1 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4 cases for printing LDP/STP memory addresses.
5 (aarch64_print_address_internal): Renamed from
6 aarch64_print_operand_address, added parameter, add Pmode check.
7 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
8 (aarch64_print_operand_address): Indirect to
9 aarch64_print_address_internal.
10 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
11 'y' operand output specifier.
13 2017-11-29 Jakub Jelinek <jakub@redhat.com>
16 * tree.c (build_simple_mem_ref_loc): Handle
17 get_addr_base_and_unit_offset returning a MEM_REF.
20 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
23 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
26 2017-11-28 David Malcolm <dmalcolm@redhat.com>
28 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
30 (layout::move_to_column): Likewise.
32 2017-11-28 Jakub Jelinek <jakub@redhat.com>
35 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
36 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
38 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
39 Martin Jambor <mjambor@suse.cz>
42 * tree.h (expr_type_first_operand_type_p): Declare
43 * tree.c (expr_type_first_operand_type_p): New function.
44 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
45 (ipa_value_from_jfunc): Adjust declaration.
46 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
47 Use it as result type for arithmetics, unless it is NULL in which case
49 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
50 ipa_get_jf_pass_through_result.
51 (propagate_vals_across_pass_through): Likewise.
52 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
53 is to propagate_vals_across_pass_through.
54 (propagate_constants_across_call): Pass PARM_TYPE to
55 propagate_scalar_across_jump_function.
56 (find_more_scalar_values_for_callers_subset): Pass parameter type to
58 (cgraph_edge_brings_all_scalars_for_node): Likewise.
59 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
60 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
61 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
62 target_type, pass it to ipa_value_from_jfunc.
63 (update_indirect_edges_after_inlining): Pass parameter type to
64 try_make_edge_direct_simple_call.
66 2017-11-28 Jeff Law <law@redhat.com>
68 * gimple-ssa-evrp-analyze.c
69 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
70 refine ranges if scev_initialized_p returns true.
71 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
73 2017-11-28 Julia Koval <julia.koval@intel.com>
75 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
76 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
77 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
78 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
79 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
80 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
81 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
82 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
83 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
84 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
85 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
86 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
87 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
88 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
89 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
90 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
91 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
92 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
93 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
94 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
95 __builtin_ia32_vpshrd_v2di_mask): New builtins.
96 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
98 2017-11-28 Julia Koval <julia.koval@intel.com>
100 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
101 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
102 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
103 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
104 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
105 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
106 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
107 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
108 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
109 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
110 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
111 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
112 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
113 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
114 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
115 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
116 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
117 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
118 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
119 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
120 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
121 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
122 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
123 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
124 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
125 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
126 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
127 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
128 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
129 __builtin_ia32_vpshld_v2di_mask): New builtins.
130 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
131 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
133 2017-11-28 Richard Biener <rguenther@suse.de>
135 PR tree-optimization/80776
136 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
138 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
140 (evrp_range_analyzer::record_ranges_from_incoming_edges):
141 If the incoming edge is an effective fallthru because the other
142 edge only reaches a __builtin_unreachable () then record ranges
143 derived from the controlling condition in SSA info.
144 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
145 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
147 2017-11-28 Olivier Hainque <hainque@adacore.com>
149 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
150 on mingw build hosts.
152 2017-11-28 Jakub Jelinek <jakub@redhat.com>
155 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
157 2017-11-28 Richard Biener <rguenther@suse.de>
160 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
161 copies generated from memcpy use a character array as reference
164 2017-11-28 Julia Koval <julia.koval@intel.com>
165 Sebastian Peryt <sebastian.peryt@intel.com>
167 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
168 c-family/c-cilkplus.o, c-family/array-notation-common.o,
169 cilk-common.o, cilk.h, cilk-common.c): Remove.
171 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
172 * builtins.c (is_builtin_name): Remove cilkplus condition.
173 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
174 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
175 cilk-builtins.def, cilkplus.def): Remove.
176 * cif-code.def (CILK_SPAWN): Remove.
177 * cilk-builtins.def: Delete.
178 * cilk-common.c: Ditto.
180 * cilkplus.def: Ditto.
181 * config/darwin.h (fcilkplus): Delete.
182 * cppbuiltin.c: Ditto.
183 * doc/extend.texi: Remove cilkplus doc.
184 * doc/generic.texi: Ditto.
185 * doc/invoke.texi: Ditto.
186 * doc/passes.texi: Ditto.
187 * gcc.c (fcilkplus): Remove.
188 * gengtype.c (cilk.h): Remove.
189 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
191 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
193 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
195 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
196 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
198 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
199 inline_read_section): Ditto.
200 * ipa-inline-analysis.c (cilk.h): Remove.
201 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
202 * lto-wrapper.c (merge_and_complain, append_compiler_options,
203 append_linker_options): Remove condition for fcilkplus.
204 * lto/lto-lang.c (cilk.h): Remove.
205 (lto_init): Remove condition for fcilkplus.
206 * omp-expand.c (expand_cilk_for_call): Delete.
207 (expand_omp_taskreg, expand_omp_for_static_chunk,
208 expand_omp_for): Remove cilkplus
210 (expand_cilk_for): Delete.
211 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
212 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
213 execute_lower_omp, diagnose_sb_0): Ditto.
214 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
215 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
216 * tree-nested.c: Ditto.
217 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
218 (dump_generic_node): Ditto.
219 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
220 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
222 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
224 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
226 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
227 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
228 for vector mode and !TARGET_SIMD.
230 2017-11-28 Jakub Jelinek <jakub@redhat.com>
232 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
234 * tree.h (SWITCH_LABELS): Remove.
235 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
236 assert SWITCH_BODY is non-NULL.
237 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
239 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
241 PR tree-optimization/80788
242 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
243 has TREE_OVERFLOW set, call drop_tree_overflow.
245 2017-11-28 Richard Biener <rguenther@suse.de>
247 PR tree-optimization/83158
248 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
251 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
254 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
255 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
257 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
259 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
260 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
262 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
265 * varasm.c (compare_constant): Take the mode of the constants into
266 account when comparing floating point constants.
268 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
270 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
271 from explicit instantiation of debug_helper.
272 * vec.h (DEFINE_DEBUG_VEC): Ditto.
274 2017-11-27 Richard Biener <rguenther@suse.de>
276 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
279 2017-11-27 Richard Biener <rguenther@suse.de>
281 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
282 (type_hash_canon): Also clear int_cst_hash_table entry for
284 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
286 2017-11-27 Tamar Christina <tamar.christina@arm.com>
288 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
289 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
290 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
292 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
294 * hash-map.h (gt_cleare_cache): Avoid UB.
296 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
298 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
299 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
300 (decide_unroll_runtime_iterations): Likewise.
301 (decide_unroll_stupid): Likewise.
303 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
306 * config/i386/i386.md: Add a loop with incssp.
308 2017-11-27 Martin Jambor <mjambor@suse.cz>
310 PR tree-optimization/81248
311 * tree-sra.c (splice_param_accesses): Remove size check.
312 (decide_one_param_reduction): Fix size check.
313 * gimple-pretty-print.c (dump_profile): Silence warning.
314 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
316 2017-11-27 Jakub Jelinek <jakub@redhat.com>
319 * dbxout.c (lastlineno): New variable.
320 (dbx_debug_hooks): Use dbxout_switch_text_section as
321 switch_text_section debug hook.
322 (dbxout_function_end): Switch to current_function_section
323 rather than function_section. If crtl->has_bb_partition,
324 output just one N_FUN, depending on in_cold_section_p.
325 (dbxout_source_line): Remember last lineno in lastlineno.
326 (dbxout_switch_text_section): New function.
327 (dbxout_function_decl): Adjust dbxout_block caller.
328 (dbx_block_with_cold_children): New function.
329 (dbxout_block): Return true if any LBRAC/RBRAC have been
330 emitted. Use dbx_block_with_cold_children at depth == 0
331 in second partition. Add PARENT_BLOCKNUM argument, pass
332 it optionally adjusted to children. Output LBRAC/RBRAC
333 around recursive call only if the block is in the current
334 partition, if not and anything was output, emit empty
336 * final.c (final_scan_insn): Compute cold_function_name
337 before calling switch_text_section debug hook. Call
338 that hook even if dwarf2out_do_frame if not emitting
342 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
345 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
347 PR rtl-optimization/82488
348 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
350 2017-11-26 Julia Koval <julia.koval@intel.com>
352 * config/i386/i386.c (processor_target_table): Add skylake_cost for
354 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
357 2017-11-26 Julia Koval <julia.koval@intel.com>
359 * config/i386/driver-i386.c (host_detect_local_cpu):
360 Detect skylake-avx512.
362 2017-11-26 Julia Koval <julia.koval@intel.com>
364 * config.gcc: Add -march=cannonlake.
365 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
366 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
367 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
368 (PTA_CANNONLAKE): New.
369 (processor_target_table): Add cannonlake.
370 (ix86_option_override_internal): Ditto.
371 (fold_builtin_cpu): Ditto.
372 (get_builtin_code_for_version): Handle cannonlake.
373 (M_INTEL_COREI7_CANNONLAKE): New.
374 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
375 * doc/invoke.texi: Add -march=cannonlake.
377 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
379 * plugin.c (add_new_plugin): Use platform-specific library extensions.
380 (try_init_one_plugin): Alternative implementation for MinGW.
381 * Makefile.in (plugin_implib): New.
382 (gengtype-lex.c): Fix broken AIX workaround.
383 * configure: Regenerate.
384 * doc/plugins.texi: Document support for MinGW.
386 2017-11-25 Jakub Jelinek <jakub@redhat.com>
388 PR rtl-optimization/81553
389 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
390 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
391 is a shift where C1 has different mode than the whole shift, use C1's
392 mode for MULT rather than the shift's mode.
395 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
396 builtins not enabled in the currently selected ISA.
398 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
400 PR tree-optimization/71026
401 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
402 (insert_reciprocals): Change to insert reciprocals before a division
403 by a square and to insert the square of a reciprocal.
404 (execute_cse_reciprocals_1): Change to consider division by a square.
405 (register_division_in): Add importance parameter.
407 2017-11-24 Richard Biener <rguenther@suse.de>
409 PR tree-optimization/82402
410 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
411 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
413 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
415 * match.pd (0-ptr): New transformation.
417 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
420 * ipa-inline.c (inline_small_functions): Set current badnes correctly
421 when skipping checking.
423 2017-11-24 Richard Biener <rguenther@suse.de>
425 PR tree-optimization/83128
426 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
427 (vn_reference_lookup_3): Likewise.
429 2017-11-24 Jakub Jelinek <jakub@redhat.com>
432 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
433 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
436 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
437 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
438 check for the latter with gimple_call_builtin_p. Do not handle
439 BUILT_IN_STPNCPY_CHK which is not a pass through call.
441 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
443 * config/arm/arm_neon.h: Fix pragma GCC push_options before
446 2017-11-23 Julia Koval <julia.koval@intel.com>
448 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
449 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
450 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
451 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
452 _mm512_maskz_expandloadu_epi16): New intrinsics.
453 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
454 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
455 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
456 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
457 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
458 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
459 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
460 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
461 _mm256_maskz_expandloadu_epi8): New intrinsics.
462 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
463 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
464 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
465 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
466 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
467 * config/i386/sse.md (VI248_VLBW): New iterator.
468 (expand<mode>_mask, expand<mode>_maskz): New patterns.
470 2017-11-23 Julia Koval <julia.koval@intel.com>
472 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
473 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
474 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
475 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
476 _mm512_mask_compressstoreu_epi16): New.
477 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
478 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
479 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
480 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
481 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
482 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
483 _mm256_mask_compressstoreu_epi8): New.
484 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
485 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
486 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
487 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
488 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
489 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
490 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
491 __builtin_ia32_compresshi128_mask,
492 __builtin_ia32_compressstoreuqi512_mask,
493 __builtin_ia32_compressstoreuhi512_mask,
494 __builtin_ia32_compressstoreuqi256_mask,
495 __builtin_ia32_compressstoreuqi128_mask,
496 __builtin_ia32_compressstoreuhi256_mask,
497 __builtin_ia32_compressstoreuhi128_mask): New builtins.
498 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
500 (ix86_expand_special_args_builtin): Handle new types.
501 (s4fma_expand): Handle new builtin array.
502 * config/i386/immintrin.h: Include new headers.
503 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
504 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
506 2017-11-23 Jakub Jelinek <jakub@redhat.com>
508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
509 fixes. Declare temp and g variables at the top in order to avoid
510 {} in most of the cases.
512 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
514 * match.pd (ptr-0): New transformation.
516 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
518 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
519 (struct addr_mode_cost_table): New.
520 (struct tune_params): Add field addr_mode_costs.
521 * config/arm/arm.c (generic_addr_mode_costs): New.
522 (arm_slowmul_tune): Initialise addr_mode_costs field.
523 (arm_fastmul_tune): Likewise.
524 (arm_strongarm_tune): Likewise.
525 (arm_xscale_tune): Likewise.
526 (arm_9e_tune): Likewise.
527 (arm_marvell_pj4_tune): Likewise.
528 (arm_v6t2_tune): Likewise.
529 (arm_cortex_tune): Likewise.
530 (arm_cortex_a8_tune): Likewise.
531 (arm_cortex_a7_tune): Likewise.
532 (arm_cortex_a15_tune): Likewise.
533 (arm_cortex_a35_tune): Likewise.
534 (arm_cortex_a53_tune): Likewise.
535 (arm_cortex_a57_tune): Likewise.
536 (arm_exynosm1_tune): Likewise.
537 (arm_xgene1_tune): Likewise.
538 (arm_cortex_a5_tune): Likewise.
539 (arm_cortex_a9_tune): Likewise.
540 (arm_cortex_a12_tune): Likewise.
541 (arm_cortex_a73_tune): Likewise.
542 (arm_v7m_tune): Likewise.
543 (arm_cortex_m7_tune): Likewise.
544 (arm_v6m_tune): Likewise.
545 (arm_fa726te_tune): Likewise.
546 (arm_mem_costs): Use table lookup to calculate cost of addressing
549 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
551 * config/arm/arm.c (arm_mem_costs): New function.
552 (arm_rtx_costs_internal): Use arm_mem_costs.
554 2017-11-23 Mark Wielaard <mark@klomp.org>
556 * dwarf2out.c (init_sections_and_labels): Use generation to create
557 unique ranges_section_label and ranges_base_label. Return generation.
558 (output_rnglists): Add generation argument. Use generation to create
559 unique ranges labels.
560 (dwarf2out_finish): Get generation from init_sections_and_labels
561 and pass generation to output_rnglists.
563 2017-11-23 Mike Stump <mikestump@comcast.net>
564 Eric Botcazou <ebotcazou@adacore.com>
566 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
567 * cfgloop.h (struct loop): Add unroll field.
568 * function.h (struct function): Add has_unroll bitfield.
569 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
570 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
571 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
572 (pass_rtl_unroll_loops::gate): Likewise.
573 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
574 for which loop->unroll==1.
575 (decide_unroll_constant_iterations): Use note for consistency and
576 take loop->unroll into account. Return early if loop->unroll is set.
577 Fix thinko in existing test.
578 (decide_unroll_runtime_iterations): Use note for consistency and
579 take loop->unroll into account.
580 (decide_unroll_stupid): Likewise.
581 * lto-streamer-in.c (input_cfg): Read loop->unroll.
582 * lto-streamer-out.c (output_cfg): Write loop->unroll.
583 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
585 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
586 (print_loop): Print loop->unroll if set.
587 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
588 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
589 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
591 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
592 loop->unroll is set and smaller than the trip count. Otherwise bypass
593 entirely the heuristics if loop->unroll is set. Remove dead note.
594 Fix off-by-one bug in other note.
595 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
596 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
598 (tree_unroll_loops_completely): Make static.
599 (pass_complete_unroll::execute): Use correct type for variable.
600 (pass_complete_unrolli::execute): Fix formatting.
601 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
603 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
605 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
606 enable when TARGET_PREFER_AVX128 is set.
608 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
610 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
612 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
613 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
615 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
617 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
618 frequencies. Fix estimation of aggregate parameters.
620 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
622 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
625 2017-11-23 Tom de Vries <tom@codesourcery.com>
627 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
629 2017-11-23 Tom de Vries <tom@codesourcery.com>
631 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
634 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
637 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
638 sibcall_value_pcrel_fdpic): Use local variable instead of
640 (calli_tbr_rel): Add missing operand 2.
641 (call_valuei_tbr_rel): Add missing operand 3.
643 2017-11-23 Jakub Jelinek <jakub@redhat.com>
646 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
647 bitpos/bitsize covering the whole destination, use store_expr only if
648 the complex mode is the same. Otherwise, use expand_normal and if
649 it returns CONCAT, subreg each part separately instead of trying to
650 subreg the whole result.
652 2017-11-23 Richard Biener <rguenther@suse.de>
654 PR tree-optimization/23094
655 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
656 come here from walking over backedges in the first iteration.
657 (vn_reference_lookup_3): Skip clobbers that store the same value.
659 2017-11-23 Richard Biener <rguenther@suse.de>
661 PR tree-optimization/81403
662 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
663 a block we need a leader relative to.
664 (phi_translate_1): For nary processing require a leader from
665 get_representative_for given we run expression simplification
666 using match-and-simplify. Remove previous fix.
668 2017-11-22 Jeff Law <law@redhat.com>
670 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
671 Use new method allocate_value_range rather than accessing the
672 vrp_value_range_pool data member directly.
673 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
674 to use extract_range_from_stmt method to avoid need for
675 extract_range_from_assignment method.
676 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
677 method rather than setting values_propgated data member directly.
678 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
679 and values propagated data members and extract_range_from_assignment
680 method. Reorder private data members to conform to standards.
681 Add new methods set_lattice_propagation_complete and
682 allocate_value_range.
684 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
686 PR rtl-optimization/83030
687 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
688 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
689 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
690 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
691 CROSSING_JUMP_P flag is already set before setting it.
692 * cfgrtl.c (fixup_partition_crossing): Likewise.
693 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
696 2017-11-22 Jakub Jelinek <jakub@redhat.com>
698 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
699 Handle the case where both arguments are using gen_const_vec_series.
701 2017-11-22 David Malcolm <dmalcolm@redhat.com>
704 * pretty-print.c (pp_format): Move quoting implementation to
705 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
706 to pass address of "quote" local.
707 (pp_begin_quote): New function.
708 (pp_end_quote): New function.
709 * pretty-print.h (printer_fn): Convert penultimate param from bool
711 (pp_begin_quote): New decl.
712 (pp_end_quote): New decl.
713 * tree-diagnostic.c (default_tree_printer): Convert penultimate
714 param from bool to bool *.
715 * tree-diagnostic.h (default_tree_printer): Likewise.
717 2017-11-22 Jeff Law <law@redhat.com>
719 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
720 Initialize vr_values.
721 (evrp_range_analyzer::try_find_new_range): Call methods attached to
722 vr_values via vr_values class instance rather than delegators.
723 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
724 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
725 (evrp_range_analyzer::push_value_range): Likewise.
726 (evrp_range_analyzer::pop_value_range): Likewise.
727 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
728 most delegators. Those remaining are exposed as public interfaces.
729 Make vr_values a pointer and private.
730 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
732 (evrp_range_analyzer::get_vr_value): New method.
733 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
734 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
735 (class evrp_dom_walker): Attach evrp_folder class, initialize
736 it in the ctor. Remove temporary delegators.
737 (evrp_dom_walker::before_dom_children): Call methods in attached
738 evrp_range_analyzer class via class instance pointer. Use
739 free value_range_constant_singleton to remove need for
740 op_with_constant_singleton_value delegator method. Do not
741 create a vrp_prop class instance for every call! Narrow
742 scope of a couple variables.
743 (evrp_dom_walker::cleanup): Call methods in attached
744 evrp_range_analyzer class via class instance pointer.
745 * vr-values.h (class vr_values): Privatize many methods and
748 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
750 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
751 * cfgexpand.c (expand_debug_expr): Remove handling for them.
752 * expr.c (expand_expr_real_2): Likewise.
753 * fold-const.c (const_unop): Likewise.
754 * optabs-tree.c (optab_for_tree_code): Likewise.
755 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
756 * tree-inline.c (estimate_operator_cost): Likewise.
757 * tree-pretty-print.c (dump_generic_node): Likewise.
758 (op_code_prio): Likewise.
759 (op_symbol_code): Likewise.
760 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
761 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
762 * internal-fn.c (direct_internal_fn_optab): New function.
763 (direct_internal_fn_array, direct_internal_fn_supported_p
764 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
765 * fold-const-call.c (fold_const_reduction): New function.
766 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
768 * tree-vect-loop.c: Include internal-fn.h.
769 (reduction_code_for_scalar_code): Rename to...
770 (reduction_fn_for_scalar_code): ...this and return an internal
772 (vect_model_reduction_cost): Take an internal_fn rather than
774 (vect_create_epilog_for_reduction): Likewise. Build calls rather
776 (vectorizable_reduction): Use internal functions rather than tree
777 codes for the reduction operation. Update calls to the functions
779 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
780 Use calls to internal functions rather than REDUC tree codes.
781 * config/aarch64/aarch64-simd.md: Update comment accordingly.
783 2017-11-22 Olivier Hainque <hainque@adacore.com>
785 * config/vxworks.c (vxworks_override_options): Pick default
786 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
787 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
788 DWARF_GNAT_ENCODINGS_DEFAULT.
789 * config/vxworksae.h: Likewise.
791 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
793 PR tree-optimization/83104
794 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
797 2017-11-22 Marek Polacek <polacek@redhat.com>
798 H.J. Lu <hongjiu.lu@intel.com>
799 Jason Merrill <jason@redhat.com>
804 * calls.c (initialize_argument_information): Call
805 warn_parameter_passing_abi target hook.
806 (store_one_arg): Use 0 for empty record size. Don't push 0 size
808 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
809 * common.opt: Update -fabi-version description.
810 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
811 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
813 (ix86_is_empty_record): New function.
814 (ix86_warn_parameter_passing_abi): New function.
815 (TARGET_EMPTY_RECORD_P): Redefine.
816 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
817 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
818 * doc/tm.texi: Regenerated.
819 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
820 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
821 * dwarf2out.c (get_ultimate_context): Move to tree.c.
822 * explow.c (hard_function_value): Call arg_int_size_in_bytes
823 instead of int_size_in_bytes.
824 * expr.c (copy_blkmode_to_reg): Likewise.
825 * function.c (aggregate_value_p): Return 0 for empty types.
826 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
827 (locate_and_pad_parm): Call arg size_in_bytes instead
829 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
830 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
831 * target.def (empty_record_p, warn_parameter_passing_abi): New target
833 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
834 (std_gimplify_va_arg_expr): Skip empty records. Call
835 arg_size_in_bytes instead size_in_bytes.
836 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
837 * tree-core.h (tree_type_common): Add empty_flag.
838 (tree_decl_common): Update comments.
839 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
841 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
842 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
844 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
845 * tree.c (default_is_empty_type): New function.
846 (default_is_empty_record): New function.
847 (arg_int_size_in_bytes): New function.
848 (arg_size_in_bytes): New function.
849 (get_ultimate_context): New function.
850 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
851 TRANSLATION_UNIT_WARN_EMPTY_P.
852 (default_is_empty_record, arg_int_size_in_bytes,
853 arg_size_in_bytes, get_ultimate_context): Declare.
855 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
857 * config/arm/arm.c (cmse_clear_registers): New function.
858 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
859 code by call to cmse_clear_registers.
860 (cmse_nonsecure_entry_clear_before_return): Likewise.
862 2017-11-22 Tamar Christina <tamar.christina@arm.com>
864 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
865 (vdot_s32, vdotq_s32): New.
866 (vdot_lane_u32, vdotq_lane_u32): New.
867 (vdot_lane_s32, vdotq_lane_s32): New.
870 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
873 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
874 for unsigned values with fewer HWIs than the precision.
875 (test_overflow): New function.
876 (wide_int_cc_tests): Call it.
878 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
880 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
881 has the same mode class as Pmode.
883 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
885 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
886 padding_bits_to_clear_ptr.
887 (cmse_nonsecure_entry_clear_before_return): Likewise.
889 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
891 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
892 auto_sbitap instead of integer bitfield to control register needing
895 2017-11-22 Jakub Jelinek <jakub@redhat.com>
897 PR tree-optimization/83044
898 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
899 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
900 that need the upper bound. Subtract offset from
901 get_addr_base_and_unit_offset only if positive and subtract it
902 before division by eltsize rather than after it.
905 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
906 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
907 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
911 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
912 expand_binop with *mul_widen_optab, make sure at least one of the
913 operands doesn't have VOIDmode.
916 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
918 PR rtl-optimization/82044
919 PR tree-optimization/82042
920 * dse.c (record_store): Check for overflow.
921 (check_mem_read_rtx): Properly check for overflow if width == -1, call
922 add_wild_read instead of clear_rhs_from_active_local_stores on
923 overflow and log it into dump_file.
925 2017-11-22 Richard Biener <rguenther@suse.de>
927 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
928 fake edges to exit when looking for a place to insert.
929 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
931 (insert_into_preds_of_block): Commit edge insertion immediately,
932 assert that doesn't require new BBs.
933 (fini_pre): Release expressions.
934 (pass_pre::execute): Shuffle things around a bit, if the fn
935 is too large do not compute AVAIL either as this is really the
938 2017-11-22 Richard Biener <rguenther@suse.de>
940 PR tree-optimization/83089
941 * tree-if-conv.c (pass_if_conversion::execute): If anything
942 changed reset SCEV and free the number of iteration estimates.
944 2017-11-21 Martin Sebor <msebor@redhat.com>
946 PR tree-optimization/82945
947 * calls.h (warn_nonstring_bound): Remove unused function.
949 2017-11-21 Martin Sebor <msebor@redhat.com>
951 PR tree-optimization/82945
952 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
953 * calls.h (maybe_warn_nonstring_arg): Declare new function.
954 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
956 (initialize_argument_information): Call maybe_warn_nonstring_arg.
957 * calls.h (get_attr_nonstring_decl): Declare new function.
958 * doc/extend.texi (attribute nonstring): Update.
959 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
960 get_attr_nonstring_decl and handle it.
961 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
962 detection of nul-termination.
963 (strlen_to_stridx): Change to a pointer.
964 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
965 (pass_strlen::execute): Same.
967 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
969 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
970 for the new option -mprefer-vector-width=[none|128|256|512].
971 * config/i386/i386.c (ix86_target_string): remove old style options
972 -mprefer-avx256 and make -mprefer-avx128 as alias.
973 (ix86_option_override_internal): Apply defaults for the
974 -mprefer-vector-width=[128|256] option.
975 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
976 Implement macros to work with -mprefer-vector-width=.
977 * config/i386/i386.opt: Implemented option
978 -mprefer-vector-width=[none|128|256|512].
979 * doc/invoke.texi: Documentation for
980 -mprefer-vector-width=[none|128|256|512].
982 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
984 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
986 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
988 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
990 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
992 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
994 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
995 * expr.c (expand_expr_real_2): Likewise.
996 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
997 fold_binary_loc): Likewise.
998 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
999 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1000 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1001 MINUS_EXPR transformations.
1002 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1003 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1004 * tree-inline.c (estimate_operator_cost): Likewise.
1005 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1006 op_symbol_code): Likewise.
1007 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1008 * vr-values.c (extract_range_from_binary_expr): Likewise.
1009 * varasm.c (initializer_constant_valid_p_1): Likewise.
1010 * tree.def: New tree code POINTER_DIFF_EXPR.
1012 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1014 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1015 integer suffix to movbe mnemonic.
1016 (*bswaphi2_movbe): Ditto.
1017 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1019 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1022 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1023 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1024 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1026 2017-11-21 Martin Liska <mliska@suse.cz>
1028 * tree-inline.c (expand_call_inline): Remove not needed
1031 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1032 Jakub Jelinek <jakub@redhat.com>
1035 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1036 Remove static keyword from f variable.
1038 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1040 PR tree-optimization/83086
1041 * gimple-ssa-store-merging.c
1042 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1043 rather than n.base_addr.
1045 2017-11-21 Martin Liska <mliska@suse.cz>
1047 PR rtl-optimization/82044
1048 PR tree-optimization/82042
1049 * dse.c (check_mem_read_rtx): Check for overflow.
1051 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1053 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1056 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1058 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1059 and its corresponding call arg location note.
1061 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1062 Andrew Burgess <andrew.burgess@embecosm.com>
1064 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1066 (arc_return_slot_offset): Likewise.
1067 (arc_eh_return_address_location): New declaration.
1068 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1069 (MUST_SAVE_REGISTER): Add exception handler case.
1070 (MUST_SAVE_RETURN_ADDR): Likewise.
1071 (arc_frame_pointer_required): Likewise.
1072 (arc_frame_pointer_needed): New function.
1073 (arc_compute_frame_size): Changed.
1074 (arc_expand_prologue): Likewise.
1075 (arc_expand_epilogue): Likewise.
1076 (arc_initial_elimination_offset): Likewise.
1077 (arc_return_slot_offset): Delete.
1078 (arc_eh_return_address_location): New function.
1079 (arc_builtin_setjmp_frame_value): Likewise.
1080 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1081 (EH_RETURN_STACKADJ_RTX): Define.
1082 (EH_RETURN_HANDLER_RTX): Likewise.
1083 * config/arc/arc.md (eh_return): Delete.
1085 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1087 * print-tree.h (debug_vec_tree): Remove prototype.
1088 * gdbinit.in (pvt): Remove macro.
1090 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1092 PR tree-optimization/83047
1093 * gimple-ssa-store-merging.c
1094 (imm_store_chain_info::output_merged_store): If the loads with the
1095 same vuse are in different basic blocks, for load_gsi pick a load
1096 location that is dominated by the other loads.
1099 * config/i386/i386.c (ix86_memmodel_check): Start
1100 -Winvalid-memory-model diagnostics with lowercase letter.
1103 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1104 set in_cold_section_p to the partition containing loc_list->first.
1105 When seeing loc_list->last_before_switch node, update secname and
1106 perform range_across_switch second partition handling only after that.
1109 * run-rtl-passes.c: Include debug.h.
1110 (run_rtl_passes): Call debug_hooks->assembly_start.
1111 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1115 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1116 OPTAB_DIRECT in calls to expand_simple_binop.
1118 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1121 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1123 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1126 * spellcheck.h (best_match::blithely_get_best_candidate): New
1129 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1131 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1132 with lower case letter.
1134 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1136 * config/i386/i386.md (bswaphi2): New expander.
1137 (*bswaphi2_movbe): New insn pattern.
1138 (bswaphi -> rorhi pepehole2): New peephole pattern.
1140 2017-11-20 Jeff Law <law@redhat.com>
1142 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1143 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1144 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1145 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1146 gimple-ssa-evrp-analyze.h.
1148 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1149 set BB_VISITED here.
1150 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1152 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1154 * tree-predcom.c: Add general comment on Store-Store chains.
1155 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1157 (get_chain_last_ref_at): Rename into...
1158 (get_chain_last_write_at): ...this.
1159 (get_chain_last_write_before_load): New function.
1160 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1161 CT_STORE_STORE when write reference is added.
1162 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1163 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1164 (initialize_root_vars_store_elim_1): Ditto.
1165 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1166 definition is created.
1167 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1168 chain by replacing it with dominant stored value.
1170 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1172 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1174 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1176 * vr-values.c (extract_range_from_binary_expr): Use a full range
1179 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1181 * config/arm/arm.md (R4_REGNUM): Define constant.
1182 (nonsecure_call_internal): Remove r4 clobber.
1183 (nonsecure_call_value_internal): Likewise.
1184 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1185 clobber and resequence match_operands.
1186 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1187 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1188 (nonsecure_call_value_reg_thumb2): Likewise.
1190 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1192 PR tree-optimization/78821
1193 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1194 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1195 to the address of the base rather than the base itself.
1196 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1197 (find_bswap_or_nop_finalize): New function.
1198 (find_bswap_or_nop): Use it.
1199 (bswap_replace): Return a tree rather than bool, change first
1200 argument from gimple * to gimple_stmt_iterator, allow inserting
1201 into an empty sequence, allow ins_stmt to be NULL - then emit
1202 all stmts into gsi. Fix up MEM_REF address gimplification.
1203 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1204 (struct store_immediate_info): Add N and INS_STMT non-static
1206 (store_immediate_info::store_immediate_info): Initialize them
1207 from newly added ctor args.
1208 (merged_store_group::apply_stores): Formatting fixes. Sort by
1210 (stmts_may_clobber_ref_p): For stores call also
1211 refs_anti_dependent_p.
1212 (gather_bswap_load_refs): New function.
1213 (imm_store_chain_info::try_coalesce_bswap): New method.
1214 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1215 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1216 (imm_store_chain_info::output_merged_store): Fail if number of
1217 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1218 and NOP_EXPR rhs_code.
1219 (pass_store_merging::process_store): Compute n and ins_stmt, if
1220 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1221 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1223 (pass_store_merging::execute): Calculate dominators.
1225 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1226 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1227 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1228 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1229 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1230 class pass_optimize_bswap, bswap_replace,
1231 pass_optimize_bswap::execute): Moved to ...
1232 * gimple-ssa-store-merging.c: ... this file.
1233 Include optabs-tree.h.
1234 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1235 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1236 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1237 anonymous namespace, remove static keywords.
1238 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1239 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1241 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1244 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1246 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1248 * vec.h (debug_helper): New function.
1249 (DEFINE_DEBUG_VEC): New macro.
1250 * hash-set.h (debug_helper): New function.
1251 (DEFINE_DEBUG_HASH_SET): New macro.
1252 * cfg.c (debug_slim (edge)): New function.
1253 Call DEFINE_DEBUG_VEC for edges.
1254 Call DEFINE_DEBUG_HASH_SET for edges.
1255 * cfghooks.c (debug_slim (basic_block)): New function.
1256 Call DEFINE_DEBUG_VEC for basic blocks.
1257 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1258 * print-tree.c (debug_slim): New function to handle trees.
1259 Call DEFINE_DEBUG_VEC for trees.
1260 Call DEFINE_DEBUG_HASH_SET for trees.
1261 (debug (vec<tree, va_gc>) &): Remove.
1262 (debug (<vec<tree, va_gc>) *): Remove.
1263 * print-rtl.c (debug_slim): New function to handle const_rtx.
1264 Call DEFINE_DEBUG_VEC for rtx_def.
1265 Call DEFINE_DEBUG_VEC for rtx_insn.
1266 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1267 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1268 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1269 (debug (vec<rtx_insn *> *ptr): Remove.
1270 (debug_insn_vector): Remove.
1271 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1273 2017-11-20 Tom de Vries <tom@codesourcery.com>
1275 PR rtl-optimization/82020
1276 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1277 IF_THEN_ELSE condition.
1279 2017-11-19 Jeff Law <law@redhat.com>
1281 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1282 of degenerates resulting from ignoring an edge.
1284 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1287 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1289 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1292 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1293 for uninitialied counts.
1295 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1298 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1300 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1303 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1305 2017-11-19 Tom de Vries <tom@codesourcery.com>
1307 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1310 2017-11-19 Tom de Vries <tom@codesourcery.com>
1312 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1315 2017-11-19 Tom de Vries <tom@codesourcery.com>
1317 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1320 2017-11-19 Tom de Vries <tom@codesourcery.com>
1322 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1324 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1326 2017-11-19 Tom de Vries <tom@codesourcery.com>
1328 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1330 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1331 semicolon after MCORE_EXPORT_NAME call.
1333 2017-11-19 Tom de Vries <tom@codesourcery.com>
1336 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1337 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1339 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1341 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1342 (cgraph_edge::clone): Cleanup updating of profile.
1343 * ipa-cp.c (update_profiling_info): Likewise.
1344 * ipa-inline-transform.c (inline_transform): Likewise.
1345 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1346 * ipa-split.c (execute_split_functions): Do not split when function
1348 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1349 * profile-count.c (profile_count::dump): Add global0.
1350 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1352 (profile_count::to_sreal_scale): Likewise.
1353 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1354 (profile_count::combine_with_ipa_count): New function.
1355 * profile-count.h (profile_guessed_global0adjusted): New.
1356 (profile_count::adjusted_zero): New.
1357 (profile_count::global0adjusted): New.
1358 (profile_count::combine_with_ipa_count): New.
1359 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1360 correct profile of return block of split functions.
1361 (copy_cfg_body): Remove unused profile_count.
1362 (copy_body): Likewise.
1363 (expand_call_inline): Update.
1364 (tree_function_versioning): Update.
1366 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1368 * hash-set.h (hash_set::empty): New.
1369 * tree-ssa-threadbackward.h: Delete.
1370 * tree-ssa-threadbackward.c (class thread_jumps): New.
1371 Move max_threaded_paths into class.
1372 (fsm_find_thread_path): Remove arguments that are now in class.
1373 (profitable_jump_thread_path): Rename to...
1374 (thread_jumps::profitable_jump_thread_path): ...this.
1375 (convert_and_register_jump_thread_path): Rename to...
1376 (thread_jumps::convert_and_register_current_path): ...this.
1377 (check_subpath_and_update_thread_path): Rename to...
1378 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1379 (register_jump_thread_path_if_profitable): Rename to...
1380 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1381 (handle_phi): Rename to...
1382 (thread_jumps::handle_phi): ...this.
1383 (handle_assignment): Rename to...
1384 (thread_jumps::handle_assignment): ...this.
1385 (fsm_find_control_statement_thread_paths): Rename to...
1386 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1387 (find_jump_threads_backwards): Rename to...
1388 (thread_jumps::find_jump_threads_backwards): ...this.
1389 Initialize path local data.
1390 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1391 from within thread_jumps class.
1392 (pass_early_thread_jumps::execute): Same.
1394 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1396 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1398 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1400 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1401 to canonical location.
1403 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1406 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1408 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1410 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1412 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1413 rcrt1.o%s/grcrt1.o%s for -static-pie.
1415 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1417 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1418 ix86_shift_rotate_cost): Break out from ...
1419 (ix86_rtx_costs): ... here.
1420 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1423 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1425 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1426 when entry block was promoted unlikely.
1427 (estimate_bb_frequencies): Increase frequency scale.
1428 * profile-count.h (profile_count): Export precision info.
1430 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1432 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1433 disturbin profile of entry block.
1435 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1437 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1440 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1442 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1443 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1445 2017-11-17 Jeff Law <law@redhat.com>
1447 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1448 from evrp_dom_walker class. Various methods moved into new class.
1449 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1450 (evrp_range_analyzer::enter): New method.
1451 (evrp_range_analyzer::leave): New method.
1452 (evrp_dom_walker): Remove delegators no longer needed by this class.
1453 Replace vr_values data member with evrp_range_analyzer
1455 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1456 method extracted from evrp_dom_walker::before_dom_children.
1457 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1458 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1460 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1461 Add private copy constructor and move assignment operators.
1462 Privatize methods and class data where trivially possible.
1463 (evrp_dom_walker::cleanup): New function, extracted from
1464 execute_early_vrp. Simplify access to class data.
1466 * vr-values.h (get_output_for_vrp): Prototype.
1467 * vr-values.c (get_output_for_vrp): New function extracted from
1468 vrp_visit_assignment_or_call and extract_range_from_stmt.
1469 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1471 2017-11-17 Luis Machado <luis.machado@linaro.org>
1473 * config/aarch64/aarch64.c
1474 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1475 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1476 tune_params::AUTOPREFETCHER_WEAK.
1478 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1481 * config/arm/arm.c (arm_valid_target_attribute_rec):
1482 Parse "arch=" and "+<ext>".
1483 (arm_valid_target_attribute_tree): Re-init global options.
1484 (arm_option_override): Make non-static.
1485 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1486 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1487 (__ARM_FEATURE_CRC32): Support undef.
1488 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1489 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1491 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1493 * gdbinit.in (break-on-diagnostic): New command.
1495 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1497 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1498 return to indirect jump for EH return if control-flow
1499 protection is enabled. Change explicit 'false' argument in
1500 pro_epilogue_adjust_stack with a value of
1502 * config/i386/i386.md (simple_return_indirect_internal):
1503 Remove SImode restriction to support 64-bit.
1505 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1507 * combine.c (added_notes_insn): New.
1508 (try_combine): Handle added_notes_insn like added_links_insn.
1509 Rewrite return value code.
1510 (distribute_notes): Set added_notes_insn to the earliest insn we added
1513 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1515 PR rtl-optimization/82621
1516 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1517 dest of one of those SETs is unused.
1519 2017-11-17 Richard Biener <rguenther@suse.de>
1522 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1523 * tree-pretty-print.c (dump_generic_node): Handle
1524 annot_expr_parallel_kind.
1525 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1526 * gimplify.c (gimple_boolify): Likewise.
1528 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1530 * config.gcc (extra_headers): Add cet.h for x86 targets.
1531 * config/i386/cet.h: New file.
1532 * doc/install.texi: Add --enable-cet/--disable-cet.
1534 2017-11-17 Richard Biener <rguenther@suse.de>
1536 PR tree-optimization/83017
1537 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1538 (gen_parallel_loop): Properly count iterations.
1539 (parallelize_loops): Handle loop->can_be_parallel independent
1540 of flag_loop_parallelize_all. Make static profitability test match
1542 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1543 * invoke.texi (parloops-min-per-thread): Document.
1545 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1547 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1548 upstreaming review comments.
1550 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1552 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1553 with fast unaligned access.
1554 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1556 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1558 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1559 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1561 2017-11-17 Richard Biener <rguenther@suse.de>
1563 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1564 folding of references.
1566 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1569 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1570 of replacing call to strncmp with corresponding call to strcmp when
1573 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1575 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1576 option prefer-avx256 for skylake-avx512 configuration.
1577 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1578 (get_builtin_code_for_version): Ditto.
1580 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1581 Monk Chiang <sh.chiang04@gmail.com>
1583 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1584 (FIXED_REGISTERS): Reserve more register numbers.
1585 (CALL_USED_REGISTERS): Likewise.
1586 (REG_ALLOC_ORDER): Likewise.
1587 (REG_CLASS_CONTENTS): Likewise.
1588 (REGISTER_NAMES): Likewise.
1590 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1591 Kito Cheng <kito.cheng@gmail.com>
1593 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1595 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1597 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1599 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1600 parameter list for vec_splats.
1602 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1604 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1606 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1608 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1611 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1612 with the required version of automake 1.11.6.
1614 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1616 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1618 (DU_C2_3_power9): Correct reservation combinations.
1619 (FP_DIV_power9, VEC_DIV_power9): New.
