TARGET_AVX512VPOPCNTDQ_P): Define.
35854 * config/i386/i386.opt: Add mavx512vpopcntdq.
35855 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35856 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35858 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35860 PR middle-end/77484
35861 * predict.def (PRED_CALL): Set to 67.
35863 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35865 * expr.c (store_field): In the bitfield case, if the value comes from
35866 a function call and is of an aggregate type returned in registers, do
35867 not modify the field mode; extract the value in all cases if the mode
35868 is BLKmode and the size is not larger than a word.
35870 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35873 * config/i386/cpuid.h: Fix undefined behavior.
35875 2017-01-04 Jeff Law <law@redhat.com>
35877 PR tree-optimization/79007
35878 PR tree-optimization/67955
35879 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35880 conservative for pt.null when flag_non_call_exceptions is on.
35882 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35884 PR translation/79019
35885 PR translation/79020
35886 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35887 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35888 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35890 * config/avr/avr.opt (maccumulate-args): Likewise.
35891 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35892 * common.opt (freport-bug): Likewise.
35893 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35894 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35895 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35896 translatable string.
35897 * config/i386/i386.c (function_value_32): Likewise.
35898 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35899 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35901 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35902 * common/config/msp430/msp430-common.c (msp430_handle_option):
35904 * symtab.c (symtab_node::verify_base): Likewise.
35905 * opts.c (set_debug_level): Likewise.
35906 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35907 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35908 missing whitespace to translatable strings.
35909 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35910 * config/sh/superh.h: Likewise.
35911 * config/i386/xopintrin.h: Likewise.
35912 * config/i386/znver1.md: Likewise.
35913 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35914 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35915 * double-int.h (struct double_int): Likewise.
35916 * double-int.c (div_and_round_double): Likewise.
35917 * wide-int.cc: Likewise.
35918 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35919 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35920 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35921 (crossjumps_occurred): ... this.
35922 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35925 PR tree-optimization/78899
35926 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35927 returning bool return struct loop *, NULL for failure and the new
35929 (versionable_outer_loop_p): Don't version outer loop if it has
35930 dont_vectorized bit set.
35931 (tree_if_conversion): When versioning outer loop, ensure
35932 tree_if_conversion is performed also on the inner loop of the
35933 non-vectorizable outer loop copy.
35934 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35935 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35936 prevent vectorization of it.
35937 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35938 the outer loop vectorization of the non-scalar version is attempted
35939 before vectorization of the inner loop in scalar version. If
35940 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35941 vectorization of its inner loop.
35942 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35943 has 2 inner loops, rename also on edges from bb whose single pred
35944 is outer_loop->header. Fix typo in function comment.
35946 2017-01-09 Martin Sebor <msebor@redhat.com>
35949 * asan.c (asan_emit_stack_protection): Increase local buffer size
35950 to avoid snprintf truncation warning.
35952 2017-01-09 Andrew Pinski <apinski@cavium.com>
35954 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35955 to reference thunderx2t99 for the tuning structure
35956 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35958 (thunderx2t99_extra_costs): This.
35959 * config/aarch64/aarch64-tune.md: Regenerate.
35960 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35961 (vulcan_addrcost_table): This.
35962 (vulcan_regmove_cost): Rename to ...
35963 (thunderx2t99_regmove_cost): This.
35964 (vulcan_vector_cost): Rename to ...
35965 (thunderx2t99_vector_cost): this.
35966 (vulcan_branch_cost): Rename to ...
35967 (thunderx2t99_branch_cost): This.
35968 (vulcan_tunings): Rename to ...
35969 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35970 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35972 2017-01-09 Martin Jambor <mjambor@suse.cz>
35976 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35977 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35978 (propagate_vr_accross_jump_function): Use the above function for all
35979 value range computations for pass-through jump functions and type
35980 converasion from explicit value range values.
35981 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35982 parameters from TYPE_ARG_TYPES.
35983 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35984 (ipa_write_node_info): Stream type of the actual argument.
35985 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35987 2017-01-09 Martin Liska <mliska@suse.cz>
35990 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35991 (lookup_compiler): Do not show error message with have_E.
35993 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35995 PR tree-optimization/78938
35996 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35997 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35998 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35999 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36002 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36004 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36007 2017-01-09 Richard Biener <rguenther@suse.de>
36009 PR tree-optimization/78997
36010 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36011 name condition properly.
36013 2017-01-09 Richard Biener <rguenther@suse.de>
36016 * dwarf2out.c (is_cxx): New overload with context.