1620 (power9-alu): Split out rotate/shift...
1621 (power9-rot): ...to here, correct dispatch resource.
1622 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1624 (power9-fp): Correct latency.
1625 (power9-sdiv): Add div/sqrt resource.
1626 (power9-ddiv): Correct latency, add div/sqrt resource.
1627 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1628 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1630 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1632 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1634 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1635 switch statement mapping KF built-ins to TF built-ins if we don't
1636 have the proper ISA 3.0 assembler support.
1638 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1640 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1641 (lower_emutls_function_body): Do not compute it.
1643 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1645 * ipa-split.c (split_bb_info): Turn time to sreal.
1646 (split_point): Likewise.
1647 (dump_split_point): Likewise.
1648 (fine_split_points): Likewise.
1649 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1651 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1653 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1656 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1658 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1660 (expensive_function_p): Remove useless assert.
1661 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1663 2017-11-16 Martin Sebor <msebor@redhat.com>
1665 PR tree-optimization/82588
1666 PR tree-optimization/82583
1667 * tree-vrp.c (check_array_ref): Handle flexible array members,
1668 string literals, and inner indices.
1669 (search_for_addr_array): Add detail to diagnostics.
1671 2017-11-16 Nathan Sidwell <nathan@acm.org>
1675 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1676 (SET_DECL_ASSEMBLER_NAME): Forward to
1677 overwrite_decl_assembler_name.
1678 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1679 (overwrite_decl_assembler_name): Declare.
1680 * tree.c (overwrite_decl_assembler_name): New.
1681 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1682 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1683 (LANG_HOOKS_INITIALIZER): Add it.
1684 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1685 * langhooks.c (lhd_set_decl_assembler_name): Use
1686 SET_DECL_ASSEMBLER_NAME.
1687 (lhd_overwrite_decl_assembler_name): Default implementation.
1689 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1690 Jackson Woodruff <jackson.woodruff@arm.com>
1692 PR tree-optimization/71026
1693 * match.pd: Canonicalize constant multiplies in division.
1695 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1697 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
1698 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
1699 Remove OPT_fomit_frame_pointer entry.
1700 * common/config/arc/arc-common.c: Likewise.
1701 * common/config/arm/arm-common.c: Likewise.
1702 * common/config/avr/avr-common.c: Likewise.
1703 * common/config/c6x/c6x-common.c: Likewise.
1704 * common/config/cr16/cr16-common.c: Likewise.
1705 * common/config/cris/cris-common.c: Likewise.
1706 * common/config/epiphany/epiphany-common.c: Likewise.
1707 * common/config/fr30/fr30-common.c: Likewise.
1708 * common/config/frv/frv-common.c: Likewise.
1709 * common/config/ia64/ia64-common.c: Likewise.
1710 * common/config/iq2000/iq2000-common.c: Likewise.
1711 * common/config/lm32/lm32-common.c: Likewise.
1712 * common/config/m32r/m32r-common.c: Likewise.
1713 * common/config/mcore/mcore-common.c: Likewise.
1714 * common/config/microblaze/microblaze-common.c: Likewise.
1715 * common/config/mips/mips-common.c: Likewise.
1716 * common/config/mmix/mmix-common.c: Likewise.
1717 * common/config/mn10300/mn10300-common.c: Likewise.
1718 * common/config/nios2/nios2-common.c: Likewise.
1719 * common/config/pa/pa-common.c: Likewise.
1720 * common/config/pdp11/pdp11-common.c: Likewise.
1721 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
1722 * common/config/riscv/riscv-common.c: Likewise.
1723 * common/config/rs6000/rs6000-common.c: Likewise.
1724 * common/config/rx/rx-common.c: Likewise.
1725 * common/config/s390/s390-common.c: Likewise.
1726 * common/config/sh/sh-common.c: Likewise.
1727 * common/config/sparc/sparc-common.c: Likewise.
1728 * common/config/tilegx/tilegx-common.c: Likewise.
1729 * common/config/tilepro/tilepro-common.c: Likewise.
1730 * common/config/v850/v850-common.c: Likewise.
1731 * common/config/visium/visium-common.c: Likewise.
1732 * common/config/xstormy16/xstormy16-common.c: Likewise.
1733 * common/config/xtensa/xtensa-common.c: Likewise.
1734 * invoke.texi (-fomit-frame-pointer): Update documentation.
1736 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1738 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
1740 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1742 * cfg.c (scale_bbs_frequencies_int,
1743 cale_bbs_frequencies_gcov_type): Remove.
1744 * cfg.h (scale_bbs_frequencies_int,
1745 cale_bbs_frequencies_gcov_type): Remove.
1747 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1749 * tree-ssa-loop-manip.c
1750 (scale_dominated_blocks_in_loop): Update to profile counts.
1751 (tree_transform_and_unroll_loop): Likewise.
1753 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1755 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
1756 scale_bbs_frequencies_int.
1758 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1760 * final.c (compute_alignments): Use counts rather than frequencies.
1762 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1764 * cfgloopanal.c: Include sreal.h
1765 (average_num_loop_insns): Use counts and sreal for accounting.
1767 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1769 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
1772 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1773 Kito Cheng <kito.cheng@gmail.com>
1775 * config/nds32/constraints.md: Provide more constraints.
1776 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
1777 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
1778 support constraints usage.
1780 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1782 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
1784 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
1785 Kito Cheng <kito.cheng@gmail.com>
1787 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
1788 * config/nds32/nds32.opt: Refine the layout.
1789 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
1790 TARGET_EXT_STRING): Support new options.
1791 * config/nds32/nds32.h: Likewise.
1792 * config/nds32/nds32.md: Likewise.
1793 * config/nds32/nds32-predicates.c: Likewise.
1794 * config/nds32/constraints.md: Likewise.
1795 * common/config/nds32/nds32-common.c: Likewise.
1797 2017-11-16 Julia Koval <julia.koval@intel.com>
1800 * config/i386/gfniintrin.h: Add sse check.
1801 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
1803 2017-11-16 Julia Koval <julia.koval@intel.com>
1805 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
1806 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
1807 (ix86_handle_option): Handle -mavx512vbmi2.
1808 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
1809 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1810 * config/i386/i386-c.c (__AVX512VBMI2__): New.
1811 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
1812 (ix86_valid_target_attribute_inner_p): Ditto.
1813 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
1814 * config/i386/i386.opt (mavx512vbmi2): New option.
1815 * doc/invoke.texi: Add new option.
1817 2017-11-16 Julia Koval <julia.koval@intel.com>
1819 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
1820 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
1821 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
1822 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
1823 _mm512_gf2p8mul_epi8): New intrinsics.
1824 * config/i386/i386-builtin-types.def
1825 (V64QI_FTYPE_V64QI_V64QI): New type.
1826 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
1827 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
1828 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
1829 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
1830 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
1831 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1833 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
1835 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
1836 explicitly as a stream of bytes.
1838 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1840 * config/rs6000/altivec.h (vec_xst_be): New #define.
1841 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
1842 and externalize from *altivec_vperm_<mode>_internal.
1843 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
1845 (XL_BE_V8HI): Likewise.
1846 (XL_BE_V4SI): Likewise.
1847 (XL_BE_V4SI): Likewise.
1848 (XL_BE_V2DI): Likewise.
1849 (XL_BE_V4SF): Likewise.
1850 (XL_BE_V2DF): Likewise.
1851 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
1852 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
1853 all array entries with these keys: VSX_BUILTIN_VEC_XL,
1854 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
1855 VSX_BUILTIN_VEC_XST_BE.
1856 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
1857 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
1859 (altivec_init_builtins): Replace conditional calls to def_builtin
1860 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
1861 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
1862 with unconditional calls. Remove calls to def_builtin for
1863 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
1864 __builtin_vec_xst_be.
1865 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
1866 to define_expand, and add alternate RTL generation for P8.
1867 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
1868 vsx_ld_elemrev_v8hi.
1869 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
1870 add alternate RTL generation for P8.
1871 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
1872 vsx_ld_elemrev_v16qi.
1873 (vsx_st_elemrev_v8hi): Convert define_insn
1874 to define_expand, and add alternate RTL generation for P8.
1875 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
1876 vsx_st_elemrev_v8hi.
1877 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
1878 add alternate RTL generation for P8.
1879 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
1880 vsx_st_elemrev_v16qi.
1882 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1885 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1886 TARGET_AVX512ER check.
1887 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1888 neither -mzeroupper nor -mno-zeroupper is used and
1889 TARGET_EMIT_VZEROUPPER is set.
1890 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1891 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1893 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1895 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1896 folding of vector compares.
1897 (fold_build_vec_cmp): New helper function.
1898 (fold_compare_helper): New helper function.
1899 (builtin_function_type): Add compare builtins to the list of functions
1900 having unsigned arguments. Cosmetic updates to comment indentation.
1901 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1902 the not+eq combination.
1904 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1906 PR tree-optimization/82726
1907 PR tree-optimization/70754
1908 * tree-predcom.c (order_drefs_by_pos): New function.
1909 (combine_chains): Move code setting has_max_use_after to...
1910 (try_combine_chains): ...here. New parameter. Sort combined chains
1911 according to position information.
1912 (tree_predictive_commoning_loop): Update call to above function.
1913 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1915 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1917 PR tree-optimization/82726
1919 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1921 PR tree-optimization/70754
1922 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1923 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1924 combined stmt before it if not NULL.
1925 (combine_chains): Process refs reversely and compute dominance point
1929 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1931 PR tree-optimization/79663
1932 * tree-predcom.c (combine_chains): Process refs in reverse order
1933 only for ZERO length chains, and add explaining comment.
1935 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1937 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1938 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1939 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1941 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1943 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1945 2017-11-15 Martin Liska <mliska@suse.cz>
1947 * tree-cfg.c (pass_warn_function_return::execute):
1948 Compare warn_return_type for greater than zero.
1950 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1954 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1955 to return true on Xeon and not on Xeon Phi.
1956 (ix86_check_avx256_register): Changed to ...
1957 (ix86_check_avx_upper_register): ... this. Add extra check for
1958 VALID_AVX512F_REG_OR_XI_MODE.
1959 (ix86_avx_u128_mode_needed): Changed
1960 ix86_check_avx256_register to ix86_check_avx_upper_register.
1961 (ix86_check_avx256_stores): Changed to ...
1962 (ix86_check_avx_upper_stores): ... this. Changed
1963 ix86_check_avx256_register to ix86_check_avx_upper_register.
1964 (ix86_avx_u128_mode_after): Changed
1965 avx_reg256_found to avx_upper_reg_found. Changed
1966 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1967 (ix86_avx_u128_mode_entry): Changed
1968 ix86_check_avx256_register to ix86_check_avx_upper_register.
1969 (ix86_avx_u128_mode_exit): Ditto.
1970 * config/i386/i386.h: (host_detect_local_cpu): New define.
1972 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1974 * config/arm/xgene1.md (xgene1): Split into automatons
1975 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1976 (xgene1_f_load): Adjust reservations and/or types.
1977 (xgene1_f_store): Likewise.
1978 (xgene1_load_pair): Likewise.
1979 (xgene1_store_pair): Likewise.
1980 (xgene1_fp_load1): Likewise.
1981 (xgene1_load1): Likewise.
1982 (xgene1_store1): Likewise.
1983 (xgene1_move): Likewise.
1984 (xgene1_alu): Likewise.
1985 (xgene1_simd): Likewise.
1986 (xgene1_bfm): Likewise.
1987 (xgene1_neon_load1): Likewise.
1988 (xgene1_neon_store1): Likewise.
1989 (xgene1_neon_logic): Likewise.
1990 (xgene1_neon_st1): Likewise.
1991 (xgene1_neon_ld1r): Likewise.
1992 (xgene1_alu_cond): Added.
1993 (xgene1_shift_reg): Likwise.
1994 (xgene1_bfx): Likewise.
1995 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1997 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2000 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2002 (can_widen_mult_without_libcall): New function.
2003 (expand_mul_overflow): If only checking unsigned mul overflow,
2004 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2005 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2006 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2007 (expand_DIVMOD): Formatting fix.
2008 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2009 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2010 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2012 PR tree-optimization/82977
2013 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2014 constructed temporary to strlen_to_stridx.put.
2016 2017-11-15 Martin Liska <mliska@suse.cz>
2018 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2019 * configure: Regenerate.
2021 2017-11-15 Martin Liska <mliska@suse.cz>
2024 * config/sh/sh-mem.cc: Use proper probability for
2027 2017-11-14 Jeff Law <law@redhat.com>
2029 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2030 the red zone for stack_clash_protection_final_dynamic_probe targets
2031 when the total dynamic stack size is zero bytes.
2033 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2034 blocks is post order.
2036 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2038 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2039 * final.c (rest_of_clean_state): Set it for the
2040 -fcompare-debug dump.
2041 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2042 class when TDF_COMPARE_DEBUG is set.
2044 * dwarf2out.c (gen_producer_string): Discard
2047 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2050 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2051 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2052 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2053 (__TGMATH_REAL_2_3): Remove macros.
2054 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2055 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2057 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2059 (remquo): Define using __TGMATH_REAL_3.
2061 2017-11-14 Jeff Law <law@redhat.com>
2063 * vr-values.c: New file with contents extracted from tree-vrp.c.
2064 * Makefile.in (OBJS): Add vr-values.o
2065 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2066 (set_value_range, set_and_canonicalize_value_range): Likewise.
2067 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2068 (value_range_constant_singleton, symbolic_range_p): Likewise.
2069 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2070 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2071 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2072 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2073 (find_case_label_range, find_case_label_index): Likewise.
2074 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2075 (range_int_cst_singleton_p, value_inside_range): Likewise.
2076 (get_single_symbol): Likewise.
2077 (switch_update): Move structure definition here.
2078 (to_remove_edges, to_update_switch_stmts): Provide externs.
2079 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2080 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2081 (vrp_val_is_min, set_value_range): Likewise.
2082 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2083 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2084 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2085 (range_is_nonnull, range_int_cst_p): Likewwise.
2086 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2087 (get_single_symbol, operand_less_p): Likewise
2088 (compare_values_warnv, compare_values): Likewise.
2089 (value_inside_range, value_range_constant_singleton): Likewise.
2090 (zero_nonzero_bitgs_from_vr): Likewise.
2091 (extract_range_from_binary_expr_1): Likewise.
2092 (overflow_comparison_p): Likewise.
2093 (to_remove_edges, to_update_switch_stmts): Likewise.
2094 (find_case_label-index, find_case_label_range): Likewise.
2095 (switch_update, set_value_range_to_nonnegative): Remove.
2096 (set_value_range_to_truthvalue): Likewise.
2097 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2098 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2099 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2100 (find_case_label_ranges, test_for_singularity): Likewise.
2101 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2102 (x_vr_values): Move to its remaining use site.
2104 2017-11-10 Jeff Law <law@redhat.com>
2106 * vr-values.h (VR_INITIALIZER): Move #define here.
2107 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2108 * Makefile.in (OBJS): Add tree-evrp.o
2109 * tree-vrp.h (assert_info): Move structure definition here.
2110 (set_value_range_to_varying): Prototype.
2111 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2112 (infer_value_range, register_edge_assert_for): Likewise.
2113 (stmt_interesting_for_vrp): Likewise.
2114 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2115 (set_value_range_to_varying): No longer static.
2116 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2117 (infer_value_range, register_edge_assert_for): Likewise.
2119 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2121 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2122 generate the XXBRD instruction.
2124 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2125 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2126 as being compatible if -mabi=ieeelongdouble.
2127 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2128 to setup float128 built-ins with hardware support.
2129 (BU_FLOAT128_HW_2): Likewise.
2130 (BU_FLOAT128_HW_3): Likewise.
2131 (BU_FLOAT128_HW_VSX_1): Likewise.
2132 (BU_FLOAT128_HW_VSX_2): Likewise.
2133 (scalar_extract_expq): Change float128 built-in functions to
2134 accommodate having both KFmode and TFmode functions. Use the
2135 KFmode variant as the default.
2136 (scalar_extract_sigq): Likewise.
2137 (scalar_test_neg_qp): Likewise.
2138 (scalar_insert_exp_q): Likewise.
2139 (scalar_insert_exp_qp): Likewise.
2140 (scalar_test_data_class_qp): Likewise.
2141 (sqrtf128_round_to_odd): Delete processing the round to odd
2142 built-in functions as special built-in functions, and define them
2143 as float128 built-ins. Use the KFmode variant as the default.
2144 (truncf128_round_to_odd): Likewise.
2145 (addf128_round_to_odd): Likewise.
2146 (subf128_round_to_odd): Likewise.
2147 (mulf128_round_to_odd): Likewise.
2148 (divf128_round_to_odd): Likewise.
2149 (fmaf128_round_to_odd): Likewise.
2150 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2151 support for KFmode and TFmode xststdcqp calls.
2152 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2153 point, switch the built-in handlers for the get/set float128
2154 exponent, get float128 mantissa, float128 test built-ins, and the
2155 float128 round to odd built-in functions. Eliminate creating the
2156 float128 round to odd built-in functions as special built-ins.
2157 (rs6000_init_builtins): Eliminate special creation of the float128
2158 round to odd built-in functions.
2159 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2160 function insns to support both TFmode and KFmode variants.
2161 (xsxsigqp_<mode>): Likewise.
2162 (xsiexpqpf_<mode>): Likewise.
2163 (xsiexpqp_<mode>): Likewise.
2164 (xststdcqp_<mode>): Likewise.
2165 (xststdcnegqp_<mode>): Likewise.
2166 (xststdcqp_<mode>): Likewise.
2168 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2170 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2171 unused path_in_freq_ptr parameter.
2172 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2174 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2176 * ipa-inline.c (edge_badness): Dump sreal frequency.
2177 (compute_inlined_call_time): Match natural implementaiton ...
2178 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2179 forgotten division by CGRAPH_FREQ_BASE.
2181 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2183 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2184 Solaris 11. Update comment.
2185 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2187 * config/sol2.h (STARTFILE_SPEC): Likewise.
2188 * configure: Regenerate.
2190 2017-11-14 Carl Love <cel@us.ibm.com>
2192 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2193 le_ and be_ prefixes to swap* variables. Remove
2194 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2197 2017-11-14 Jason Merrill <jason@redhat.com>
2199 Support GTY((cache)) on hash_map.
2200 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2201 (ggc_cache_remove): Override it instead of ggc_mx.
2202 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2203 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2204 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2205 (simple_cache_map_traits): Override maybe_mx.
2206 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2207 (hash_map): Friend gt_cleare_cache.
2208 (gt_cleare_cache): New.
2209 * tree.h (tree_cache_traits): New hash_map traits class.
2210 (tree_cache_map): New typedef.
2212 2017-11-14 Richard Biener <rguenther@suse.de>
2214 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2215 paramter and handling.
2216 (cleanup_control_flow_bb): Likewise.
2217 (cleanup_control_flow_pre): New helper performing a DFS walk
2218 to call cleanup_control_flow_bb in PRE order.
2219 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2220 via cleanup_control_flow_pre.
2222 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2224 * config/aarch64/aarch64-simd.md
2225 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2226 (*aarch64_simd_bsl<mode>_alt): Likewise.
2227 (aarch64_simd_bsldi_internal): New.
2228 (aarch64_simd_bsldi_alt): Likewise.
2230 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2232 * tracer.c (better_p): Do not compare frequencies.
2233 * reg-stack.c (better_edge): Likewise.
2234 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2237 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2239 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2240 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2241 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2242 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2243 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2244 * predict.c (maybe_hot_frequency_p): Inline to ...
2245 (maybe_hot_count_p): ... here; rewrite to counts.
2246 (counts_to_freqs): Rename to ...
2247 (update_max_bb_count): ... this one.
2248 (expensive_function_p): Use counts.
2249 (estimate_bb_frequencies): Update.
2250 (rebuild_frequencies): Update.
2251 * predict.h (counts_to_freqs): Rename to ...
2252 (update_max_bb_count): ... this one.
2253 * profile.c (compute_branch_probabilities): Add debug info
2254 * tree-inline.c (expand_call_inline): Update debug info.
2255 (optimize_inline_calls): Use update_max_bb_count..
2256 (tree_function_versioning): Use update_max_bb_count..
2257 * value-prof.c (gimple_value_profile_transformations):
2258 Do not use update_max_bb_count.
2260 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2262 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2263 always use frequencies.
2265 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2267 * bb-reorder.c: Remove frequencies from comments.
2268 (better_edge_p): Use profile counts.
2269 (find_traces): Dump profile counts.
2270 (rotate_loop): Use profile counts.
2271 (find_traces_1_round): Likewise.
2272 (connect_better_edge_p): Use counts instead of probabilities for
2274 (copy_bb_p): Drop early check for non-0 frequency.
2275 (sanitize_hot_paths): Update comments.
2277 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2279 * ipa-split.c (struct split_point): Add count.
2280 (consider_split): Do not compute incoming frequency; compute incoming
2281 count and store it to split_point.
2282 (split_function): Set count of the call to split part correctly.
2284 2017-11-13 Carl Love <cel@us.ibm.com>
2286 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2288 2017-11-13 Tom Tromey <tom@tromey.com>
2290 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2292 2017-11-13 Carl Love <cel@us.ibm.com>
2294 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2295 Add support for builtins:
2296 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2297 vector {un,}signed {char,int,short},
2298 vector {un,}signed {char,int,short}) arguments.
2299 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2300 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2301 Add BU_P9V_AV_2 expansions for the builtins.
2302 * config/rs6000/altivec.h (vec_first_match_index,
2303 vec_first_mismatch_index, vec_first_match_or_eos_index,
2304 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2305 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2306 new extern declaration.
2307 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2308 * config/rs6000/vsx.md (first_match_index_<mode>,
2309 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2310 first_mismatch_or_eos_index_<mode>): Add define expand.
2311 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2312 * doc/extend.texi: Update the built-in documenation file for the new
2315 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2317 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2318 into the min/max operations for _Float<N> and _Float<N>X types.
2320 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2323 * dwarf2out.c (do_eh_frame): New static variable.
2324 (dwarf2out_begin_prologue): Set it.
2325 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2327 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2329 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2331 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2333 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2336 * debug.h (dwarf2out_do_eh_frame): Declare.
2337 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2338 (dwarf2out_do_frame): Use it.
2339 (dwarf2out_do_cfi_asm): Likewise.
2340 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2341 (dwarf2out_assembly_start): Likewise.
2342 (dwarf2out_begin_prologue): Fix comment.
2343 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2344 if the target needs either debug or unwind DWARF2 info.
2345 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2346 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2347 -ftrapping-math, -ftrapv and -fwrapv.
2349 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2351 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2352 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2353 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2354 (estimate_edge_size_and_time): Likewise.
2355 (ipa_merge_fn_summary_after_inlining): Likewise.
2356 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2357 (compute_uninlined_call_time): Use sreal_frequency.
2358 (compute_inlined_call_time): Likewise.
2359 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2360 * profile-count.c: Include sreal.h.
2361 (profile_count::to_sreal_scale): New.
2362 * profile-count.h: Forward declare sreal.
2363 (profile_count::to_sreal_scale): Declare.
2365 2017-11-13 Nathan Sidwell <nathan@acm.org>
2367 * diagnostic.c (maybe_line_and_column): New.
2368 (diagnostic_get_location_text): Use it.
2369 (diagnostic_report_current_module): Likewise.
2370 (test_diagnostic_get_location_text): Add tests.
2372 2017-11-13 Luis Machado <luis.machado@linaro.org>
2374 * doc/md.texi (Specifying processor pipeline description): Fix
2375 incorrect latency for the div instruction example.
2377 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2379 PR tree-optimization/78821
2380 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2381 that bit_not_p is the same.
2382 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2383 (split_group): Count precisely bit_not_p bits in each statement.
2384 (invert_op): New function.
2385 (imm_store_chain_info::output_merged_store): Use invert_op to
2386 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2387 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2389 2017-11-13 Martin Liska <mliska@suse.cz>
2391 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2392 (struct source_info): Likewise.
2393 (add_branch_counts): Likewise.
2394 (add_line_counts): Likewise.
2395 (function_summary): Likewise.
2396 (output_intermediate_line): Likewise.
2397 (generate_results): Likewise.
2399 2017-11-13 Martin Liska <mliska@suse.cz>
2401 * gcov.c (struct block_info): Remove typedef for block_t.
2402 (struct line_info): Likewise.
2403 (line_info::has_block): Likewise.
2404 (EXIT_BLOCK): Likewise.
2405 (unblock): Likewise.
2406 (circuit): Likewise.
2407 (get_cycles_count): Likewise.
2408 (process_file): Likewise.
2409 (read_graph_file): Likewise.
2410 (solve_flow_graph): Likewise.
2411 (find_exception_blocks): Likewise.
2412 (add_line_counts): Likewise.
2413 (accumulate_line_info): Likewise.
2414 (output_line_details): Likewise.
2416 2017-11-13 Martin Liska <mliska@suse.cz>
2418 * gcov.c (struct arc_info): Remove typedef for arc_t.
2419 (struct line_info): Likewise.
2420 (add_branch_counts): Likewise.
2421 (output_branch_count): Likewise.
2422 (function_info::~function_info): Likewise.
2423 (circuit): Likewise.
2424 (output_intermediate_line): Likewise.
2425 (read_graph_file): Likewise.
2426 (solve_flow_graph): Likewise.
2427 (find_exception_blocks): Likewise.
2428 (add_line_counts): Likewise.
2429 (accumulate_line_info): Likewise.
2430 (output_line_details): Likewise.
2431 (output_function_details): Likewise.
2433 2017-11-13 Martin Liska <mliska@suse.cz>
2435 * gcov.c (struct function_info): Remove typedef for function_t.
2436 (struct source_info): Likewise.
2437 (source_info::get_functions_at_location): Likewise.
2438 (solve_flow_graph): Likewise.
2439 (find_exception_blocks): Likewise.
2440 (add_line_counts): Likewise.
2441 (output_intermediate_file): Likewise.
2442 (process_file): Likewise.
2443 (generate_results): Likewise.
2444 (release_structures): Likewise.
2445 (read_graph_file): Likewise.
2446 (read_count_file): Likewise.
2447 (accumulate_line_counts): Likewise.
2448 (output_lines): Likewise.
2450 2017-11-13 Martin Liska <mliska@suse.cz>
2452 * gcov.c (function_info::function_info): Remove num_counts
2453 and add vector<gcov_type>.
2454 (function_info::~function_info): Use the vector.
2455 (process_file): Likewise.
2456 (read_graph_file): Likewise.
2457 (read_count_file): Likewise.
2458 (solve_flow_graph): Likewise.
2460 2017-11-13 Martin Liska <mliska@suse.cz>
2462 * gcov.c (function_info::is_artificial): New function.
2463 (process_file): Erase all artificial early.
2464 (generate_results): Skip as all artificial are already
2467 2017-11-13 Martin Liska <mliska@suse.cz>
2469 * gcov.c (read_graph_file): Store to global vector of functions.
2470 (read_count_file): Iterate the vector.
2471 (process_file): Likewise.
2472 (generate_results): Likewise.
2473 (release_structures): Likewise.
2475 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2477 PR tree-optimization/82954
2478 * gimple-ssa-store-merging.c
2479 (imm_store_chain_info::coalesce_immediate_stores): If
2480 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2482 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2484 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2485 Upddate call to ENDIAN_LANE_N.
2486 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2487 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2488 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2489 and use aarch64_endian_lane_rtx.
2490 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2492 2017-11-12 Tom de Vries <tom@codesourcery.com>
2494 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2496 2017-11-12 Tom de Vries <tom@codesourcery.com>
2498 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2500 2017-11-12 Tom de Vries <tom@codesourcery.com>
2502 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2504 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2505 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2506 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2508 2017-11-11 Martin Sebor <msebor@redhat.com>
2511 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2514 * prefic.c (translate_name): Replace strncpy with memcpy to
2515 avoid -Wstringop-truncation.
2517 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2519 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2521 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2523 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2525 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2527 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2530 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2532 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2533 enable generating XXBRH if the value is in a vector register.
2534 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2535 value is in a vector register.
2536 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2537 register to register bswap64's instead of doing the GPR sequence
2538 used on previous machines.
2539 (bswapdi2_xxbrd): New insn.
2540 (bswapdi2_reg): Disallow on ISA 3.0.
2541 (register to register bswap64 splitter): Do not split the insn on
2542 ISA 3.0 systems that use XXBRD.
2544 2017-11-10 Martin Sebor <msebor@redhat.com>
2547 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2549 (find_subframework_file): Same.
2551 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2553 * auto-profile.c (afdo_indirect_call): Drop frequency.
2554 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2555 (cgraph_node::create_edge): Drop frequency argument.
2556 (cgraph_node::create_indirect_edge): Drop frequency argument.
2557 (cgraph_edge::make_speculative): Drop frequency arguments.
2558 (cgraph_edge::resolve_speculation): Do not update frequencies
2559 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2560 (cgraph_node::dump): Check consistency in IPA mode.
2561 (cgraph_edge::maybe_hot_p): Use IPA counter.
2562 (cgraph_edge::verify_count_and_frequency): Rename to ...
2563 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2564 (cgraph_node::verify_node): Update.
2565 * cgraph.h (struct cgraph_edge): Drop frequency.
2566 (cgraph_edge::frequency): New function.
2567 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2569 (cgraph_edge::rebuild_edges): Likewise.
2570 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2571 (duplicate_thunk_for_node): Do not pass frequency.
2572 (cgraph_node::create_clone): Scale only counts.
2573 (cgraph_node::create_virtual_clone): Do not pass frequency.
2574 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2575 (cgraph_node::create_version_clone): Do not pass frequency.
2576 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2577 (cgraph_node::expand_thunk): Do not pass frequency.
2578 (cgraph_node::create_wrapper): Do not pass frequency.
2579 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2581 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2582 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2583 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2584 (gather_caller_stats): Use frequency function.
2585 (ipcp_cloning_candidate_p): Use frequency function.
2586 (ipcp_propagate_stage): Use frequency function.
2587 (get_info_about_necessary_edges): Use frequency function.
2588 (update_profiling_info): Update only IPA profile.
2589 (update_specialized_profile): Use frequency functoin.
2590 (perhaps_add_new_callers): Update only IPA profile.
2591 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2592 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2593 (dump_ipa_call_summary): Use frequency function.
2594 (estimate_edge_size_and_time): Use frequency function.
2595 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2596 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2597 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2598 (update_noncloned_counts): ... ths one; scale counts only.
2599 (clone_inlined_nodes): Do not scale frequency.
2600 (inline_call): Do not pass frequency.
2601 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2602 (compute_inlined_call_time): Use IPA profile.
2603 (want_inline_small_function_p): Use IPA profile.
2604 (want_inline_self_recursive_call_p): Use IPA profile.
2605 (edge_badness): Use IPA profile.
2606 (lookup_recursive_calls): Use IPA profile.
2607 (recursive_inlining): Do not pass frequency.
2608 (resolve_noninline_speculation): Do not update frequency.
2609 (inline_small_functions): Collect max of IPA profile.
2610 (dump_overall_stats): Dump IPA porfile.
2611 (dump_inline_stats): Dump IPA porfile.
2612 (ipa_inline): Collect IPA stats.
2613 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2614 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2615 (ipa_propagate_frequency): Use frequency function.
2616 (ipa_profile): Cleanup.
2617 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2618 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2619 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2620 (input_node): Do not stream frequency.
2621 (input_edge): Do not stream frequency.
2622 (merge_profile_summaries): Scale only IPA profiles.
2623 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2624 * predict.c (drop_profile): Do not recompute frequency.
2625 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2626 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2627 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2628 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2629 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2630 * tree-inline.c (copy_bb): Do not scale frequency.
2631 (expand_call_inline): Do not scale frequency.
2632 (tree_function_versioning): Do not scale frequency.
2633 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2635 2017-11-10 Julia Koval <julia.koval@intel.com>
2637 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2638 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2639 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2640 (_mm256_maskz_gf2p8affine_epi64_epi8)
2641 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2642 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2643 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2644 (__builtin_ia32_vgf2p8affineqb_v32qi)
2645 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2646 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2648 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2652 (arm_option_override): Refactor.
2653 (arm_option_reconfigure_globals): New.
2654 (arm_options_perform_arch_sanity_checks): New.
2655 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2657 (arm_options_perform_arch_sanity_checks): Likewise
2659 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2661 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2662 (power9-qpmul): New.
2663 * rs6000/rs6000.md ("type" attr): Add qmul.
2664 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2665 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2666 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2668 2017-11-10 Martin Sebor <msebor@redhat.com>
2671 * builtins.c (compute_objsize): Handle arrays that
2672 compute_builtin_object_size likes to fail for. Make extern.
2673 * builtins.h (compute_objsize): Declare.
2674 (check_strncpy_sizes): New function.
2675 (expand_builtin_strncpy): Call check_strncpy_sizes.
2676 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2677 -Wstringop-truncation.
2678 (gimple_fold_builtin_strncat): Same.
2679 * gimple.c (gimple_build_call_from_tree): Set call location.
2680 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2681 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2682 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2683 (handle_builtin_strlen): Use strlen_to_stridx.
2684 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2686 Use strlen_to_stridx.
2687 (pass_strlen::execute): Release strlen_to_stridx.
2688 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2689 (-Wstringop-truncation): Document new option.
2691 2017-11-10 Martin Liska <mliska@suse.cz>
2693 PR gcov-profile/82702
2694 * gcov.c (main): Handle intermediate files in a different
2696 (get_gcov_intermediate_filename): New function.
2697 (output_gcov_file): Remove support of intermediate files.
2698 (generate_results): Allocate intermediate file.
2699 (release_structures): Clean-up properly fn_end.
2700 (output_intermediate_file): Start iterating with line 1.
2702 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2704 PR tree-optimization/82929
2705 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2706 ops_swapped_p non-static data member.
2707 (store_immediate_info::store_immediate_info): Clear it.
2708 (imm_store_chain_info::coalesce_immediate_stores): If swapping
2709 ops set ops_swapped_p.
2710 (count_multiple_uses): Handle ops_swapped_p.
2712 2017-11-10 Martin Liska <mliska@suse.cz>
2714 * coverage.c (coverage_init): Stream information about
2715 support of has_unexecuted_blocks.
2716 * doc/gcov.texi: Document that.
2717 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
2718 * gcov.c (read_graph_file): Likewise.
2719 (output_line_beginning): Fix a small issue with
2722 2017-11-10 Bin Cheng <bin.cheng@arm.com>
2724 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
2725 reference of trivial component.
2727 2017-11-10 Jakub Jelinek <jakub@redhat.com>
2730 * gimple-ssa-store-merging.c
2731 (pass_store_merging::terminate_all_aliasing_chains): For
2732 gimple_store_p stmts also call refs_output_dependent_p.
2734 PR rtl-optimization/82913
2735 * compare-elim.c (try_merge_compare): Punt if def_insn is not
2738 2017-11-09 Jeff Law <law@redhat.com>
2740 * vr-values.h: New file with vr_values class.
2741 * tree-vrp.c: Include vr-values.h
2742 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
2743 data objects into the vr_values class.
2744 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
2745 (get_value_range): Make it a member function within vr_values class.
2746 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
2747 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
2748 (op_with_constant_singleton_value_range): Likewise.
2749 (extract_range_for_var_from_comparison_expr): Likewise.
2750 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
2751 (extract_range_from_binary_expr): Likewise.
2752 (extract_range_from_unary_expr): Likewise.
2753 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
2754 (check_for_binary_op_overflow, extract_range_basic): Likewise.
2755 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
2756 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
2757 (compare_name_with_value, compare_names): Likewise.
2758 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
2759 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
2760 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
2761 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
2762 (extract_range_from_phi_node): Likewise.
2763 (simplify_truth_ops_using_ranges): Likewise.
2764 (simplify_div_or_mod_using_ranges): Likewise.
2765 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
2766 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
2767 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
2768 (simplify_float_conversion_using_ranges): Likewise.
2769 (simplify_internal_call_using_ranges): Likewise.
2770 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
2771 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
2772 poitner via x_vr_values for calls into gimple folder.
2773 (vrp_initialize_lattice): Make this the vr_values ctor.
2774 (vrp_free_lattice): Make this the vr_values dtor.
2775 (set_vr_value): New function.
2776 (class vrp_prop): Add vr_values data member. Add various member
2777 functions as well as member functions that delegate to vr_values.
2778 (check_array_ref): Make a member function within vrp_prop class.
2779 (search_for_addr_array, vrp_initialize): Likewise.
2780 (vrp_finalize): Likewise. Revamp to avoid direct access to
2781 vr_value, values_propagated, etc.
2782 (check_array_bounds): Extract vrp_prop class instance pointer from
2783 walk info structure. Use it to call member functions.
2784 (check_all_array_refs): Make a member function within vrp_prop class.
2785 Smuggle class instance pointer via walk info structure.
2786 (x_vr_values): New local static.
2787 (vrp_valueize): Use x_vr_values to get class instance.
2788 (vr_valueize_1): Likewise.
2789 (class vrp_folder): Add vr_values data member. Add various member
2790 functions as well as member functions that delegate to vr_values.
2791 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
2792 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
2793 class instance from vr_values. Use it to call member functions.
2794 (vrp_dom_walker): Add vr_values data member.
2795 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
2796 instance via x_vr_values.
2797 (identify_jump_threads): Accept vr_values as argument. Store
2798 it into the walker structure.
2799 (evrp_dom_walker): Add vr_values class data member. Add various
2801 (evrp_dom_walker::try_find_new_range): Use vr_values data
2802 member to access the memory allocator.
2803 (evrp_dom_walker::before_dom_children): Store vr_values class
2804 instance into the vrp_folder class.
2805 (evrp_dom_walker::push_value_range): Rework to avoid direct
2806 access to num_vr_values and vr_value.
2807 (evrp_dom_walker::pop_value_range): Likewise.
2808 (execute_early_vrp): Remove call to vrp_initialize_lattice.
2809 Use vr_values to get to dump_all_value_ranges member function.
2810 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
2811 and simplify_cond_using_ranges_2 via vrp_prop class instance.
2812 Pass vr_values class instance down to identify_jump_threads.
2813 Remove call to vrp_free_lattice.
2814 (debug_all_value_ranges): Remove.
2816 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
2817 (vrp_folder): Likewise.
2819 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
2820 Get it from the existing bitmap instead.
2821 (vrp_intersect_ranges_1): Likewise.
2823 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2825 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
2827 (store_immediate_info::store_immediate_info): Add bitnotp argument,
2828 set bit_not_p to it.
2829 (imm_store_chain_info::coalesce_immediate_stores): Break group
2830 if bit_not_p is different.
2831 (count_multiple_uses, split_group,
2832 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
2833 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
2834 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
2835 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
2836 to store_immediate_info ctor.
2838 2017-11-09 Jim Wilson <jimw@sifive.com>
2840 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
2841 (scan_prog_file): Likewise.
2843 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2845 * bb-reorder.c (max_entry_frequency): Remove.
2846 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
2847 connect_traces, push_to_next_round_p): Remove prototypes.
2848 (find_traces_1_round): Use counts only.
2849 (push_to_next_round_p): Likewise.
2850 (find_traces): Likewise.
2851 (rotate_loop): Likewise.
2852 (find_traces_1_round): Likewise.
2853 (connect_traces): Likewise.
2854 (edge_order): Likewise.
2856 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
2858 * config/arm/arm.c (output_return_instruction): Add comments to
2859 indicate requirement for cmse_nonsecure_entry return to account
2860 for the size of clearing instruction output here.
2861 (thumb_exit): Likewise.
2862 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
2863 return in hardfloat mode.
2865 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
2867 * config/rs6000/rs6000.c (machine_function): Add a bool,
2868 "toc_is_wrapped_separately".
2869 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
2870 if it wasn't explicitly set or unset, we are optimizing for speed, and
2871 doing separate shrink-wrapping.
2872 (rs6000_get_separate_components): Enable the TOC component if
2873 saving the TOC register in the prologue.
2874 (rs6000_components_for_bb): Handle the TOC component.
2875 (rs6000_emit_prologue_components): Store the TOC register where needed.
2876 (rs6000_set_handled_components): Mark TOC as handled, if handled.
2877 (rs6000_emit_prologue): Don't save the TOC if that is already done.
2879 2017-11-09 Martin Jambor <mjambor@suse.cz>
2881 * ipa-param-manipulation.c: New file.
2882 * ipa-param-manipulation.h: Likewise.
2883 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2884 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2885 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2886 (ipa_parm_adjustment): Likewise.
2887 (ipa_parm_adjustment_vec): Likewise.
2888 (ipa_get_vector_of_formal_parms): Moved declaration to
2889 ipa-param-manipulation.h.
2890 (ipa_get_vector_of_formal_parm_types): Likewise.
2891 (ipa_modify_formal_parameters): Likewise.
2892 (ipa_modify_call_arguments): Likewise.
2893 (ipa_combine_adjustments): Likewise.
2894 (ipa_dump_param_adjustments): Likewise.
2895 (ipa_modify_expr): Likewise.
2896 (ipa_get_adjustment_candidate): Likewise.
2897 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2898 ipa-param-manipulation.c.
2899 (ipa_get_vector_of_formal_parm_types): Likewise.
2900 (ipa_modify_formal_parameters): Likewise.
2901 (ipa_modify_call_arguments): Likewise.
2902 (ipa_modify_expr): Likewise.
2903 (get_ssa_base_param): Likewise.
2904 (ipa_get_adjustment_candidate): Likewise.
2905 (index_in_adjustments_multiple_times_p): Likewise.
2906 (ipa_combine_adjustments): Likewise.
2907 (ipa_dump_param_adjustments): Likewise.
2908 * tree-sra.c: Also include ipa-param-manipulation.h
2909 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2912 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2913 Alan Hayward <alan.hayward@arm.com>
2914 David Sherwood <david.sherwood@arm.com>
2916 * doc/sourcebuild.texi (vect_masked_store): Document.
2918 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2919 Alan Hayward <alan.hayward@arm.com>
2920 David Sherwood <david.sherwood@arm.com>
2922 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2924 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2925 Alan Hayward <alan.hayward@arm.com>
2926 David Sherwood <david.sherwood@arm.com>
2928 * doc/sourcebuild.texi (vect_variable_length): Document.
2930 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2931 Alan Hayward <alan.hayward@arm.com>
2932 David Sherwood <david.sherwood@arm.com>
2934 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2936 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2942 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2943 Alan Hayward <alan.hayward@arm.com>
2944 David Sherwood <david.sherwood@arm.com>
2946 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2947 previously undocumented selectors.