36017 (is_naming_typedef_decl): Use it.
36019 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36021 * invoke.texi (Option Summary): Correct spacing in option lists
36022 and add line breaks to fix over-long lines.
36024 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36026 PR middle-end/17660
36028 * extend.texi (Common Variable Attributes): Add xref to GCC
36029 Internals manual to explain mode attribute keywords.
36031 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36034 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36035 and Preprocessor Options.
36036 (Options for Linking): Document -pthread here....
36037 (RS/6000 and PowerPC Options): ...not here.
36038 (Solaris 2 Options): ...or here.
36039 * doc/cppopts.texi: Document -pthread.
36041 2017-01-08 Martin Sebor <msebor@redhat.com>
36043 PR middle-end/77708
36044 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36045 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36046 New member functions.
36047 (format_directive): Used them.
36049 (pass_sprintf_length::handle_gimple_call): Same.
36050 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36051 to avoid truncation for any argument.
36052 (extract_affine_mul): Same.
36053 * tree.c (get_file_function_name): Same.
36055 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36057 PR middle-end/77484
36058 * predict.def (PRED_INDIR_CALL): Set to 86.
36060 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36062 PR preprocessor/54124
36063 * doc/cppopts.texi: Reformat -d subtable to list the full name
36064 of the options. Add cross-reference to the docs for the general
36065 compiler -d options.
36066 * doc/invoke.texi (Developer Options): Add cross-reference to the
36067 preprocessor-specific -d option documentation.
36069 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36071 PR preprocessor/13498
36072 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36073 redudant material, and reflect new command-line options.
36074 (System Headers): Likewise.
36076 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36078 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36079 -isystem, and -idirafter. Copy-edit.
36080 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36081 default for -ftrack-macro-expansion. Delete obsolete and
36082 badly-formatted implementation details about -fdebug-cpp output.
36083 * doc/cppwarnopts.texi: Copy-edit.
36085 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36088 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36089 that the transition from a max line width >= 1<<10 to narrower
36090 lines works correctly.
36092 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36094 * doc/options.texi (PerFunction): New.
36095 * opt-functions.awk (switch_flags): Map both Optimization and
36096 PerFunction to CL_OPTIMIZATION.
36097 * opth-gen.awk: Test for PerFunction flag along with
36099 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36100 it only when the latter is present. Skip those that don't in
36101 the hash function generator.
36102 * common.opt (fvar-tracking): Mark as PerFunction instead of
36104 (fvar-tracking-assignments): Likewise.
36105 (fvar-tracking-assignments-toggle): Likewise.
36106 (fvar-tracking-uninit): Likewise.
36108 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36110 PR translation/79018
36111 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36114 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36117 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36118 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36119 TARGET_LONG_JUMP_TABLE_OFFSETS.
36120 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36121 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36122 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36123 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36124 * config/m68k/m68k.md (tablejump expander): Likewise.
36125 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36126 TARGET_LONG_JUMP_TABLE_OFFSETS.
36127 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36128 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36130 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36131 David Holsgrove <david.holsgrove@xilinx.com>
36133 * common/config/microblaze/microblaze-common.c
36134 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36135 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36137 * config/microblaze/microblaze.c (microblaze_must_save_register)
36138 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36140 (microblaze_eh_return): New function.
36141 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36142 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36143 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36144 * config/microblaze/microblaze.md (eh_return): New pattern.
36146 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36148 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36149 GCC_DIAGNOSTIC_STRINGIFY): Define.
36151 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36153 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36155 * config/arm/arm.md (<mcrr>): New.
36157 * config/arm/arm.c (arm_arch5te): New.
36158 (arm_option_override): Set arm_arch5te.
36159 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36161 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36162 (arm_mcrr_qualifiers): ... this. New.
36163 (MRRC_QUALIFIERS): Define to...
36164 (arm_mrrc_qualifiers): ... this. New.
36165 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36167 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36168 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36169 (MRRCI, mrrc, MRRC): New.
36170 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36171 VUNSPEC_MRRC2): New.
36173 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36175 * config/arm/arm.md (<mcr>): New.
36177 * config/arm/arm.c (arm_coproc_builtin_available): Add
36178 support for mcr, mrc, mcr2 and mrc2.
36179 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36180 (arm_mcr_qualifiers): ... this. New.
36181 (MRC_QUALIFIERS): Define to ...
36182 (arm_mrc_qualifiers): ... this. New.
36183 (MCR_QUALIFIERS): Define to ...
36184 (arm_mcr_qualifiers): ... this. New.