2948 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2950 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2952 * doc/rtl.texi (const_vector): Say that elements can be
2953 const_wide_ints too.
2954 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2955 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2956 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2957 * optabs.c (expand_vector_broadcast): Likewise.
2959 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2960 Alan Hayward <alan.hayward@arm.com>
2961 David Sherwood <david.sherwood@arm.com>
2963 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2964 scaled index even if the unscaled address was invalid.
2965 Don't increase the complexity of using a scale in that case.
2967 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2968 Alan Hayward <alan.hayward@arm.com>
2969 David Sherwood <david.sherwood@arm.com>
2971 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2972 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2973 UNITS_PER_WORD bytes.
2974 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2975 based on REGMODE_NATURAL_SIZE of the inner mode.
2976 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2977 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2978 * expmed.c (lowpart_bit_field_p): Divide the value up into
2979 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2980 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2981 whether something is likely to occupy more than one register.
2983 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2986 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2987 profile_count::adjust_for_ipa_scaling.
2988 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2989 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2991 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2993 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2995 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2996 (split_group): Add total_orig and total_new arguments, estimate the
2997 number of statements related to the store group without store merging
2998 and with store merging.
2999 (imm_store_chain_info::output_merged_store): Adjust split_group
3000 callers, punt if estimated number of statements with store merging
3001 is not smaller than estimated number of statements without it.
3003 (handled_load): Remove has_single_use checks.
3004 (pass_store_merging::process_store): Likewise.
3006 2017-11-09 Richard Biener <rguenther@suse.de>
3008 PR tree-optimization/82902
3009 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3011 2017-11-09 Martin Liska <mliska@suse.cz>
3014 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3017 2017-11-09 Martin Liska <mliska@suse.cz>
3019 PR tree-optimization/82669
3020 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3022 2017-11-09 Martin Liska <mliska@suse.cz>
3024 PR gcov-profile/48463
3025 * coverage.c (coverage_begin_function): Output also end locus
3026 of a function and information whether the function is
3028 * gcov-dump.c (tag_function): Parse and print the information.
3029 * gcov.c (INCLUDE_MAP): Add include.
3030 (INCLUDE_SET): Likewise.
3031 (struct line_info): Move earlier in the source file because
3032 of vector<line_info> in function_info structure.
3033 (line_info::line_info): Likewise.
3034 (line_info::has_block): Likewise.
3035 (struct source_info): Add new member index.
3036 (source_info::get_functions_at_location): New function.
3037 (function_info::group_line_p): New function.
3038 (output_intermediate_line): New function.
3039 (output_intermediate_file): Use the mentioned function.
3040 (struct function_start): New.
3041 (struct function_start_pair_hash): Likewise.
3042 (process_file): Add code that identifies group functions.
3043 Assign lines either to global or function scope.
3044 (generate_results): Skip artificial functions.
3045 (find_source): Assign index for each source file.
3046 (read_graph_file): Read new flag artificial and end_line.
3047 (add_line_counts): Assign it either to global of function scope.
3048 (accumulate_line_counts): Isolate core of the function to
3049 accumulate_line_info and call it for both function and global
3051 (accumulate_line_info): New function.
3052 (output_line_beginning): Fix GNU coding style.
3053 (print_source_line): New function.
3054 (output_line_details): Likewise.
3055 (output_function_details): Likewise.
3056 (output_lines): Iterate both source (global) scope and function
3058 (struct function_line_start_cmp): New class.
3059 * doc/gcov.texi: Reflect changes in documentation.
3061 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3064 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3065 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3066 and similarly for not instead of neg.
3068 2017-11-08 Andi Kleen <ak@linux.intel.com>
3070 * config/i386/i386.opt: Add -mforce-indirect-call.
3071 * config/i386/predicates.md: Check for flag_force_indirect_call.
3072 * doc/invoke.texi: Document -mforce-indirect-call
3074 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3076 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3078 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3079 config/riscv/riscv.c (predict.h): New include.
3080 (riscv_slow_unaligned_access_p): No longer static.
3081 (riscv_block_move_straight): Add require.
3082 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3084 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3087 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3088 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3089 nonimmediate_operand predicate for operand 1 instead of
3092 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3094 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3096 * config/aarch64/constraints.md (Uml): New constraint.
3097 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3100 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3102 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3103 of two vec_duplicates into a vec_concat.
3105 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3107 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3108 Simplify vec_merge of vec_duplicate and vec_concat.
3109 * config/aarch64/constraints.md (Utq): New constraint.
3110 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3113 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3115 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3116 Simplify vec_merge of vec_duplicate and const_vector.
3117 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3119 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3120 mode iterator. Update predicate on operand 1 to
3121 handle non-const_vec constants. Delete constraints.
3122 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3124 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3126 PR tree-optimization/78821
3127 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3129 (store_operand_info::store_operand_info): Initialize it to false.
3130 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3131 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3132 store in the group, and if chain_info is non-NULL, to ignore altogether
3134 (compatible_load_p): Fail if bit_not_p does not match.
3135 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3136 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3137 (pass_store_merging::process_store): Adjust
3138 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3139 call terminate_all_aliasing_chains newly when adding a store into
3140 a chain with non-NULL chain_info.
3142 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3144 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3146 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3148 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3150 (aarch64_layout_frame): Initialise emit_frame_chain.
3151 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3152 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3154 2017-11-08 Martin Liska <mliska@suse.cz>
3156 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3157 of gimple_call_internal_p.
3159 2017-11-07 Tom de Vries <tom@codesourcery.com>
3161 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3163 2017-11-07 Tom de Vries <tom@codesourcery.com>
3165 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3168 2017-11-08 Martin Liska <mliska@suse.cz>
3171 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3173 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3175 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3177 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3180 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3181 (*cmp<mode>_ccz_1): New insn with $k alternative.
3184 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3185 RTX_COMM_COMPARE as commutative as well.
3186 (ix86_binary_operator_ok): Formatting fix.
3187 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3188 *<code><mode>3<mask_name><round_saeonly_name>,
3189 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3190 *<s>mul<mode>3_highpart<mask_name>,
3191 *vec_widen_umult_even_v16si<mask_name>,
3192 *vec_widen_umult_even_v8si<mask_name>,
3193 *vec_widen_umult_even_v4si<mask_name>,
3194 *vec_widen_smult_even_v16si<mask_name>,
3195 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3196 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3197 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3198 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3199 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3200 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3201 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3202 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3203 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3204 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3205 ix86_binary_operator_ok. Formatting fixes.
3206 (*<plusminus_insn><mode>3<mask_name><round_name>,
3207 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3210 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3212 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3213 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3214 eq and ne if TARGET_ISEL.
3215 (cmp): New code_iterator.
3216 (UNS, UNSU_, UNSIK): New code_attrs.
3217 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3218 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3220 ("eq<mode>3"): ... this.
3221 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3223 ("ne<mode>3"): ... this.
3225 2017-11-07 Julia Koval <julia.koval@intel.com>
3228 * common/config/i386/i386-common.c
3229 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3230 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3231 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3232 * config/i386/i386.opt: Ditto.
3233 * config/i386/i386.c (ix86_target_string): Ditto.
3234 (ix86_option_override_internal): Ditto.
3235 (ix86_init_mpx_builtins): Move MPX to args2.
3236 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3237 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3238 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3239 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3240 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3241 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3242 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3244 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3247 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3248 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3249 (zero-extendsidi peephole2): Remove peephole.
3251 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3254 * stor-layout.c: Include attribs.h.
3255 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3256 explicit lookup of "aligned" attribute.
3258 2017-11-07 Andrew Waterman <andrew@sifive.com>
3260 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3261 (riscv_expand_block_move): Likewise.
3262 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3264 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3265 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3266 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3267 (riscv_adjust_block_mem): Likewise.
3268 (riscv_block_move_loop): Likewise.
3269 (riscv_expand_block_move): Likewise.
3270 * config/riscv/riscv.md (movmemsi): New pattern.
3272 2017-11-07 Michael Clark <michaeljclark@mac.com>
3274 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3275 (MUSL_DYNAMIC_LINKER): Likewise.
3277 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3279 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3280 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3283 2017-11-07 Richard Biener <rguenther@suse.de>
3285 * match.pd: Fix build.
3287 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3288 Jackson Woodruff <jackson.woodruff@arm.com>
3290 PR tree-optimization/71026
3291 * match.pd: Canonicalize negate in division.
3293 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3296 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3298 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3300 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3301 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3303 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3305 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3306 non-scalar integral types.
3307 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3308 (-(A-B), -(~A)): New transformations.
3310 2017-11-07 Tom de Vries <tom@codesourcery.com>
3312 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3313 semicolon after "do {} while (0)".
3314 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3315 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3316 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3317 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3318 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3320 2017-11-07 Tom de Vries <tom@codesourcery.com>
3322 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3323 after "do {} while (0)".
3324 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3325 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3326 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3327 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3328 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3330 2017-11-07 Tom de Vries <tom@codesourcery.com>
3333 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3335 (arm_rtx_costs_internal): Add missing semicolon after
3336 HANDLE_NARROW_SHIFT_ARITH call.
3338 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3340 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3341 disable isel if it was not set explicitly.
3343 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3345 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3346 (add_type_attribute) likewise.
3348 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3350 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3351 of struct ix86_frame.
3352 (ix86_initial_elimination_offset): Likewise.
3353 (ix86_expand_split_stack_prologue): Likewise.
3355 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3357 * tree-vrp.h (enum value_range_type): Update stale comment.
3359 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3360 Alan Hayward <alan.hayward@arm.com>
3361 David Sherwood <david.sherwood@arm.com>
3363 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3364 (aarch64_expand_vec_perm_const): Take the number of units too.
3365 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3366 (aarch64_expand_vec_perm_const): Likewise.
3367 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3368 (vec_perm<mode>): Update accordingly.
3370 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3371 Alan Hayward <alan.hayward@arm.com>
3372 David Sherwood <david.sherwood@arm.com>
3374 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3375 Take the number of units too.
3376 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3377 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3378 but check for a vector mode before rather than after the call.
3379 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3380 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3381 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3382 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3383 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3384 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3385 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3386 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3387 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3388 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3389 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3390 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3391 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3392 (aarch64_sqdmull2_n<mode>): Update accordingly.
3394 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3395 Alan Hayward <alan.hayward@arm.com>
3396 David Sherwood <david.sherwood@arm.com>
3398 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3399 the number of units too.
3400 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3401 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3402 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3403 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3404 (vec_store_lanesxi<mode>): Update accordingly.
3406 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3407 Alan Hayward <alan.hayward@arm.com>
3408 David Sherwood <david.sherwood@arm.com>
3410 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3411 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3412 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3413 of units rather than the mode.
3414 * config/aarch64/iterators.md (nunits): New mode attribute.
3415 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3416 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3417 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3418 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3419 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3420 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3421 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3422 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3423 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3424 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3425 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3426 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3427 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3428 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3429 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3430 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3431 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3432 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3433 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3434 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3435 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3436 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3437 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3438 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3439 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3440 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3441 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3442 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3443 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3444 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3445 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3446 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3447 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3448 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3449 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3450 (aarch64_simd_vec_setv2di): Likewise.
3452 2017-11-06 Carl Love <cel@us.ibm.com>
3454 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3456 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3457 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3459 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3461 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3462 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3463 Add power 8 macro expansions.
3464 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3465 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3466 power 8 instructions. (VSX_XXBR): Add iterator.
3468 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3470 * config/arm/arm.md (predicable_short_it): Change default to "no",
3471 improve documentation, remove uses that are identical to the default.
3472 (enabled_for_depr_it): Rename to enabled_for_short_it.
3473 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3474 Remove default uses.
3475 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3476 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3477 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3478 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3480 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3483 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3484 float128 helper macros, which are no longer used after deleting
3485 the old 'q' built-in functions, and moving the round to odd
3486 built-in functions to being special built-in functions.
3487 (BU_FLOAT128_2): Likewise.
3488 (BU_FLOAT128_1_HW): Likewise.
3489 (BU_FLOAT128_2_HW): Likewise.
3490 (BU_FLOAT128_3_HW): Likewise.
3491 (FABSQ): Delete old 'q' built-in functions.
3492 (COPYSIGNQ): Likewise.
3493 (SQRTF128_ODD): Move round to odd built-in functions to be
3494 special built-in functions, so that we can handle
3495 -mabi=ieeelongdouble.
3496 (TRUNCF128_ODD): Likewise.
3497 (ADDF128_ODD): Likewise.
3498 (SUBF128_ODD): Likewise.
3499 (MULF128_ODD): Likewise.
3500 (DIVF128_ODD): Likewise.
3501 (FMAF128_ODD): Likewise.
3502 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3503 built-in names to 'f128'.
3504 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3505 old 'q' built-in functions, as the machine independent code for
3506 'f128' built-in functions handles this.
3507 (rs6000_expand_builtin): Add expansion for float128 round to odd
3508 functions, keying off on -mabi=ieeelongdouble of whether to use
3509 the KFmode or TFmode variant.
3510 (rs6000_init_builtins): Initialize the _Float128 round to odd
3512 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3513 _Float128 'q' built-in functions are now mapped into the new
3514 'f128' built-in functions.
3516 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3518 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3519 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3521 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3524 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3525 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3526 * toplev.c: Include "ipa-fnsummary.h".
3527 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3529 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3531 PR tree-optimization/82838
3532 * gimple-ssa-store-merging.c
3533 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3534 on a separate gimple_seq which is then appended to seq.
3536 2017-11-06 Jeff Law <law@redhat.com>
3539 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3540 (get_probe_interval): New functions.
3541 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3542 (ix86_adjust_stack_and_probe): Likewise.
3543 (output_adjust_stack_and_probe): Likewise.
3544 (ix86_emit_probe_stack_range): Likewise.
3545 (ix86_expand_prologue): Likewise.
3547 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3549 PR tree-optimization/82816
3550 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3551 if the modes of the two types are the same.
3552 (convert_plusminus_to_widen): Likewise.
3554 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3556 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3558 (usadv16qi): New define_expand.
3559 (usadv8hi): New define_expand.
3561 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3564 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3566 (inline_transform): Likewise.
3567 * predict.c (counts_to_freqs): Remove useless conditional.
3568 * profile-count.h (profile_count::apply_scale): Move sanity check.
3569 * tree-inline.c (copy_bb): Always scale.
3570 (copy_cfg_body): Likewise.
3572 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3575 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3577 (*cmp_ite0): Add enabled_for_depr_it attribute.
3578 (*cmp_ite1): Likewise.
3580 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3582 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3585 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3587 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3588 return the value by pointer.
3589 (extract_range_from_multiplicative_op_1): Update accordingly.
3590 Return as soon as an operation fails.
3592 2017-11-05 Tom de Vries <tom@codesourcery.com>
3595 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3596 (DEF_SANITIZER_BUILTIN): ... here.
3597 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3598 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3600 2017-11-05 Tom de Vries <tom@codesourcery.com>
3603 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3605 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3606 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3607 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3609 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3610 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3611 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3613 2017-11-05 Tom de Vries <tom@codesourcery.com>
3616 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3619 2017-11-04 Michael Clark <michaeljclark@mac.com>
3621 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3622 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3624 (*addsi3_extended): Likewise.
3625 (*addsi3_extended2): Likewise.
3626 (<optab>si3): Likewise.
3627 (<optab>di3): Likewise.
3628 (<optab><mode>3): Likewise.
3629 (<*optabe>si3_internal): Likewise.
3630 (zero_extendqi<SUPERQI:mode>2): Likewise.
3631 (*add<mode>hi3): Likewise.
3632 (*xor<mode>hi3): Likewise.
3633 (<optab>di3): Likewise.
3634 (*<optab>si3_extend): Likewise.
3635 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3636 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3637 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3639 2017-11-04 Andrew Waterman <andrew@sifive.com>
3641 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3642 TARGET_STRICT_ALIGN based upon -mtune argument.
3644 2017-11-04 Andrew Waterman <andrew@sifive.com>
3646 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3648 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3650 * config/i386/i386.c (choose_basereg): Use optional scratch
3651 register and add assertion.
3652 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3653 needed, and don't allocate stack.
3654 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3655 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3656 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3658 2017-11-03 Jeff Law <law@redhat.com>
3660 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3661 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3662 to probe at the start of a noreturn function.
3664 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3666 PR tree-optimization/78821
3667 * gimple-ssa-store-merging.c: Update the file comment.
3668 (MAX_STORE_ALIAS_CHECKS): Define.
3669 (struct store_operand_info): New type.
3670 (store_operand_info::store_operand_info): New constructor.
3671 (struct store_immediate_info): Add rhs_code and ops data members.
3672 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3673 and op1r arguments to the ctor, initialize corresponding data members.
3674 (struct merged_store_group): Add load_align_base and load_align
3676 (merged_store_group::merged_store_group): Initialize them.
3677 (merged_store_group::do_merge): Update them.
3678 (merged_store_group::apply_stores): Pick the constant for
3679 encode_tree_to_bitpos from one of the two operands, or skip
3680 encode_tree_to_bitpos if neither operand is a constant.
3681 (class pass_store_merging): Add process_store method decl. Remove
3682 bool argument from terminate_all_aliasing_chains method decl.
3683 (pass_store_merging::terminate_all_aliasing_chains): Remove
3684 var_offset_p argument and corresponding handling.
3685 (stmts_may_clobber_ref_p): New function.
3686 (compatible_load_p): New function.
3687 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3688 if there is overlap and rhs_code is not INTEGER_CST. For
3689 non-overlapping stores terminate group if rhs is not mergeable.
3690 (get_alias_type_for_stmts): Change first argument from
3691 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3692 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3693 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3695 (get_location_for_stmts): Change first argument from
3696 auto_vec<gimple *> & to vec<gimple *> &.
3697 (struct split_store): Remove orig_stmts data member, add orig_stores.
3698 (split_store::split_store): Create orig_stores rather than orig_stmts.
3699 (find_constituent_stmts): Renamed to ...
3700 (find_constituent_stores): ... this. Change second argument from
3701 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
3702 to info structures rather than the statements.
3703 (split_group): Rename ALLOW_UNALIGNED argument to
3704 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
3705 it. Adjust find_constituent_stores caller.
3706 (imm_store_chain_info::output_merged_store): Handle rhs_code other
3707 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
3708 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
3709 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
3710 (mem_valid_for_store_merging): New function.
3711 (handled_load): New function.
3712 (pass_store_merging::process_store): New method.
3713 (pass_store_merging::execute): Use process_store method. Adjust
3714 terminate_all_aliasing_chains caller.
3716 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3718 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
3719 Return true for more constants, symbols and label references.
3720 (aarch64_valid_floating_const): Remove unused function.
3722 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3725 * config/aarch64/aarch64.c (aarch64_layout_frame):
3726 Undo forcing of LR at bottom of frame.
3728 2017-11-03 Jeff Law <law@redhat.com>
3731 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
3732 for int_registers_saved.
3734 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
3735 extracted from tree-ssa-dom.c.
3736 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
3737 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
3738 (record_equivalences_from_incoming_edge): Add additional argument
3739 to single_pred_edge_ignoring_loop_edges call.
3740 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
3741 (uncprop_dom_walker::before_dom_children): Add additional argument
3742 to single_pred_edge_ignoring_loop_edges call.
3743 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
3744 single_pred_edge_ignoring_loop_edges rather than open coding.
3745 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
3747 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
3749 * match.pd (-(-A)): Rewrite.
3751 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3753 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
3754 (rs6000_emit_int_cmove): New declaration.
3755 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
3756 (rs6000_emit_sISEL): Delete.
3757 (rs6000_emit_int_cmove): Make non-static.
3758 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
3759 instead of rs6000_emit_sISEL.
3761 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
3763 * asan.c (create_cond_insert_point): Maintain profile.
3764 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
3766 * basic-block.h (struct basic_block_def): Remove frequency.
3767 (EDGE_FREQUENCY): Use to_frequency
3768 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
3770 (find_traces): Update to use to_frequency.
3771 (find_traces_1_round): Likewise; use only IPA counts.
3772 (bb_to_key): Likewise.
3773 (connect_traces): Use IPA counts only.
3774 (copy_bb_p): Update to use to_frequency.
3775 (fix_up_crossing_landing_pad): Likewise.
3776 (sanitize_hot_paths): Likewise.
3777 * bt-load.c (basic_block_freq): Likewise.
3778 * cfg.c (init_flow): Set count_max to uninitialized.
3779 (check_bb_profile): Remove frequencies; check counts.
3780 (dump_bb_info): Do not dump frequencies.
3781 (update_bb_profile_for_threading): Update counts only.
3782 (scale_bbs_frequencies_int): Likewise.
3783 (MAX_SAFE_MULTIPLIER): Remove.
3784 (scale_bbs_frequencies_gcov_type): Update counts only.
3785 (scale_bbs_frequencies_profile_count): Update counts only.
3786 (scale_bbs_frequencies): Update counts only.
3787 * cfg.h (struct control_flow_graph): Add count-max.
3788 (update_bb_profile_for_threading): Update prototype.
3789 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
3790 (find_many_sub_basic_blocks): Likewise.
3791 * cfgcleanup.c (try_forward_edges): Likewise.
3792 (try_crossjump_to_edge): Likewise.
3793 * cfgexpand.c (expand_gimple_cond): Likewise.
3794 (expand_gimple_tailcall): Likewise.
3795 (construct_init_block): Likewise.
3796 (construct_exit_block): Likewise.
3797 * cfghooks.c (verify_flow_info): Check consistency of counts.
3798 (dump_bb_for_graph): Do not dump frequencies.
3799 (split_block_1): Do not update frequencies.
3800 (split_edge): Do not update frequencies.
3801 (make_forwarder_block): Do not update frequencies.
3802 (duplicate_block): Do not update frequencies.
3803 (account_profile_record): Do not update frequencies.
3804 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
3805 for global heuristics.
3806 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
3807 (expected_loop_iterations_unbounded): Use counts only.
3808 * cfgloopmanip.c (scale_loop_profile): Simplify.
3809 (create_empty_loop_on_edge): Simplify
3811 (duplicate_loop_to_header_edge): Simplify
3812 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
3813 (update_br_prob_note): Take care of removing note when profile
3815 (relink_block_chain): Do not dump frequency.
3816 (rtl_account_profile_record): Use to_frequency.
3817 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
3818 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
3819 (cgraph_update_edges_for_call_stmt_node): Likewise.
3820 (cgraph_edge::verify_count_and_frequency): Update.
3821 (cgraph_node::verify_node): Temporarily disable frequency verification.
3822 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
3823 to_cgraph_frequency.
3824 (cgraph_edge::rebuild_edges): Convert to ipa counts.
3825 * cgraphunit.c (init_lowered_empty_function): Do not initialize
3827 (cgraph_node::expand_thunk): Update profile.
3828 * except.c (dw2_build_landing_pads): Do not update frequency.
3829 * final.c (compute_alignments): Use to_frequency.
3830 (dump_basic_block_info): Do not dump frequency.
3831 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
3832 (dump_gimple_bb_header): Do not dump frequency.
3833 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
3835 * gimple-streamer-in.c (input_bb): Do not stream frequency.
3836 * gimple-streamer-out.c (output_bb): Do not stream frequency.
3837 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
3838 (init_before_recovery): Do not update frequency.
3839 (sched_create_recovery_edges): Do not update frequency.
3840 * hsa-gen.c (convert_switch_statements): Do not update frequency.
3841 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
3842 (ipa_cp_c_finalize): Set max_count to uninitialized.
3843 * ipa-fnsummary.c (get_minimal_bb): Use counts.
3844 (param_change_prob): Use counts.
3845 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
3847 * ipa-split.c (consider_split): Use to_frequency.
3848 (split_function): Use to_frequency.
3849 * ira-build.c (loop_compare_func): Likewise.
3850 (mark_loops_for_removal): Likewise.
3851 (mark_all_loops_for_removal): Likewise.
3852 * loop-doloop.c (doloop_modify): Do not update frequency.
3853 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
3855 * lto-streamer-in.c (input_function): Update count_max.
3856 * omp-expand.c (expand_omp_taskreg): Update count_max.
3857 * omp-simd-clone.c (simd_clone_adjust): Update profile.
3858 * predict.c (maybe_hot_frequency_p): Use to_frequency.
3859 (maybe_hot_count_p): Use ipa counts only.
3860 (maybe_hot_bb_p): Simplify.
3861 (maybe_hot_edge_p): Simplify.
3862 (probably_never_executed): Do not take frequency argument.
3863 (probably_never_executed_bb_p): Do not pass frequency.
3864 (probably_never_executed_edge_p): Likewise.
3865 (combine_predictions_for_bb): Check that profile is nonzero.
3866 (propagate_freq): Do not set frequency.
3867 (drop_profile): Simplify.
3868 (counts_to_freqs): Simplify.
3869 (expensive_function_p): Use to_frequency.
3870 (propagate_unlikely_bbs_forward): Simplify.
3871 (determine_unlikely_bbs): Simplify.
3872 (estimate_bb_frequencies): Add hack to silence graphite issues.
3873 (compute_function_frequency): Use ipa counts.
3874 (pass_profile::execute): Update.
3875 (rebuild_frequencies): Use counts only.
3876 (force_edge_cold): Use counts only.
3877 * profile-count.c (profile_count::dump): Dump new count types.
3878 (profile_count::differs_from_p): Check compatiblity.
3879 (profile_count::to_frequency): New function.
3880 (profile_count::to_cgraph_frequency): New function.
3881 * profile-count.h (struct function): Declare.
3882 (enum profile_quality): Add profile_guessed_local and
3883 profile_guessed_global0.
3884 (class profile_proability): Decrease number of bits to 29;
3885 update from_reg_br_prob_note and to_reg_br_prob_note.
3886 (class profile_count: Update comment; decrease number of bits
3887 to 61. Check compatibility.
3888 (profile_count::compatible_p): New private member function.
3889 (profile_count::ipa_p): New member function.
3890 (profile_count::operator<): Handle global zero correctly.
3891 (profile_count::operator>): Handle global zero correctly.
3892 (profile_count::operator<=): Handle global zero correctly.
3893 (profile_count::operator>=): Handle global zero correctly.
3894 (profile_count::nonzero_p): New member function.
3895 (profile_count::force_nonzero): New member function.
3896 (profile_count::max): New member function.
3897 (profile_count::apply_scale): Handle IPA scalling.
3898 (profile_count::guessed_local): New member function.
3899 (profile_count::global0): New member function.
3900 (profile_count::ipa): New member function.
3901 (profile_count::to_frequency): Declare.
3902 (profile_count::to_cgraph_frequency): Declare.
3903 * profile.c (OVERLAP_BASE): Delete.
3904 (compute_frequency_overlap): Delete.
3905 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3906 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3907 * sched-ebb.c (rank): Use counts only.
3908 * shrink-wrap.c (handle_simple_exit): Use counts only.
3909 (try_shrink_wrapping): Use counts only.
3910 (place_prologue_for_one_component): Use counts only.
3911 * tracer.c (find_best_predecessor): Use to_frequency.
3912 (find_trace): Use to_frequency.
3913 (tail_duplicate): Use to_frequency.
3914 * trans-mem.c (expand_transaction): Do not update frequency.
3915 * tree-call-cdce.c: Do not update frequency.
3916 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3917 (gimple_merge_blocks): Likewise.
3918 (gimple_split_edge): Likewise.
3919 (gimple_duplicate_sese_region): Likewise.
3920 (gimple_duplicate_sese_tail): Likewise.
3921 (move_sese_region_to_fn): Likewise.
3922 (gimple_account_profile_record): Likewise.
3923 (insert_cond_bb): Likewise.
3924 * tree-complex.c (expand_complex_div_wide): Likewise.
3925 * tree-eh.c (lower_resx): Update profile.
3926 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3928 (initialize_cfun): Do not initialize frequencies
3929 (freqs_to_counts): Delete.
3930 (copy_cfg_body): Ignore count parameter.
3931 (copy_body): Update.
3932 (expand_call_inline): Update count_max.
3933 (optimize_inline_calls): Update count_max.
3934 (tree_function_versioning): Update count_max.
3935 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3936 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3938 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3939 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3940 (try_peel_loop): Likewise.
3941 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3943 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3944 (tree_transform_and_unroll_loop): Do not use frequencies
3945 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3946 Use reliable prediction only.
3947 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3948 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3949 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3950 probability scaling.
3951 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3952 not update frequency
3953 (any_remaining_duplicated_blocks): Likewise.
3954 (update_profile): Likewise.
3955 (estimated_freqs_path): Delete.
3956 (freqs_to_counts_path): Delete.
3957 (clear_counts_path): Delete.
3958 (ssa_fix_duplicate_block_edges): Likewise.
3959 (duplicate_thread_path): Likewise.
3960 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3961 * tree-tailcall.c (decrease_profile): Do not update frequency.
3962 (eliminate_tail_call): Likewise.
3963 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3964 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3965 (optimize_mask_stores): Likewise.
3966 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3967 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3968 (ubsan_expand_ptr_ifn): Update profile.
3969 * value-prof.c (gimple_ic): Simplify.
3970 * value-prof.h (gimple_ic): Update prototype.
3971 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3972 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3974 (want_inline_self_recursive_call_p): Likewise.
3975 (resolve_noninline_speculation): Only cummulate defined counts.
3976 (inline_small_functions): Use nonzero_p.
3977 (ipa_inline): Do not access freed node.
3979 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3981 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3982 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3984 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3986 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3987 non-legitimate address.
3989 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3991 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3992 (*lt0_<mode>di, *lt0_<mode>si): New.
3994 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3996 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3997 TARGET_PAIRED_FLOAT.
3999 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4000 Jim Wilson <jim.wilson@linaro.org>
4002 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4003 * config/aarch64/aarch64-tune.md: Regenerated.
4004 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4005 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4007 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4009 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4010 (arc_expand_prologue): Restore blink for millicode.
4011 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4013 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4016 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4017 gen_vec_duplicate after forcing the scalar into a register.
4019 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4021 * combine (try_combine): Print the insns input to try_combine to the
4024 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4027 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4028 Remove second argument from aarch64_process_target_attr call.
4029 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4031 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4033 (aarch64_handle_attr_arch): Remove second argument.
4034 (aarch64_handle_attr_cpu): Ditto.
4035 (aarch64_handle_attr_tune): Ditto.
4036 (aarch64_handle_attr_isa_flags): Ditto.
4037 (aarch64_process_one_target_attr): Ditto.
4038 (aarch64_process_target_attr): Ditto.
4039 (aarch64_option_valid_attribute_p): Remove second argument.
4040 on aarch64_process_target_attr call.
4042 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4044 * diagnostic.c: Include "selftest-diagnostic.h".
4045 (selftest::assert_location_text): New function.
4046 (selftest::test_diagnostic_get_location_text): New function.
4047 (selftest::diagnostic_c_tests): Call it.
4049 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4051 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4052 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4053 (class selftest::test_diagnostic_context): Move to...
4054 * selftest-diagnostic.c: New file.
4055 * selftest-diagnostic.h: New file.
4057 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4059 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4060 offset range for FT32B.
4061 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4062 * config/ft32/ft32.md: Add TARGET_NOPM.
4063 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4064 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4066 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4068 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4070 2017-11-02 Jeff Law <law@redhat.com>
4072 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4073 virtual keyword on FINAL OVERRIDE members.
4075 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4076 virtuals together. Add virtual destructor.
4077 (substitute_and_fold_engine): Similarly.
4079 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4081 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4083 2017-11-02 Richard Biener <rguenther@suse.de>
4085 PR tree-optimization/82795
4086 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4088 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4090 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4092 (gcc_GAS_CHECK_FEATURE): Remove.
4093 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4094 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4095 * configure: Regenerate.
4097 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4099 * config/arc/arc.c (hwloop_optimize): Account for empty
4102 2017-11-02 Richard Biener <rguenther@suse.de>
4105 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4106 Truncate ARRAY_REF index and element size.
4108 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4110 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4112 2017-11-01 Jeff Law <law@redhat.com>
4114 * tree-ssa-ccp.c (ccp_folder): New class derived from
4115 substitute_and_fold_engine.
4116 (ccp_folder::get_value): New member function.
4117 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4118 (ccp_fold_stmt): Remove prototype.
4119 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4120 * tree-ssa-copy.c (copy_folder): New class derived from
4121 substitute_and_fold_engine.
4122 (copy_folder::get_value): Renamed from get_value.
4123 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4124 * tree-vrp.c (vrp_folder): New class derived from
4125 substitute_and_fold_engine.
4126 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4127 (vrp_folder::get_value): New member function.
4128 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4129 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4130 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4131 provide a class interface to folder/substitute routines.
4132 (ssa_prop_fold_stmt_fn): Remove typedef.
4133 (ssa_prop_get_value_fn): Likewise.
4134 (subsitute_and_fold): Remove prototype.
4135 (replace_uses_in): Likewise.
4136 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4137 Renamed from replace_uses_in. Call the virtual member function
4138 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4139 (substitute_and_fold_dom_walker): Remove initialization of
4140 data member entries for calbacks. Add substitute_and_fold_engine
4141 member and initialize it.
4142 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4143 member functions for get_value, replace_phi_args_in c
4144 replace_uses_in, and fold_stmt calls.
4145 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4146 substitute_and_fold. Remove assert. Update ctor call.
4148 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4149 (ssa_prop_visit_phi_fn): Likewise.
4150 (class ssa_propagation_engine): New class to provide an interface
4152 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4154 (ssa_prop_visit_phi): Likewise.
4155 (ssa_propagation_engine::simulate_stmt): Moved into class.
4156 Call visit_phi/visit_stmt from the class rather than via
4157 file scoped static variables.
4158 (ssa_propagation_engine::simulate_block): Moved into class.
4159 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4160 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4161 set file scoped statics for the visit_stmt/visit_phi callbacks.
4162 * tree-complex.c (complex_propagate): New class derived from
4163 ssa_propagation_engine.
4164 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4165 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4166 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4168 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4169 ssa_propagation_engine.
4170 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4171 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4172 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4173 * tree-ssa-copy.c (copy_prop): New class derived from
4174 ssa_propagation_engine.
4175 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4176 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4177 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4178 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4179 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4180 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4181 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4183 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4185 PR rtl-optimization/82778
4186 PR rtl-optimization/82597
4187 * compare-elim.c (struct comparison): Add in_a_setter field.
4188 (find_comparison_dom_walker::before_dom_children): Remove killed
4189 bitmap and df_simulate_find_defs call, instead walk the defs.
4190 Compute last_setter and initialize in_a_setter. Merge definitions
4191 with first initialization for a few variables.
4192 (try_validate_parallel): Use insn_invalid_p instead of
4193 recog_memoized. Return insn rather than just the pattern.
4194 (try_merge_compare): Fix up comment. Don't uselessly test if
4195 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4197 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4201 Alan Hayward <alan.hayward@arm.com>
4202 David Sherwood <david.sherwood@arm.com>
4204 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4205 aarch64_hard_regno_nregs to get the number of registers
4208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4209 Alan Hayward <alan.hayward@arm.com>
4210 David Sherwood <david.sherwood@arm.com>
4212 * config/aarch64/constraints.md (Upl): Rename to...
4214 * config/aarch64/aarch64.md
4215 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4218 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4219 Alan Hayward <alan.hayward@arm.com>
4220 David Sherwood <david.sherwood@arm.com>
4222 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4223 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4226 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4227 Alan Hayward <alan.hayward@arm.com>
4228 David Sherwood <david.sherwood@arm.com>
4230 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4231 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4232 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4234 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4235 of a single element.
4236 * config/aarch64/iterators.md: Add a comment above the permute
4237 unspecs to say that they are generated directly by
4238 aarch64_expand_vec_perm_const.
4239 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4241 2017-11-01 Nathan Sidwell <nathan@acm.org>
4243 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4245 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4247 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4248 and medany code models, and describe what they do.
4250 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4252 Revert accidental duplicate:
4254 * combine.c (can_change_dest_mode): Reject changes in
4255 REGMODE_NATURAL_SIZE.
4257 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4259 PR rtl-optimization/64682
4260 PR rtl-optimization/69567
4261 PR rtl-optimization/69737
4262 PR rtl-optimization/82683
4263 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4264 register mentioned in the note, drop the note, unless it came from I3,
4265 in which case it should go to I3 again.
4267 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4269 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4270 and return false if not.
4271 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4273 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4275 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4276 range is known to be empty.
4278 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4279 Alan Hayward <alan.hayward@arm.com>
4280 David Sherwood <david.sherwood@arm.com>
4282 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4283 and CONST_VECTOR_NUNITS instead of computing the number of units from
4284 the byte sizes of the vector and element.
4285 (simplify_binary_operation_1): Likewise.
4286 (simplify_const_binary_operation): Likewise.
4287 (simplify_ternary_operation): Likewise.
4289 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4290 Alan Hayward <alan.hayward@arm.com>
4291 David Sherwood <david.sherwood@arm.com>
4293 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4294 (int_mem_offset): ...this new function.
4295 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4296 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4297 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4300 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4301 Alan Hayward <alan.hayward@arm.com>
4302 David Sherwood <david.sherwood@arm.com>
4304 * lower-subreg.c (interesting_mode_p): New function.
4305 (compute_costs, find_decomposable_subregs, decompose_register)
4306 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4307 (resolve_clobber, dump_choices): Use it.
4309 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4310 Alan Hayward <alan.hayward@arm.com>
4311 David Sherwood <david.sherwood@arm.com>
4313 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4315 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4316 Alan Hayward <alan.hayward@arm.com>
4317 David Sherwood <david.sherwood@arm.com>
4319 * alias.c (find_base_value, find_base_term): Only process integer
4320 truncations. Check the precision rather than the size.
4322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4323 Alan Hayward <alan.hayward@arm.com>
4324 David Sherwood <david.sherwood@arm.com>
4326 * machmode.h (is_narrower_int_mode): New function
4327 * optabs.c (expand_float, expand_fix): Use it.
4328 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4331 Alan Hayward <alan.hayward@arm.com>
4332 David Sherwood <david.sherwood@arm.com>
4334 * rtl.h (narrower_subreg_mode): New function.
4335 * ira-color.c (update_costs_from_allocno): Use it.
4337 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4338 Alan Hayward <alan.hayward@arm.com>
4339 David Sherwood <david.sherwood@arm.com>
4341 * optabs-query.h (convert_optab_p): New function, split out from...
4342 (convert_optab_handler): ...here.
4343 (widening_optab_handler): Delete.
4344 (find_widening_optab_handler): Remove permit_non_widening parameter.
4345 (find_widening_optab_handler_and_mode): Likewise. Provide an
4346 override that operates on mode class wrappers.
4347 * optabs-query.c (widening_optab_handler): Delete.
4348 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4349 parameter. Assert that the two modes are the same class and that
4350 the "from" mode is narrower than the "to" mode. Use
4351 convert_optab_handler instead of widening_optab_handler.
4352 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4353 instead of widening_optab_handler.
4354 * expr.c (expand_expr_real_2): Update calls to
4355 find_widening_optab_handler.
4356 * optabs.c (expand_widen_pattern_expr): Likewise.
4357 (expand_binop_directly): Take the insn_code as a parameter.
4358 (expand_binop): Only call find_widening_optab_handler for
4359 conversion optabs; use optab_handler otherwise. Update calls
4360 to find_widening_optab_handler and expand_binop_directly.
4361 Use convert_optab_handler instead of widening_optab_handler.
4362 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4363 find_widening_optab_handler and use scalar_mode rather than
4365 (convert_plusminus_to_widen): Likewise.
4367 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4368 Alan Hayward <alan.hayward@arm.com>
4369 David Sherwood <david.sherwood@arm.com>
4371 * machmode.h (fixed_size_mode): New class.
4372 * rtl.h (get_pool_mode): Return fixed_size_mode.
4373 * gengtype.c (main): Add fixed_size_mode.
4374 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4375 (get_raw_arg_mode): Likewise.
4376 * doc/tm.texi: Regenerate.
4377 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4378 * targhooks.c (default_get_reg_raw_mode): Likewise.
4379 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4380 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4381 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4382 (msp430_get_raw_result_mode): Likewise.
4383 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4384 * dbxout.c (dbxout_parms): Require fixed-size modes.
4385 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4386 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4387 * omp-low.c (lower_oacc_reductions): Likewise.
4388 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4389 (simplify_subreg): Update accordingly.
4390 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4391 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4392 that aren't fixed-size.
4393 (get_pool_mode): Return a fixed_size_mode.
4394 (output_constant_pool_2): Take a fixed_size_mode.
4396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4397 Alan Hayward <alan.hayward@arm.com>
4398 David Sherwood <david.sherwood@arm.com>
4400 * doc/rtl.texi (vec_series): Document.
4401 (const): Say that the operand can be a vec_series.
4402 * rtl.def (VEC_SERIES): New rtx code.
4403 * rtl.h (const_vec_series_p_1): Declare.
4404 (const_vec_series_p): New function.
4405 * emit-rtl.h (gen_const_vec_series): Declare.
4406 (gen_vec_series): Likewise.
4407 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4408 (gen_vec_series): Likewise.
4409 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4410 * simplify-rtx.c (simplify_unary_operation): Handle negations
4412 (simplify_binary_operation_series): New function.
4413 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4414 (test_vector_ops_series): New function.
4415 (test_vector_ops): Call it.
4416 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4417 gen_const_vec_series.
4418 (altivec_lvsr): Likewise.
4419 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4422 Alan Hayward <alan.hayward@arm.com>
4423 David Sherwood <david.sherwood@arm.com>
4425 * doc/rtl.texi (const): Update description of address constants.
4426 Say that vector constants are allowed too.
4427 * common.md (E, F): Use CONSTANT_P instead of checking for
4429 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4430 checking for CONST_VECTOR.
4431 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4433 * expr.c (expand_expr_real_2): Check for vector modes instead
4434 of checking for CONST_VECTOR.
4435 * rtl.h (const_vec_p): New function.
4436 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4437 (unwrap_const_vec_duplicate): Handle them here too.
4439 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4440 David Malcolm <dmalcolm@redhat.com>
4441 Alan Hayward <alan.hayward@arm.com>
4442 David Sherwood <david.sherwood@arm.com>
4444 * rtl.h (vec_duplicate_p): New function.
4445 * selftest-rtl.c (assert_rtx_eq_at): New function.
4446 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4447 (assert_rtx_eq_at): Declare.
4448 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4449 * selftest-run-tests.c (selftest::run_tests): Call it.
4450 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4451 (simplify_unary_operation_1): Recursively handle vector duplicates.
4452 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4454 (simplify_subreg): Handle subregs of vector duplicates.
4455 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4456 (selftest::simplify_rtx_c_tests): New functions.