36185 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36187 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36188 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36189 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36190 VUNSPEC_MRC2): New.
36192 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36194 * config/arm/arm.md (*ldc): New.
36198 * config/arm/arm.c (arm_coproc_builtin_available): Add
36199 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36200 (arm_coproc_ldc_stc_legitimate_address): New.
36201 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36202 'qualifier_const_pointer'.
36203 (LDC_QUALIFIERS): Define to...
36204 (arm_ldc_qualifiers): ... this. New.
36205 (STC_QUALIFIERS): Define to...
36206 (arm_stc_qualifiers): ... this. New.
36207 * config/arm/arm-protos.h
36208 (arm_coproc_ldc_stc_legitimate_address): New.
36209 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36210 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36211 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36212 stc2, stcl, stc2l): New.
36213 * config/arm/constraints.md (Uz): New.
36214 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36215 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36216 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36217 VUNSPEC_STC2L): New.
36219 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36221 * config/arm/arm.md (<cdp>): New.
36222 * config/arm/arm.c (neon_const_bounds): Rename this ...
36223 (arm_const_bounds): ... this.
36224 (arm_coproc_builtin_available): New.
36225 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36226 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36227 (CDP_QUALIFIERS): Define to...
36228 (arm_cdp_qualifiers): ... this. New.
36230 (arm_expand_builtin_args): Add case for 6 arguments.
36231 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36232 (arm_const_bounds): ... this.
36233 (arm_coproc_builtin_available): New.
36234 * config/arm/arm_acle.h (__arm_cdp): New.
36236 * config/arm/arm_acle_builtins.def (cdp): New.
36238 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36239 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36240 'arm_const_bounds'.
36241 * config/arm/types.md (coproc): New.
36242 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36243 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36244 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36245 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36247 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36249 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36250 (UBINOP_QUALIFIERS): New.
36252 (acle_builtin_data): New. Change comment.
36253 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36254 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36255 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36256 arm_acle_builtins.def.
36257 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36258 (arm_init_acle_builtins): New.
36259 (CRC32_BUILTIN): Remove.
36260 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36261 crc32cb, crc32ch and crc32cw.
36262 (arm_init_crc32_builtins): Remove.
36263 (arm_init_builtins): Use arm_init_acle_builtins rather
36264 than arm_init_crc32_builtins.
36265 (arm_expand_acle_builtin): New.
36266 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36267 * config/arm/arm_acle_builtins.def: New.
36269 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36271 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36272 (arm_builtin_datum): ... this.
36273 (arm_init_neon_builtin): Rename to ...
36274 (arm_init_builtin): ... this. Add a new parameters PREFIX
36275 and USE_SIG_IN_NAME.
36276 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36277 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36278 'arm_builtin_datum'.
36279 (arm_init_vfp_builtins): Likewise.
36280 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36281 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36282 (arm_expand_neon_args): Rename to ...
36283 (arm_expand_builtin_args): ... this. Rename builtin_arg
36284 enum values and differentiate between ARG_BUILTIN_MEMORY
36285 and ARG_BUILTIN_NEON_MEMORY.
36286 (arm_expand_neon_builtin_1): Rename to ...
36287 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36288 values, arm_expand_builtin_args and add bool parameter NEON.
36289 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36290 (arm_expand_vfp_builtin): Likewise.
36291 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36293 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36295 PR middle-end/77484
36296 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36297 * predict.c (tree_estimate_probability_bb): Reverse direction of
36298 polymorphic call predictor.
36300 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36302 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36303 (determine_pass_name_match): ...this new function and...
36304 (should_skip_pass_p): ...this new function.
36306 2017-01-06 Nathan Sidwell <nathan@acm.org>
36308 * ipa-visibility.c (function_and_variable_visibility): Reformat
36309 comments and long lines. Remove extrneous if.
36310 * symtab.c (symtab_node::make_decl_local): Fix code format.
36311 (symtab_node::set_section_for_node): Fix comment typo.
36313 2017-01-06 Martin Liska <mliska@suse.cz>
36316 * lra-constraints.c: Rename invariant to lra_invariant.
36317 * predict.c (set_even_probabilities): Initialize e to NULL.
36319 2017-01-05 Martin Sebor <msebor@redhat.com>
36321 PR tree-optimization/78910
36322 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36323 (format_integer): Correct off-by-one error in the handling
36324 of precision with negative numbers in signed conversions..
36326 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36328 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36330 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36332 PR tree-optimization/71016
36333 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36334 factor_out_conditional_conversion. Formatting fix.