4458 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4459 Alan Hayward <alan.hayward@arm.com>
4460 David Sherwood <david.sherwood@arm.com>
4462 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4463 (gen_vec_duplicate): Likewise.
4464 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4466 (gen_const_vector): ...here.
4467 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4468 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4469 whose elements are all equal.
4470 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4471 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4472 (simplify_relational_operation): Likewise.
4473 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4475 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4476 (aarch64_expand_vector_init): Likewise.
4477 * config/arm/arm.c (neon_vdup_constant): Likewise.
4478 (neon_expand_vector_init): Likewise.
4479 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4480 (arm_block_set_unaligned_vect): Likewise.
4481 (arm_block_set_aligned_vect): Likewise.
4482 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4483 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4484 (expand_vec_perm_even_odd_pack): Likewise.
4485 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4486 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4487 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4488 gen_const_vec_duplicate.
4489 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4490 * config/mips/mips.c (mips_gen_const_int_vector): Use
4491 gen_const_vec_duplicate.
4492 (mips_expand_vector_init): Use CONST0_RTX.
4493 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4494 (define_split): Use gen_const_vec_duplicate.
4495 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4496 (define_split): Use gen_const_vec_duplicate.
4497 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4498 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4499 * config/spu/spu.c (spu_const): Likewise.
4501 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4502 Alan Hayward <alan.hayward@arm.com>
4503 David Sherwood <david.sherwood@arm.com>
4505 * combine.c (can_change_dest_mode): Reject changes in
4506 REGMODE_NATURAL_SIZE.
4508 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4510 * configure.ac (--enable-libssp): New.
4511 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4512 trying to determine target-specific default. Adjust indentation.
4513 * configure: Regenerated.
4514 * doc/install.texi (Configuration): Expand --disable-libssp
4517 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4519 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4522 2017-10-31 Martin Jambor <mjambor@suse.cz>
4525 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4527 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4529 * auto-profile.c (autofdo_source_profile::read): Use
4530 UNKNOWN_LOCATION rather than 0.
4531 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4532 (warning_at): ...this overload.
4533 (warning_at_rich_loc_n): Rename to...
4534 (warning_n): ...this overload.
4535 (error_at_rich_loc): Rename to...
4536 (error_at): ...this overload.
4537 (pedwarn_at_rich_loc): Rename to...
4538 (pedwarn): ...this overload.
4539 (permerror_at_rich_loc): Rename to...
4540 (permerror): ...this overload.
4541 (inform_at_rich_loc): Rename to...
4542 (inform): ...this overload.
4543 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4544 (diagnostic_n_impl_richloc): Rename to...
4545 (diagnostic_n_impl): ...this rich_location *-based decl.
4546 (inform_at_rich_loc): Rename to...
4547 (inform): ...this, and add an assertion.
4548 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4549 (warning_at_rich_loc): Rename to...
4550 (warning_at): ...this, and add an assertion.
4551 (warning_at_rich_loc_n): Rename to...
4552 (warning_n): ...this, and add an assertion.
4553 (warning_n): Update location_t-based implementation for removal of
4554 location_t-based diagnostic_n_impl.
4555 (pedwarn_at_rich_loc): Rename to...
4556 (pedwarn): ...this, and add an assertion.
4557 (permerror_at_rich_loc): Rename to...
4558 (permerror): ...this, and add an assertion.
4559 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4560 (error_at_rich_loc): Rename to...
4561 (error_at): ...this, and add an assertion.
4562 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4563 (driver::do_spec_on_infiles): Likewise.
4564 * substring-locations.c (format_warning_va): Update for renaming
4565 of inform_at_rich_loc.
4567 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4569 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4570 _Float<N>X built-in functions so that the variant without the
4571 "__builtin_" prefix is only enabled for the GNU C and Objective C
4572 languages when they are in non-strict ANSI/ISO mode.
4573 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4574 * target.def (floatn_builtin_p): Add a target hook to control
4575 whether _Float<N> and _Float<N>X built-in functions without the
4576 "__builtin_" prefix are enabled, and return true for C and
4577 Objective C in the default hook. Include langhooks.h in
4579 * targhooks.h (default_floatn_builtin_p): Likewise.
4580 * targhooks.c (default_floatn_builtin_p): Likewise.
4581 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4582 floatn_builtin_p target hook.
4583 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4585 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4586 Eric Botcazou <ebotcazou@adacore.com>
4588 PR rtl-optimization/81803
4589 * lra-constraints.c (curr_insn_transform): Also reload the whole
4590 register for a strict subreg no wider than a word if this is for
4591 a WORD_REGISTER_OPERATIONS target.
4593 2017-10-31 Jason Merrill <jason@redhat.com>
4595 * gdbinit.in: Skip over inlines from timevar.h.
4597 2017-10-31 Martin Liska <mliska@suse.cz>
4599 * doc/gcov.texi: Document new option.
4600 * gcov.c (print_usage): Likewise print it.
4601 (process_args): Support the argument.
4602 (format_count): New function.
4603 (format_gcov): Use the function.
4605 2017-10-31 Martin Liska <mliska@suse.cz>
4607 * gcov.c (struct name_map): do not use typedef.
4608 Define operator== and operator<.
4609 (name_search): Remove.
4610 (name_sort): Remove.
4611 (main): Do not allocate names.
4612 (process_file): Add vertical space.
4613 (generate_results): Use std::find.
4614 (release_structures): Do not release memory.
4615 (find_source): Use std::find.
4617 2017-10-31 Martin Liska <mliska@suse.cz>
4619 * gcov.c (struct line_info): Remove it's typedef.
4620 (line_info::line_info): Add proper ctor.
4621 (line_info::has_block): Do not use a typedef.
4622 (struct source_info): Do not use typedef.
4623 (circuit): Likewise.
4624 (get_cycles_count): Likewise.
4625 (output_intermediate_file): Iterate via vector iterator.
4626 (add_line_counts): Use std::vector methods.
4627 (accumulate_line_counts): Likewise.
4628 (output_lines): Likewise.
4630 2017-10-31 Martin Liska <mliska@suse.cz>
4632 * gcov.c (struct source_info): Remove typedef.
4633 (source_info::source_info): Add proper ctor.
4634 (accumulate_line_counts): Use struct, not it's typedef.
4635 (output_gcov_file): Likewise.
4636 (output_lines): Likewise.
4637 (main): Do not allocate an array.
4638 (output_intermediate_file): Use size of vector container.
4639 (process_file): Resize the vector.
4640 (generate_results): Do not preallocate, use newly added vector
4642 (release_structures): Do not release sources.
4643 (find_source): Use vector methods.
4644 (add_line_counts): Do not use typedef.
4646 2017-10-31 Martin Liska <mliska@suse.cz>
4648 * doc/gcov.texi: Document that.
4649 * gcov.c (add_line_counts): Mark lines with a non-executed
4651 (output_line_beginning): Handle such lines.
4652 (output_lines): Pass new argument.
4653 (output_intermediate_file): Print it in intermediate format.
4655 2017-10-31 Martin Liska <mliska@suse.cz>
4657 * color-macros.h: New file.
4658 * diagnostic-color.c: Factor out color related to macros to
4660 * doc/gcov.texi: Document -k option.
4661 * gcov.c (INCLUDE_STRING): Include string.h.
4662 (print_usage): Add -k option.
4663 (process_args): Parse it.
4664 (pad_count_string): New function.
4665 (output_line_beginning): Likewise.
4666 (DEFAULT_LINE_START): New macro.
4667 (output_lines): Support color output.
4669 2017-10-31 Martin Liska <mliska@suse.cz>
4671 PR gcov-profile/82633
4672 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4673 their interaction with GCOV infrastructure.
4674 * configure.ac: Add -fkeep-{inline,static}-functions to
4676 * configure: Regenerate.
4678 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4681 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4683 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4686 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4687 into a register if it does not fit into an immediate offset field.
4689 2017-10-31 Olivier Hainque <hainque@adacore.com>
4691 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4693 2017-10-31 Julia Koval <julia.koval@intel.com>
4695 * config.gcc: Add gfniintrin.h.
4696 * config/i386/gfniintrin.h: New.
4697 * config/i386/i386-builtin-types.def
4698 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4699 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4700 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4701 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4702 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4703 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
4704 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
4705 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
4706 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
4707 V64QI_FTYPE_V64QI_V64QI_INT): New types.
4708 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4709 * config/i386/immintrin.h: Include gfniintrin.h.
4710 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
4712 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
4714 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
4716 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4718 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
4719 (arm_ashldi3_1bit): Remove pattern.
4720 (ashrdi3): Remove shift by 1 expansion.
4721 (arm_ashrdi3_1bit): Remove pattern.
4722 (lshrdi3): Remove shift by 1 expansion.
4723 (arm_lshrdi3_1bit): Remove pattern.
4724 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
4725 cost of ashldi3 by 1.
4726 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
4727 (<shift>di3_neon): Likewise.
4729 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4731 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
4732 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
4733 and (*aarch64_simd_mov<VQ:mode>).
4734 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
4735 pattern alternative.
4736 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
4737 attributes to match pattern alternative.
4739 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
4741 * config.gcc (powerpc*-*-*): Add emmintrin.h.
4742 * config/rs6000/emmintrin.h: New file.
4743 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
4745 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
4747 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
4748 * (movdi_vfp_cortexa8): Remove pattern.
4750 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4752 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
4754 (Specific, alpha*-dec-osf5.1): Remove.
4755 (Specific, mips-sgi-irix5): Remove.
4756 (Specific, mips-sgi-irix6): Remove.
4758 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4761 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
4762 arguments to clear_bit_region_be.
4764 2017-10-30 Jim Wilson <wilson@tuliptree.org>
4766 * gimplify.c: Include memmodel.h.
4768 2017-10-30 Martin Jambor <mjambor@suse.cz>
4770 * omp-grid.c (grid_attempt_target_gridification): Also insert a
4771 condition whether loop should be executed at all.
4773 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
4775 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4776 gimple folding of vec_madd() intrinsics.
4777 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
4778 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
4779 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
4781 2017-10-30 Richard Biener <rguenther@suse.de>
4783 PR tree-optimization/82762
4785 2017-10-23 Richard Biener <rguenther@suse.de>
4787 PR tree-optimization/82129
4789 2017-08-01 Richard Biener <rguenther@suse.de>
4791 PR tree-optimization/81181
4792 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4793 (compute_antic): ... end of iteration here.
4795 2017-10-30 Joseph Myers <joseph@codesourcery.com>
4797 * doc/invoke.texi (C Dialect Options): Document -std=c17,
4798 -std=iso9899:2017 and -std=gnu17.
4799 * doc/standards.texi (C Language): Document C17 support.
4800 * doc/cpp.texi (Overview): Mention -std=c17.
4801 (Standard Predefined Macros): Document C11 and C17 values of
4802 __STDC_VERSION__. Do not refer to C99 support as incomplete.
4803 * doc/extend.texi (Inline): Do not list individual options for
4804 standards newer than C99.
4805 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
4807 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
4810 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4812 * asan.c (asan_finish_file): Align asan globals array by shadow
4815 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4818 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
4819 (struct store_immediate_info): Add bitregion_start and bitregion_end
4821 (store_immediate_info::store_immediate_info): Add brs and bre
4822 arguments and initialize bitregion_{start,end} from those.
4823 (struct merged_store_group): Add bitregion_start, bitregion_end,
4824 align_base and mask fields. Drop unnecessary struct keyword from
4825 struct store_immediate_info. Add do_merge method.
4826 (clear_bit_region_be): Use memset instead of loop storing zeros.
4827 (merged_store_group::do_merge): New method.
4828 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
4829 stores as long as the surrounding bitregions have no gaps.
4830 (merged_store_group::merge_overlapping): Use do_merge.
4831 (merged_store_group::apply_stores): Test that bitregion_{start,end}
4832 is byte aligned, rather than requiring that start and width are
4833 byte aligned. Drop unnecessary struct keyword from
4834 struct store_immediate_info. Allocate and populate also mask array.
4835 Make start of the arrays relative to bitregion_start rather than
4836 start and size them according to bitregion_{end,start} difference.
4837 (struct imm_store_chain_info): Drop unnecessary struct keyword from
4838 struct store_immediate_info.
4839 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
4840 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
4841 struct keyword from struct store_immediate_info.
4842 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
4843 between stores as long as the surrounding bitregions have no gaps.
4845 (struct split_store): Add orig non-static data member.
4846 (split_store::split_store): Initialize orig to false.
4847 (find_constituent_stmts): Return store_immediate_info *, non-NULL
4848 if there is exactly a single original stmt. Change stmts argument
4849 to pointer from reference, if NULL, don't push anything to it. Add
4850 first argument, use it to optimize skipping over orig stmts that
4851 are known to be before bitpos already. Simplify.
4852 (split_group): Return unsigned int count how many stores are or
4853 would be needed rather than a bool. Add allow_unaligned argument.
4854 Change split_stores argument from reference to pointer, if NULL,
4855 only do a dry run computing how many stores would be produced.
4856 Rewritten algorithm to use both alignment and misalign if
4857 !allow_unaligned and handle bitfield stores with gaps.
4858 (imm_store_chain_info::output_merged_store): Set start_byte_pos
4859 from bitregion_start instead of start. Compute allow_unaligned
4860 here, if true, do 2 split_group dry runs to compute which one
4861 produces fewer stores and prefer aligned if equal. Punt if
4862 new count is bigger or equal than original before emitting any
4863 statements, rather than during that. Remove no longer needed
4864 new_ssa_names tracking. Replace num_stmts with
4865 split_stores.length (). Use 32-bit stack allocated entries
4866 in split_stores auto_vec. Try to reuse original store lhs/rhs1
4867 if possible. Handle bitfields with gaps.
4868 (pass_store_merging::execute): Ignore bitsize == 0 stores.
4869 Compute bitregion_{start,end} for the stores and construct
4870 store_immediate_info with that. Formatting fixes.
4872 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
4875 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
4876 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
4878 2017-10-29 Jim Wilson <wilson@tuliptree.org>
4880 * gimplify.c: Include tm_p.h.
4882 * common.opt (gcoff): Re-add as ignored option.
4883 (gcoff1, gcoff2, gcoff3): Likewise.
4885 * Makefile.in (OBJS): Delete sdbout.o.
4886 (GTFILES): Delete $(srcdir)/sdbout.c.
4887 * debug.h: Delete sdb_debug_hooks.
4888 * final.c: Delete sdbout.h include.
4889 (final_scan_insn): Delete SDB_DEBUG check.
4890 (rest_of_clean_state): Likewise.
4891 * output.h: Delete sdb_begin_function_line.
4894 * toplev.c: Delete sdbout.h include.
4895 (process_options): Delete SDB_DEBUG check.
4896 * tree-core.h (tree_type_common): Delete pointer field of
4898 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4899 TYPE_SYMTAB_POINTER.
4900 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4901 (TYPE_SYMTAB_IS_POINTER): Delete.
4902 (TYPE_SYMTAB_IS_DIE): Renumber.
4903 * xcoffout.c: Refer to former sdbout.c file.
4904 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4906 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4907 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4908 (Debugging Options): Delete -gcoff.
4909 (-gstabs): Delete SDB reference.
4911 (-gcoff@var{level}): Delete.
4912 * doc/passes.texi (Debugging information output): Delete SDB and
4913 sdbout.c references.
4914 * doc/tm.texi: Regenerate.
4915 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4916 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4918 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4919 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4920 (SDB_DEBUGGING_INFO): Delete.
4921 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4922 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4923 * target.def (output_source_filename): Delete COFF reference.
4925 * common.opt (gcoff): Delete.
4926 (gxcoff+): Update Negative chain.
4927 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4929 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4930 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4931 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4933 (expand_function_start): Change sdb reference to past tense.
4934 (expand_function_end): Change sdb reference to past tense.
4935 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4936 * opts.c (debug_type_names): Delete coff entry.
4937 (common_handle_option): Delete OPT_gcoff case.
4938 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4940 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4941 * config/cris/cris.h: Delete SDB reference in comment.
4942 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4943 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4944 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4945 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4947 (ix86_expand_prologue): Likewise.
4948 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4949 * config/ia64/ia64.h: Likewise.
4950 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4951 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4953 * config/mmix/mmix.h: Likewise.
4954 * config/nds32/nds32.c: Likewise.
4955 * config/stormy/storym16.h: Likewise.
4956 * config/visium/visium.h: Likewise.
4957 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4959 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4961 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4962 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4963 FRAME_POINTER_REGNUM point at high end of local var area.
4965 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4967 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4968 Move comment around. Do not reset best_edge for a copiable
4969 destination if the copy would cause a partition change.
4970 (better_edge_p): Remove redundant check.
4972 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4974 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4976 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4978 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4979 for math functions that have _Float<N> and _Float<N>X variants.
4980 (mathfn_built_in_2): Add support for math functions that have
4981 _Float<N> and _Float<N>X variants.
4982 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4983 (expand_builtin_mathfn_ternary): Add support for fma with
4984 _Float<N> and _Float<N>X variants.
4985 (expand_builtin): Likewise.
4986 (fold_builtin_3): Likewise.
4987 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4988 create math function _Float<N> and _Float<N>X variants as external
4990 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4991 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4992 the __builtin_ prefix and if not strict ansi, without the prefix.
4993 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4994 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4995 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4996 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4997 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4998 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4999 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5000 function signatures for fma _Float<N> and _Float<N>X variants.
5001 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5002 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5003 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5004 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5005 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5006 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5007 * gencfn-macros.c (print_case_cfn): Add support for math functions
5008 that have _Float<N> and _Float<N>X variants.
5009 (print_define_operator_list): Likewise.
5010 (fltfn_suffixes): Likewise.
5012 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5013 for math functions that have _Float<N> and _Float<N>X variants.
5014 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5015 and _Float<N>X variants.
5016 (COPYSIGN): Likewise.
5019 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5020 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5022 (integer_valued_read_call_p): Likewise.
5023 * fold-const-call.c (fold_const_call_ss): Likewise.
5024 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5025 _Float<N> and _Float<N>X variants.
5026 (fold_const_call_ssss): Add support for fma _Float<N> and
5027 _Float<N>X variants.
5028 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5029 support for copysign and fma _Float<N> and _Float<N>X variants.
5030 (backprop::process_builtin_call_use): Likewise.
5031 * tree-call-cdce.c (can_test_argument_range); Add support for
5032 sqrt _Float<N> and _Float<N>X variants.
5033 (edom_only_function): Likewise.
5034 (get_no_error_domain): Likewise.
5035 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5036 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5037 copysign _Float<N> and _Float<N>X variants.
5038 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5039 handled by machine independent code.
5040 (FMAF128): Likewise.
5041 * doc/cpp.texi (Common Predefined Macros): Document defining
5042 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5043 fma _Float<N> and _Float<N>X variants.
5045 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5048 * config/i386/i386-modes.def (CCFPU): Remove definition.
5049 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5050 (ix86_cc_modes_compatible): Ditto.
5051 (ix86_expand_carry_flag_compare): Ditto.
5052 (ix86_expand_int_movcc): Ditto.
5053 (ix86_expand_int_addcc): Ditto.
5054 (ix86_reverse_condition): Ditto.
5055 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5056 Return true/false for unordered/ordered fp comparisons.
5057 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5058 (ix86_prepare_fp_compare_args): Update for rename.
5059 (ix86_expand_fp_compare): Update for rename. Generate unordered
5060 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5061 (ix86_expand_sse_compare_and_jump): Ditto.
5062 * config/i386/predicates.md (fcmov_comparison_operator):
5063 Remove CCFPU mode handling.
5064 (ix86_comparison_operator): Ditto.
5065 (ix86_carry_flag_operator): Ditto.
5066 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5067 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5068 (*cmpu<mode>_cc_i387): Ditto.
5069 (FPCMP): Remove mode iterator.
5070 (unord): Remove mode attribute.
5071 (unord_subst): New define_subst transformation
5072 (unord): New define_subst attribute.
5074 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5075 (*cmpi<unord>xf_i387): Ditto.
5076 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5077 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5078 using unord_subst transformation.
5079 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5080 (round_saeonly): Also handle CCFP mode.
5081 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5082 Remove UNSPEC_SAHF unspec handling.
5084 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5086 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5088 2017-10-27 Jeff Law <law@redhat.com>
5090 * gimple-ssa-sprintf.c: Include domwalk.h.
5091 (class sprintf_dom_walker): New class, derived from dom_walker.
5092 (sprintf_dom_walker::before_dom_children): New function.
5093 (struct call_info): Moved into sprintf_dom_walker class
5094 (compute_formath_length, handle_gimple_call): Likewise.
5095 (sprintf_length::execute): Call the dominator walker rather
5096 than walking the statements.
5098 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5099 gimple statement locations.
5100 (check_array_bounds): Corresponding changes. Get the statement's
5101 location directly from wi->stmt.
5103 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5106 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5108 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5110 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5111 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5113 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5116 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5117 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5118 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5119 maybe_get_pool_constant.
5120 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5123 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5125 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5126 2.26 caveat. Update gas and gld versions.
5127 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5130 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5132 * cgraph.h (set_malloc_flag): Declare.
5133 * cgraph.c (set_malloc_flag_1): New function.
5134 (set_malloc_flag): Likewise.
5135 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5136 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5138 (read_ipa_call_summary): Add support for reading is_return_callee.
5139 (write_ipa_call_summary): Stream is_return_callee.
5140 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5141 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5142 ipa-prop.h, ipa-fnsummary.h.
5143 (pure_const_names): Change to static.
5144 (malloc_state_e): Define.
5145 (malloc_state_names): Define.
5146 (funct_state_d): Add field malloc_state.
5147 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5148 (check_retval_uses): New function.
5149 (malloc_candidate_p): Likewise.
5150 (analyze_function): Add support for malloc attribute.
5151 (pure_const_write_summary): Stream malloc_state.
5152 (pure_const_read_summary): Add support for reading malloc_state.
5153 (dump_malloc_lattice): New function.
5154 (propagate_malloc): New function.
5155 (warn_function_malloc): New function.
5156 (ipa_pure_const::execute): Call propagate_malloc and
5157 ipa_free_fn_summary.
5158 (pass_local_pure_const::execute): Add support for malloc attribute.
5159 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5160 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5162 2017-10-27 Martin Liska <mliska@suse.cz>
5164 PR gcov-profile/82457
5165 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5166 for fork-like functions to be properly instrumented.
5168 2017-10-27 Richard Biener <rguenther@suse.de>
5171 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5172 info when we redirected EH.
5174 2017-10-26 Michael Collison <michael.collison@arm.com>
5176 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5178 (<optab>_trunchf<GPI:mode>2: New pattern.
5179 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5180 * config/aarch64/iterators.md (wv): New mode attribute.
5181 (vf, VF): New mode attributes.
5182 (vgp, VGP): New mode attributes.
5183 (s): Update attribute with SImode and DImode prefixes.
5185 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5187 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5188 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5189 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5190 (nios2_option_overide): Initialize it. Don't allow R0-relative
5191 addressing with PIC.
5192 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5193 (nios2_symbolic_constant_p): Likewise.
5194 (nios2_legitimate_address_p): Likewise.
5195 (nios2_r0rel_section_name_p): New.
5196 (nios2_symbol_ref_in_r0rel_data_p): New.
5197 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5198 (r0rel_constant_p): New.
5199 (nios2_print_operand_address): Handle r0rel_constant_p.
5200 (nios2_cdx_narrow_form_p): Likewise.
5201 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5202 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5203 (Nios II Options): Document -mr0rel-sec.
5205 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5207 * config/nios2/nios2.c: Include xregex.h.
5208 (nios2_gprel_sec_regex): New.
5209 (nios2_option_overide): Initialize it. Don't allow GP-relative
5210 addressing with PIC.
5211 (nios2_small_section_name_p): Check for regex match.
5212 * config/nios2/nios2.opt (mgprel-sec=): New option.
5213 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5214 (Nios II Options): Document -mgprel-sec.
5216 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5218 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5220 2017-10-26 Tom de Vries <tom@codesourcery.com>
5222 PR tree-optimization/82707
5223 * gimple.c (gimple_copy): Fix unsharing of
5224 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5226 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5228 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5229 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5230 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5231 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5232 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5233 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5234 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5235 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5236 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5237 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5238 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5239 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5240 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5241 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5242 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5243 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5244 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5246 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5248 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5250 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5251 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5252 warning to rs6000.c. Remove the Undocumented flag, since it has
5254 (-mabi=ibmlongdouble): Likewise.
5255 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5256 already set, set the default format for long double.
5257 (rs6000_debug_reg_global): Print whether long double is IBM or
5259 (rs6000_option_override_internal): Rework setting long double
5260 format. Only warn if the user is changing the long double default
5261 and they did not use -Wno-psabi.
5262 * doc/invoke.texi (PowerPC options): Update the documentation for
5263 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5265 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5266 Alan Hayward <alan.hayward@arm.com>
5267 David Sherwood <david.sherwood@arm.com>
5269 * rtl.h (wider_subreg_mode): New function.
5270 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5271 rather than an unsigned int *.
5272 * ira-color.c (regno_max_ref_width): Replace with...
5273 (regno_max_ref_mode): ...this new variable.
5274 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5275 Use wider_subreg_mode.
5276 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5277 rather than an unsigned int *.
5278 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5279 (process_alt_operands): Likewise.
5280 (invariant_p): Likewise.
5281 * lra-spills.c (assign_mem_slot): Likewise.
5282 (add_pseudo_to_slot): Likewise.
5283 * lra.c (collect_non_operand_hard_regs): Likewise.
5284 (add_regs_to_insn_regno_info): Likewise.
5285 * reload1.c (regno_max_ref_width): Replace with...
5286 (regno_max_ref_mode): ...this new variable.
5287 (reload): Update accordingly. Update call to
5288 ira_sort_regnos_for_alter_reg.
5289 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5290 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5291 (scan_paradoxical_subregs): Likewise.
5293 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5295 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5296 (aarch64_frame): Add emit_frame_chain boolean.
5297 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5298 Move eh_return case to aarch64_layout_frame.
5299 (aarch64_layout_frame): Initialize emit_frame_chain.
5300 (aarch64_expand_prologue): Use emit_frame_chain.
5302 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5304 * config/aarch64/aarch64.c (aarch64_layout_frame):
5305 Ensure LR is always stored at the bottom of the callee-saves.
5306 Remove rarely used frame layout which saves callee-saves at top of
5307 frame, so the store of LR can be used as a valid probe in all cases.
5309 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5311 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5312 Improve unaligned TImode/TFmode base/offset split.
5314 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5315 Alan Hayward <alan.hayward@arm.com>
5316 David Sherwood <david.sherwood@arm.com>
5318 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5319 * combine.c (find_single_use_1): Likewise.
5320 (expand_field_assignment): Likewise.
5321 (move_deaths): Likewise.
5322 * lra-constraints.c (simplify_operand_subreg): Likewise.
5323 (curr_insn_transform): Likewise.
5324 * lra.c (collect_non_operand_hard_regs): Likewise.
5325 (add_regs_to_insn_regno_info): Likewise.
5326 * rtlanal.c (reg_referenced_p): Likewise.
5327 (covers_regno_no_parallel_p): Likewise.
5329 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5331 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5332 Don't print any bits outside the precision of the value.
5333 * wide-int.cc (test_printing): Add some new tests.
5335 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5337 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5338 supports -xbrace_comment option.
5339 * configure: Regenerate.
5340 * config.in: Regenerate.
5341 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5342 (ASM_CPU_SPEC): Use it.
5344 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5346 * target.def (static_rtx_alignment): New hook.
5347 * targhooks.h (default_static_rtx_alignment): Declare.
5348 * targhooks.c (default_static_rtx_alignment): New function.
5349 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5350 * doc/tm.texi: Regenerate.
5351 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5352 instead of targetm.constant_alignment. Remove call to
5354 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5355 (cris_preferred_mininum_alignment): New function, split out from...
5356 (cris_constant_alignment): ...here.
5357 (cris_static_rtx_alignment): New function.
5358 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5360 (ix86_constant_alignment): ...here.
5361 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5362 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5363 (mmix_static_rtx_alignment): New function.
5364 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5365 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5367 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5370 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5371 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5373 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5375 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5376 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5377 mark as a sign-extending load.
5378 (local_pic_load_u): Define.
5380 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5383 * fold-const.c (operand_equal_for_comparison_p): Also return true
5384 if ARG0 is a simple variant of ARG1 with narrower precision.
5385 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5387 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5389 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5391 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5392 scatter_static, scatter_per_elt.
5393 * x86-tune-costs.h: Add new cost entries.
5395 2017-10-25 Richard Biener <rguenther@suse.de>
5397 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5398 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5399 class pass_fre): Move to ...
5400 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5401 class pass_fre): ... here and adjust for statistics.
5403 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5406 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5407 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5408 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5409 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5412 2017-10-25 Richard Biener <rguenther@suse.de>
5414 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5415 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5416 eliminate_push_avail, eliminate_insert): Move inside...
5417 (class eliminate_dom_walker): ... this class in preparation
5419 (fini_eliminate): Remove by merging with ...
5420 (eliminate): ... this function. Adjust for class changes.
5421 (pass_pre::execute): Remove fini_eliminate call.
5422 (pass_fre::execute): Likewise.
5424 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5427 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5428 (VPERMI2, VPERMI2I): New mode iterators.
5429 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5430 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5432 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5433 mode iterator. Remove 3 old define_insn patterns.
5434 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5435 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5436 VPERMI2 mode iterator, remove the other two expanders.
5437 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5438 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5439 instructions, remove the other two patterns.
5440 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5441 mode iterator, remove the other two patterns.
5442 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5443 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5444 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5445 and adjust argument order accordingly.
5446 (ix86_expand_vec_perm): Adjust caller.
5447 (expand_vec_perm_1): Likewise.
5448 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5449 (expand_vec_perm_vpermt2_vpshub2): ... this.
5450 (ix86_expand_vec_perm_const_1): Adjust caller.
5451 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5454 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5455 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5456 (vec_shl_<mode>): Remove unused expander.
5457 (avx512bw_<shift_insn><mode>3): New define_insn.
5458 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5459 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5461 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5464 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5467 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5469 PR rtl-optimization/82396
5470 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5471 (autopref_multipass_init): Simplify initialization.
5472 (autopref_rank_data): Simplify sort order.
5473 * gcc/sched-int.h (autopref_multipass_data_): Remove
5474 multi_mem_insn_p, min_offset and max_offset.
5476 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5479 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5480 Check special value of flag_omit_frame_pointer.
5481 (aarch64_can_eliminate): Likewise.
5482 (aarch64_override_options_after_change_1): Simplify handling of
5483 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5485 2017-10-24 Richard Biener <rguenther@suse.de>
5487 PR tree-optimization/82697
5488 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5489 zero for conditional load and unconditional store.
5491 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5493 * doc/install.texi: Document bootstrap-cet.
5495 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5498 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5499 ENDBR instruction at function entrance if function is only
5502 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5505 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5506 patterns to better describe from which operation the CF is computed.
5507 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5508 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5509 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5510 is 0, use _0 suffixed expanders instead of emitting a comparison
5513 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5515 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5516 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5518 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5521 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5522 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5523 * loop-iv.c (iv_get_reaching_def): Likewise.
5524 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5525 variable is promoted and the partition contains undefined values.
5527 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5529 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5531 (nios2_address_cost): Define.
5532 (nios2_legitimize_address): Recognize (exp + constant) directly.
5533 (TARGET_ADDRESS_COST): Define.
5535 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5537 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5538 (nios2_symbolic_memory_operand_p): Declare.
5539 (nios2_split_large_constant): Declare.
5540 (nios2_split_symbolic_memory_operand): Declare.
5541 * config/nios2/nios2.c: Adjust includes.
5542 (nios2_symbolic_constant_allowed): New.
5543 (nios2_symbolic_constant_p): New.
5544 (nios2_plus_symbolic_constant_p): New.
5545 (nios2_valid_addr_expr_p): Recognize addresses involving
5547 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5548 (nios2_symbolic_memory_operand_p): New.
5549 (nios2_large_constant_p): New.
5550 (nios2_split_large_constant): New.
5551 (nios2_split_plus_large_constant): New.
5552 (nios2_split_symbolic_memory_operand): New.
5553 (nios2_legitimize_address): Code refactoring. Handle addresses
5554 involving symbolic constants.
5555 (nios2_emit_move_sequence): Likewise.
5556 (nios2_print_operand): Improve error output.
5557 (nios2_print_operand_address): Handle LO_SUM.
5558 (nios2_cdx_narrow_form_p): Likewise.
5559 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5560 operands involving symbolic constants.
5561 (movhi_internal, movsi_internal): Likewise.
5562 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5563 (extendhisi2, extendqi<mode>2): Likewise.
5565 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5567 * tree-pass.h (PROP_rtl_split_insns): Define.
5568 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5570 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5572 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5574 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5577 * target.def (const_not_ok_for_debug_p): Default to
5578 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5579 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5580 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5581 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5582 which targetm.const_not_ok_for_debug_p returned true.
5583 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5585 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5587 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5588 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5589 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5591 (ix86_const_not_ok_for_debug_p): New function.
5592 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5593 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5595 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5598 * system.h: Conditionally include "unique-ptr.h" if
5599 INCLUDE_UNIQUE_PTR is defined.
5600 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5601 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5603 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5605 * config/rl78/rl78.md: New define_expand "subdi3".
5607 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5610 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5611 DF_REF_INSN if DF_REF_INSN_INFO is false.
5613 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5615 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5616 xmm_move instead of sse_move.
5617 (sse_store_index): New function.
5618 (ix86_register_move_cost): Be more sensible about mismatch stall;
5619 model AVX moves correctly; make difference between sse->integer and
5621 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5622 moves; make difference between SSE and AVX.
5623 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5624 and zmm_move. Increase size of sse load and store tables;
5625 add unaligned load and store tables; add ssemmx_to_integer.
5626 * x86-tune-costs.h: Update all entries according to real
5627 move latencies from Agner Fog's manual and chip documentation.
5629 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5632 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5633 * config/i386/constraints.md (Wf): New constraint.
5634 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5635 (cmp<dwi>_doubleword): Removed.
5636 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5637 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5638 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5639 expand with cmp<dwi>_doubleword. For LTU and GEU use
5640 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5642 * common.opt (gcolumn-info): Enable by default.
5643 * doc/invoke.texi (gcolumn-info): Document new default.
5645 2017-10-23 Richard Biener <rguenther@suse.de>
5647 PR tree-optimization/82672
5648 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5649 Fold the stmt if we propagated into it.
5651 2017-10-23 Richard Biener <rguenther@suse.de>
5653 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5654 (bitmap_remove_expr_from_set): ... this. All callers call this
5655 for non-constant values.
5656 (bitmap_set_subtract): Rename to...
5657 (bitmap_set_subtract_expressions): ... this. Adjust and
5659 (bitmap_set_contains_value): Remove superfluous check.
5660 (bitmap_set_replace_value): Inline into single caller ...
5661 (bitmap_value_replace_in_set): ... here and simplify.
5662 (dependent_clean): Merge into ...
5663 (clean): ... this using an overload. Adjust.
5664 (prune_clobbered_mems): Adjust.
5665 (compute_antic_aux): Likewise.
5666 (compute_partial_antic_aux): Likewise.
5668 2017-10-23 Richard Biener <rguenther@suse.de>
5670 PR tree-optimization/82129
5672 2017-08-01 Richard Biener <rguenther@suse.de>
5674 PR tree-optimization/81181
5675 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5676 (compute_antic): ... end of iteration here.
5678 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5680 * target.def (starting_frame_offset): New hook.
5681 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5682 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5683 * doc/tm.texi.in: Regenerate.
5684 * hooks.h (hook_hwi_void_0): Declare.
5685 * hooks.c (hook_hwi_void_0): New function.
5686 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5687 STARTING_FRAME_OFFSET.
5688 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5689 * reload1.c (reload): Likewise.
5690 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5691 instead of STARTING_FRAME_OFFSET.
5692 * function.c (try_fit_stack_local): Likewise.
5693 (assign_stack_local_1): Likewise
5694 (instantiate_virtual_regs): Likewise.
5695 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5696 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
5697 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
5698 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
5699 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
5700 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
5701 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
5702 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
5703 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
5704 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
5705 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
5706 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
5707 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
5708 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
5709 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
5710 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
5711 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
5712 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
5713 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
5714 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
5715 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
5716 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
5717 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
5718 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
5719 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
5720 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
5721 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
5722 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
5723 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
5724 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
5725 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
5726 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
5727 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
5728 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
5729 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
5730 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
5731 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
5732 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
5733 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
5734 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
5735 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
5736 * config/avr/avr.c (avr_starting_frame_offset): Make static and
5737 return a HOST_WIDE_INT.
5738 (avr_builtin_setjmp_frame_value): Use it instead of
5739 STARTING_FRAME_OFFSET.
5740 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5741 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
5742 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
5744 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5745 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
5746 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
5747 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5748 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
5749 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
5750 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5751 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
5752 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
5753 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5754 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
5755 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
5757 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5758 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
5759 * config/mips/mips.c (mips_compute_frame_info): Refer to
5760 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5761 (mips_starting_frame_offset): New function.
5762 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5763 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
5764 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
5765 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
5766 and return a HOST_WIDE_INT.
5767 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5768 (mmix_initial_elimination_offset): Refer to
5769 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
5770 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
5771 * config/pa/pa.c (pa_starting_frame_offset): New function.
5772 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
5773 (pa_expand_prologue): Likewise.
5774 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5775 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
5776 !FRAME_GROWS_DOWNWARD handling to...
5777 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5778 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
5779 !FRAME_GROWS_DOWNWARD handling to...
5780 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5781 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
5782 !FRAME_GROWS_DOWNWARD handling to...
5783 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5784 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
5786 (rs6000_starting_frame_offset): New function.
5787 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
5788 !FRAME_GROWS_DOWNWARD handling to...
5789 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5790 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
5791 !FRAME_GROWS_DOWNWARD handling to...
5792 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5793 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
5794 !FRAME_GROWS_DOWNWARD handling to...
5795 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
5796 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
5797 (rs6000_starting_frame_offset): New function.
5798 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
5799 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
5800 * config/vax/vax.c (vax_starting_frame_offset): New function.
5801 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
5802 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5803 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
5804 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
5805 (TARGET_STARTING_FRAME_OFFSET): Redefine.
5806 * system.h (STARTING_FRAME_OFFSET): Poison.
5808 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5810 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
5811 SCALAR_TYPE_MODE instead of TYPE_MODE.
5813 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5814 Alan Hayward <alan.hayward@arm.com>
5815 David Sherwood <david.sherwood@arm.com>
5817 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
5819 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5820 Alan Hayward <alan.hayward@arm.com>
5821 David Sherwood <david.sherwood@arm.com>
5823 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
5825 2017-10-23 Richard Biener <rguenther@suse.de>
5827 PR tree-optimization/82129
5828 * tree-ssa-pre.c (bitmap_set_and): Remove.
5829 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
5830 canonicalizing expressions in the set to those with lowest
5831 ID rather than taking that from the first edge.
5833 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5835 * combine.c (rtx_equal_for_field_assignment_p): Use
5836 byte_lowpart_offset.
5838 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5839 Alan Hayward <alan.hayward@arm.com>
5840 David Sherwood <david.sherwood@arm.com>
5842 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
5843 to a SUBREG_PROMOTED_VAR.
5845 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5846 Alan Hayward <alan.hayward@arm.com>
5847 David Sherwood <david.sherwood@arm.com>
5849 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
5850 (expand_debug_source_expr): Likewise.
5851 * combine.c (combine_simplify_rtx): Likewise.
5852 * cse.c (fold_rtx): Likewise.
5853 * optabs.c (expand_float): Likewise.
5854 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5855 (simplify_binary_operation_1): Likewise.
5857 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5858 Alan Hayward <alan.hayward@arm.com>
5859 David Sherwood <david.sherwood@arm.com>
5861 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
5862 (record_promoted_value): Likewise.
5863 * expr.c (expand_expr_real_2): Likewise.
5864 * ree.c (update_reg_equal_equiv_notes): Likewise.
5865 (combine_set_extension): Likewise.
5866 * rtlanal.c (low_bitmask_len): Likewise.
5867 * simplify-rtx.c (neg_const_int): Likewise.
5868 (simplify_binary_operation_1): Likewise.
5870 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5871 Alan Hayward <alan.hayward@arm.com>
5872 David Sherwood <david.sherwood@arm.com>
5874 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
5875 * regcprop.c (maybe_mode_change): Likewise.
5876 * reload1.c (alter_reg): Likewise.
5878 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5880 * inchash.h (inchash::hash::add_wide_int): New function.
5881 * lto-streamer-out.c (hash_tree): Use it.
5883 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5885 * inchash.h (inchash::hash::add_wide_int): Rename to...
5886 (inchash::hash::add_hwi): ...this.
5887 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5888 (polymorphic_call_target_hasher::hash): Likewise.
5889 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5890 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5891 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5892 * lto-streamer-out.c (hash_tree): Likewise.
5893 * optc-save-gen.awk: Likewise.
5894 * tree.c (add_expr): Likewise.
5896 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5899 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5900 for ordered inequality comparisons even with TARGET_IEEE_FP.
5902 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5905 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5906 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5907 Expand with cmp<dwi>_doubleword.
5909 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5911 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5912 List CET intrinsics.
5913 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5914 specific to -fcf-protection option.
5916 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5918 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5919 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5920 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5921 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5922 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5923 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5924 (extra_objs): Add cet.o for Linux/x86 targets.
5925 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5926 * config/i386/cet.c: New file.
5927 * config/i386/cetintrin.h: Likewise.
5928 * config/i386/t-cet: Likewise.
5929 * config/i386/cpuid.h (bit_SHSTK): New.
5930 (bit_IBT): Likewise.
5931 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5932 pass IBT and SHSTK bits.
5933 * config/i386/i386-builtin-types.def
5934 (VOID_FTYPE_UNSIGNED_PVOID): New.
5935 (VOID_FTYPE_UINT64_PVOID): Likewise.
5936 * config/i386/i386-builtin.def: Add CET intrinsics.
5937 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5938 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5939 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5940 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5942 * config/i386/i386.c (rest_of_insert_endbranch): New.
5943 (pass_data_insert_endbranch): Likewise.
5944 (pass_insert_endbranch): Likewise.
5945 (make_pass_insert_endbranch): Likewise.
5946 (ix86_notrack_prefixed_insn_p): Likewise.
5947 (ix86_target_string): Add -mibt, -mshstk flags.
5948 (ix86_option_override_internal): Add flag_cf_protection
5950 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5951 (ix86_print_operand): Add 'notrack' prefix output.
5952 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5953 (ix86_expand_builtin): Expand CET intrinsics.
5954 (x86_output_mi_thunk): Add 'endbranch' instruction.
5955 * config/i386/i386.h (TARGET_IBT): New.
5956 (TARGET_IBT_P): Likewise.
5957 (TARGET_SHSTK): Likewise.
5958 (TARGET_SHSTK_P): Likewise.
5959 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5960 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5961 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5962 (builtin_setjmp_setup): New pattern.
5963 (builtin_longjmp): Likewise.
5964 (rdssp<mode>): Likewise.
5965 (incssp<mode>): Likewise.
5966 (saveprevssp): Likewise.
5967 (rstorssp): Likewise.
5968 (wrss<mode>): Likewise.
5969 (wruss<mode>): Likewise.
5970 (setssbsy): Likewise.
5971 (clrssbsy): Likewise.
5972 (nop_endbr): Likewise.
5973 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5975 * config/i386/immintrin.h: Include <cetintrin.h>.
5976 * config/i386/linux-common.h
5977 (file_end_indicate_exec_stack_and_cet): New prototype.
5978 (TARGET_ASM_FILE_END): New.
5980 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5982 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5983 latencies instead of having separate table; make difference between
5984 integer and float costs.
5985 * i386.h (processor_costs): Remove scalar_stmt_cost,
5986 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5987 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5989 * x86-tune-costs.h: Remove entries which has been removed in
5990 procesor_costs from all tables; make cond_taken_branch_cost
5991 and cond_not_taken_branch_cost COST_N_INSNS based.
5993 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5995 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5997 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5999 * config/i386/i386.md (isa): Remove fma_avx512f.
6000 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6001 <avx512>_fmadd_<mode>_mask3<round_name>,
6002 <avx512>_fmsub_<mode>_mask<round_name>,
6003 <avx512>_fmsub_<mode>_mask3<round_name>,
6004 <avx512>_fnmadd_<mode>_mask<round_name>,
6005 <avx512>_fnmadd_<mode>_mask3<round_name>,
6006 <avx512>_fnmsub_<mode>_mask<round_name>,
6007 <avx512>_fnmsub_<mode>_mask3<round_name>,
6008 <avx512>_fmaddsub_<mode>_mask<round_name>,
6009 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6010 <avx512>_fmsubadd_<mode>_mask<round_name>,
6011 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6012 (*vec_widen_umult_even_v16si<mask_name>,
6013 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6014 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6016 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6018 * extend.texi: Add 'nocf_check' documentation.
6019 * gimple.texi: Add second parameter to
6020 gimple_build_call_from_tree.
6021 * invoke.texi: Add -fcf-protection documentation.
6022 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6024 2017-10-20 Richard Biener <rguenther@suse.de>
6026 PR tree-optimization/82473
6027 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6028 the largest input type.
6030 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6032 * c-attribs.c (handle_nocf_check_attribute): New function.
6033 (c_common_attribute_table): Add 'nocf_check' handling.
6034 * gimple-parser.c: Add second argument NULL to
6035 gimple_build_call_from_tree.
6036 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6037 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6039 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6040 * common.opt: Add fcf-protection flag.
6041 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6042 * flag-types.h: Add enum cf_protection_level.
6043 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6044 Add 'nocf_check' attribute propagation to gimple call.
6045 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6046 (gimple_build_call_from_tree): Update prototype.
6047 (gimple_call_nocf_check_p): New function.
6048 (gimple_call_set_nocf_check): Likewise.
6049 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6050 * ipa-icf.c: Add nocf_check attribute in statement hash.
6051 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6052 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6053 * toplev.c (process_options): Add flag_cf_protection handling.
6055 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6057 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6059 2017-10-20 Richard Biener <rguenther@suse.de>
6061 PR tree-optimization/82603
6062 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6063 remove false predicated stores.
6065 2017-10-20 Richard Biener <rguenther@suse.de>
6067 * graphite-isl-ast-to-gimple.c
6068 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6069 Remove return value and simplify, dump copied stmt after lhs
6071 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6072 Reduce dump verbosity.
6073 (gsi_insert_earliest): Likewise.
6074 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6075 * graphite.c (print_global_statistics): Adjust dumping.
6076 (print_graphite_scop_statistics): Likewise.
6077 (print_graphite_statistics): Do not dump loops here.
6078 (graphite_transform_loops): But here.
6080 2017-10-20 Nicolas Roche <roche@adacore.com>
6082 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6083 * configure: Regenerate.
6085 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6088 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6089 functions when optimizing replace GIMPLE_RETURN stmts with
6090 calls to __builtin_unreachable ().
6093 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6094 for -fsanitize=thread link of executables.
6095 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6096 link of executables.
6099 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6101 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6102 define_insns for logical vector shifts to use VI248_AVX512BW
6103 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6104 condition, useless isa and prefix attributes. Change the first
6105 2 of these define_insns to ...
6106 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6107 define_insn for avx512vl.
6108 (<shift_insn><mode>3): ... and this, new define_insn without
6109 masking for non-avx512vl.
6112 * config/i386/sse.md (*andnot<mode>3,
6113 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6114 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6115 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6116 not applied use empty suffix even for TARGET_AVX512VL.
6117 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6118 is applied, supply evex,evex or evex,evex,evex instead of just
6121 2017-10-20 Julia Koval <julia.koval@intel.com>
6123 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6124 (OPTION_MASK_ISA_GFNI_UNSET): New.
6125 (ix86_handle_option): Handle OPT_mgfni.
6126 * config/i386/cpuid.h (bit_GFNI): New.
6127 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6128 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6129 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6130 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6131 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6132 * config/i386/i386.opt: Add mgfni.
6134 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6136 * config/msp430/msp430.c (msp430_option_override): Disable
6137 -fdelete-null-pointer-checks.
6138 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6140 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6142 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6143 of x87 and SSE instructions.
6145 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6147 * asan.c (create_cond_insert_point): Do not update edge count.
6148 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6149 (afdo_propagate_circuit): Likewise.
6150 (afdo_calculate_branch_prob): Likewise.
6151 (afdo_annotate_cfg): Likewise.
6152 * basic-block.h (struct edge_def): Remove count.
6153 (edge_def::count): New accessor.
6154 * bb-reorder.c (rotate_loop): Update.
6155 (find_traces_1_round): Update.
6156 (connect_traces): Update.
6157 (sanitize_hot_paths): Update.
6158 * cfg.c (unchecked_make_edge): Update.
6159 (make_single_succ_edge): Update.
6160 (check_bb_profile): Update.
6161 (dump_edge_info): Update.
6162 (update_bb_profile_for_threading): Update.
6163 (scale_bbs_frequencies_int): Update.
6164 (scale_bbs_frequencies_gcov_type): Update.
6165 (scale_bbs_frequencies_profile_count): Update.
6166 (scale_bbs_frequencies): Update.
6167 * cfganal.c (connect_infinite_loops_to_exit): Update.
6168 * cfgbuild.c (compute_outgoing_frequencies): Update.
6169 (find_many_sub_basic_blocks): Update.
6170 * cfgcleanup.c (try_forward_edges): Update.
6171 (try_crossjump_to_edge): Update
6172 * cfgexpand.c (expand_gimple_cond): Update
6173 (expand_gimple_tailcall): Update
6174 (construct_exit_block): Update
6175 * cfghooks.c (verify_flow_info): Update
6176 (redirect_edge_succ_nodup): Update
6177 (split_edge): Update
6178 (make_forwarder_block): Update
6179 (duplicate_block): Update
6180 (account_profile_record): Update
6181 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6182 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6183 * cfgloopmanip.c (scale_loop_profile): Update.
6185 (lv_adjust_loop_entry_edge): Update.
6186 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6187 (force_nonfallthru_and_redirect): Update.
6188 (purge_dead_edges): Update.
6189 (rtl_flow_call_edges_add): Update.
6190 * cgraphunit.c (init_lowered_empty_function): Update.
6191 (cgraph_node::expand_thunk): Update.
6192 * gimple-pretty-print.c (dump_probability): Update.
6193 (dump_edge_probability): Update.
6194 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6195 * haifa-sched.c (sched_create_recovery_edges): Update.
6196 * hsa-gen.c (convert_switch_statements): Update.
6197 * ifcvt.c (dead_or_predicable): Update.
6198 * ipa-inline-transform.c (inline_transform): Update.
6199 * ipa-split.c (split_function): Update.
6200 * ipa-utils.c (ipa_merge_profiles): Update.
6201 * loop-doloop.c (add_test): Update.
6202 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6203 * lto-streamer-in.c (input_cfg): Update.
6204 (input_function): Update.
6205 * lto-streamer-out.c (output_cfg): Update.
6206 * modulo-sched.c (sms_schedule): Update.
6207 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6208 * predict.c (maybe_hot_edge_p): Update.
6209 (unlikely_executed_edge_p): Update.
6210 (probably_never_executed_edge_p): Update.
6211 (dump_prediction): Update.
6212 (drop_profile): Update.
6213 (propagate_unlikely_bbs_forward): Update.
6214 (determine_unlikely_bbs): Update.
6215 (force_edge_cold): Update.
6216 * profile.c (compute_branch_probabilities): Update.
6217 * reg-stack.c (better_edge): Update.
6218 * shrink-wrap.c (handle_simple_exit): Update.
6219 * tracer.c (better_p): Update.
6220 * trans-mem.c (expand_transaction): Update.
6221 (split_bb_make_tm_edge): Update.
6222 * tree-call-cdce.c: Update.
6223 * tree-cfg.c (gimple_find_sub_bbs): Update.
6224 (gimple_split_edge): Update.
6225 (gimple_duplicate_sese_region): Update.
6226 (gimple_duplicate_sese_tail): Update.
6227 (gimple_flow_call_edges_add): Update.
6228 (insert_cond_bb): Update.
6229 (execute_fixup_cfg): Update.
6230 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6231 * tree-complex.c (expand_complex_div_wide): Update.
6232 * tree-eh.c (lower_resx): Update.
6233 (unsplit_eh): Update.
6234 (cleanup_empty_eh_move_lp): Update.
6235 * tree-inline.c (copy_edges_for_bb): Update.
6236 (freqs_to_counts): Update.
6237 (copy_cfg_body): Update.
6238 * tree-ssa-dce.c (remove_dead_stmt): Update.
6239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6240 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6241 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6242 (unloop_loops): Update.
6243 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6244 * tree-ssa-loop-split.c (connect_loops): Update.
6245 (split_loop): Update.
6246 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6247 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6248 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6249 * tree-ssa-reassoc.c (branch_fixup): Update.
6250 * tree-ssa-tail-merge.c (replace_block_by): Update.
6251 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6252 (compute_path_counts): Update.
6253 (update_profile): Update.
6254 (recompute_probabilities): Update.
6255 (update_joiner_offpath_counts): Update.
6256 (estimated_freqs_path): Update.
6257 (freqs_to_counts_path): Update.
6258 (clear_counts_path): Update.
6259 (ssa_fix_duplicate_block_edges): Update.
6260 (duplicate_thread_path): Update.
6261 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6262 (case_bit_test_cmp): Update.
6263 (collect_switch_conv_info): Update.
6264 (gen_inbound_check): Update.
6265 (do_jump_if_equal): Update.
6266 (emit_cmp_and_jump_insns): Update.
6267 * tree-tailcall.c (decrease_profile): Update.
6268 (eliminate_tail_call): Update.
6269 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6270 (vect_do_peeling): Update.
6271 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6272 * ubsan.c (ubsan_expand_null_ifn): Update.
6273 (ubsan_expand_ptr_ifn): Update.
6274 * value-prof.c (gimple_divmod_fixed_value): Update.
6275 (gimple_mod_pow2): Update.
6276 (gimple_mod_subtract): Update.
6277 (gimple_ic): Update.
6278 (gimple_stringop_fixed_value): Update.
6280 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6283 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6285 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6287 PR rtl-optimization/82395
6288 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6289 based on non_spilled_static_chain_regno_p.
6291 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6293 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6294 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6295 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6297 2017-10-19 Martin Sebor <msebor@redhat.com>
6299 PR tree-optimization/82596
6300 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6302 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6304 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6305 (get_mem_refs_of_builtin_call): Likewise.
6306 * builtins.c (expand_builtin_apply): Adjust call to
6307 allocate_dynamic_stack_space.
6308 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6309 the third argument to allocate_dynamic_stack_space, otherwise -1.
6310 (expand_builtin): Deal with all alloca variants.
6311 (is_inexpensive_builtin): Likewise.
6312 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6313 * calls.c (special_function_p): Deal with all alloca variants.
6314 (initialize_argument_information): Adjust call to
6315 allocate_dynamic_stack_space.
6316 (expand_call): Likewise.
6317 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6318 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6319 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6320 use it for the stack usage computation.
6321 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6322 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6323 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6324 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6325 (in_loop_p): Remove first argument and useless check.
6326 (pass_walloca::execute): Remove useless test and adjust call to above.
6327 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6328 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6329 (gimplify_call_expr): Deal with all alloca variants.
6330 * hsa-gen.c (gen_hsa_alloca): Likewise.
6331 (gen_hsa_insns_for_call): Likewise.
6332 * ipa-pure-const.c (special_builtin_state): Likewise.
6333 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6334 * tree-object-size.c (alloc_object_size): Likewise.
6335 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6336 (call_may_clobber_ref_p_1): Likewise.
6337 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6338 (ccp_fold_stmt): Likewise.
6339 (optimize_stack_restore): Likewise.
6340 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6341 (mark_all_reaching_defs_necessary_1): Likewise.
6342 (propagate_necessity): Likewise.
6343 (eliminate_unnecessary_stmts): Likewise.
6344 * tree.c (build_common_builtin_nodes): Build
6345 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6346 (build_alloca_call_expr): New function.
6347 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6348 (CASE_BUILT_IN_ALLOCA): Likewise.
6349 (build_alloca_call_expr): Declare.
6350 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6352 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6355 * dwarf2out.c (new_die_raw): New static inline function.
6356 (new_die): Use it to create the DIE.
6357 (add_AT_external_die_ref): Likewise.
6358 (clone_die): Likewise.
6359 (clone_as_declaration): Likewise.
6360 (dwarf2out_vms_debug_main_pointer): Likewise.
6361 (base_type_die): Likewise. Remove early return for corner cases.
6362 Do not call add_pubtype on the DIE here.
6363 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6364 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6365 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6366 native order exists for base types, attach the DIE manually and call
6367 add_pubtype on it. Do not equate a reverse order DIE to the type.
6369 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6371 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6372 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6374 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6376 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6378 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6379 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6380 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6381 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6382 * opts.c (sanitizer_opts): Add builtin.
6383 * ubsan.c (instrument_builtin): New function.
6384 (pass_ubsan::execute): Call it.
6385 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6386 * doc/invoke.texi: Document -fsanitize=builtin.
6388 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6389 builtins, store max (log2 (align), 0) into uchar field instead of
6390 align into uptr field.
6391 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6392 store uchar 0 field instead of uptr 0 field.
6393 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6394 instead of passing one address of struct with 2 locations pass
6395 two addresses of structs with 1 location each.
6396 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6397 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6398 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6399 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6400 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6401 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6402 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6403 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6405 2017-10-19 Martin Liska <mliska@suse.cz>
6408 * file-find.c (remove_prefix): Remove.
6409 * file-find.h (remove_prefix): Likewise.
6410 * gcc-ar.c: Remove smartness of lookup.
6412 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6414 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6415 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6416 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6418 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6421 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6422 (setcc + and to xor + setcc): New peephole2.
6424 2017-10-19 Tom de Vries <tom@codesourcery.com>
6426 * doc/sourcebuild.texi (Test Directives, Variants of
6427 dg-require-support): Add dg-require-stack-size.
6429 2017-10-19 Martin Liska <mliska@suse.cz>
6432 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6433 that have a large alignment.
6434 (gimplify_target_expr): Likewise.
6436 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6438 PR rtl-optimization/82602
6439 * ira.c (rtx_moveable_p): Return false for volatile asm.
6441 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6444 * config/i386/i386-modes.def (CCGZ): New CC mode.
6445 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6446 * config/i386/predicates.md (ix86_comparison_operator):
6448 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6449 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6450 with double-word subtraction.
6451 (put_condition_code): Handle CCGZmode.
6453 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6455 * wide-int.cc (debug (const wide_int &)): New.
6456 (debug (const wide_int *)): New.
6457 (debug (const widest_int &)): New.
6458 (debug (const widest_int *)): New.
6460 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6463 * lra-constraints.c (curr_insn_transform): Use non-input operand
6464 instead of output one for matched reload.
6466 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6468 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6469 (tree-ssa-loop-ivopts.h): New header file.
6470 (struct builtin_info): New fields.
6471 (classify_builtin_1): Compute and record base and offset parts for
6472 memset builtin partition by calling strip_offset.
6473 (offset_cmp, fuse_memset_builtins): New functions.
6474 (finalize_partitions): Fuse adjacent memset partitions by calling
6476 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6477 Expose the interface.
6478 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6480 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6482 PR tree-optimization/82574
6483 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6484 that data reference must be executed exactly once per iteration
6485 against the outermost loop in nest.
6486 (classify_partition): Update call to above function.
6488 2017-10-18 Richard Biener <rguenther@suse.de>
6490 PR tree-optimization/82591
6491 * graphite.c (graphite_transform_loops): Move code gen message
6493 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6494 Here. Handle scop_to_isl_ast failing.
6495 (scop_to_isl_ast): Limit the number of ISL operations.
6497 2017-10-18 Richard Biener <rguenther@suse.de>
6499 * graphite-isl-ast-to-gimple.c
6500 (translate_isl_ast_to_gimple::set_rename): Simplify.
6501 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6502 (graphite_copy_stmts_from_block): ... here.
6503 (copy_bb_and_scalar_dependences): Simplify.
6504 (add_parameters_to_ivs_params): Canonicalize.
6505 (generate_entry_out_of_ssa_copies): Simplify.
6506 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6507 by passing in ISL dimension.
6508 (parameter_index_in_region_1): Rename to ...
6509 (parameter_index_in_region): ... this.
6510 (extract_affine): Adjust assert, pass down parameter index.
6511 (add_param_constraints): Use range-info when available.
6512 (build_scop_context): Adjust.
6513 * sese.c (new_sese_info): Adjust.
6514 (free_sese_info): Likewise.
6515 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6516 Remove unused typedefs.
6517 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6519 2017-10-18 Martin Liska <mliska@suse.cz>
6521 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6523 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6525 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6526 (s390_sched_init): Do not reset s390_sched_state if we entered the
6527 current basic block via a fallthru edge and all others are unlikely.
6529 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6531 * config/s390/s390.c (NUM_SIDES): New variable.
6532 (LONGRUNNING_THRESHOLD): New variable.
6533 (LATENCY_FACTOR): New variable.
6534 (s390_sched_score): Decrease score for long-running instructions on
6536 (s390_sched_variable_issue): Perform bookkeeping for long-running
6539 2017-10-18 Richard Biener <rguenther@suse.de>
6541 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6542 Simplify with removal of the parameter rename map.
6543 (set_rename): Likewise.
6544 (should_copy_to_new_region): Likewise.
6545 (graphite_copy_stmts_from_block): Likewise.
6546 (copy_bb_and_scalar_dependences): Remove initialization of
6547 unused copied_bb_map.
6549 (copy_internal_parameters): Likewise.
6550 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6551 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6552 Use INTEGRAL_TYPE_P.
6553 (parameter_index_in_region_1): Rename to ...
6554 (assign_parameter_index_in_region): ... this. Assert we have
6555 a parameter we handle.
6556 (scan_tree_for_params): Adjust.
6557 * sese.h (parameter_rename_map_t): Remove.
6558 (struct sese_info_t): Remove unused parameter_rename_map and
6559 copied_bb_map members.
6560 * sese.c (new_sese_info): Adjust.
6561 (free_sese_info): Likewise.
6563 2017-10-18 Martin Liska <mliska@suse.cz>
6566 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6567 on an abnormal edge.
6569 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6571 * doc/invoke.texi (ffunction-sections and fdata-sections):
6574 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6576 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6577 the use statement can throw internally.
6579 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6581 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6582 any RTX present on the RHS of a SET.
6583 * compare-elim.c (try_eliminate_compare): Restore comment.
6585 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6587 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6588 may be also called on tcc_type nodes.
6589 * langhooks.c (lhd_tree_size): Likewise.
6591 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6593 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6594 format_warning_at_substring.
6595 (maybe_warn): Convert source_range * param to a location_t. Pass
6596 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6597 (format_directive): Remove code to extract source_ranges and
6598 source_range * in favor of just a location_t.
6599 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6601 * substring-locations.c (format_warning_va): Convert
6602 source_range * param to a location_t.
6603 (format_warning_at_substring): Likewise.
6604 * substring-locations.h (format_warning_va): Likewise.
6605 (format_warning_at_substring): Likewise.
6607 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6609 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6611 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6612 and scatter/gather ops.
6614 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6615 vec_gather_load and vec_scatter_store.
6616 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6617 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6618 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6619 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6620 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6621 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6623 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6625 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6626 Detect FTST instruction and handle its register pops. Only pop
6627 second operand if can_pop_second_op is true.
6628 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6629 set can_pop_second_op to false in the compare_for_stack_reg call.
6631 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6632 output_fp_compare for stack register operands.
6633 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6634 instructions here. Do not emit stack register pops here. Assert
6635 that FCOMPP pops next to top stack register. Rewrite function.
6637 2017-10-17 Nathan Sidwell <nathan@acm.org>
6640 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6641 use DECL_ASSEMBLER_NAME_RAW.
6644 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6647 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6648 Wilco Dijkstra <wilco.dijkstra@arm.com>
6650 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6651 converstion to Pmode from the buf_addr.
6653 2017-10-17 Richard Biener <rguenther@suse.de>
6655 * graphite-dependences.c (scop_get_reads_and_writes): Change
6656 output parameters to references.
6658 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6660 PR 71026/tree-optimization
6661 * fold-const.c (distribute_real_division): Removed.
6662 (fold_binary_loc): Remove calls to distribute_real_divison.
6664 2017-10-17 Richard Biener <rguenther@suse.de>
6666 * graphite-scop-detection.c
6667 (scop_detection::stmt_has_simple_data_refs_p): Always use
6668 the full nest as region.
6669 (try_generate_gimple_bb): Likewise.
6670 * sese.c (scalar_evolution_in_region): Simplify now that
6671 SCEV can handle instantiation in regions.
6672 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6673 in the non-loop part of a function if requested.
6675 2017-10-17 Richard Biener <rguenther@suse.de>
6677 PR tree-optimization/82563
6678 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6680 (graphite_regenerate_ast_isl): Call it.
6681 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6683 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6685 PR tree-optimization/82549
6686 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6687 Formatting fixes. Instead of calling make_bit_field_ref with negative
6690 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6692 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6693 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6694 _mm_maskz_reduce_ss): New.
6695 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6696 __builtin_ia32_reducess_mask): Ditto..
6697 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
6698 * config/i386/sse.md (reduces<mode>): Renamed to ...
6699 (reduces<mode><mask_scalar_name>): ... this.
6700 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
6702 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
6703 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
6705 2017-10-16 David Malcolm <dmalcolm@redhat.com>
6707 * Makefile.in (OBJS): Add unique-ptr-tests.o.
6708 * selftest-run-tests.c (selftest::run_tests): Call
6709 selftest::unique_ptr_tests_cc_tests.
6710 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
6711 * unique-ptr-tests.cc: New file.
6713 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
6716 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6718 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6719 (make_hard_regno_born, make_hard_regno_dead): Update
6720 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
6722 2017-10-16 Jeff Law <law@redhat.com>
6724 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6726 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
6728 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
6730 2017-10-16 Olivier Hainque <hainque@adacore.com>
6732 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
6733 with_cpu if we were configured for an e500v2 target cpu name.
6735 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
6737 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
6738 * doc/invoke.texi: Document +nodsp as a valid extension for
6741 2017-10-16 Martin Liska <mliska@suse.cz>
6743 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
6744 (test_set_range): Likewise.
6745 (test_range_functions): Rename to ...
6746 (test_bit_in_range): ... this.
6747 (sbitmap_c_tests): Add new test.
6749 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6751 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
6753 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
6754 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
6756 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6758 * config/aarch64/aarch64-builtins.c
6759 (aarch64_types_quadopu_lane_qualifiers): New.
6760 (TYPES_QUADOPU_LANE): New.
6761 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
6762 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
6763 (aarch64_<sur>dot_laneq<vsi2qi>): New.
6764 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
6765 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
6766 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
6767 (Vdottype, DOTPROD): New.
6768 (sur): Add SDOT and UDOT.
6770 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6772 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
6773 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
6774 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
6776 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
6777 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
6778 Enable TARGET_DOTPROD.
6779 (cortex-a75.cortex-a55): Likewise.
6780 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
6782 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6784 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
6785 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
6787 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
6789 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
6790 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
6791 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
6792 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
6793 * config/arm/types.md (neon_dot, neon_dot_q): New.
6794 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
6796 2017-10-16 Tamar Christina <tamar.christina@arm.com>
6798 * config/arm/arm.h (TARGET_DOTPROD): New.
6799 * config/arm/arm.c (arm_arch_dotprod): New.
6800 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
6801 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
6802 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
6803 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
6804 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
6805 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
6806 * doc/invoke.texi (armv8.2-a): Document dotprod
6808 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6810 * i386.c (ix86_vec_cost): New function.
6811 (ix86_rtx_costs): Handle vector operations better.
6812 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6813 * x86-tune-costs.h: Add new costs to all tables.
6815 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
6817 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
6819 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
6820 divsd, sqrtss and sqrtsd
6821 * x86-tune-costs.h: Add new entries to all costs.
6822 (znver1_cost): Fix to match real instruction latencies.
6824 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6825 Michael Collison <michael.collison@arm.com>
6827 * compare-elim.c: Include emit-rtl.h.
6828 (can_merge_compare_into_arith): New function.
6829 (try_validate_parallel): Likewise.
6830 (try_merge_compare): Likewise.
6831 (try_eliminate_compare): Call the above when no previous clobber
6833 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
6836 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6840 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
6841 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
6845 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
6846 to be any operand_equal_p operands. For & (B - 1) require
6847 B to be power of 2. Recognize
6848 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
6850 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6853 * optabs.c (expand_memory_blockage): Fix call of
6854 targetm.have_memory_blockage.
6856 2017-10-14 Jakub Jelinek <jakub@redhat.com>
6859 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
6860 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
6861 objects to extra_objs instead of overwriting it.
6863 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6865 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
6866 Use any_fp_register_operand as operand[3] predicate. Simplify
6867 equality test for operands[2] and operands[4] memory location.
6868 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
6869 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
6870 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
6871 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
6872 any_fp_register_operand as operand[1] predicate. Simplify
6873 equality test for operands[0] and operands[3] memory location.
6874 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
6875 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
6876 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
6878 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
6880 * target-insns.def: Add memory_blockage.
6881 * optabs.c (expand_memory_blockage): New function.
6882 (expand_asm_memory_barrier): Rename ...
6883 (expand_asm_memory_blockage): ... to this.
6884 (expand_mem_thread_fence): Call expand_memory_blockage
6885 instead of expand_asm_memory_barrier.
6886 (expand_mem_singnal_fence): Ditto.
6887 (expand_atomic_load): Ditto.
6888 (expand_atomic_store): Ditto.
6889 * doc/md.texi (Standard Pattern Names For Generation):
6890 Document memory_blockage instruction pattern.
6892 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6894 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6895 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6896 * config/rl78/rl78.md: New define_expand "adddi3".
6898 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6900 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6903 2017-10-13 Jeff Law <law@redhat.com>
6905 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6907 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6910 * internal-fn.c (expand_mul_overflow): If both operands have
6911 the same highpart of -1 or 0 and the topmost bit of lowpart
6912 is different, overflow is if res <= 0 rather than res < 0.
6914 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6916 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6917 TARGET_P9_VECTOR code for unaligned_load case.
6919 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6921 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6923 2017-10-13 Nathan Sidwell <nathan@acm.org>
6925 * tree-core.h (tree_contains_struct): Make bool.
6926 * tree.c (tree_contains_struct): Likewise.
6927 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6928 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6929 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6930 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6931 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6933 2017-10-13 Richard Biener <rguenther@suse.de>
6935 * graphite-isl-ast-to-gimple.c
6936 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6937 parameters and dominance check.
6938 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6939 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6940 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6941 Do not update SSA form here or do intermediate IL verification.
6942 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6943 (graphite_initialize): Remove check on the number of loops in
6944 the function and inline into graphite_transform_loops.
6945 (graphite_finalize): Inline into graphite_transform_loops.
6946 (graphite_transform_loops): Perform SSA update and IL verification
6948 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6950 2017-10-13 Richard Biener <rguenther@suse.de>
6952 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6953 graphite_expression_type_precision): Avoid global constructor
6955 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6956 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6957 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6958 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6959 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6960 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6962 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6965 * config/i386/i386.h (ix86_red_zone_size): New.
6966 * config/i386/i386.md (push peephole2s): Replace
6967 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6969 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6970 Alan Hayward <alan.hayward@arm.com>
6971 David Sherwood <david.sherwood@arm.com>
6973 * combine.c (can_change_dest_mode): Reject changes in
6974 REGMODE_NATURAL_SIZE.
6976 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6977 Alan Hayward <alan.hayward@arm.com>
6978 David Sherwood <david.sherwood@arm.com>
6980 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6981 (expand_debug_source_expr): Likewise.
6982 * combine.c (combine_simplify_rtx): Likewise.
6983 * cse.c (fold_rtx): Likewise.
6984 * fwprop.c (canonicalize_address): Likewise.
6985 * targhooks.c (default_shift_truncation_mask): Likewise.
6987 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6992 (widened_mode): Likewise.
6993 (expand_unop): Likewise.
6994 * ree.c (transform_ifelse): Likewise.
6995 (merge_def_and_ext): Likewise.
6996 (combine_reaching_defs): Likewise.
6997 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6999 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7000 Alan Hayward <alan.hayward@arm.com>
7001 David Sherwood <david.sherwood@arm.com>
7003 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7004 * combine.c (gen_lowpart_for_combine): Likewise.
7005 * dwarf2out.c (rtl_for_decl_location): Likewise.
7006 * final.c (alter_subreg): Likewise.
7007 * rtlhooks.c (gen_lowpart_general): Likewise.
7008 (gen_lowpart_if_possible): Likewise.
7010 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7011 Alan Hayward <alan.hayward@arm.com>
7012 David Sherwood <david.sherwood@arm.com>
7014 * calls.c (expand_call): Use subreg_lowpart_offset.
7015 * cse.c (cse_insn): Likewise.
7016 * regcprop.c (copy_value): Likewise.
7017 (copyprop_hardreg_forward_1): Likewise.
7019 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7022 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7023 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7024 =Q constraints to +Q and into insn condition add check
7025 that operands[0] and operands[1] are equal.
7026 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7027 =Q constraints to +Q and into insn condition add check
7028 that operands[0] is equal to either operands[1] or operands[2].
7031 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7032 instead of handling MINUS_EXPR twice (once for each argument),
7033 canonicalize operand order and handle just once, use rtype where
7034 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7037 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7038 any values of __C while still being pattern recognizable as a simple
7041 2017-10-13 Richard Biener <rguenther@suse.de>
7043 PR tree-optimization/82451
7045 2017-10-02 Richard Biener <rguenther@suse.de>
7047 PR tree-optimization/82355
7048 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7049 a mapping for the enclosing loop but avoid generating one for
7051 (copy_bb_and_scalar_dependences): Remove premature codegen
7052 error on PHIs in blocks duplicated into multiple places.
7053 * graphite-scop-detection.c
7054 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7055 in the region use it as loop and nest to analyze the DR in.
7056 (try_generate_gimple_bb): Likewise.
7057 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7058 (add_loop_constraints): For blocks in a loop not in the region
7059 create a dimension with a single iteration.
7060 * sese.h (gbb_loop_at_index): Remove assert.
7062 * cfgloop.c (loop_preheader_edge): For the loop tree root
7063 return the single successor of the entry block.
7064 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7065 Reset the SCEV hashtable and niters.
7066 * graphite-scop-detection.c
7067 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7068 assert that we only have POLYNOMIAL_CHREC that vary in loops
7069 contained in the region.
7070 (scop_detection::graphite_can_represent_expr): Adjust.
7071 (scop_detection::stmt_has_simple_data_refs_p): For loops
7072 not in the region set loop to NULL. The nest is now the
7073 entry edge to the region.
7074 (try_generate_gimple_bb): Likewise.
7075 * sese.c (scalar_evolution_in_region): Adjust for
7076 instantiate_scev change.
7077 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7078 Make nest parameter the edge into the region.
7079 (create_data_ref): Likewise.
7080 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7081 entry edge into a region and adjust instantiate_scev calls.
7082 (create_data_ref): Likewise.
7083 (graphite_find_data_references_in_stmt): Likewise.
7084 (find_data_references_in_stmt): Pass the loop preheader edge
7085 from the nest argument.
7086 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7087 parameter the edge into the region.
7088 (instantiate_parameters): Use the loop preheader edge as entry.
7089 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7091 (get_instantiated_value_entry): Make instantiate_below parameter
7092 the edge into the region.
7093 (instantiate_scev_name): Likewise. Adjust dominance checks,
7094 when we cannot use loop-based instantiation instantiate by
7095 walking use-def chains.
7096 (instantiate_scev_poly): Adjust.
7097 (instantiate_scev_binary): Likewise.
7098 (instantiate_scev_convert): Likewise.
7099 (instantiate_scev_not): Likewise.
7100 (instantiate_array_ref): Remove.
7101 (instantiate_scev_3): Likewise.
7102 (instantiate_scev_2): Likewise.
7103 (instantiate_scev_1): Likewise.
7104 (instantiate_scev_r): Do not blindly handle N-operand trees.
7105 Do not instantiate array-refs. Handle all constants and invariants.
7106 (instantiate_scev): Make instantiate_below parameter
7107 the edge into the region.
7108 (resolve_mixers): Use the loop preheader edge for the region
7109 parameter to instantiate_scev_r.
7110 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7112 2017-10-13 Richard Biener <rguenther@suse.de>
7114 PR tree-optimization/82525
7115 * graphite-isl-ast-to-gimple.c
7116 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7118 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7119 Fail code generation when we cannot represent the isl integer.
7120 (binary_op_to_tree): Elide modulo operations that are no-ops
7121 in the type we code generate. Remove now superfluous code
7124 2017-10-13 Richard Biener <rguenther@suse.de>
7126 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7127 (scop_detection::harmful_loop_in_region): Remove premature
7128 IV type restriction.
7129 (scop_detection::graphite_can_represent_scev): We can handle
7130 pointer IVs just fine.
7132 2017-10-13 Alan Modra <amodra@gmail.com>
7134 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7135 "Clobbers and Scratch Registers". Add paragraph on
7136 alternative to clobbers for scratch registers and OpenBLAS
7139 2017-10-13 Alan Modra <amodra@gmail.com>
7141 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7142 example of a memory input for a string of known length. Move
7143 commentary out of table. Add a number of new examples
7144 covering array memory inputs.
7146 2017-10-12 Martin Liska <mliska@suse.cz>
7148 PR tree-optimization/82493
7149 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7150 (test_range_functions): New function.
7151 (sbitmap_c_tests): Likewise.
7152 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7153 * selftest.h (sbitmap_c_tests): New function.
7155 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7157 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7159 * config/rs6000/amo.h: Fix spacing issue.
7161 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7164 * config/i386/i386.md (*ashl<mode>3_mask_1,
7165 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7166 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7169 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7171 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7172 (profile_probability): Set max_probability
7173 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7175 * profile-count.c (profile_probability::differs_from_p): Do not
7176 rely on max_probaiblity == 10000
7178 2017-10-12 Jeff Law <law@redhat.com>
7180 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7183 2017-10-12 Martin Sebor <msebor@redhat.com>
7187 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7188 (handle_alias_pairs): Call it.
7189 * common.opt (-Wattribute-alias): New option.
7190 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7191 * doc/invoke.texi (-Wattribute-alias): Document.
7193 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7196 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7198 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7200 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7201 (make_hard_regno_born, make_hard_regno_dead): Update
7202 bb_killed_pseudos and bb_gen_pseudos.
7204 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7206 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7208 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7210 * config/alpha/alpha.c (alpha_split_conditional_move):
7211 Use std::swap instead of manually swapping.
7212 (alpha_stdarg_optimize_hook): Ditto.
7213 (alpha_canonicalize_comparison): Ditto.
7215 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7217 * tree-loop-distribution.c (struct builtin_info): New struct.
7218 (struct partition): Refactor fields into struct builtin_info.
7219 (partition_free): Free struct builtin_info.
7220 (build_size_arg_loc, build_addr_arg_loc): Delete.
7221 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7222 information from struct builtin_info.
7223 (find_single_drs): New function refactored from classify_partition.
7224 Also moved builtin validity checks to this function.
7225 (compute_access_range, alloc_builtin): New functions.
7226 (classify_builtin_st, classify_builtin_ldst): New functions.
7227 (classify_partition): Refactor code into functions find_single_drs,
7228 classify_builtin_st and classify_builtin_ldst.
7229 (distribute_loop): Don't do runtime alias check when distributing
7231 (find_seed_stmts_for_distribution): New function.
7232 (pass_loop_distribution::execute): Refactor code finding seed
7233 stmts into above function. Support distribution for the innermost
7234 two-level loop nest. Adjust dump information.
7236 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7238 * tree-loop-distribution.c: Adjust the general comment.
7239 (NUM_PARTITION_THRESHOLD): New macro.
7240 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7241 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7242 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7243 in call to build_partition_graph.
7244 (finalize_partitions): New parameter. Make loop distribution more
7245 conservative by fusing more partitions.
7246 (distribute_loop): Don't do runtime alias check in case of loop nest
7248 (find_seed_stmts_for_distribution): New function.
7249 (prepare_perfect_loop_nest): New function.
7250 (pass_loop_distribution::execute): Refactor code finding seed stmts
7251 and loop nest into above functions. Support loop nest distribution.
7252 Adjust dump information accordingly.
7254 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7256 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7257 and set PTYPE_SEQUENTIAL for merged partition.
7259 2017-10-12 Richard Biener <rguenther@suse.de>
7261 PR tree-optimization/69728
7263 2017-09-19 Richard Biener <rguenther@suse.de>
7265 PR tree-optimization/69728
7266 * graphite-sese-to-poly.c (schedule_error): New global.
7267 (add_loop_schedule): Handle empty domain by failing the
7269 (build_original_schedule): Handle schedule_error.
7271 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7272 domain by returning an unchanged schedule.
7274 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7276 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7277 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7279 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7281 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7284 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7287 * expr.c (store_field): Don't optimize away bitsize == 0 store
7288 from CALL_EXPR with addressable return type.
7290 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7292 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7293 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7294 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7295 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7296 TARGET_ISEL instead of TARGET_ISEL<sel>.
7298 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7300 * config/rs6000/rs6000.c
7301 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7303 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7305 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7307 (reg_or_cint_operand): Fix comment.
7308 (reg_or_zero_operand): New predicate.
7309 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7310 * config/rs6000/rs6000.c (output_isel): Delete.
7311 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7312 instead of reg_or_cint_operand. Output instruction directly (not via
7314 (isel_unsigned_<mode>): Ditto.
7315 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7316 gpc_reg_operand. Add an instruction alternative for this. Output
7317 instruction directly.
7318 (*isel_reversed_unsigned_<mode>): Ditto.
7320 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7322 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7323 (TARGET_CANONICALIZE_COMPARISON): Define.
7325 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7328 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7329 Check whether the dest is REG before adding REG_EQUIV note.
7331 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7334 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7336 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7337 (make_hard_regno_born, make_hard_regno_dead): Update
7338 bb_killed_pseudos and bb_gen_pseudos.
7340 2017-10-11 Nathan Sidwell <nathan@acm.org>
7342 * incpath.h (enum incpath_kind): Name enum, prefix values.
7343 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7344 * incpath.c (heads, tails): Use INC_MAX.
7345 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7346 (merge_include_chains, split_quote_chain,
7347 register_include_chains): Update incpath_kind names.
7348 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7349 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7351 (add_framework_path, darwin_register_objc_includes): Likewise.
7352 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7354 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7356 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7357 Do not use float_operator operator predicate.
7358 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7359 * config/i386/predicates.md (float_operator): Remove predicate.
7361 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7363 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7364 (*jccxf_i387): Ditto.
7365 (*jcc<mode>_i387): Ditto.
7366 (*jccu<mode>_i387): Ditto.
7367 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7368 (*jcc_*_i387 splitters): Remove.
7369 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7370 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7371 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7374 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7376 * profile-count.h (slow_safe_scale_64bit): New function.
7377 (safe_scale_64bit): New inline.
7378 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7379 * profile-count.c: Include wide-int.h
7380 (slow_safe_scale_64bit): New.
7382 2017-10-11 Nathan Sidwell <nathan@acm.org>
7384 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7385 HAS_DECL_ASSEMBLER_NAME_P.
7386 * gimple-expr.c (gimple_decl_printable_name: Check
7387 HAS_DECL_ASSEMBLER_NAME_P too.
7388 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7389 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7390 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7391 * passes.c (rest_of_decl_compilation): Check
7392 HAS_DECL_ASSEMBLER_NAME_P too.
7393 * recog.c (verify_changes): Likewise.
7394 * tree-pretty-print.c (dump_decl_name): Likewise.
7395 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7397 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7398 (SET_DECL_ASSEMBLER_NAME): Use it.
7399 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7400 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7401 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7403 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7405 * config.gcc (i386, x86_64): Add extra objects.
7406 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7407 (ix86_min_insn_size): Declare.
7408 (ix86_issue_rate): Declare.
7409 (ix86_adjust_cost): Declare.
7410 (ia32_multipass_dfa_lookahead): Declare.
7411 (ix86_macro_fusion_p): Declare.
7412 (ix86_macro_fusion_pair_p): Declare.
7413 (ix86_bd_has_dispatch): Declare.
7414 (ix86_bd_do_dispatch): Declare.
7415 (ix86_core2i7_init_hooks): Declare.
7416 (ix86_atom_sched_reorder): Declare.
7417 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7418 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7419 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7420 (rip_relative_addr_p): Rename to ...
7421 (ix86_rip_relative_addr_p): ... this one; export.
7422 (memory_address_length): Update.
7423 (ix86_issue_rate): Move to x86-tune-sched.c.