36335 (factor_out_conditional_conversion): Add cond_stmt argument.
36336 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36337 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36340 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36342 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36343 read-rtl-function.o, and selftest-rtl.o.
36344 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36345 (selftest::aarch64_test_loading_full_dump): New function.
36346 (selftest::aarch64_run_selftests): New function.
36347 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36348 selftest::aarch64_run_selftests.
36349 * config/i386/i386.c
36350 (selftest::ix86_test_loading_dump_fragment_1): New function.
36351 (selftest::ix86_test_loading_call_insn): New function.
36352 (selftest::ix86_test_loading_full_dump): New function.
36353 (selftest::ix86_test_loading_unspec): New function.
36354 (selftest::ix86_run_selftests): Call the new functions.
36355 * emit-rtl.c (maybe_set_max_label_num): New function.
36356 * emit-rtl.h (maybe_set_max_label_num): New decl.
36357 * function.c (instantiate_decls): Guard call to
36358 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36359 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36361 * gensupport.c (gen_reader::gen_reader): Pass "false"
36362 for new "compact" param of rtx_reader.
36363 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36364 rather than an empty string for NULL strings.
36365 * read-md.c: Potentially include config.h rather than bconfig.h.
36366 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36367 (have_error): New global, copied from errors.c.
36368 (md_reader::read_name): Rename to...
36369 (md_reader::read_name_1): ...this, adding "out_loc" param,
36370 and converting "missing name or number" to returning false, rather
36372 (md_reader::read_name): Reimplement in terms of read_name_1.
36373 (md_reader::read_name_or_nil): New function.
36374 (md_reader::read_string): Handle "(nil)" by returning NULL.
36375 (md_reader::md_reader): Add new param "compact".
36376 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36377 (md_reader::read_file): New method.
36378 * read-md.h (md_reader::md_reader): Add new param "compact".
36379 (md_reader::read_file): New method.
36380 (md_reader::is_compact): New accessor.
36381 (md_reader::read_name): Convert return type from void to file_location.
36382 (md_reader::read_name_or_nil): New decl.
36383 (md_reader::read_name_1): New decl.
36384 (md_reader::m_compact): New field.
36385 (noop_reader::noop_reader): Pass "false" for new "compact" param
36387 (rtx_reader::rtx_reader): Add new "compact" param.
36388 (rtx_reader::read_rtx_operand): Make virtual and convert return
36389 type from void to rtx.
36390 (rtx_reader::read_until): New decl.
36391 (rtx_reader::handle_any_trailing_information): New virtual function.
36392 (rtx_reader::postprocess): New virtual function.
36393 (rtx_reader::finalize_string): New virtual function.
36394 (rtx_reader::m_in_call_function_usage): New field.
36395 (rtx_reader::m_reuse_rtx_by_id): New field.
36396 * read-rtl-function.c: New file.
36397 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36398 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36399 (selftest::verify_three_block_rtl_cfg): New decl.
36400 * read-rtl-function.h: New file.
36401 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36402 For host, include function.h, memmodel.h, and emit-rtl.h.
36403 (one_time_initialization): New function.
36404 (struct compact_insn_name): New struct.
36405 (compact_insn_names): New array.
36406 (find_code): Handle insn codes in compact dumps.
36407 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36408 (bind_subst_iter_and_attr): Likewise.
36409 (add_condition_to_string): Likewise.
36410 (add_condition_to_rtx): Likewise.
36411 (apply_attribute_uses): Likewise.
36412 (add_current_iterators): Likewise.
36413 (apply_iterators): Likewise.
36414 (initialize_iterators): Guard usage of apply_subst_iterator with
36415 #ifdef GENERATOR_FILE.
36416 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36417 (md_reader::read_mapping): Likewise.
36418 (add_define_attr_for_define_subst): Likewise.
36419 (add_define_subst_attr): Likewise.
36420 (read_subst_mapping): Likewise.
36421 (check_code_iterator): Likewise.
36422 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36424 (one_time_initialization): New function.
36425 (rtx_reader::read_until): New method.
36426 (read_flags): New function.
36427 (parse_reg_note_name): New function.
36428 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36429 Handle reuse_rtx ids.
36430 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36431 Add parsing support for RTL dumps, mirroring the special-cases in
36432 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36433 values, and calling handle_any_trailing_information.
36434 (rtx_reader::read_rtx_operand): Convert return type from void
36435 to rtx, returning return_rtx. Handle case 'e'. Call
36436 finalize_string on XSTR and XTMPL fields.