7424 (ix86_flags_dependent): Move to x86-tune-sched.c.
7425 (ix86_agi_dependent): Move to x86-tune-sched.c.
7426 (exact_dependency_1): Move to x86-tune-sched.c.
7427 (exact_store_load_dependency): Move to x86-tune-sched.c.
7428 (ix86_adjust_cost): Move to x86-tune-sched.c.
7429 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7430 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7431 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7432 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7433 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7434 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7435 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7436 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7437 (min_insn_size): Rename to ...
7438 (ix86_min_insn_size): ... this one; export.
7439 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7440 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7441 (core2i7_first_cycle_multipass_backtrack): Move to
7442 x86-tune-sched-core.c.
7443 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7444 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7445 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7446 (ix86_avoid_jump_mispredicts): Update.
7447 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7448 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7449 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7450 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7451 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7452 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7453 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7454 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7455 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7456 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7457 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7458 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7459 (BIG): Move to ix86-tune-sched-bd.c.
7460 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7461 (enum insn_path): Move to ix86-tune-sched-bd.c.
7462 (get_mem_group): Move to ix86-tune-sched-bd.c.
7463 (is_cmp): Move to ix86-tune-sched-bd.c.
7464 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7465 (is_branch): Move to ix86-tune-sched-bd.c.
7466 (is_prefetch): Move to ix86-tune-sched-bd.c.
7467 (init_window): Move to ix86-tune-sched-bd.c.
7468 (allocate_window): Move to ix86-tune-sched-bd.c.
7469 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7470 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7471 (process_end_window): Move to ix86-tune-sched-bd.c.
7472 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7473 (find_constant): Move to ix86-tune-sched-bd.c.
7474 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7475 (has_immediate): Move to ix86-tune-sched-bd.c.
7476 (get_insn_path): Move to ix86-tune-sched-bd.c.
7477 (get_insn_group): Move to ix86-tune-sched-bd.c.
7478 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7479 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7480 (add_insn_window): Move to ix86-tune-sched-bd.c.
7481 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7482 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7483 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7484 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7485 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7486 (do_dispatch): Move to ix86-tune-sched-bd.c.
7487 (has_dispatch): Move to ix86-tune-sched-bd.c.
7488 * config/i386/t-i386: Add new object files.
7489 * config/i386/x86-tune-costs.h: New file.
7490 * config/i386/x86-tune-sched-atom.c: New file.
7491 * config/i386/x86-tune-sched-bd.c: New file.
7492 * config/i386/x86-tune-sched-core.c: New file.
7493 * config/i386/x86-tune-sched.c: New file.
7495 2017-10-11 Liu Hao <lh_mouse@126.com>
7497 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7498 the generic version below instead.
7499 (should_colorize): Recognize Windows consoles as terminals
7501 * pretty-print.c [__MINGW32__] (write_all): New function.
7502 [__MINGW32__] (find_esc_head): Likewise.
7503 [__MINGW32__] (find_esc_terminator): Likewise.
7504 [__MINGW32__] (eat_esc_sequence): Likewise.
7505 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7507 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7510 2017-10-11 Richard Biener <rguenther@suse.de>
7512 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7513 Properly call analyze_scalar_evolution with the loop of the stmt.
7515 2017-10-11 Richard Biener <rguenther@suse.de>
7517 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7518 * tree-core.h (tree_base): Add chrec_var union member.
7519 * tree.h (CHREC_VAR): Remove.
7520 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7521 * tree-chrec.h (build_polynomial_chrec): Adjust.
7522 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7523 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7525 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7527 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7528 * match.pd: ... here.
7529 ((T) X == (T) Y): Relax condition.
7531 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7533 PR tree-optimization/82472
7534 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7536 (break_alias_scc_partitions): Update postorder number.
7538 2017-10-11 Martin Liska <mliska@suse.cz>
7541 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7543 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7545 2017-10-11 Martin Liska <mliska@suse.cz>
7547 * print-rtl.c (print_insn): Move declaration of idbuf
7548 to same scope as name.
7550 2017-10-11 Martin Liska <mliska@suse.cz>
7555 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7558 2017-10-11 Martin Liska <mliska@suse.cz>
7561 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7564 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7566 * config.gcc (default_gnu_indirect_function): Default to yes for
7567 arm*-*-linux* with glibc.
7569 2017-10-11 Richard Biener <rguenther@suse.de>
7571 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7572 default-defs and types we do not want to analyze.
7573 (interpret_loop_phi): Replace unreachable code with an assert.
7574 (compute_scalar_evolution_in_loop): Remove and inline ...
7575 (analyze_scalar_evolution_1): ... here, replacing condition with
7576 what makes the intent clearer. Remove handling of cases
7577 get_scalar_evolution now handles.
7579 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7581 PR rtl-optimization/81434
7582 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7583 comment for main loop. In sched_group_found if, also add checks for
7584 pass and min_cost_group.
7586 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7588 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7589 (rs6000_insn_cost): New function.
7590 * config/rs6000/rs6000.md (cost): New attribute.
7592 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7593 H.J. Lu <hongjiu.lu@intel.com>
7597 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7598 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7599 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7600 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7601 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7602 OPTION_MASK_ISA_MMX and some other bit set require both
7603 mmx and the ISAs without the mmx bit.
7604 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7605 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7606 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7607 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7608 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7609 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7610 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7611 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7612 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7613 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7614 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7615 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7616 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7617 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7618 Add OPTION_MASK_ISA_MMX.
7620 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7622 * config.gcc (armv7*-*-freebsd*): New target.
7623 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7625 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7627 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7628 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7631 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7633 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7634 a template parameter.
7635 (WIDE_INT_REF_FOR): Update accordingly.
7636 * tree.h (wi::int_traits <const_tree>): Delete.
7637 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7638 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7639 (wi::tree_to_wide_ref): New typedef.
7640 (wi::to_wide): New function.
7641 * calls.c (get_size_range): Use wi::to_wide when operating on
7643 * cgraph.c (cgraph_node::create_thunk): Likewise.
7644 * config/i386/i386.c (ix86_data_alignment): Likewise.
7645 (ix86_local_alignment): Likewise.
7646 * dbxout.c (stabstr_O): Likewise.
7647 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7648 * expr.c (const_vector_from_tree): Likewise.
7649 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7650 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7651 (fold_negate_expr_1, int_const_binop_1, const_binop)
7652 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7653 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7654 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7655 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7656 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7657 (fold_not_const, round_up_loc): Likewise.
7658 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7659 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7660 (alloca_call_type): Likewise.
7661 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7662 * godump.c (go_output_typedef): Likewise.
7663 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7664 * internal-fn.c (get_min_precision): Likewise.
7665 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7666 * ipa-polymorphic-call.c
7667 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7668 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7669 (ipa_modify_call_arguments): Likewise.
7670 * match.pd: Likewise.
7671 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7672 * print-tree.c (print_node_brief, print_node): Likewise.
7673 * stmt.c (expand_case): Likewise.
7674 * stor-layout.c (layout_type): Likewise.
7675 * tree-affine.c (tree_to_aff_combination): Likewise.
7676 * tree-cfg.c (group_case_labels_stmt): Likewise.
7677 * tree-data-ref.c (dr_analyze_indices): Likewise.
7678 (prune_runtime_alias_test_list): Likewise.
7679 * tree-dump.c (dequeue_and_dump): Likewise.
7680 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7681 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7682 * tree-pretty-print.c (dump_generic_node): Likewise.
7683 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7684 (simple_iv_with_niters): Likewise.
7685 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7686 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7687 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7688 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7689 (refine_value_range_using_guard, number_of_iterations_ne_max)
7690 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7691 (get_cst_init_from_scev, record_nonwrapping_iv)
7692 (scev_var_range_cant_overflow): Likewise.
7693 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7694 * tree-ssa-pre.c (compute_avail): Likewise.
7695 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7696 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
7697 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
7698 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
7699 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
7700 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
7701 (dump_case_nodes, try_switch_expansion): Likewise.
7702 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
7703 (vect_do_peeling): Likewise.
7704 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
7705 * tree-vect-stmts.c (vectorizable_load): Likewise.
7706 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
7707 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
7708 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
7709 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
7710 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
7711 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
7712 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
7713 (evrp_dom_walker::before_dom_children): Likewise.
7714 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
7715 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
7716 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
7717 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
7718 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
7719 (get_range_pos_neg): Likewise.
7720 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
7721 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
7722 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
7723 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
7724 * config/avr/avr.c (avr_fold_builtin): Likewise.
7725 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
7726 * config/msp430/msp430.c (msp430_attr): Likewise.
7727 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
7728 * config/powerpcspe/powerpcspe-c.c
7729 (altivec_resolve_overloaded_builtin): Likewise.
7730 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
7731 (rs6000_expand_ternop_builtin): Likewise.
7732 * config/rs6000/rs6000-c.c
7733 (altivec_resolve_overloaded_builtin): Likewise.
7734 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
7735 (rs6000_expand_ternop_builtin): Likewise.
7736 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
7738 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7740 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
7741 when copying loop nest with only one inner loop.
7743 2017-10-10 Richard Biener <rguenther@suse.de>
7745 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
7746 blocks if SCEV is active.
7747 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
7749 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
7750 (scev_initialize): Assert we are not yet initialized.
7752 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7754 * tree-loop-distribution.c (generate_loops_for_partition): Remove
7755 inner loop's exit stmt by making it always exit the loop, otherwise
7756 we would generate an infinite empty loop.
7758 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7760 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
7761 renaming variables in new preheader if it's deleted.
7763 2017-10-10 Bin Cheng <bin.cheng@arm.com>
7765 * tree-loop-distribution.c (struct partition): Remove unused field
7766 loops of the structure.
7767 (partition_alloc, partition_free): Ditto.
7768 (build_rdg_partition_for_vertex): Ditto.
7770 2017-10-09 Jeff Law <law@redhat.com>
7772 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
7773 return type to match prototype and documentation.
7775 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7777 * config/rs6000/rs6000.c (processor_costs): Move to ...
7778 * config/rs6000/rs6000.h: ... here.
7779 (rs6000_cost): Declare.
7781 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
7783 * except.c (setjmp_fn): New global variable.
7784 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
7785 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
7786 if DONT_USE_BUILTIN_SETJMP is defined.
7788 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7790 * target.def (insn_cost): New hook.
7791 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
7792 * doc/tm.texi: Regenerate.
7793 * rtlanal.c (insn_cost): Use the new hook.
7795 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7797 * combine.c (combine_validate_cost): Compute the new insn_cost,
7798 not just pattern_cost.
7799 (try_combine): Adjust comment.
7801 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
7803 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
7805 * combine.c (uid_insn_cost): Adjust comment.
7806 (combine_validate_cost): Adjust comment. Use pattern_cost instead
7808 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
7809 * dse.c (find_shift_sequence): Ditto.
7810 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
7811 (bb_valid_for_noce_process_p): Use pattern_cost.
7812 * rtl.h (insn_rtx_cost): Delete.
7813 (pattern_cost): New prototype.
7814 (insn_cost): New prototype.
7815 * rtlanal.c (insn_rtx_cost): Rename to...
7816 (pattern_cost): ... this.
7819 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
7821 * config/i386/i386.md (*jcc_2): Remove insn pattern.
7822 (*jcc<mode>_0_r_i387): Ditto.
7823 (*jccxf_r_i387): Ditto.
7824 (*jcc<mode>_r_i387): Ditto.
7825 (*jccu<mode>_r_i387): Ditto.
7826 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
7827 (*jcc): Rename from *jcc_1.
7829 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7831 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
7832 deferred rescans after the lvx/stvx recombination pre-pass.
7834 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7836 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
7837 memory operation instruction support.
7838 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
7839 (rs6000-ibm-aix[789]*): Likewise.
7840 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
7841 Document new functions.
7843 2017-10-09 Richard Biener <rguenther@suse.de>
7845 PR tree-optimization/82397
7846 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
7847 equality only for semantically equal trees.
7849 2017-10-09 Richard Biener <rguenther@suse.de>
7851 PR tree-optimization/82449
7852 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
7853 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
7854 allow constant addresses.
7855 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
7858 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7860 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
7863 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7866 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7869 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7872 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
7874 2017-10-09 Jakub Jelinek <jakub@redhat.com>
7877 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
7878 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
7880 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7882 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7883 (WI_BINARY_PREDICATE_RESULT): Likewise.
7884 (wi::binary_traits::operator_result): New type.
7885 (wi::binary_traits::predicate_result): Likewise.
7886 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7887 (generic_wide_int::operator==, generic_wide_int::operator!=)
7888 (generic_wide_int::operator&, generic_wide_int::and_not)
7889 (generic_wide_int::operator|, generic_wide_int::or_not)
7890 (generic_wide_int::operator^, generic_wide_int::operator+
7891 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7893 (operator~, unary operator-, operator==, operator!=, operator&)
7894 (operator|, operator^, operator+, binary operator-, operator*): New
7896 * expr.c (get_inner_reference): Use wi::bit_and_not.
7897 * fold-const.c (fold_binary_loc): Likewise.
7898 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7899 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7900 (bit_value_binop): Likewise.
7901 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7902 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7903 (extract_range_from_binary_expr_1): Likewise.
7904 (masked_increment): Likewise.
7905 (simplify_bit_ops_using_ranges): Likewise.
7907 2017-10-09 Martin Jambor <mjambor@suse.cz>
7910 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7911 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7912 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7914 (hsa_fixup_mov_insn_type): New function.
7915 (hsa_op_with_type::get_in_type): Use it.
7916 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7917 immediates in an assert.
7918 (hsa_op_with_type::extend_int_to_32bit): New method.
7919 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7920 types. Convert to dest type if necessary.
7921 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7922 (reg_for_gimple_ssa): Pass false as min32int to
7923 hsa_type_for_scalar_tree_type.
7924 (gen_hsa_addr): Fixup type when creating addresable temporary.
7925 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7926 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7927 necessary. Call hsa_fixup_mov_insn_type.
7928 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7929 extend operands and convert to dest type if necessary.
7930 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7931 to dest type if necessary.
7932 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7933 if conversion nt necessary and size matches.
7934 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7935 to dest type if necessary.
7936 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7937 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7939 (gen_hsa_clrsb): Likewise.
7940 (gen_hsa_ffs): Likewise.
7941 (gen_hsa_divmod): Extend operands and convert to dest type if
7943 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7945 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7947 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7948 Remove empty default arguments. Use a brace block as output
7950 (conditional return): Ditto.
7952 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7953 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7955 (group_ending_nop): Ditto.
7956 (doloop_end): Ditto.
7957 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7958 (splitters for those): Ditto.
7960 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7962 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7963 a conditional jump (and the compare for it) so that pc_rtx is the
7965 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7966 for the deleted and renamed ctr<mode>_internal[234] patterns.
7967 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7968 Delete second conditional return pattern.
7969 (ctr<mode>_internal2): Delete this second bdnz pattern.
7970 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7971 (ctr<mode>_internal4): Delete this second bdz pattern.
7973 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7975 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7976 (always_initialized_rtx_for_ssa_name_p): New predicate.
7977 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7978 (finish_out_of_ssa): Free new field of SA.
7979 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7980 * tree-ssa-coalesce.c: Include tree-ssa.h.
7981 (get_parm_default_def_partitions): Remove extern keyword.
7982 (get_undefined_value_partitions): New function.
7983 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7984 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7986 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7988 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7990 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7992 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7994 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7995 for targets that preffer 128bit.
7997 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7999 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8001 2017-10-08 Olivier Hainque <hainque@adacore.com>
8003 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8004 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8006 (thumb_set_return_address): Likewise.
8008 2017-10-08 Olivier Hainque <hainque@adacore.com>
8010 * common/config/arm/arm-common.c (arm_except_unwind_info):
8011 Handle DWARF2_UNWIND_INFO.
8013 2017-10-07 Michael Collison <michael.collison@arm.com>
8015 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8018 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8020 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8021 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8022 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8023 defined, force the creation of a new block for a dispatch label.
8025 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8027 * invoke.texi (Wsuggest-attribute=cold): Document.
8028 * common.opt (Wsuggest-attribute=cold): New
8029 * ipa-pure-const.c (warn_function_cold): New function.
8030 * predict.c (compute_function_frequency): Use it.
8031 * predict.h (warn_function_cold): Declare.
8033 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8035 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8038 2017-10-06 Martin Liska <mliska@suse.cz>
8040 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8041 keyword for member functions.
8042 (struct sanopt_tree_couple): New struct.
8043 (struct sanopt_tree_couple_hash): New function.
8044 (struct sanopt_ctx): Add new hash_map.
8045 (has_dominating_ubsan_ptr_check): New function.
8046 (record_ubsan_ptr_check_stmt): Likewise.
8047 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8048 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8049 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8051 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8054 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8055 aarch64_simd_valid_immediate on CONST_VECTORs.
8056 (aarch64_reg_or_bic_imm): Likewise.
8058 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8060 PR rtl-optimization/82396
8061 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8063 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8065 * graphite-dependences.c (scop_get_reads): Move code to...
8066 (scop_get_must_writes): Move code to...
8067 (scop_get_may_writes): Move code to...
8068 (scop_get_reads_and_writes): ... here.
8069 (scop_get_dependences): Call scop_get_reads_and_writes.
8071 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8073 PR tree-optimization/82434
8074 * fold-const.h (can_native_encode_type_p,
8075 can_native_encode_string_p): Remove.
8076 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8077 don't encode anything, just return what would be otherwise returned.
8078 (native_encode_fixed, native_encode_complex, native_encode_vector):
8080 (native_encode_string): Likewise. Inline by hand
8081 can_native_encode_string_p.
8082 (can_native_encode_type_p): Remove.
8083 (can_native_encode_string_p): Remove.
8084 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8085 STRING_CSTs using can_native_encode_string_p, test all
8086 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8087 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8088 argument from native_encode_expr.
8089 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8090 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8093 2017-10-06 Richard Biener <rguenther@suse.de>
8095 PR tree-optimization/82397
8096 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8097 operand_equal_p but rely on data_ref_compare_tree for detecting
8099 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8100 to match up with dr_group_sort_cmp.
8102 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8105 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8107 * config/s390/s390-builtin-types.def: Regenerate.
8109 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8112 * config/s390/s390-builtin-types.def: Regenerate.
8113 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8114 Change flag from B_VXE to B_VX.
8115 (s390_vec_min_dbl): Remove B_VXE flag.
8117 2017-10-06 Richard Biener <rguenther@suse.de>
8119 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8120 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8121 translate_isl_ast_to_gimple::is_valid_rename,
8122 translate_isl_ast_to_gimple::get_rename,
8123 translate_isl_ast_to_gimple::get_def_bb_for_const,
8124 translate_isl_ast_to_gimple::get_new_name,
8125 translate_isl_ast_to_gimple::collect_all_ssa_names,
8126 translate_isl_ast_to_gimple::copy_loop_phi_args,
8127 translate_isl_ast_to_gimple::collect_all_ssa_names,
8128 translate_isl_ast_to_gimple::copy_loop_phi_args,
8129 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8130 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8131 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8132 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8133 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8134 translate_isl_ast_to_gimple::copy_cond_phi_args,
8135 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8136 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8137 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8138 translate_isl_ast_to_gimple::rename_uses,
8139 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8140 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8141 (set_rename_for_each_def): Likewise.
8142 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8143 here. Handle rewriting SCEV analyzable uses here.
8144 (copy_bb_and_scalar_dependences): Generate code for PHI
8146 (graphite_regenerate_ast_isl): Adjust.
8147 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8148 (add_write, add_read): New functions.
8149 (build_cross_bb_scalars_def): Use it and simplify.
8150 (build_cross_bb_scalars_use): Likewise.
8151 (graphite_find_cross_bb_scalar_vars): Inline into...
8152 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8153 simulating out-of-SSA. Compute liveout and add dependencies.
8154 (build_scops): Force an empty entry block.
8155 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8157 (sese_build_liveouts): Declare.
8158 (sese_trivially_empty_bb_p): Likewise.
8159 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8160 compute liveout and debug_liveout.
8161 (sese_bad_liveouts_use): Remove.
8162 (sese_reset_debug_liveouts_bb): Likewise.
8163 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8164 (sese_build_liveouts): Build liveout and debug_liveout and store
8166 (new_sese_info): Adjust.
8167 (free_sese_info): Likewise.
8168 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8169 do not build liveout here.
8170 (move_sese_in_condition): Adjust region entry.
8171 (scev_analyzable_p): Match up with chrec_apply requirements.
8172 (sese_trivially_empty_bb_p): New.
8173 * tree-into-ssa.c (get_reaching_def): Properly support generating
8174 default-defs for incremental rewrite of anonymous names.
8176 2017-10-06 Richard Biener <rguenther@suse.de>
8178 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8179 precision do not perform modulo reduction.
8181 2017-10-06 Richard Biener <rguenther@suse.de>
8183 PR tree-optimization/82436
8184 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8185 conservatively choose the vectorization factor when checking
8186 whether we can perform the required load permutation.
8187 (vect_transform_slp_perm_load): Assert when we may not fail.
8189 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8191 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8192 message for incompatible -msdata=* and -mcall-* options.
8194 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8196 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8197 rate for post-reload scheduling.
8199 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8201 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8203 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8205 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8206 to improve monte carlo in scimark.
8208 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8210 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8211 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8212 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8213 pentium4_cost, nocona_cost): Set reassociation width to 1.
8214 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8215 width to 2 for fp operations and 1 otherwise.
8216 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8218 (atom_cost): Set reassociation width to 2.
8219 (slm_cost, generic_cost): Set fp reassociation width
8220 to 2 and 1 otherwise.
8221 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8222 (core_cost): Set fp reassociation width to 4 and vector to 2.
8223 (ix86_reassociation_width): Rewrite using cost table; special case
8224 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8225 and TARGET_AVX128_OPTIMAL.
8226 * config/i386/i386.h (processor_costs): Add
8227 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8228 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8229 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8230 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8231 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8232 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8234 2017-10-05 Nathan Sidwell <nathan@acm.org>
8236 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8238 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8240 * config/arm/arm.c (arm_test_fpu_data): New.
8241 (arm_run_selftests): Call arm_test_fpu_data.
8243 2017-10-04 Nathan Sidwell <nathan@acm.org>
8245 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8247 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8249 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8251 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8252 check type for aarch64_simd_valid_immediate.
8253 (aarch64_output_simd_mov_immediate): Update prototype.
8254 (aarch64_simd_valid_immediate): Update prototype.
8255 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8256 support for ORR-immediate.
8257 (and<mode>3): modified pattern to add support for BIC-immediate.
8258 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8259 now checks for valid immediate for BIC and ORR based on new enum
8261 (aarch64_output_simd_mov_immediate): Function now used to output
8262 BIC/ORR imm as well based on new enum argument.
8263 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8265 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8266 (aarch64_reg_or_bic_imm): Likewise.
8268 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8270 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8271 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8273 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8277 PR rtl-optimization/82396
8278 * haifa-sched.c (autopref_multipass_init): Simplify
8280 (autopref_rank_data): Simplify sort order.
8281 * sched-int.h (autopref_multipass_data_): Remove
8282 multi_mem_insn_p, min_offset and max_offset.
8284 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8286 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8288 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8290 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8293 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8295 * doc/sourcebuild.texi: Document vect_long_mult.
8297 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8299 PR tree-optimization/82413
8300 * fold-const.c (build_range_check): Use widest_int when comparing
8301 the maximum ETYPE value with HIGH.
8303 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8305 PR rtl-optimization/82396
8306 * haifa-sched.c (autopref_multipass_init): Simplify
8308 (autopref_rank_data): Simplify sort order.
8309 * sched-int.h (autopref_multipass_data_): Remove
8310 multi_mem_insn_p, min_offset and max_offset.
8312 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8314 PR tree-optimization/82381
8315 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8316 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8319 PR tree-optimization/82374
8320 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8321 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8322 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8323 current_function_decl to the new decl.
8325 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8327 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8328 helper macro for IEEE float128 hardware built-in functions.
8329 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8331 (TRUNCF128_ODD): Likewise.
8332 (ADDF128_ODD): Likewise.
8333 (SUBF128_ODD): Likewise.
8334 (MULF128_ODD): Likewise.
8335 (DIVF128_ODD): Likewise.
8336 (FMAF128_ODD): Likewise.
8337 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8338 UNSPEC_TRUNC_ROUND_TO_ODD.
8339 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8340 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8341 floating point round to odd instructions.
8342 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8343 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8344 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8345 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8346 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8347 (trunc<mode>sf2_hw): Change the truncate with round to odd
8348 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8349 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8350 to odd hardware instructions.
8351 (sub<mode>3_odd): Likewise.
8352 (mul<mode>3_odd): Likewise.
8353 (div<mode>3_odd): Likewise.
8354 (sqrt<mode>2_odd): Likewise.
8355 (fma<mode>4_odd): Likewise.
8356 (fms<mode>4_odd): Likewise.
8357 (nfma<mode>4_odd): Likewise.
8358 (nfms<mode>4_odd): Likewise.
8359 (trunc<mode>df2_odd): Change the truncate with round to odd
8360 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8362 * doc/extend.texi (PowerPC built-in functions): Update documentation
8363 for existing IEEE float128-bit built-in functions. Add built-in
8364 functions that generate the IEEE 128-bit floating point round to
8367 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8369 PR rtl-optimization/77729
8370 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8371 to (X&(C1&~C2))|C2 transformations.
8373 2017-10-03 Martin Jambor <mjambor@suse.cz>
8375 PR tree-optimization/82363
8376 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8377 mismatch, mark lacc written regardless of racc.
8379 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8381 PR tree-optimization/82381
8382 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8383 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8384 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8385 fallthrough into reassoc_stmt_dominates_stmt_p.
8388 * combine.c (combine_instructions): Don't combine in unreachable
8391 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8395 function to not use the have_cpu variable. Do not set cpu_index,
8396 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8398 (rs6000_valid_attribute_p): Remove duplicate initializations of
8399 old_optimize and func_optimize.
8400 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8401 (rs6000_activate_target_options): Make global.
8402 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8405 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8407 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8408 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8409 Set *poffset to 0 and *psize and *pmax_size to -1 if
8410 *poffset + *psize overflows in HOST_WIDE_INT.
8412 PR tree-optimization/82387
8413 PR tree-optimization/82388
8414 PR tree-optimization/82389
8415 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8416 instead of live_bytes non-NULL.
8418 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8421 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8424 2017-10-02 Richard Biener <rguenther@suse.de>
8426 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8427 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8428 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8430 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8432 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8433 requested precision matches the type's.
8434 * calls.c (alloc_max_size): Calculate the new candidate size as
8435 a widest_int and use wi::to_widest when comparing it with the
8436 current candidate size.
8437 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8438 zero rather than integer_zero_node.
8439 * match.pd: Check for a no-op conversion before using wi::add
8440 rather than after. Use tree_to_uhwi when summing small shift
8441 counts into an unsigned int.
8443 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8444 Alan Hayward <alan.hayward@arm.com>
8445 David Sherwood <david.sherwood@arm.com>
8448 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8449 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8450 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8451 POINTER_AND_FP_REGS.
8453 2017-10-02 Richard Biener <rguenther@suse.de>
8455 PR tree-optimization/82355
8456 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8457 a mapping for the enclosing loop but avoid generating one for
8459 (copy_bb_and_scalar_dependences): Remove premature codegen
8460 error on PHIs in blocks duplicated into multiple places.
8461 * graphite-scop-detection.c
8462 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8463 in the region use it as loop and nest to analyze the DR in.
8464 (try_generate_gimple_bb): Likewise.
8465 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8466 (add_loop_constraints): For blocks in a loop not in the region
8467 create a dimension with a single iteration.
8468 * sese.h (gbb_loop_at_index): Remove assert.
8470 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8472 * omp-expand.c (adjust_context_scope): New function.
8473 (expand_parallel_call): Call adjust_context_scope.
8475 2017-10-01 Jeff Law <law@redhat.com>
8477 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8478 dom_opt_dom_walker class with direct access to private members.
8479 Add comments. Call test_for_singularity.
8480 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8481 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8482 m_dummy_cond anymore.
8483 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8485 (pass_dominator:execute): Build the dummy_cond here and pass it
8486 to the dom_opt_dom_walker ctor.
8487 (test_for_singularity): New function.
8489 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8490 Maya Rashish <coypu@sdf.org>
8492 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8493 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8494 (alpha*-*-netbsd*) Use nbsd_tm_file.
8495 (arm*-*-netbsdelf*) Likewise.
8496 (i[34567]86-*-netbsdelf*) Likewise.
8497 (x86_64-*-netbsd*) Likewise.
8498 (mips*-*-netbsd*) Likewise.
8499 (powerpc-*-netbsd*) Likewise.
8500 (sh*-*-netbsd*) Likewise.
8501 (sparc-*-netbsdelf*) Likewise.
8502 (sparc64-*-netbsd*) Likewise.
8503 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8505 (vax-*-netbsdelf*) Likewise.
8506 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8507 (UINT_FAST8_TYPE) Likewise.
8508 (INT_FAST16_TYPE) Check CHAR_FAST16.
8509 (UINT_FAST16_TYPE) Likewise.
8511 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8514 * config/i386/i386.md
8515 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8516 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8517 *divmodsi4_zext_2): New define_insn_and_split.
8518 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8519 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8520 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8521 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8522 New define_insn_and_split.
8523 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8524 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8525 operands[1] having DImode when mode is SImode.
8527 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8528 always SImode for DIV and MOD in REG_EQUAL notes.
8530 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8533 * match.pd: Fix handling of NaNs in pattern.
8535 2017-09-29 Jeff Law <law@redhat.com>
8537 * sbitmap.c (bitmap_bit_in_range_p): New function.
8538 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8539 * tree-ssa-dse.c (live_bytes_read): New function.
8540 (dse_classify_store): Ignore reads of dead bytes.
8542 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8543 typos and whitespace errors.
8544 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8545 * config/s390/s390.c (s390_emit_prologue): Likewise.
8547 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8550 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8551 with a symbol for LRA.
8553 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8555 PR rtl-optimization/82338
8556 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8558 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8560 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8561 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8562 (qsort_chk): Declare.
8563 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8564 (qsort_chk): New function.
8566 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8568 PR tree-optimization/82337
8569 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8570 phi definition if the PHI result appears in an abnormal PHI.
8571 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8572 basis appears in an abnormal PHI.
8574 2017-09-29 Richard Biener <rguenther@suse.de>
8576 * graphite-isl-ast-to-gimple.c
8577 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8578 (binary_op_to_tree): Use it.
8579 (get_rename_from_scev): Likewise.
8580 (copy_loop_phi_nodes): Likewise.
8581 (copy_bb_and_scalar_dependences): Likewise.
8582 (translate_pending_phi_nodes): Likewise.
8584 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8587 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8588 for movabsq $(i32 << shift), r64.
8590 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8592 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8593 index when encoding %esp as %rsp to avoid 0x67 prefix.
8595 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8597 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8598 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8600 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8602 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8603 Extensions with more than 16 double VFP registers.
8604 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8605 to_clear_mask and all code related to it. Replace the remaining
8606 entry by a sbitmap and adapt code accordingly.
8608 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8610 * brig-builtins.def: Change pure attributes to const.
8612 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8614 * config.gcc (default_gnu_indirect_function): Default to yes for
8615 sparc*-*-linux* with glibc.
8617 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8619 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8620 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8621 when creating .init_array and .fini_array sections with priority
8624 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8627 * config/aarch64/aarch64.c
8628 (aarch64_builtin_support_vector_misalignment): Always return false
8629 when misalignment is unknown.
8631 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8633 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8634 this function to return false if the definition used by the swap
8635 instruction is artificial, or if the memory address from which the
8636 constant value is loaded is not represented by a base address held
8637 in a register or if the base address register is a frame or stack
8638 pointer. Additionally, return false if the base address of the
8639 loaded constant is a SYMBOL_REF but is not considered to be a
8641 (replace_swapped_load_constant): New function.
8642 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8643 loaded constant vector with a load of a swapped constant vector.
8645 2017-09-27 Carl Love <cel@us.ibm.com>
8647 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8648 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8649 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8652 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8654 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8655 first, always call autopref_rank_data otherwise.
8657 2017-09-27 Richard Biener <rguenther@suse.de>
8659 * graphite-scop-detection.c (find_scop_parameters): Move
8660 loop bound handling ...
8661 (gather_bbs::before_dom_children): ... here, avoiding the need
8662 to build scop_info->loop_nest.
8663 (record_loop_in_sese): Remove.
8664 * sese.h (sese_info_t::loop_nest): Remove.
8665 * sese.c (new_sese_info): Do not allocate loop_nest.
8666 (free_sese_info): Do not free loop_nest.
8668 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8671 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8672 lhs from calls if the lhs has addressable type.
8674 2017-09-27 Richard Biener <rguenther@suse.de>
8676 * graphite.h (scop::max_alias_set): New member.
8677 * graphite-scop-detection.c: Remove references to non-existing
8678 --param in comments.
8679 (build_alias_sets): Record the maximum alias set used for drs.
8680 (build_scops): Support zero as unlimited for
8681 --param graphite-max-arrays-per-scop.
8682 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8684 (build_poly_sr_1): ... here. Compute alias set based on the
8685 maximum alias set used for drs rather than
8686 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8688 2017-09-27 Richard Biener <rguenther@suse.de>
8690 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8691 --param loop-block-tile-size=0 to disable tiling.
8693 2017-09-27 Richard Biener <rguenther@suse.de>
8695 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8696 (graphite-max-nb-scop-params): Document special value zero.
8697 * domwalk.h (dom_walker::STOP): New symbolical constant.
8698 (dom_walker::dom_walker): Add optional parameter for bb to
8700 (dom_walker::~dom_walker): Declare.
8701 (dom_walker::before_dom_children): Document STOP return value.
8702 (dom_walker::m_user_bb_to_rpo): New member.
8703 (dom_walker::m_bb_to_rpo): Likewise.
8704 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
8705 mapping here if not provided by the user.
8706 (dom_walker::~dom_walker): Free bb to RPO mapping if not
8707 provided by the user.
8708 (dom_walker::STOP): Define.
8709 (dom_walker::walk): Do not compute bb to RPO mapping here.
8710 Support STOP return value from before_dom_children to stop
8712 * graphite-optimize-isl.c (optimize_isl): If the schedule
8713 is the same still generate code if -fgraphite-identity
8714 or -floop-parallelize-all are given.
8715 * graphite-scop-detection.c: Include cfganal.h.
8716 (gather_bbs::gather_bbs): Get and pass through bb to RPO
8718 (gather_bbs::before_dom_children): Return STOP for BBs
8720 (build_scops): Compute bb to RPO mapping and pass it to
8721 the domwalk. Treat --param graphite-max-nb-scop-params=0
8722 as not limiting the number of params.
8723 * graphite.c (graphite_initialize): Remove limit on the
8724 number of basic-blocks in a function.
8725 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
8726 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
8727 default value of 10.
8729 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8731 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
8732 Adjust code to eliminate needing to do the shift right 32-bits
8733 operation after XSCVDPSPN.
8735 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8737 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
8738 ((X / Y) != 0 -> X >= Y): Likewise.
8740 2017-09-26 Carl Love <cel@us.ibm.com>
8742 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
8743 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
8744 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
8745 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
8746 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
8747 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
8748 definitions and overloading.
8749 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
8750 statement for P9V_BUILTIN_XST_LEN_R.
8751 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
8752 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
8753 define_expand and define_insn for the instructions and builtins.
8754 * doc/extend.texi: Update the built-in documentation file for the new
8756 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
8757 define_insn for the instructions
8759 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
8762 * gcc/config/netbsd-protos.h: New file.
8763 * gcc/config/netbsd.c: New file.
8764 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
8765 * gcc/config/t-netbsd: New file.
8766 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
8767 (tmake_file) Add t-netbsd.
8768 (extra_objs) Add netbsd.o.
8770 2017-09-26 Janus Weil <janus@gcc.gnu.org>
8774 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
8776 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8778 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
8779 sign extension from a vector register to a GPR by doing a 32-bit
8780 direct move and then an EXTSW.
8781 (extendsi<mode>2 splitter): Likewise.
8782 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
8783 right or vector extract after doing XSCVDPSPN. Use
8784 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
8786 (movdi_from_sf_zero_ext): Likewise.
8787 (reload_gpr_from_vsxsf): Likewise.
8788 (p8_mfvsrd_4_disf): Delete, no longer used.
8789 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
8790 then needing to move the SFmode to a GPR to use the XSCVDPSP
8791 instruction instead of FRSP and XSCVDPSPN.
8792 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
8793 it is adjacent to the other XSCVSPDP insns.
8794 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
8795 SFmode to be in traditional Altivec registers.
8796 (vsx_xscvdpspn): Eliminate useless alternative constraint.
8797 (vsx_xscvspdpn): Likewise.
8798 (vsx_xscvspdpn_scalar): Likewise.
8800 2017-09-26 Martin Jambor <mjambor@suse.cz>
8802 * tree-sra.c (compare_access_positions): Put integral types first,
8803 stabilize sorting of integral types, remove conditions putting
8804 non-full-precision integers last.
8805 (sort_and_splice_var_accesses): Disable scalarization if a
8806 non-integert would be represented by a non-full-precision integer.
8808 2017-09-26 Joseph Myers <joseph@codesourcery.com>
8810 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
8811 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
8812 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
8813 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
8814 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
8815 conditionals inside the function instead of around it. Call
8816 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
8817 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
8819 2017-09-26 Richard Biener <rguenther@suse.de>
8821 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
8823 (scop_detection::build_scop_breadth): ... this. Removed.
8824 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
8825 (scop_detection::harmful_stmt_in_bb): Likewise.
8826 (scop_detection::graphite_can_represent_stmt): Likewise.
8827 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
8828 (scop_detection::can_represent_loop): Remove recursion, fold in ...
8829 (scop_detection::can_represent_loop_1): ... this. Removed.
8830 (scop_detection::harmful_loop_in_region): Simplify after inlining
8831 the above and remove more quadraticness.
8832 (build_scops): Adjust.
8833 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
8836 2017-09-26 Jakub Jelinek <jakub@redhat.com>
8839 * config/i386/i386.c (ix86_print_operand_address_as): Only test
8840 REGNO (base) == SP_REG if base is a REG.
8843 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
8844 if it is different SSA_NAME.
8845 (optimize_range_tests_cmp_bitwise): New function.
8846 (optimize_range_tests): Call it.
8848 2017-09-26 Richard Biener <rguenther@suse.de>
8850 PR tree-optimization/82321
8851 * graphite.c (canonicalize_loop_closed_ssa): Properly check
8852 for the def being inside the loop.
8854 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8856 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
8858 * config/s390/s390-builtins.def: Fix constraint on op4.
8860 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8862 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
8863 independent expanders.
8864 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
8865 ("vec_ordered", "vec_unordered"): New expanders.
8867 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8869 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
8872 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8874 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
8875 vec_unpacks_lo_v16qi.
8876 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
8878 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8880 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
8881 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8882 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8884 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8886 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8888 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8890 ("*vec_slb<mode>"): New insn pattern.
8891 ("vec_shr_<mode>"): New expander.
8892 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8893 and force the shift count operand to V16QImode.
8894 ("vec_srb<mode>"): Set shift count mode to V16QI.
8896 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8898 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8899 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8900 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8902 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8905 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8907 2017-09-26 Richard Biener <rguenther@suse.de>
8909 PR tree-optimization/82320
8910 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8913 2017-09-25 Jeff Law <law@redhat.com>
8915 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8916 prototype for new argument.
8917 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8918 mostly extracted from rs6000_emit_allocate_stack.
8919 (rs6000_emit_probe_stack_range_stack_clash): New function.
8920 (rs6000_emit_allocate_stack): Call
8921 rs6000_emit_probe_stack_range_stack_clash as needed.
8922 (rs6000_emit_probe_stack_range): Add additional argument
8923 to call to gen_probe_stack_range{si,di}.
8924 (output_probe_stack_range): New.
8925 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8926 (output_probe_stack_range_stack_clash): New.
8927 (rs6000_emit_prologue): Emit notes into dump file as requested.
8928 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8929 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8930 Add additional operand and pass it to output_probe_stack_range.
8932 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8934 PR tree-optimization/82163
8935 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8936 (checking_verify_loop_closed_ssa): New parameter.
8937 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8938 (check_loop_closed_ssa_stmt): Delete.
8939 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8940 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8941 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8944 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8946 * brig-builtins.def: Treat HSAIL barrier builtins as
8947 setjmp/longjump style functions.
8949 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8951 * target.def (constant_alignment): New hook.
8952 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8953 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8954 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8955 * doc/tm.texi: Regenerate.
8956 * targhooks.h (default_constant_alignment): Declare.
8957 (constant_alignment_word_strings): Likewise.
8958 * targhooks.c (default_constant_alignment): New function.
8959 (constant_alignment_word_strings): Likewise.
8960 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8961 instead of CONSTANT_ALIGNMENT.
8962 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8963 (force_const_mem): Likewise.
8964 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8965 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8966 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8967 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8968 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8970 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8971 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8972 constant_alignment_word_strings.
8973 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8974 (CONSTANT_ALIGNMENT): Likewise.
8975 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8976 (arm_constant_alignment): New function.
8977 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8978 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8979 constant_alignment_word_strings.
8980 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8981 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8982 constant_alignment_word_strings.
8983 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8984 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8985 (cris_constant_alignment): New function.
8986 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8987 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8988 (epiphany_constant_alignment): New function.
8989 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8990 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8991 constant_alignment_word_strings.
8992 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8993 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8994 constant_alignment_word_strings.
8995 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8996 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8997 constant_alignment_word_strings.
8998 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8999 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9000 * config/i386/i386.c (ix86_constant_alignment): Make static.
9001 Use the same interface as the target hook.
9002 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9003 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9004 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9005 constant_alignment_word_strings.
9006 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9007 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9008 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9009 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9010 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9011 constant_alignment_word_strings.
9012 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9013 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9014 constant_alignment_word_strings.
9015 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9016 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9017 constant_alignment_word_strings.
9018 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9019 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9021 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9022 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9023 * config/mips/mips.c (mips_constant_alignment): New function.
9024 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9025 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9026 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9027 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9028 (mmix_constant_alignment): Make static. Use the same interface
9030 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9031 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9032 constant_alignment_word_strings.
9033 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9034 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9035 constant_alignment_word_strings.
9036 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9037 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9038 constant_alignment_word_strings.
9039 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9040 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9041 (rs6000_constant_alignment): New function.
9042 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9043 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9044 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9045 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9046 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9047 (rs6000_constant_alignment): New function.
9048 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9049 * config/s390/s390.c (s390_constant_alignment): New function.
9050 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9051 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9052 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9053 constant_alignment_word_strings.
9054 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9055 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9056 (sparc_constant_alignment): New function.
9057 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9058 * config/spu/spu.c (spu_constant_alignment): New function.
9059 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9060 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9061 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9062 constant_alignment_word_strings.
9063 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9064 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9065 constant_alignment_word_strings.
9066 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9067 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9068 constant_alignment_word_strings.
9069 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9070 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9071 (visium_constant_alignment): New function.
9072 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9073 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9074 (xtensa_constant_alignment): New function.
9075 * system.h (CONSTANT_ALIGNMENT): Poison.
9077 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9079 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9080 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9081 (rs6000_builtin_valid_without_lhs): New helper function.
9082 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9083 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9085 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9087 * target.h (vec_perm_indices): Use unsigned short rather than
9089 (auto_vec_perm_indices): Likewise.
9090 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9091 Use unsigned int rather than unsigned char.
9092 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9094 2017-09-25 Richard Biener <rguenther@suse.de>
9096 * cfgloop.h (sort_sibling_loops): Declare.
9097 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9098 (sort_sibling_loops): New function sorting the sibling loop list
9100 * graphite.c (graphite_transform_loops): Sort sibling loops.
9102 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9104 * target.def (vec_perm_const_ok): Change sel parameter to
9106 * optabs-query.c (can_vec_perm_p): Update accordingly.
9107 * doc/tm.texi: Regenerate.
9108 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9109 auto_vec_perm_indices and remove separate nelt field.
9110 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9111 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9112 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9113 (aarch64_expand_vec_perm_const): Update accordingly.
9114 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9115 to vec_perm_indices.
9116 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9117 auto_vec_perm_indices and remove separate nelt field.
9118 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9119 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9120 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9122 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9123 to vec_perm_indices.
9124 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9125 sel to vec_perm_indices.
9126 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9127 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9128 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9130 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9133 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9136 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9137 on the FUNCTION_DECL function context if it has a DIE that is a
9140 2017-09-25 Richard Biener <rguenther@suse.de>
9142 PR tree-optimization/82285
9143 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9146 2017-09-25 Tom de Vries <tom@codesourcery.com>
9150 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9153 2017-09-25 Richard Biener <rguenther@suse.de>
9155 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9156 ISL errors other than isl_error_quota happen. Dump if the
9157 schedule is the same.
9158 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9159 errors instead of aborting inside ISL.
9161 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9164 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9165 of libgcc_eh for m64.
9166 * config/i386/darwin64.h: Likewise.
9168 2017-09-25 Richard Biener <rguenther@suse.de>
9171 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9172 attribute for incomplete types nor twice for complete ones.
9174 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9177 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9178 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9181 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9184 * config/i386/i386.opt (mprefer-avx256): Use
9185 ix86_target_flags variable.
9186 * config/i386/i386.c (ix86_target_string): Move
9187 -mprefer-avx256 to flag2_opts.
9189 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9192 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9193 and x != -1 | y != -1 into (x & y) != -1.
9195 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9197 * config.gcc: Add new case statement to set
9198 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9199 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9200 s390x-*-linux* case statements. Added aarch64 to the list of
9201 supported architectures.
9203 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9205 PR tree-optimization/82289
9206 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9207 STMT_VINFO_RELEVANT_P.
9209 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9210 Alan Hayward <alan.hayward@arm.com>
9211 David Sherwood <david.sherwood@arm.com>
9213 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9214 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9215 (extract_range_from_binary_expr_1): Don't call
9216 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9218 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9219 Alan Hayward <alan.hayward@arm.com>
9220 David Sherwood <david.sherwood@arm.com>
9222 * target.def (preferred_vector_alignment): New hook.
9223 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9225 * doc/tm.texi: Regenerate.
9226 * targhooks.h (default_preferred_vector_alignment): Declare.
9227 * targhooks.c (default_preferred_vector_alignment): New function.
9228 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9230 (DR_TARGET_ALIGNMENT): New macro.
9231 (aligned_access_p): Update commentary.
9232 (vect_known_alignment_in_bytes): New function.
9233 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9235 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9236 Calculate the misalignment based on the target alignment rather than
9238 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9239 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9240 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9241 the target alignment, rather than masking the element misalignment
9242 with the number of elements in a vector. Also use the target
9243 alignment when calculating the maximum number of peels.
9244 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9245 instead of TYPE_ALIGN_UNIT.
9246 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9247 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9248 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9249 (vect_create_data_ref_ptr): Likewise.
9250 (vect_setup_realignment): Realign by ANDing with
9251 -DR_TARGET_MISALIGNMENT.
9252 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9253 the number of peels based on DR_TARGET_ALIGNMENT.
9254 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9255 with the guaranteed alignment boundary when deciding whether
9257 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9258 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9259 (ensure_base_align): Remove stmt_info parameter. Get the
9260 target base alignment from DR_TARGET_ALIGNMENT.
9261 (vectorizable_store): Update call accordingly. Interpret
9262 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9264 (vectorizable_load): Likewise.
9266 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9267 Alan Hayward <alan.hayward@arm.com>
9268 David Sherwood <david.sherwood@arm.com>
9270 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9271 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9272 (vect_enhance_data_refs_alignment): Likewise.
9274 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9276 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9277 error. Only quit immediately if parsing is complete.
9278 (BEGIN): Initialize fatal_err and parse_done.
9279 (begin fpu, end fpu): Check number of arguments.
9280 (begin arch, end arch): Likewise.
9281 (begin cpu, end cpu): Likewise.
9282 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9283 (optalias): Likewise.
9285 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9287 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9288 * config/arm/arm-isa.h: Delete. Move definitions to ...
9289 * arm-cpus.in: ... here. Use new feature and fgroup values.
9290 * config/arm/arm.c (arm_option_override): Use lower case for feature
9292 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9293 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9294 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9296 (print_isa_bits_for): New function.
9297 (gen_isa): New function.
9298 (gen_comm_data): Use print_isa_bits_for.
9299 (define feature): New keyword.
9300 (define fgroup): New keyword.
9301 * config/arm/t-arm (TM_H): Remove.
9302 (GTM_H): Add arm-isa.h.
9303 (arm-isa.h): Add rule to generate file.
9304 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9305 case for feature bit names.
9307 2017-09-22 Richard Biener <rguenther@suse.de>
9309 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9311 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9312 print of no dependency loops ...
9313 * graphite.c (graphite_transform_loops): ... here.
9314 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9316 (same_close_phi_node, remove_duplicate_close_phi,
9317 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9318 (canonicalize_loop_closed_ssa): ... here and simplify.
9319 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9320 (optimize_isl): Use dump_printf_loc to tell when we stopped
9321 optimizing because of an ISL timeout.
9323 2017-09-22 Richard Biener <rguenther@suse.de>
9325 PR tree-optimization/82291
9326 * tree-if-conv.c (predicate_mem_writes): Make sure to
9327 remove writes in blocks predicated with false.
9329 2017-09-22 Richard Biener <rguenther@suse.de>
9331 * sese.c: Include cfganal.h.
9332 (if_region_set_false_region): Remove.
9333 (create_if_region_on_edge): Likewise.
9334 (move_sese_in_condition): Re-implement without destroying
9337 2017-09-22 Richard Biener <rguenther@suse.de>
9339 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9340 Verify both BBs contain loop PHI nodes before dispatching to
9342 (graphite_regenerate_ast_isl): Do not recompute dominators,
9343 do not verify three times. Restructure for clarity.
9344 * graphite-scop-detection.c (same_close_phi_node,
9345 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9346 defined_in_loop_p, canonicalize_loop_closed_ssa,
9347 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9348 checking and SSA rewrite, move to ...
9349 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9350 (graphite_initialize): Do not pass in ctx, do not reset the
9351 SCEV cache, compute only dominators.
9352 (graphite_transform_loops): Allocate ISL ctx after
9353 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9354 Maintain post-dominators only around build_scops.
9355 * sese.c (if_region_set_false_region): Make static. Free
9356 and recompute dominators.
9357 (move_sese_in_condition): Assert we don't get called with
9358 post-dominators computed.
9359 * sese.h (if_region_set_false_region): Remove.
9361 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9363 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9364 mode attribute for TARGET_AVX512VL.
9366 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9368 * config/i386/i386.opt (mprefer-avx256): New option.
9369 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9371 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9372 for TARGET_PREFER_AVX256.
9373 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9375 2017-09-21 Jeff Law <law@redhat.com>
9377 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9378 Fix dump output if the only stack space is for pushed registers.
9380 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9382 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9385 2017-09-21 Martin Sebor <msebor@redhat.com>
9388 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9389 code (in C++) or code that triggers warnings.
9391 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9393 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9395 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9397 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9398 * sched-rgn.c: Ditto.
9399 * sel-sched-ir.c: Ditto.
9401 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9403 * toplev.h (set_random_seed): Adjust return type.
9404 * toplev.c (init_local_tick): Move eager initialization of random_seed
9405 to get_random_seed. Adjust comment.
9406 (init_random_seed): Inline to get_random_seed, delete.
9407 (get_random_seed): Initialize random_seed lazily.
9408 (set_random_seed): Do not return previous value.
9409 (print_switch_value): Do not call get_random_seed.
9411 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9413 * cgraph.c (delete_function_version): New, broken out from...
9414 (cgraph_node::delete_function_version): ...here. Rename to
9415 cgraph_node::delete_function_version_by_decl. Update all uses.
9416 (cgraph_node::remove): Call delete_function_version.
9418 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9421 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9422 VAR_DECLs to which addressable non-volatile parameters are mapped
9423 and for id->retvar after the return value assignment. Clear
9424 id->retval and id->retbnd after inlining.
9426 2017-09-21 Richard Biener <rguenther@suse.de>
9428 PR tree-optimization/82276
9429 PR tree-optimization/82244
9430 * tree-vrp.c (build_assert_expr_for): Set
9431 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9433 (remove_range_assertions): Revert earlier change.
9435 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9438 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9440 2017-09-21 Richard Biener <rguenther@suse.de>
9442 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9443 Restore valid IL after code generation errors.
9444 * graphite.c (graphite_transform_loops): Diagnose code
9445 generation issues as MSG_MISSED_OPTIMIZATION and continue
9446 with processing SCOPs.
9448 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9449 Alan Hayward <alan.hayward@arm.com>
9450 David Sherwood <david.sherwood@arm.com>
9452 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9453 rather than choosing between plus_constant and gen_rtx_<CODE>.
9454 * expr.c (emit_push_insn): Likewise.
9455 (expand_expr_real_2): Likewise.
9457 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9458 Alan Hayward <alan.hayward@arm.com>
9459 David Sherwood <david.sherwood@arm.com>
9461 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9463 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9464 Alan Hayward <alan.hayward@arm.com>
9465 David Sherwood <david.sherwood@arm.com>
9467 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9468 calling tree_to_uhwi.
9470 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9471 Alan Hayward <alan.hayward@arm.com>
9472 David Sherwood <david.sherwood@arm.com>
9474 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9475 INTEGER_CST rather than a negative test for ADDR_EXPR.
9477 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9478 Alan Hayward <alan.hayward@arm.com>
9479 David Sherwood <david.sherwood@arm.com>
9481 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9482 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9484 2017-09-21 Richard Biener <rguenther@suse.de>
9486 PR tree-optimization/71351
9487 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9488 graphite_create_new_loop_guard): Remove, fold remaining parts
9490 (translate_isl_ast_node_for): ... here and simplify.
9492 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9495 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9496 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9497 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9498 alternative always use QI mode, for -Os imov (=R,R) alternative
9499 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9502 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9503 Jeff Law <law@redhat.com>
9505 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9506 (allocate_stack_space): New function, partially extracted from
9508 (s390_emit_prologue): Track offset to most recent stack probe.
9509 Code to allocate space moved into allocate_stack_space.
9510 Dump actions when no stack is allocated.
9511 (s390_prologue_plus_offset): New function.
9512 (s390_emit_stack_probe): Likewise.
9514 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9516 * common.opt (Wa, Wl, Wp, g, gz=): Add
9518 (gno-column-info): Remove.
9519 (gcolumn-info): Drop RejectNegative.
9521 (gno-record-gcc-switches): Remove.
9522 (grecord-gcc-switches): Drop RejectNegative.
9523 (gno-split-dwarf): Remove.
9524 (gsplit-dwarf): Drop RejectNegative.
9525 (gno-strict-dwarf): Remove.
9526 (gstrict-dwarf): Drop RejectNegative.
9527 * config/darwin.opt (gfull, gused): Add RejectNegative.
9528 * dwarf2out.c (gen_producer_string): Drop
9529 gno-record-gcc-switches handler.
9530 * optc-gen.awk: Add g to prefixes with negative forms.
9531 * opts-common.c (remapping_prefix_p): New.
9532 (find_opt): Check it.
9533 (generate_canonical_option): Test g prefix.
9534 (option_map): Add -gno- mapping.
9535 (add_misspelling_candidates): Check remapping_prefix_p.
9537 2017-09-20 Jeff Law <law@redhat.com>
9539 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9540 thinko in stack clash protection support.
9542 * explow.c (compute_stack_clash_protection_loop_data): Use
9543 CONST_INT_P instead of explicit test. Verify object is a
9544 CONST_INT_P before looking at INTVAL.
9545 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9546 instead of explicit test.
9548 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9551 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9552 address instead of to r1 and r11.
9554 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9556 * config.gcc: Support "knm".
9557 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9558 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9560 * config/i386/i386.c (m_KNM): Define.
9561 (processor_target_table): Add "knm".
9563 (ix86_option_override_internal): Add "knm".
9564 (ix86_issue_rate): Add PROCESSOR_KNM.
9565 (ix86_adjust_cost): Ditto.
9566 (ia32_multipass_dfa_lookahead): Ditto.
9567 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9568 (fold_builtin_cpu): Add M_INTEL_KNM.
9569 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9570 (processor_type): Add PROCESSOR_KNM.
9571 * config/i386/x86-tune.def: Add m_KNM.
9572 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9574 2017-09-20 Richard Biener <rguenther@suse.de>
9576 PR tree-optimization/80213
9577 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9578 are allowed in empty BBs as well.
9579 (canonicalize_loop_closed_ssa): Also look for other complex
9581 (scop_detection::get_sese): Include the loop-closed PHI block
9583 (scop_detection::merge_sese): Remove code adding extra blocks.
9584 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9585 (build_scops): Assert the final returned scop is invalid.
9587 2017-09-20 Richard Biener <rguenther@suse.de>
9589 PR tree-optimization/82264
9590 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9592 (vn_phi_lookup): Likewise.
9593 (vn_phi_insert): Likewise.
9595 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9597 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9598 that fits into uhwi or shwi, add DW_AT_const_value regardless
9599 of early_dwarf without going through RTL, using add_AT_unsigned
9602 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9603 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9604 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9605 (DEBUG_MACRO_SECTION): Likewise.
9606 (DEBUG_LTO_MACRO_SECTION): Likewise.
9607 (DEBUG_STR_DWO_SECTION): Likewise.
9608 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9609 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9610 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9611 (DEBUG_LTO_LINE_STR_SECTION): Define.
9612 (init_sections_and_labels): Initialize debug_line_str_section
9613 variable. Initialize debug_loc_section for -gdwarf-5 to
9614 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9616 2017-09-20 Richard Biener <rguenther@suse.de>
9618 * graphite-sese-to-poly.c (extract_affine): Properly handle
9619 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9621 2017-09-20 Richard Biener <rguenther@suse.de>
9623 PR tree-optimization/81373
9624 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9625 Force SESE live-out defs to be handled even if they are
9628 2017-09-19 Jeff Law <law@redhat.com>
9630 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9631 nothing for stack adjustments with REG_STACK_CHECK.
9632 * sched-deps.c (parse_add_or_inc): Reject insns with
9633 REG_STACK_CHECK from dependency breaking.
9634 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9635 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9636 * reg-notes.def (STACK_CHECK): New note.
9638 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9639 (ix86_expand_prologue): Dump stack clash info as needed.
9640 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9642 * function.c (dump_stack_clash_frame_info): New function.
9643 * function.h (dump_stack_clash_frame_info): Prototype.
9644 (enum stack_clash_probes): New enum.
9646 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9647 flag_stack_clash_protection.
9648 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9649 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9650 (arm_frame_pointer_required): Likewise.
9651 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9652 (ia64_expand_prologue): Likewise.
9653 * config/mips/mips.c (mips_expand_prologue): Likewise.
9654 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9655 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9656 (sparc_flat_expand_prologue): Likewise.
9657 * config/spu/spu.c (spu_expand_prologue): Likewise.
9659 * explow.c: Include "params.h".
9660 (anti_adjust_stack_and_probe_stack_clash): New function.
9661 (get_stack_check_protect): Likewise.
9662 (compute_stack_clash_protection_loop_data): Likewise.
9663 (emit_stack_clash_protection_loop_start): Likewise.
9664 (emit_stack_clash_protection_loop_end): Likewise.
9665 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9666 Use anti_adjust_stack_and_probe_stack_clash.
9667 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9668 (emit_stack_clash_protection_loop_start): Likewise.
9669 (emit_stack_clash_protection_loop_end): Likewise.
9670 * rtl.h (get_stack_check_protect): Prototype.
9671 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9672 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9673 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9675 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9677 * doc/tm.texi: Rebuilt.
9678 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9679 get_stack_check_protect.
9680 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9681 * config/arm/arm.c (arm_expand_prologue): Likewise.
9682 (arm_frame_pointer_required): Likewise.
9683 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9684 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9685 * config/mips/mips.c (mips_expand_prologue): Likewise.
9686 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9687 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9688 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9689 (sparc_flat_expand_prologue): Likewise.
9691 * common.opt (-fstack-clash-protection): New option.
9692 * flag-types.h (enum stack_check_type): Note difference between
9693 -fstack-check= and -fstack-clash-protection.
9694 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9695 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9696 * toplev.c (process_options): Issue warnings/errors for cases
9697 not handled with -fstack-clash-protection.
9698 * doc/invoke.texi (-fstack-clash-protection): Document new option.
9699 (-fstack-check): Note additional problem with -fstack-check=generic.
9700 Note that -fstack-check is primarily for Ada and refer users
9701 to -fstack-clash-protection for stack-clash-protection.
9702 Document new params for stack clash protection.
9704 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
9706 * config/i386/i386.c (ix86_split_long_move): Do not handle
9707 address used for LEA in a special way.
9709 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
9711 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
9713 2017-09-19 Martin Sebor <msebor@redhat.com>
9716 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
9717 of incompatible types.
9719 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
9721 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9722 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
9723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9724 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
9726 2017-09-19 Richard Biener <rguenther@suse.de>
9728 PR tree-optimization/82244
9729 * tree-vrp.c (remove_range_assertions): Do not propagate
9730 a constant to abnormals but replace the assert with a copy.
9732 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
9734 PR rtl-optimization/57878
9735 PR rtl-optimization/68988
9736 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
9737 avoidance test involving non_reload_pseudos. Move frequency test
9738 below the general fragmentation avoidance test.
9740 2017-09-19 Richard Biener <rguenther@suse.de>
9742 PR tree-optimization/69728
9743 * graphite-sese-to-poly.c (schedule_error): New global.
9744 (add_loop_schedule): Handle empty domain by failing the
9746 (build_original_schedule): Handle schedule_error.
9748 2017-09-19 Richard Biener <rguenther@suse.de>
9750 * graphite-scop-detection.c (scop_detection::can_represent_loop):
9751 Do not iterate to sibling loops but only to siblings of inner
9754 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
9757 * config/m68k/m68k.md (moveq feeding equality comparison): Check
9758 that the registers are different.
9760 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
9762 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
9763 to processor_model and "amdfam17h" to arch_names_table.
9764 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
9766 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9769 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
9771 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9772 Alan Hayward <alan.hayward@arm.com>
9773 David Sherwood <david.sherwood@arm.com>
9775 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
9777 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
9779 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
9780 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
9781 vect_schedule_slp_instance.
9782 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
9783 Update call to vect_slp_analyze_node_operations. Simplify return
9785 (vect_slp_analyze_bb_1): Update call accordingly.
9786 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
9787 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
9788 (vect_schedule_slp): Update call accordingly.
9790 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9791 Alan Hayward <alan.hayward@arm.com>
9792 David Sherwood <david.sherwood@arm.com>
9794 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
9795 with types that aren't in fact scalar.
9797 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9799 * tree-vect-slp.c (vect_record_max_nunits): New function,
9801 (vect_build_slp_tree_1): ...here.
9802 (vect_build_slp_tree_2): Call it for phis too.
9804 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9806 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
9807 to vect_get_vec_def_for_operand when getting the mask operand.
9809 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9810 Alan Hayward <alan.hayward@arm.com>
9811 David Sherwood <david.sherwood@arm.com>
9813 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
9816 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9817 Alan Hayward <alan.hayward@arm.com>
9818 David Sherwood <david.sherwood@arm.com>
9820 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
9821 calculation for vector booleans.
9823 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
9824 Alan Hayward <alan.hayward@arm.com>
9825 David Sherwood <david.sherwood@arm.com>
9827 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
9829 (vect_transform_stmt): ...here.
9830 (vect_analyze_stmt): Use it instead of calling
9831 vectorizable_live_operation directly.
9833 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
9835 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
9836 non-SIMT targets in acc vector loops.
9838 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
9840 * configure.ac: Add arc and check if assembler supports gdwarf2.
9841 * configure: Regenerate.
9843 2017-09-18 Richard Biener <rguenther@suse.de>
9845 PR tree-optimization/82220
9846 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
9847 epilogue niters from the min_profitable_iters compute.
9849 2017-09-18 Jakub Jelinek <jakub@redhat.com>
9852 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
9853 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
9854 (ix86_init_pic_reg): Revert 2017-09-01 changes.
9856 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
9859 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
9860 switching to a new text section.
9862 2017-09-18 Richard Biener <rguenther@suse.de>
9864 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
9866 (build_alias_set): Reject aliases with no access function.
9868 2017-09-18 Richard Biener <rguenther@suse.de>
9870 PR tree-optimization/79622
9871 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
9873 (build_cross_bb_scalars_use): Likewise.
9875 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
9877 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
9879 2017-09-18 Alan Modra <amodra@gmail.com>
9882 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9883 stack_pointer_rtx for count 0. Update comments. Break up
9884 large rtl expression.
9886 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9888 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9889 Increase to 20 bytes.
9890 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9891 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9892 or avx version of the stub.
9894 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9897 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9898 compute the minimum stack alignment. Also update preferred stack
9899 boundary for leaf functions.
9901 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9903 PR tree-optimization/82228
9904 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9907 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9909 * common/config/nds32/nds32-common.c
9910 (nds32_option_optimization_table): Refine formatting.
9911 (nds32_option_optimization_table): Use -fsched-pressure and
9912 -fomit-frame-pointer for specific optimization level.
9914 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9916 * config/nds32/nds32.c: Refine formatting and comments.
9917 * config/nds32/nds32.h: Likewise.
9918 * config/nds32/nds32.md: Likewise.
9919 * config/nds32/nds32-cost.c: Likewise.
9920 * config/nds32/nds32-isr.c: Likewise.
9921 * config/nds32/nds32-md-auxiliary.c: Likewise.
9922 * config/nds32/nds32-multiple.md: Likewise.
9923 * config/nds32/nds32-predicates.c: Likewise.
9925 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9926 Jakub Jelinek <jakub@redhat.com>
9928 Add support for -std=c++2a.
9929 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9931 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9933 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9936 * doc/extend.texi (Common Function Attributes): Add
9937 references to ARM, AArch64, and S/390 specific attributes.
9938 (Function Specific Option Pragmas): Add AArch64 and S/390
9939 to list of back ends that support the target pragma.
9941 2017-09-15 Nathan Sidwell <nathan@acm.org>
9943 * doc/standards.texi: Fix C++17 description. Update URLs for
9946 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9948 * common.opt (Wcast-align=strict): New warning option.
9949 * doc/invoke.texi: Document -Wcast-align=strict.
9951 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9953 * cgraph.h (cgraph_thunk_info): Add comments.
9954 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9955 assert for VIRTUAL_* arguments stricter.
9957 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9959 PR tree-optimization/71026
9960 * match.pd: Move RDIV patterns from fold-const.c
9961 * fold-const.c (distribute_real_division): Removed.
9962 (fold_binary_loc): Remove calls to distribute_real_divison.
9964 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9966 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9967 c++1z and gnu++1z as deprecated. Change other references to
9968 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9969 Change -Wc++1z-compat to -Wc++17-compat.
9970 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9971 * dwarf2out.c (highest_c_language): Handle C++17.
9972 (gen_compile_unit_die): Likewise.
9974 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9976 PR rtl-optimization/82192
9977 * combine.c (make_extraction): Don't look through non-paradoxical
9978 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9981 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9982 Alan Hayward <alan.hayward@arm.com>
9983 David Sherwood <david.sherwood@arm.com>
9985 * target.def (function_arg_offset): New hook.
9986 * targhooks.h (default_function_arg_offset): Declare.
9987 * targhooks.c (default_function_arg_offset): New function.
9988 * function.c (locate_and_pad_parm): Use
9989 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9990 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9991 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9992 * doc/tm.texi: Regenerate.
9993 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9994 * config/spu/spu.c (spu_function_arg_offset): New function.
9995 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9996 * system.h (FUNCTION_ARG_OFFSET): Poison.
9998 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9999 Alan Hayard <alan.hayward@arm.com>
10000 David Sherwood <david.sherwood@arm.com>
10002 * target.def (truly_noop_truncation): New hook.
10003 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10004 than TRULY_NOOP_TRUNCATION.
10005 * hooks.h (hook_bool_uint_uint_true): Declare.
10006 * hooks.c (hook_bool_uint_uint_true): New function.
10007 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10008 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10009 * doc/tm.texi: Regenerate.
10010 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10011 rather than TRULY_NOOP_TRUNCATION in comments.
10012 (simplify_comparison): Likewise.
10013 (record_truncated_value): Likewise.
10014 * expmed.c (extract_bit_field_1): Likewise.
10015 (extract_split_bit_field): Likewise.
10016 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10017 instead of TRULY_NOOP_TRUNCATION.
10018 * function.c (assign_parm_setup_block): Likewise.
10019 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10020 * rtlhooks.c: Include target.h.
10021 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10022 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10023 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10024 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10025 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10026 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10027 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10028 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10029 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10030 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10031 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10032 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10033 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10034 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10035 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10036 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10037 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10038 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10039 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10040 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10041 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10042 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10043 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10044 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10045 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10046 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10047 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10048 rather than TRULY_NOOP_TRUNCATION in comments.
10049 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10050 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10051 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10052 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10053 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10054 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10055 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10056 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10057 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10058 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10059 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10060 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10061 rather than TRULY_NOOP_TRUNCATION in comments.
10062 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10063 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10064 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10065 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10066 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10067 TRULY_NOOP_TRUNCATION condition.
10068 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10069 (TRULY_NOOP_TRUNCATION): Delete.
10070 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10071 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10072 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10073 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10074 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10075 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10076 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10077 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10078 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10079 rather than TRULY_NOOP_TRUNCATION in comments.
10080 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10081 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10082 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10083 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10084 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10085 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10087 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10090 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10091 (*cmp_ior): Likewise.
10092 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10093 (*ior_scc_scc_cmp): Likewise.
10094 (*and_scc_scc): Likewise.
10095 (*and_scc_scc_cmp): Likewise.
10097 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10098 Alan Hayard <alan.hayward@arm.com>
10099 David Sherwood <david.sherwood@arm.com>
10101 * target.def (can_change_mode_class): New hook.
10102 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10103 (hard_regno_nregs): Likewise.
10104 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10105 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10106 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10107 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10108 (LOAD_EXTEND_OP): Update accordingly.
10109 * doc/tm.texi: Regenerate.
10110 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10111 CANNOT_CHANGE_MODE_CLASS.
10112 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10113 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10114 * combine.c (simplify_set): Update accordingly.
10115 * emit-rtl.c (validate_subreg): Likewise.
10116 * recog.c (general_operand): Likewise.
10117 * regcprop.c (mode_change_ok): Likewise.
10118 * reload1.c (choose_reload_regs): Likewise.
10119 (inherit_piecemeal_p): Likewise.
10120 * rtlanal.c (simplify_subreg_regno): Likewise.
10121 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10122 instead of CANNOT_CHANGE_MODE_CLASS.
10123 (reload_cse_simplify_operands): Likewise.
10124 * reload.c (push_reload): Use targetm.can_change_mode_class
10125 instead of CANNOT_CHANGE_MODE_CLASS.
10126 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10127 REG_CANNOT_CHANGE_MODE_P.
10128 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10129 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10130 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10131 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10132 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10133 (arm_can_change_mode_class): New function.
10134 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10135 than CANNOT_CHANGE_MODE_CLASS in comments.
10136 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10137 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10138 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10139 (ix86_can_change_mode_class): ...this new function, inverting the
10140 sense of the return value.
10141 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10142 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10143 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10144 (ia64_can_change_mode_class): New function.
10145 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10146 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10147 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10148 (m32c_can_change_mode_class): ...this new function, inverting the
10149 sense of the return value.
10150 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10151 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10152 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10153 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10154 (mips_can_change_mode_class): ...this new function, inverting the
10155 sense of the return value.
10156 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10157 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10158 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10159 (msp430_can_change_mode_class): New function.
10160 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10161 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10162 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10163 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10164 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10165 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10166 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10167 (pa_cannot_change_mode_class): Replace with...
10168 (pa_can_change_mode_class): ...this new function, inverting the
10169 sense of the return value.
10170 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10171 than CANNOT_CHANGE_MODE_CLASS in comments.
10172 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10173 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10174 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10175 (pdp11_cannot_change_mode_class): Replace with...
10176 (pdp11_can_change_mode_class): ...this new function, inverting the
10177 sense of the return value.
10178 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10179 * config/powerpcspe/powerpcspe-protos.h
10180 (rs6000_cannot_change_mode_class_ptr): Delete.
10181 * config/powerpcspe/powerpcspe.c
10182 (rs6000_cannot_change_mode_class_ptr): Delete.
10183 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10184 (rs6000_option_override_internal): Assign to
10185 targetm.can_change_mode_class instead of
10186 rs6000_cannot_change_mode_class_ptr.
10187 (rs6000_cannot_change_mode_class): Replace with...
10188 (rs6000_can_change_mode_class): ...this new function, inverting the
10189 sense of the return value.
10190 (rs6000_debug_cannot_change_mode_class): Replace with...
10191 (rs6000_debug_can_change_mode_class): ...this new function.
10192 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10193 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10194 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10195 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10196 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10198 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10199 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10200 (rs6000_option_override_internal): Assign to
10201 targetm.can_change_mode_class instead of
10202 rs6000_cannot_change_mode_class_ptr.
10203 (rs6000_cannot_change_mode_class): Replace with...
10204 (rs6000_can_change_mode_class): ...this new function, inverting the
10205 sense of the return value.
10206 (rs6000_debug_cannot_change_mode_class): Replace with...
10207 (rs6000_debug_can_change_mode_class): ...this new function.
10208 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10209 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10210 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10211 (s390_can_change_mode_class): ...this new function, inverting the
10212 sense of the return value.
10213 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10214 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10215 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10216 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10217 (sh_cannot_change_mode_class): Replace with...
10218 (sh_can_change_mode_class): ...this new function, inverting the
10219 sense of the return value.
10220 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10221 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10222 (sparc_can_change_mode_class): New function.
10223 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10224 * config/spu/spu.c (spu_can_change_mode_class): New function.
10225 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10226 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10227 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10228 (visium_can_change_mode_class): New function.
10229 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10231 2017-09-15 Richard Biener <rguenther@suse.de>
10233 PR tree-optimization/82217
10234 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10235 but not undefined case.
10237 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10240 * postreload.c (reload_cse_simplify_operands): Skip
10241 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10243 2017-09-15 Richard Biener <rguenther@suse.de>
10245 PR tree-optimization/68823
10246 * graphite-scop-detection.c (build_alias_set): If we have a
10247 possible dependence check whether we can handle them by just
10248 looking at the DRs DR_ACCESS_FNs.
10249 (build_scops): If build_alias_set fails, fail the SCOP.
10251 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10253 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10254 to support float128 built-in functions that require the ISA 3.0
10256 (BU_FLOAT128_3_HW): Likewise.
10257 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10258 built-in functions.
10259 (FMAF128): Likewise.
10261 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10262 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10263 floating point instructions.
10264 (rs6000_invalid_builtin): Likewise.
10265 (rs6000_builtin_mask_names): Likewise.
10266 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10267 (RS6000_BTM_FLOAT128_HW): Likewise.
10268 (RS6000_BTM_COMMON): Likewise.
10269 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10271 * doc/extend.texi (RS/6000 built-in functions): Document the
10272 IEEE 128-bit floating point square root and fused multiply-add
10273 built-in functions.
10275 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10277 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10278 reg (r2) isn't in the set of registers defined in the prologue.
10280 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10281 Alan Hayward <alan.hayward@arm.com>
10282 David Sherwood <david.sherwood@arm.com>
10284 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10285 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10286 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10287 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10288 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10290 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10291 max_vectorization_factor.
10292 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10294 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10295 Alan Hayward <alan.hayward@arm.com>
10296 David Sherwood <david.sherwood@arm.com>
10298 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10299 (vect_worthwhile_without_simd_p): Declare.
10300 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10301 (vectorizable_reduction): Use it.
10302 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10303 (vectorizable_operation): Likewise.
10305 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10306 Alan Hayward <alan.hayward@arm.com>
10307 David Sherwood <david.sherwood@arm.com>
10309 * tree-vectorizer.h (vect_get_num_copies): New function.
10310 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10311 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10312 (vectorizable_induction): Likewise.
10313 (vectorizable_live_operation): Likewise.
10314 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10315 (vectorizable_bswap): Likewise.
10316 (vectorizable_call): Likewise.
10317 (vectorizable_conversion): Likewise.
10318 (vectorizable_assignment): Likewise.
10319 (vectorizable_shift): Likewise.
10320 (vectorizable_operation): Likewise.
10321 (vectorizable_store): Likewise.
10322 (vectorizable_load): Likewise.
10323 (vectorizable_condition): Likewise.
10324 (vectorizable_comparison): Likewise.
10325 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10327 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10328 Alan Hayward <alan.hayward@arm.com>
10329 David Sherwood <david.sherwood@arm.com>
10331 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10332 of vect_init_vector.
10334 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10335 Alan Hayward <alan.hayward@arm.com>
10336 David Sherwood <david.sherwood@arm.com>
10338 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10339 an inline wrapper that provides a location.
10340 (gimple_build_vector): Likewise.
10341 * gimple-fold.c (gimple_build_vector_from_val): New function.
10342 (gimple_build_vector): Likewise.
10343 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10344 functions to build the initial value. Always return a gimple value.
10345 (get_initial_defs_for_reduction): Likewise. Only compute
10347 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10348 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10349 (vectorizable_induction): Use gimple_build_vector rather than
10352 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10353 Alan Hayward <alan.hayward@arm.com>
10354 David Sherwood <david.sherwood@arm.com>
10356 * target.h (vec_perm_indices): New typedef.
10357 (auto_vec_perm_indices): Likewise.
10358 * optabs-query.h: Include target.h
10359 (can_vec_perm_p): Take a vec_perm_indices *.
10360 * optabs-query.c (can_vec_perm_p): Likewise.
10361 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10362 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10363 * tree-vect-generic.c (lower_vec_perm): Likewise.
10364 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10365 (vect_grouped_load_supported): Likewise.
10366 (vect_shift_permute_load_chain): Likewise.
10367 (vect_permute_store_chain): Use auto_vec_perm_indices.
10368 (vect_permute_load_chain): Likewise.
10369 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10370 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10371 Update uses of can_vec_perm_p.
10372 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10373 mode with a number of elements. Take a vec_perm_indices *.
10374 (vect_create_epilog_for_reduction): Update accordingly.
10375 Use auto_vec_perm_indices.
10376 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10377 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10378 (vect_transform_slp_perm_load): Likewise.
10379 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10380 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10381 (vect_gen_perm_mask_checked): Likewise.
10382 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10383 (vect_gen_perm_mask_checked): Likewise.
10384 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10385 (vectorizable_store): Likewise.
10386 (vectorizable_load): Likewise.
10387 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10388 (vectorizable_bswap): Likewise.
10390 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10391 Alan Hayward <alan.hayward@arm.com>
10392 David Sherwood <david.sherwood@arm.com>
10394 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10395 * tree.c (build_vector): Likewise.
10396 (build_vector_from_ctor): Update accordingly.
10397 (build_vector_from_val): Likewise.
10398 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10399 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10400 * tree-vect-generic.c (add_rshift): Likewise.
10401 (expand_vector_divmod): Likewise.
10402 (optimize_vector_constructor): Likewise.
10403 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10404 (vect_transform_slp_perm_load): Likewise.
10405 (vect_schedule_slp_instance): Likewise.
10406 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10407 (vectorizable_call): Likewise.
10408 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10409 * expmed.c (make_tree): Likewise.
10410 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10411 a vector passed to build_vector.
10412 (fold_convert_const): Likewise.
10413 (exact_inverse): Likewise.
10414 (fold_ternary_loc): Likewise.
10415 (fold_relational_const): Likewise.
10416 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10417 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10418 (const_unop): Likewise. Store the reduction accumulator in a
10419 variable rather than an array.
10420 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10421 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10422 the new vector, rather than constructing it after the input arrays.
10423 (native_interpret_vector): Use auto_vec<tree> when building
10424 a vector passed to build_vector. Add elements in order.
10425 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10426 auto_vec<tree> when building a vector passed to build_vector.
10427 (vect_create_epilog_for_reduction): Likewise.
10428 (vectorizable_induction): Likewise.
10429 (get_initial_def_for_reduction): Likewise. Fix indentation of
10431 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10433 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10434 passed to build_vector.
10436 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10437 Alan Hayward <alan.hayward@arm.com>
10438 David Sherwood <david.sherwood@arm.com>
10440 * tree-core.h (tree_base::u): Add an "nelts" field.
10441 (tree_vector): Use VECTOR_CST_NELTS as the length.
10442 * tree.c (tree_size): Likewise.
10443 (make_vector): Initialize VECTOR_CST_NELTS.
10444 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10445 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10446 TYPE_VECTOR_SUBPARTS.
10447 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10448 as the number of units, setting it from VECTOR_CST_NELTS.
10449 (const_vector_from_tree): Likewise.
10450 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10451 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10452 (fold_negate_expr_1): Likewise.
10453 (fold_convert_const): Likewise.
10454 (const_binop): Likewise. Differentiate the number of output and
10456 (const_unop): Likewise.
10457 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10458 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10459 in cases that did the opposite.
10461 2017-09-14 Richard Biener <rguenther@suse.de>
10463 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10466 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10468 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10470 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10473 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10474 if and where to split a bb, except for splitting before debug insn
10475 sequences followed by non-label real insn. Delete debug insns
10476 in between basic blocks.
10478 * combine.c (make_compound_operation_int): Formatting fixes.
10480 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10481 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10482 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10483 * config/sol2.h (LINK_EH_SPEC): Likewise.
10484 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10485 * config/s390/linux.h (LINK_SPEC): Likewise.
10486 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10487 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10488 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10489 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10490 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10491 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10492 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10493 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10495 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10497 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10499 (ENDFILE_LINUX_SPEC): Likewise.
10500 (LINK_EH_SPEC): Likewise.
10501 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10502 (LINK_OS_LINUX_SPEC32): Likewise.
10503 (LINK_OS_LINUX_SPEC64): Likewise.
10504 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10505 (LINK_OS_LINUX_SPEC): Likewise.
10507 2017-09-13 Martin Liska <mliska@suse.cz>
10509 PR middle-end/82154
10510 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10511 CASE_HIGH is NULL_TREE.
10513 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10514 Alan Hayward <alan.hayward@arm.com>
10515 David Sherwood <david.sherwood@arm.com>
10517 * target.def (secondary_memory_needed): New hook.
10518 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10519 instead of SECONDARY_MEMORY_NEEDED.
10520 (secondary_memory_needed_mode): Likewise.
10521 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10522 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10523 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10524 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10525 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10526 * doc/tm.texi: Regenerate.
10527 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10528 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10529 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10530 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10531 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10532 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10533 mode argument first and change the reg_class arguments to reg_class_t.
10534 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10535 Make static. Update the call to inline_secondary_memory_needed.
10536 (ix86_register_move_cost): Update the call to
10537 inline_secondary_memory_needed.
10538 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10539 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10541 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10542 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10544 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10545 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10546 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10547 and match hook interface. Add comment from mips.h.
10548 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10549 * config/mmix/mmix.md (truncdfsf2): Refer to
10550 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10552 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10553 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10554 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10555 (pa_secondary_memory_needed): New function.
10556 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10557 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10558 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10559 (pdp11_secondary_memory_needed): Make static and match hook interface.
10560 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10561 * config/powerpcspe/powerpcspe-protos.h
10562 (rs6000_secondary_memory_needed_ptr): Delete.
10563 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10565 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10566 (rs6000_option_override_internal): Assign to
10567 targetm.secondary_memory_needed rather than
10568 rs6000_secondary_memory_needed_ptr.
10569 (rs6000_secondary_memory_needed): Match hook interface.
10570 (rs6000_debug_secondary_memory_needed): Likewise.
10571 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10572 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10573 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10574 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10575 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10576 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10578 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10579 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10580 (rs6000_option_override_internal): Assign to
10581 targetm.secondary_memory_needed rather than
10582 rs6000_secondary_memory_needed_ptr.
10583 (rs6000_secondary_memory_needed): Match hook interface.
10584 (rs6000_debug_secondary_memory_needed): Likewise.
10585 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10586 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10587 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10588 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10589 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10590 (sparc_secondary_memory_needed): New function.
10591 * lra-constraints.c (check_and_process_move): Refer to
10592 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10594 (curr_insn_transform): Likewise.
10595 (process_alt_operands): Use targetm.secondary_memory_needed
10596 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10597 (check_secondary_memory_needed_p): Likewise.
10598 (choose_split_class): Likewise.
10599 * reload.c: Unconditionally include code that was previously
10600 conditional on SECONDARY_MEMORY_NEEDED.
10601 (push_secondary_reload): Use targetm.secondary_memory_needed
10602 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10603 (push_reload): Likewise.
10604 * reload1.c: Unconditionally include code that was previously
10605 conditional on SECONDARY_MEMORY_NEEDED.