36437 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36438 "(nil)" values were omitted. Call the postprocess vfunc on the
36440 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36441 class ctor. Initialize m_in_call_function_usage. Call
36442 one_time_initialization.
36443 * rtl-tests.c (selftest::test_uncond_jump): Call
36444 set_new_first_and_last_insn.
36445 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36446 * selftest-rtl.c: New file.
36447 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36448 (selftest::get_insn_by_uid): New decl.
36449 * selftest-run-tests.c (selftest::run_tests): Call
36450 read_rtl_function_c_tests.
36451 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36452 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36455 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36457 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36458 operands in a special way. Assert that pos+len <= mode precision.
36460 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36462 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36463 3 argument Alias with unlimited for the negative form.
36464 (fno-vect-cost-model): Removed.
36466 2017-01-05 Martin Liska <mliska@suse.cz>
36468 * hsa-gen.c (gen_hsa_divmod): New function.
36469 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36471 2017-01-05 Martin Liska <mliska@suse.cz>
36474 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36477 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36479 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36480 small constant length operands.
36482 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36484 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36485 between loop iterations.
36487 2017-01-05 Martin Liska <mliska@suse.cz>
36490 * gimplify.c (gimplify_decl_expr): Compare to
36491 asan_poisoned_variables instread of checking flags.
36492 (gimplify_target_expr): Likewise.
36493 (gimplify_expr): Likewise.
36494 (gimplify_function_tree): Conditionally initialize
36495 asan_poisoned_variables.
36497 2017-01-04 Jeff Law <law@redhat.com>
36499 PR tree-optimizatin/78812
36500 * rtl.h (contains_mem_rtx_p): Prototype.
36501 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36502 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36503 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36504 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36505 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36507 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36509 * input.c (assert_char_at_range): Default-initialize actual_range.
36511 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36513 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36514 to match the caller.
36516 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36518 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36519 insns after final jump in test to emit dummy move.
36521 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36523 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36524 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36526 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36528 * multiple_target.c (create_dispatcher_calls): Init e_next.
36529 * tree-ssa-loop-split.c (split_loop): Init border.
36530 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36533 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36538 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36539 (altivec_register_operand): Do not return true if the operand
36540 contains a SUBREG mixing SImode and SFmode.
36541 (vsx_register_operand): Likewise.
36542 (vsx_reg_sfsubreg_ok): New predicate.
36543 (vfloat_operand): Do not return true if the operand contains a
36544 SUBREG mixing SImode and SFmode.
36545 (vint_operand): Likewise.
36546 (vlogical_operand): Likewise.
36547 (gpc_reg_operand): Likewise.
36548 (int_reg_operand): Likewise.
36549 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36550 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36551 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36553 (rs6000_emit_move_si_sf_subreg): New helper function.
36554 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36555 fixup SUBREGs involving SImode and SFmode.
36556 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36557 numbers for the new peephole2 optimization.
36558 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36559 the GLIBC math library that do AND/IOR/XOR operations on single
36560 precision floating point.
36561 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36562 target macros to say whether we need to avoid SUBREGs mixing
36564 (TARGET_ALLOW_SF_SUBREG): Likewise.
36565 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36566 (UNSPEC_SI_FROM_SF): Likewise.
36567 (iorxor): Change spacing.
36568 (and_ior_xor): New iterator for AND, IOR, and XOR.
36569 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36570 (movdi_from_sf_zero_ext): Likewise.
36571 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36572 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36573 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36574 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36575 (fms<mode>4): Likewise.
36576 (fnma<mode>4): Likewise.
36577 (fnms<mode>4): Likewise.
36578 (nfma<mode>4): Likewise.
36579 (nfms<mode>4): Likewise.
36581 2017-01-04 Marek Polacek <polacek@redhat.com>
36584 * doc/invoke.texi: Document -Wpointer-compare.
36586 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36588 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36591 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36592 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36595 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36598 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36599 documentation of the powerpc_popcntb_ok attribute.
36600 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36601 code to issue warning messages if a requested CPU configuration is
36602 not supported by the binary (assembler and loader) toolchain.
36603 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36604 made to define a built-in function that has been disabled.
36605 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36606 made to define a built-in function that has been disabled.
36607 (altivec_init_builtins): Add comment explaining why definition
36608 of the DST built-in functions is not preceded by an assertion
36609 check. Add assertions to prevent ICE if attempts are made to
36610 define an altivec predicate or an abs* built-in function that has
36612 (htm_init_builtins): Add comment explaining why definition of the
36613 htm built-in functions is not preceded by an assertion check.
36615 2017-01-04 Jeff Law <law@redhat.com>
36617 PR tree-optimizatin/67955
36618 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36619 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36620 the points-to solution does not include pt_null. Use DECL_PT_UID
36623 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36625 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36626 Use gen_int_mode instead of gen_lopwart for const_int operands.
36628 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36630 PR tree-optimization/71563
36631 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36632 out of range value - has low bits known to be zero.
36634 2017-01-04 Alan Modra <amodra@gmail.com>
36636 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36637 * configure: Regenerate.
36638 * config.in: Regenerate.
36640 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36643 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36644 a substring of the message, but strcmp with the whole message. Ifdef
36645 ENABLE_NLS, translate the message first using dgettext.
36647 2017-01-03 Jeff Law <law@redhat.com>
36649 PR tree-optimizatin/78856
36650 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36651 (mark_threaded_blocks): Remove code to truncate thread paths that
36652 cross multiple loop headers. Instead invalidate the cached loop
36653 iteration information and handle case of a thread path walking
36654 into an irreducible region.
36656 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36659 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36660 assertions. Add support for doing the signbit if the IEEE 128-bit
36661 floating point value is in a GPR.
36662 * config/rs6000/rs6000.md (Fsignbit): Delete.
36663 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36664 Update the length attribute if the value is in a GPR.
36665 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36666 the sign or zero extension instruction, since the value is always 0/1.
36667 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36670 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36671 extracting SImode to a GPR register so that we can generate a
36672 store, limit the vector to be in a traditional Altivec register
36673 for the vextuwrx instruction.
36675 2017-01-03 Ian Lance Taylor <iant@google.com>
36677 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36679 2017-01-03 Martin Sebor <msebor@redhat.com>
36681 PR tree-optimization/78696
36682 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36683 precision. Use MPFR for %f for greater fidelity. Correct handling
36685 (pass_sprintf_length::compute_format_length): Set width and precision
36686 specified by asrerisk to void_node for vararg functions.
36687 (try_substitute_return_value): Adjust dump output.
36689 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36691 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36693 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36695 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36696 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36697 -mlra/-mno-lra was passed to the compiler.
36699 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36701 PR rtl-optimization/65618
36702 * emit-rtl.c (try_split): Move initialization of "before" and
36703 "after" to just before the call to emit_insn_after_setloc.
36705 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36707 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36709 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36711 * dwarf2out.c (gen_enumeration_type_die): When
36712 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36714 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36716 PR tree-optimization/78965
36717 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36718 Change first argument from const call_info & to call_info &. For %n
36719 set info.nowrite to false.
36721 PR middle-end/78901
36722 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36723 possibly throwing calls.
36725 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36726 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36727 and fns handling, rather than in a separate case SSA_NAME.
36729 2017-01-02 Jeff Law <law@redhat.com>
36731 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36732 fixes for first_period and second_period variables.
36734 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36737 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36738 (*insvqi_1): New insn pattern.
36739 (*insvqi_1_mem_rex64): Ditto.
36740 (*insvqi_2): Ditto.
36741 (*insvqi_3): Rename from *insvqi.
36743 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36745 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36747 * doc/cfg.texi (Edges): Remove reference to Java.
36748 (Maintaining the CFG): Ditto.
36750 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36752 PR middle-end/77674
36753 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36754 transparent aliases.
36756 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36758 PR middle-end/77484
36759 * predict.def (PRED_CALL): Update hitrate.
36760 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36761 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36762 into direct/indirect/polymorphic variants.
36764 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36766 Update copyright years.
36768 * gcc.c (process_command): Update copyright notice dates.
36769 * gcov-dump.c (print_version): Ditto.
36770 * gcov.c (print_version): Ditto.
36771 * gcov-tool.c (print_version): Ditto.
36772 * gengtype.c (create_file): Ditto.
36773 * doc/cpp.texi: Bump @copying's copyright year.
36774 * doc/cppinternals.texi: Ditto.
36775 * doc/gcc.texi: Ditto.
36776 * doc/gccint.texi: Ditto.
36777 * doc/gcov.texi: Ditto.
36778 * doc/install.texi: Ditto.
36779 * doc/invoke.texi: Ditto.
36781 Copyright (C) 2017 Free Software Foundation, Inc.
36783 Copying and distribution of this file, with or without modification,
36784 are permitted in any medium without royalty provided the copyright
36785 notice and this notice are preserved.