10606 (choose_reload_regs): Use targetm.secondary_memory_needed
10607 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10608 (gen_reload): Likewise.
10609 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10611 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10612 Alan Hayward <alan.hayward@arm.com>
10613 David Sherwood <david.sherwood@arm.com>
10615 * target.def (secondary_memory_needed_mode): New hook:
10616 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10617 * targhooks.h (default_secondary_memory_needed_mode): New function.
10618 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10619 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10620 * doc/tm.texi: Regenerate.
10621 * lra-constraints.c (check_and_process_move): Use
10622 targetm.secondary_memory_needed_mode instead of
10623 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10624 (curr_insn_transform): Likewise.
10625 * reload.c (get_secondary_mem): Likewise.
10626 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10627 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10629 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10630 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10631 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10632 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10633 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10635 * config/powerpcspe/powerpcspe-protos.h
10636 (rs6000_secondary_memory_needed_mode): Delete.
10637 * config/powerpcspe/powerpcspe.c
10638 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10639 (rs6000_secondary_memory_needed_mode): Make static.
10640 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10641 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10643 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10645 (rs6000_secondary_memory_needed_mode): Make static.
10646 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10647 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10648 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10649 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10650 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10652 (sparc_secondary_memory_needed_mode): New function.
10653 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10655 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10657 * config/aarch64/constraints.md (Umq): New constraint.
10658 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10660 (mov<mode>): Update condition.
10662 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10664 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10665 when bitposition is the same.
10667 2017-09-13 Richard Biener <rguenther@suse.de>
10669 * dwarf2out.c (output_die_symbol): Remove.
10670 (output_die): Do not output a DIEs symbol.
10672 2017-09-13 Richard Biener <rguenther@suse.de>
10674 PR middle-end/82128
10675 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10676 default-def to avoid breaking iterator update with the weird
10677 interaction with cgraph_update_edges_for_call_stmt_node.
10679 2017-09-13 Richard Biener <rguenther@suse.de>
10681 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10682 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10683 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10684 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10686 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10688 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10689 Disable pc relative literal load irrespective of
10690 TARGET_FIX_ERR_A53_84341 for default.
10692 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10694 * config/sparc/sparc.c (output_return): Output the source location of
10695 the insn in the delay slot, if any.
10696 (output_sibcall): Likewise.
10698 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
10701 * common.opt (-static-pie): New alias.
10702 (shared): Negate static-pie.
10703 (-no-pie): Update help text.
10705 (static-pie): New option.
10706 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
10707 -static-pie support.
10708 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
10709 (LINK_EH_SPEC): Likewise.
10710 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10711 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10712 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
10713 * gcc.c (LINK_COMMAND_SPEC): Likewise.
10714 (init_gcc_specs): Likewise.
10715 (init_spec): Likewise.
10716 (display_help): Update help message for -pie.
10717 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
10720 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
10722 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
10723 (movdi_aarch64): Likewise.
10724 (movti_aarch64): Likewise.
10726 2017-09-12 Simon Wright <simon@pushface.org>
10729 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
10730 calculation of the minor version, always output as 0.
10732 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10735 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10736 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
10737 on it early, rather than manual conversion late. For
10738 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
10739 instead of performing manual conversion.
10741 2017-09-12 Carl Love <cel@us.ibm.com>
10743 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
10744 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
10746 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10747 VMULOSW): Add definitions.
10748 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10749 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10750 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10751 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10752 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10754 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10756 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
10758 (movti_aarch64): Likewise.
10759 (movdf_aarch64): Likewise.
10760 (movtf_aarch64): Likewise.
10761 (load_pairdi): Likewise.
10762 (store_pairdi): Likewise.
10763 (load_pairdf): Likewise.
10764 (store_pairdf): Likewise.
10765 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
10766 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
10767 (ldr_got_small_<mode>): Likewise.
10768 (ldr_got_small_28k_<mode>): Likewise.
10769 (ldr_got_tiny): Likewise.
10770 * config/aarch64/iterators.md (ldst_sz): New.
10771 (ldpstp_sz): Likewise.
10772 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
10774 (thunderx_load): Split load_8 to load_16.
10775 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
10777 (thunderx2t99_storepair_basic): Split store_8 to store_16.
10778 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
10779 (xgene1_store_pair): Split store_8 to store_16.
10780 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
10781 (falkor_st_0_st_sd): Split store_8 to store_16.
10783 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
10785 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
10786 and store1/2/3/4 to store_4/8/12/16.
10787 * config/aarch64/aarch64.md: Update for rename.
10788 * config/arm/arm.md: Likewise.: Likewise.
10789 * config/arm/arm.c: Likewise.
10790 * config/arm/thumb1.md: Likewise.
10791 * config/arm/thumb2.md: Likewise.
10792 * config/arm/vfp.md: Likewise.
10793 * config/arm/arm-generic.md: Likewise.
10794 * config/arm/arm1020e.md: Likewise.
10795 * config/arm/arm1026ejs.md: Likewise.
10796 * config/arm/arm1136jfs.md: Likewise.
10797 * config/arm/arm926ejs.md: Likewise.
10798 * config/arm/cortex-a15.md: Likewise.
10799 * config/arm/cortex-a17.md: Likewise.
10800 * config/arm/cortex-a5.md: Likewise.
10801 * config/arm/cortex-a53.md: Likewise.
10802 * config/arm/cortex-a57.md: Likewise.
10803 * config/arm/cortex-a7.md: Likewise.
10804 * config/arm/cortex-a8.md: Likewise.
10805 * config/arm/cortex-a9.md: Likewise.
10806 * config/arm/cortex-m4.md: Likewise.
10807 * config/arm/cortex-m7.md: Likewise.
10808 * config/arm/cortex-r4.md: Likewise.
10809 * config/arm/exynos-m1.md: Likewise.
10810 * config/arm/fa526.md: Likewise.
10811 * config/arm/fa606te.md: Likewise.
10812 * config/arm/fa626te.md: Likewise.
10813 * config/arm/fa726te.md: Likewise.
10814 * config/arm/fmp626.md: Likewise.
10815 * config/arm/iwmmxt.md: Likewise.
10816 * config/arm/ldmstm.md: Likewise.
10817 * config/arm/marvell-pj4.md: Likewise.
10818 * config/arm/xgene1.md: Likewise.
10819 * config/aarch64/thunderx.md: Likewise.
10820 * config/aarch64/thunderx2t99.md: Likewise.
10821 * config/aarch64/falkor.md: Likewise.
10823 2017-09-12 Martin Liska <mliska@suse.cz>
10825 * attribs.c (private_lookup_attribute): New function.
10826 * attribs.h (private_lookup_attribute): Declared here.
10827 (lookup_attribute): Called from this place.
10829 2017-09-12 Richard Biener <rguenther@suse.de>
10831 PR tree-optimization/82157
10832 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
10833 stmts with side-effects.
10835 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10836 Alan Hayward <alan.hayward@arm.com>
10837 David Sherwood <david.sherwood@arm.com>
10839 * target.def (hard_regno_nregs): New hook.
10840 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
10841 * targhooks.h (default_hard_regno_nregs): Declare.
10842 * targhooks.c (default_hard_regno_nregs): New function.
10843 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
10844 (TARGET_HARD_REGNO_NREGS): ...this hook.
10845 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
10846 (CLASS_MAX_NREGS): Likewise.
10847 * doc/tm.texi: Regenerate.
10848 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
10849 instead of HARD_REGNO_NREGS.
10850 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
10851 HARD_REGNO_NREGS in the comment.
10852 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
10853 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
10854 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
10855 Return an unsigned int.
10856 (TARGET_HARD_REGNO_NREGS): Redefine.
10857 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
10858 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
10859 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10860 (arc_hard_regno_nregs): New function.
10861 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
10862 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
10863 (arm_hard_regno_nregs): New function.
10864 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
10865 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
10866 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
10867 (TARGET_HARD_REGNO_NREGS): Redefine.
10868 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
10869 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
10870 (HARD_REGNO_NREGS): Delete.
10871 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
10872 (cr16_hard_regno_nregs): New function.
10873 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
10874 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
10875 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
10876 (cris_hard_regno_nregs): New function.
10877 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
10878 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
10879 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10880 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
10881 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10882 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10883 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10884 (frv_hard_regno_nregs): Make static. Take and return an
10886 (frv_class_max_nregs): Remove outdated copy of documentation.
10887 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10888 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10889 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10890 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10891 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10892 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10893 (TARGET_HARD_REGNO_NREGS): Redefine.
10894 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10895 (CLASS_MAX_NREGS): Update comment.
10896 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10897 (ia64_hard_regno_nregs): New function.
10898 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10899 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10900 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10901 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10902 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10904 (m32c_hard_regno_nregs): Likewise. Make static.
10905 (TARGET_HARD_REGNO_NREGS): Redefine.
10906 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10907 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10908 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10909 (m68k_hard_regno_nregs): New function.
10910 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10911 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10912 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10913 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10914 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10915 Take and return an unsigned int.
10916 (TARGET_HARD_REGNO_NREGS): Redefine.
10917 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10918 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10919 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10920 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10921 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10922 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10923 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10924 (msp430_hard_regno_nregs): Make static. Take and return an
10926 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10927 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10928 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10929 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10930 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10931 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10932 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10933 (TARGET_HARD_REGNO_NREGS): Redefine.
10934 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10935 (PA_HARD_REGNO_NREGS): ...this.
10936 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10937 (PA_HARD_REGNO_NREGS): ...this.
10938 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10939 (pa_hard_regno_nregs): New function.
10940 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10941 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10942 (pdp11_hard_regno_nregs): New function.
10943 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10944 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10945 (rs6000_hard_regno_nregs_hook): New function.
10946 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10947 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10948 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10949 Take and return an unsigned int. Move earlier in file.
10950 (TARGET_HARD_REGNO_NREGS): Redefine.
10951 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10952 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10953 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10954 (rl78_hard_regno_nregs): Make static. Take and return an
10956 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10957 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10958 (rs6000_hard_regno_nregs_hook): New function.
10959 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10960 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10961 (TARGET_HARD_REGNO_NREGS): Redefine.
10962 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10963 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10964 instead of HARD_REGNO_NREGS.
10965 (s390_hard_regno_nregs): New function.
10966 (s390_hard_regno_mode_ok): Add comment from s390.h.
10967 (TARGET_HARD_REGNO_NREGS): Redefine.
10968 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10969 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10970 (sh_hard_regno_nregs): New function.
10971 (sh_pass_in_reg_p): Use it.
10972 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10973 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10974 (sparc_hard_regno_nregs): New function.
10975 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10976 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10977 (spu_function_arg_advance): Use it, supplying a valid register number.
10978 (TARGET_HARD_REGNO_NREGS): Redefine.
10979 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10980 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10981 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10982 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10983 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10984 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10985 (CLASS_MAX_NREGS): Remove copy of old documentation.
10986 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10987 (visium_hard_regno_nregs): New function.
10988 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10989 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10990 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10991 xtensa_hard_regno_nregs): New function.
10992 * system.h (HARD_REGNO_NREGS): Poison.
10994 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10996 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10997 hard_regno_nregs instead of HARD_REGNO_NREGS.
10998 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10999 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11000 (c6x_expand_epilogue): Likewise.
11001 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11002 (frv_read_iacc_argument): Likewise.
11003 * config/sh/sh.c: Include regs.h.
11004 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11005 (regs_used): Likewise.
11006 (output_stack_adjust): Likewise.
11007 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11008 * expmed.c: Include regs.h.
11009 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11010 * ree.c: Include regs.h.
11011 (combine_reaching_defs): Use hard_regno_nregs instead of
11013 (add_removable_extension): Likewise.
11015 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11017 * regs.h (hard_regno_nregs): Turn into a function.
11018 (end_hard_regno): Update accordingly.
11019 * caller-save.c (setup_save_areas): Likewise.
11020 (save_call_clobbered_regs): Likewise.
11021 (replace_reg_with_saved_mem): Likewise.
11022 (insert_restore): Likewise.
11023 (insert_save): Likewise.
11024 * combine.c (can_change_dest_mode): Likewise.
11025 (move_deaths): Likewise.
11026 (distribute_notes): Likewise.
11027 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11028 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11029 (rs6000_split_multireg_move): Likewise.
11030 (rs6000_register_move_cost): Likewise.
11031 (rs6000_memory_move_cost): Likewise.
11032 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11033 (rs6000_split_multireg_move): Likewise.
11034 (rs6000_register_move_cost): Likewise.
11035 (rs6000_memory_move_cost): Likewise.
11036 * cselib.c (cselib_reset_table): Likewise.
11037 (cselib_lookup_1): Likewise.
11038 * emit-rtl.c (set_mode_and_regno): Likewise.
11039 * function.c (aggregate_value_p): Likewise.
11040 * ira-color.c (setup_profitable_hard_regs): Likewise.
11041 (check_hard_reg_p): Likewise.
11042 (calculate_saved_nregs): Likewise.
11043 (assign_hard_reg): Likewise.
11044 (improve_allocation): Likewise.
11045 (calculate_spill_cost): Likewise.
11046 * ira-emit.c (modify_move_list): Likewise.
11047 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11048 (ira_hard_reg_in_set_p): Likewise.
11049 * ira.c (setup_reg_mode_hard_regset): Likewise.
11050 (clarify_prohibited_class_mode_regs): Likewise.
11051 (check_allocation): Likewise.
11052 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11053 (lra_setup_reg_renumber): Likewise.
11054 (setup_try_hard_regno_pseudos): Likewise.
11055 (spill_for): Likewise.
11056 (assign_hard_regno): Likewise.
11057 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11058 * lra-constraints.c (in_class_p): Likewise.
11059 (lra_constraint_offset): Likewise.
11060 (simplify_operand_subreg): Likewise.
11061 (lra_constraints): Likewise.
11062 (split_reg): Likewise.
11063 (split_if_necessary): Likewise.
11064 (invariant_p): Likewise.
11065 (inherit_in_ebb): Likewise.
11066 * lra-lives.c (process_bb_lives): Likewise.
11067 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11068 (get_hard_regs): Likewise.
11069 (do_remat): Likewise.
11070 * lra-spills.c (assign_spill_hard_regs): Likewise.
11071 * mode-switching.c (create_pre_exit): Likewise.
11072 * postreload.c (reload_combine_recognize_pattern): Likewise.
11073 * recog.c (peep2_find_free_register): Likewise.
11074 * regcprop.c (kill_value_regno): Likewise.
11075 (set_value_regno): Likewise.
11076 (copy_value): Likewise.
11077 (maybe_mode_change): Likewise.
11078 (find_oldest_value_reg): Likewise.
11079 (copyprop_hardreg_forward_1): Likewise.
11080 * regrename.c (check_new_reg_p): Likewise.
11081 (regrename_do_replace): Likewise.
11082 * reload.c (push_reload): Likewise.
11083 (combine_reloads): Likewise.
11084 (find_dummy_reload): Likewise.
11085 (operands_match_p): Likewise.
11086 (find_reloads): Likewise.
11087 (find_equiv_reg): Likewise.
11088 (reload_adjust_reg_for_mode): Likewise.
11089 * reload1.c (count_pseudo): Likewise.
11090 (count_spilled_pseudo): Likewise.
11091 (find_reg): Likewise.
11092 (clear_reload_reg_in_use): Likewise.
11093 (free_for_value_p): Likewise.
11094 (allocate_reload_reg): Likewise.
11095 (choose_reload_regs): Likewise.
11096 (reload_adjust_reg_for_temp): Likewise.
11097 (emit_reload_insns): Likewise.
11098 (delete_output_reload): Likewise.
11099 * rtlanal.c (subreg_get_info): Likewise.
11100 * sched-deps.c (sched_analyze_reg): Likewise.
11101 * sel-sched.c (init_regs_for_mode): Likewise.
11102 (mark_unavailable_hard_regs): Likewise.
11103 (choose_best_reg_1): Likewise.
11104 (verify_target_availability): Likewise.
11105 * valtrack.c (dead_debug_insert_temp): Likewise.
11106 * var-tracking.c (track_loc_p): Likewise.
11107 (emit_note_insn_var_location): Likewise.
11108 * varasm.c (make_decl_rtl): Likewise.
11109 * reginfo.c (choose_hard_reg_mode): Likewise.
11110 (init_reg_modes_target): Refer directly to
11111 this_target_regs->x_hard_regno_nregs.
11113 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11115 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11116 instead of hard_regno_nregs.
11118 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11120 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11121 end_hard_regno instead of hard_regno_nregs.
11122 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11123 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11124 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11125 * ira-color.c (improve_allocation): Likewise.
11126 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11127 * lra-lives.c (mark_regno_live): Likewise.
11128 (mark_regno_dead): Likewise.
11129 * lra-remat.c (operand_to_remat): Likewise.
11130 * lra.c (collect_non_operand_hard_regs): Likewise.
11131 * postreload.c (reload_combine_note_store): Likewise.
11132 (move2add_valid_value_p): Likewise.
11133 * reload.c (regno_clobbered_p): Likewise.
11135 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11137 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11139 * config/v850/v850.c (v850_reorg): Likewise.
11140 * reload.c (refers_to_regno_for_reload_p): Likewise.
11141 (find_equiv_reg): Likewise.
11142 * reload1.c (reload_reg_reaches_end_p): Likewise.
11144 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11146 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11148 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11149 * config/arm/arm.c (output_move_neon): Likewise.
11150 (arm_attr_length_move_neon): Likewise.
11151 (neon_split_vcombine): Likewise.
11152 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11153 (c6x_mark_reg_written): Likewise.
11154 (c6x_dwarf_register_span): Likewise.
11155 * config/i386/i386.c (ix86_save_reg): Likewise.
11156 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11157 (rws_access_reg): Likewise.
11158 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11159 * mode-switching.c (create_pre_exit): Likewise.
11160 * ree.c (combine_reaching_defs): Likewise.
11161 (add_removable_extension): Likewise.
11162 * regcprop.c (find_oldest_value_reg): Likewise.
11163 (copyprop_hardreg_forward_1): Likewise.
11164 * reload.c (reload_inner_reg_of_subreg): Likewise.
11165 (push_reload): Likewise.
11166 (combine_reloads): Likewise.
11167 (find_dummy_reload): Likewise.
11168 (reload_adjust_reg_for_mode): Likewise.
11169 * reload1.c (find_reload_regs): Likewise.
11170 (forget_old_reloads_1): Likewise.
11171 (reload_reg_free_for_value_p): Likewise.
11172 (reload_adjust_reg_for_temp): Likewise.
11173 (emit_reload_insns): Likewise.
11174 (delete_output_reload): Likewise.
11175 * sel-sched.c (choose_best_reg_1): Likewise.
11176 (choose_best_pseudo_reg): Likewise.
11178 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11179 Alan Hayward <alan.hayward@arm.com>
11180 David Sherwood <david.sherwood@arm.com>
11182 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11183 * target.def (slow_unaligned_access): New hook.
11184 * targhooks.h (default_slow_unaligned_access): Declare.
11185 * targhooks.c (default_slow_unaligned_access): New function.
11186 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11187 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11188 * doc/tm.texi: Regenerate.
11189 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11190 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11191 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11193 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11194 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11196 (rs6000_slow_unaligned_access): New function.
11197 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11198 (expand_block_compare): Likewise.
11199 (expand_strn_compare): Likewise.
11200 (rs6000_rtx_costs): Likewise.
11201 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11202 (riscv_slow_unaligned_access): Likewise.
11203 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11204 (riscv_slow_unaligned_access_p): ...this and make static.
11205 (riscv_option_override): Update accordingly.
11206 (riscv_slow_unaligned_access): New function.
11207 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11208 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11209 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11210 (rs6000_slow_unaligned_access): New function.
11211 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11212 (rs6000_rtx_costs): Likewise.
11213 * config/rs6000/rs6000-string.c (expand_block_compare)
11214 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11215 of SLOW_UNALIGNED_ACCESS.
11216 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11217 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11218 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11219 of SLOW_UNALIGNED_ACCESS.
11220 * expmed.c (simple_mem_bitfield_p): Likewise.
11221 * expr.c (alignment_for_piecewise_move): Likewise.
11222 (emit_group_load_1): Likewise.
11223 (emit_group_store): Likewise.
11224 (copy_blkmode_from_reg): Likewise.
11225 (emit_push_insn): Likewise.
11226 (expand_assignment): Likewise.
11227 (store_field): Likewise.
11228 (expand_expr_real_1): Likewise.
11229 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11230 * lra-constraints.c (simplify_operand_subreg): Likewise.
11231 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11232 * gimple-ssa-store-merging.c: Likewise in block comment at start
11234 * tree-ssa-strlen.c: Include target.h.
11235 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11236 of SLOW_UNALIGNED_ACCESS.
11237 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11239 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11241 PR rtl-optimization/82185
11242 * expmed.c (emit_store_flag_int): Only test tem if it has been
11245 2017-09-12 Richard Biener <rguenther@suse.de>
11247 PR middle-end/82149
11248 * match.pd ((FTYPE) N CMP CST): Fix typo.
11250 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11252 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11254 (mips_near_type_p): Add 'short_call' attribute as a synonym
11256 * doc/extend.texi (short_call): Document new function attribute.
11258 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11261 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11262 assertion check that in the condition.
11263 (get_atomic_generic_size): Likewise. Before testing if parameter
11264 has pointer type, if it has array type, call for C++
11265 default_conversion to perform array-to-pointer conversion.
11267 2017-09-12 Richard Biener <rguenther@suse.de>
11269 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11270 for operations we cannot scalarize.
11272 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11274 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11275 vectors heap vectors. Clean up comments.
11276 Make visited_bbs a reference.
11277 (profitable_jump_thread_path): Make GC
11278 vectors heap vectors. Clean up comments.
11280 (convert_and_register_jump_thread_path): Make GC vectors heap
11282 (check_subpath_and_update_thread_path): Same. Clean up comments.
11283 Make visited_bbs a reference.
11284 (handle_phi): Abstract common code to to
11285 register_jump_thread_path_if_profitable.
11286 Rename VAR_BB to DEF_BB.
11288 Make GC vectors heap vectors.
11289 Make visited_bbs a reference.
11290 (handle_assignment): Same.
11291 (register_jump_thread_path_if_profitable): New.
11292 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11294 Make GC vectors heap vectors. Clean up comments.
11295 Make visited_bbs a reference.
11296 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11297 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11300 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11303 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11304 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11306 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11308 Revert r251800 and r251799.
11310 2017-09-11 Martin Jambor <mjambor@suse.cz>
11313 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11314 arguments in advance.
11315 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11316 use it to find predecessor edges.
11317 (naive_outof_ssa): Collect vector of predecessors.
11319 2017-09-08 Jason Merrill <jason@redhat.com>
11321 PR c++/70029 - ICE with ref-qualifier and -flto
11322 * langhooks.h (struct lang_hooks_for_types): Add
11323 copy_lang_qualifiers.
11324 * attribs.c (build_type_attribute_qual_variant): Use it.
11325 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11327 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11328 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11330 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11333 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11334 (*mulsi3_sp64): New instruction.
11335 (mulsi3): New expander.
11337 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11339 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11341 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11343 * sancov.c: Include memmodel.h.
11345 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11348 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11351 2017-09-07 Carl Love <cel@us.ibm.com>
11353 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11354 the sldi instruction.
11356 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11358 * sancov.c: Include tm_p.h.
11360 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11363 * output.h (switch_to_other_text_partition): New declaration.
11364 * varasm.c (switch_to_other_text_partition): New function.
11365 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11366 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11367 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11368 to the other text partition before emitting LCL label and switch back
11369 after emitting the word after it.
11371 2017-09-07 Richard Biener <rguenther@suse.de>
11373 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11374 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11375 critical edges here, after loop init.
11376 (pass_data_pre): Remove PROP_no_crit_edges flags.
11377 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11378 for valueization of call args to avoid leaking VN_TOP.
11379 (visit_use): Assert we do not visit default defs.
11380 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11381 Use error_mark_node to more easily detect leaking VN_TOP.
11382 All default-defs are varying, not VN_TOP. Mark them visited.
11383 (run_scc_vn): Make code match comment.
11385 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11387 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11388 OPTION_MASK_FLOAT128_KEYWORD.
11389 (POWERPC_MASKS): Likewise.
11390 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11391 support for the -mfloat128-type option, and make -mfloat128
11392 default on PowerPC Linux systems. Define or undefine
11393 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11394 Define __float128 to be __ieee128 if IEEE 128-bit support is
11395 enabled, or undefine it.
11396 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11397 Delete defining __FLOAT128_TYPE__.
11398 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11399 -mfloat128-type option and make -mfloat128 default on PowerPC
11401 (TARGET_FLOAT128_TYPE): Likewise.
11402 (-mfloat128-type): Likewise.
11403 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11404 Delete the -mfloat128-type option and make -mfloat128 default on
11405 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11406 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11407 128-bit floating point is enabled. Change tests from using
11408 -mfloat128-type to -mfloat128.
11409 (rs6000_mangle_type): Use the correct mangling for the __float128
11410 type even if normal long double is restricted to 64-bits.
11411 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11413 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11414 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11415 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11417 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11418 documentation for -mfloat128.
11420 2017-09-06 Olivier Hainque <hainque@adacore.com>
11422 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11424 2017-09-06 Wish Wu <wishwu007@gmail.com>
11425 Jakub Jelinek <jakub@redhat.com>
11427 * asan.c (initialize_sanitizer_builtins): Add
11428 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11429 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11430 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11431 BT_FN_VOID_UINT64_PTR variables.
11432 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11433 (BT_FN_VOID_UINT16_UINT16): Likewise.
11434 (BT_FN_VOID_UINT32_UINT32): Likewise.
11435 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11436 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11437 (BT_FN_VOID_UINT64_PTR): Likewise.
11438 * common.opt (flag_sanitize_coverage): New variable.
11439 (fsanitize-coverage=trace-pc): Remove.
11440 (fsanitize-coverage=): Add.
11441 * flag-types.h (enum sanitize_coverage_code): New enum.
11442 * fold-const.c (fold_range_test): Disable non-short-circuit
11443 optimization if flag_sanitize_coverage.
11444 (fold_truth_andor): Likewise.
11445 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11446 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11447 (coverage_sanitizer_opts): New array.
11448 (get_closest_sanitizer_option): Add OPTS argument, handle also
11449 OPT_fsanitize_coverage_.
11450 (parse_sanitizer_options): Adjusted to also handle
11451 OPT_fsanitize_coverage_.
11452 (common_handle_option): Add OPT_fsanitize_coverage_.
11453 * sancov.c (instrument_comparison, instrument_switch): New function.
11454 (sancov_pass): Add trace-cmp support.
11455 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11456 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11457 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11458 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11459 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11460 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11461 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11462 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11463 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11464 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11466 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11468 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11469 error. Only quit immediately if parsing is complete.
11470 (BEGIN): Initialize fatal_err and parse_done.
11471 (begin fpu, end fpu): Check number of arguments.
11472 (begin arch, end arch): Likewise.
11473 (begin cpu, end cpu): Likewise.
11474 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11475 (optalias): Likewise.
11477 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11479 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11480 * config/arm/arm-isa.h: Delete. Move definitions to ...
11481 * arm-cpus.in: ... here. Use new feature and fgroup values.
11482 * config/arm/arm.c (arm_option_override): Use lower case for feature
11484 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11485 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11486 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11488 (print_isa_bits_for): New function.
11489 (gen_isa): New function.
11490 (gen_comm_data): Use print_isa_bits_for.
11491 (define feature): New keyword.
11492 (define fgroup): New keyword.
11493 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11494 (arm-isa.h): Add rule to generate file.
11495 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11496 case for feature bit names.
11498 2017-09-06 Richard Biener <rguenther@suse.de>
11500 * tree-ssa-pre.c (NECESSARY): Remove.
11501 (create_expression_by_pieces): Do not touch pass-local flags.
11502 (insert_into_preds_of_block): Likewise.
11503 (do_pre_regular_insertion): Likewise.
11504 (eliminate_insert): Likewise.
11505 (eliminate_dom_walker::before_dom_children): Likewise.
11506 (fini_eliminate): Do not look at inserted_exprs.
11507 (mark_operand_necessary): Remove.
11508 (remove_dead_inserted_code): Replace with simple work-list
11509 algorithm based on inserted_exprs and SSA uses.
11510 (pass_pre::execute): Re-order fini_eliminate and
11511 remove_dead_inserted_code.
11513 2017-09-06 Olivier Hainque <hainque@adacore.com>
11515 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11516 for VxWorks 7. Adjust surrounding comments.
11518 2017-09-06 Richard Biener <rguenther@suse.de>
11520 * gimple-ssa-strength-reduction.c
11521 (find_candidates_dom_walker::before_dom_children): Also allow
11524 2017-09-06 Richard Biener <rguenther@suse.de>
11526 PR tree-optimization/82108
11527 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11528 for gap in the non-permutation SLP case.
11530 2017-09-06 Martin Jambor <mjambor@suse.cz>
11532 PR tree-optimization/82078
11533 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11534 add_access_to_work_queue...
11535 (build_accesses_from_assign): ...here.
11536 (propagate_all_subaccesses): Make sure racc is the group
11537 representative, if there is one.
11539 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11541 PR middle-end/82095
11542 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11545 2017-09-06 Richard Biener <rguenther@suse.de>
11547 * gimple-ssa-strength-reduction.c
11548 (find_candidates_dom_walker::before_doom_children): Use a
11549 type and not a mode check.
11551 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11554 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11556 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11558 2017-09-05 Jeff Law <law@redhat.com>
11560 PR tree-optimization/64910
11561 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11562 cases where we have 3 or more operands.
11564 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11566 PR middle-end/81768
11567 * omp-low.c (lower_omp_for): Recompute tree invariant if
11568 gimple_omp_for_initial/final is ADDR_EXPR.
11570 PR middle-end/81768
11571 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11572 into gimple val before gimplification fo the COND_EXPR.
11574 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11576 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11577 REGION_COPY argument.
11578 (thread_through_all_blocks): Remove unused argument to
11579 duplicate_thread_path.
11581 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11582 Alan Hayward <alan.hayward@arm.com>
11583 David Sherwood <david.sherwood@arm.com>
11585 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11586 Take a scalar_mode rather than a machine_mode.
11587 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11588 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11589 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11590 (aarch64_gen_adjusted_ldpstp): Likewise.
11591 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11593 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11598 Take a scalar_int_mode instead of a machine_mode.
11599 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11600 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11601 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11602 (aarch64_simd_attr_length_rglist): Delete.
11603 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11604 a scalar_int_mode instead of a machine_mode.
11605 (aarch64_add_offset): Likewise.
11606 (aarch64_internal_mov_immediate): Likewise
11607 (aarch64_add_constant_internal): Likewise.
11608 (aarch64_add_constant): Likewise.
11609 (aarch64_movw_imm): Likewise.
11610 (aarch64_rtx_arith_op_extract_p): Likewise.
11611 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11612 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11613 Remove assert that the mode isn't a vector.
11614 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11615 (aarch64_expand_mov_immediate): Update calls after above changes.
11616 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11617 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11618 (aarch64_move_imm): Likewise.
11619 (aarch64_can_const_movi_rtx_p): Likewise.
11620 (aarch64_strip_extend): Likewise.
11621 (aarch64_extr_rtx_p): Likewise.
11622 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11623 a CONST_INT when the mode parameter is VOIDmode.
11624 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11626 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11628 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11629 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11630 (bitwise_type_for_mode): Update accordingly.
11632 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11634 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11635 * stor-layout.c (mode_for_size_tree): Likewise.
11636 (mode_for_array): Update accordingly.
11637 (layout_decl): Likewise.
11638 (compute_record_mode): Likewise. Only set the mode once.
11640 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11642 * target.def (get_mask_mode): Change return type to opt_mode.
11644 * doc/tm.texi: Regenerate.
11645 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11646 * targhooks.c (default_get_mask_mode): Likewise.
11647 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11648 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11649 targetm.get_mask_mode.
11650 * tree.c (build_truth_vector_type): Likewise.
11652 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11654 * machmode.h (mode_for_vector): Return an opt_mode.
11655 * stor-layout.c (mode_for_vector): Likewise.
11656 (mode_for_int_vector): Update accordingly.
11657 (layout_type): Likewise.
11658 * config/i386/i386.c (emit_memmov): Likewise.
11659 (ix86_expand_set_or_movmem): Likewise.
11660 (ix86_expand_vector_init): Likewise.
11661 (ix86_get_mask_mode): Likewise.
11662 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11664 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11665 * expmed.c (extract_bit_field_1): Likewise.
11666 * expr.c (expand_expr_real_2): Likewise.
11667 * optabs-query.c (can_vec_perm_p): Likewise.
11668 (can_vec_mask_load_store_p): Likewise.
11669 * optabs.c (expand_vec_perm): Likewise.
11670 * targhooks.c (default_get_mask_mode): Likewise.
11671 * tree-vect-stmts.c (vectorizable_store): Likewise.
11672 (vectorizable_load): Likewise.
11673 (get_vectype_for_scalar_type_and_size): Likewise.
11675 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11677 * machmode.h (mode_for_int_vector): New function.
11678 * stor-layout.c (mode_for_int_vector): Likewise.
11679 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11680 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11681 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11682 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11683 (s390_expand_vcond): Likewise.
11685 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11687 * machmode.h (opt_machine_mode): New type.
11688 (opt_mode<T>): Allow construction from anything that can be
11690 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11691 (mode_for_size): Return an opt_machine_mode.
11692 * stor-layout.c (mode_for_size): Likewise.
11693 (mode_for_size_tree): Update call accordingly.
11694 (bitwise_mode_for_mode): Likewise.
11695 (make_fract_type): Likewise.
11696 (make_accum_type): Likewise.
11697 * caller-save.c (replace_reg_with_saved_mem): Update call
11699 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11700 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11701 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11702 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
11703 * expmed.c (extract_bit_field_1): Likewise.
11704 * reload.c (get_secondary_mem): Likewise.
11705 * varasm.c (assemble_integer): Likewise.
11706 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
11709 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11711 * machmode.h (decimal_float_mode_for_size): New function.
11712 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
11713 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
11714 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
11715 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
11716 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
11717 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
11719 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11721 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
11722 (get_builtin_sync_mode): Likewise.
11723 (expand_ifn_atomic_compare_exchange): Likewise.
11724 (expand_builtin_atomic_clear): Likewise.
11725 (expand_builtin_atomic_test_and_set): Likewise.
11726 (fold_builtin_atomic_always_lock_free): Likewise.
11727 * calls.c (compute_argument_addresses): Likewise.
11728 (emit_library_call_value_1): Likewise.
11729 (store_one_arg): Likewise.
11730 * combine.c (combine_instructions): Likewise.
11731 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
11732 * config/arm/arm.c (arm_function_value): Likewise.
11733 (aapcs_allocate_return_reg): Likewise.
11734 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
11735 * config/i386/i386.c (construct_container): Likewise.
11736 (ix86_gimplify_va_arg): Likewise.
11737 (ix86_expand_sse_cmp): Likewise.
11738 (emit_memmov): Likewise.
11739 (emit_memset): Likewise.
11740 (expand_small_movmem_or_setmem): Likewise.
11741 (ix86_expand_pextr): Likewise.
11742 (ix86_expand_pinsr): Likewise.
11743 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
11744 * config/microblaze/microblaze.c (microblaze_block_move_straight):
11746 * config/mips/mips.c (mips_function_value_1) Likewise.
11747 (mips_block_move_straight): Likewise.
11748 (mips_expand_ins_as_unaligned_store): Likewise.
11749 * config/powerpcspe/powerpcspe.c
11750 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11751 (rs6000_darwin64_record_arg_flush): Likewise.
11752 * config/rs6000/rs6000.c
11753 (rs6000_darwin64_record_arg_advance_flush): Likewise.
11754 (rs6000_darwin64_record_arg_flush): Likewise.
11755 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
11756 (sparc_function_value_1): Likewise.
11757 * config/spu/spu.c (adjust_operand): Likewise.
11758 (spu_emit_branch_or_set): Likewise.
11759 (arith_immediate_p): Likewise.
11760 * emit-rtl.c (gen_lowpart_common): Likewise.
11761 * expr.c (expand_expr_real_1): Likewise.
11762 * function.c (assign_parm_setup_block): Likewise.
11763 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
11764 * reload1.c (alter_reg): Likewise.
11765 * stor-layout.c (mode_for_vector): Likewise.
11766 (layout_type): Likewise.
11768 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11770 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
11771 (spu_convert_move): Likewise.
11772 * lower-subreg.c (resolve_simple_move): Likewise.
11774 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11777 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
11778 define_insn to a define_expand.
11779 (altivec_vsum2sws_direct): New define_insn.
11780 (altivec_vsumsws): Convert from a define_insn to a define_expand.
11782 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11784 * config/arm/arm.c (arm_option_params_internal): Improve setting of
11787 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
11792 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
11793 realign stack if stack alignment needed is less than incoming
11796 2017-09-05 Marek Polacek <polacek@redhat.com>
11799 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
11802 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
11804 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
11806 2017-09-05 Richard Biener <rguenther@suse.de>
11808 PR tree-optimization/82084
11809 * fold-const.c (can_native_encode_string_p): Handle wide characters.
11811 2017-09-05 Richard Biener <rguenther@suse.de>
11813 PR tree-optimization/82102
11814 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
11816 2017-09-05 Martin Liska <mliska@suse.cz>
11818 PR tree-optimization/82032
11819 * tree-cfg.c (generate_range_test): New function.
11820 * tree-cfg.h (generate_range_test): Declared here.
11821 * tree-cfgcleanup.c (convert_single_case_switch): New function.
11822 (cleanup_control_expr_graph): Use it.
11823 * tree-switch-conversion.c (try_switch_expansion): Remove
11825 (emit_case_nodes): Use generate_range_test.
11827 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11830 * config/i386/i386.md (*<btsc><mode>_mask): Add
11831 TARGET_USE_BT to insn constraint.
11832 (*btr<mode>_mask): Ditto.
11834 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
11836 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
11837 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
11839 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
11842 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
11843 TARGET_NEON and TARGET_IWMMXT.
11844 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
11845 TARGET_NEON and TARGET_IWMMXT.
11846 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
11848 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
11850 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
11851 (ix86_rewrite_tls_address): Ditto.
11852 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
11853 (ix86_rewrite_tls_address_1): Ditto.
11854 (ix86_rewrite_tls_address): Ditto.
11855 * config/i386/predicates.md (tls_address_pattern): New predicate.
11856 * config/i386/i386.md (TLS address splitter): New splitter.
11858 2017-09-04 Richard Biener <rguenther@suse.de>
11860 PR tree-optimization/82084
11861 * fold-const.h (can_native_encode_string_p): Declare.
11862 * fold-const.c (can_native_encode_string_p): Factor out from ...
11863 (native_encode_string): ... here.
11864 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
11865 vectorizing stores from constants we later cannot handle.
11867 2017-09-04 Marek Polacek <polacek@redhat.com>
11870 * doc/invoke.texi: Update -Wtautological-compare documentation.
11872 2017-09-04 Jeff Law <law@redhat.com>
11874 PR tree-optimization/64910
11875 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
11876 swap the first and last operand if the last is a constant.
11878 2017-09-04 Marek Polacek <polacek@redhat.com>
11881 * convert.c (do_narrow): When sanitizing signed integer overflows,
11882 bail out for signed types.
11883 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11885 2017-09-04 Richard Biener <rguenther@suse.de>
11887 PR tree-optimization/82060
11888 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11889 Move devirtualization after stmt folding and before EH/AB/noreturn
11890 cleanup to get the stmt refs canonicalized. Use a bool instead
11891 of gimple_modified_p since that doesn't work for NOPs. Schedule
11892 NOPs generated by folding for removal.
11894 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11895 Alan Hayward <alan.hayward@arm.com>
11896 David Sherwood <david.sherwood@arm.com>
11898 * coretypes.h (pad_direction): New enum.
11899 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11900 (FUNCTION_ARG_PADDING): Likewise.
11901 * target.def (function_arg_padding): New hook.
11902 * targhooks.h (default_function_arg_padding): Declare.
11903 * targhooks.c (default_function_arg_padding): New function.
11904 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11905 (TARGET_FUNCTION_ARG_PADDING): ...this.
11906 * doc/tm.texi: Regenerate.
11907 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11908 instead of direction.
11909 (compute_argument_addresses): Likewise.
11910 (load_register_parameters): Likewise.
11911 (emit_library_call_value_1): Likewise.
11912 (store_one_arg): Use targetm.calls.function_arg_padding instead
11913 of FUNCTION_ARG_PADDING.
11914 (must_pass_in_stack_var_size_or_pad): Likewise.
11915 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11916 (emit_group_store): Likewise.
11917 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11918 instead of FUNCTION_ARG_PADDING.
11919 (emit_push_insn): Likewise, and propagate enum change throughout
11921 * function.h (direction): Delete.
11922 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11924 * function.c (assign_parm_find_stack_rtl): Likewise.
11925 (assign_parm_setup_block_p): Likewise.
11926 (assign_parm_setup_block): Likewise.
11927 (gimplify_parameters): Likewise.
11928 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11929 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11931 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11932 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11933 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11934 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11935 (aarch64_function_arg_padding): ...this new function.
11936 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11937 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11938 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11939 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11940 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11941 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11942 (arm_pad_arg_upward): Replace with...
11943 (arm_function_arg_padding): ...this new function.
11944 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11946 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11947 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11948 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11949 (ia64_hpux_function_arg_padding): Replace with...
11950 (ia64_function_arg_padding): ...this new function. Use pad_direction
11951 instead of direction. Check for TARGET_HPUX.
11952 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11953 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11954 (iq2000_function_arg_padding): New function.
11955 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11956 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11957 (mips_function_arg_padding): ...this new function.
11958 (mips_pad_reg_upward): Update accordingly.
11959 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11960 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11961 targetm.calls.function_arg_padding.
11962 (FUNCTION_ARG_PADDING): Delete.
11963 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11964 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11965 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11966 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11967 (nios2_block_reg_padding): Return pad_direction instead of direction.
11968 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11969 instead of direction.
11970 (nios2_function_arg_padding): Likewise. Make static.
11971 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11972 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11973 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11974 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11975 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11976 (pa_function_arg_padding): Make static. Return pad_direction instead
11978 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11979 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11980 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11981 instead of direction. Use targetm.calls.function_arg_padding.
11982 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11983 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11984 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11985 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11986 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11988 (function_arg_padding): Rename to...
11989 (rs6000_function_arg_padding): ...this. Make static. Return
11990 pad_direction instead of direction.
11991 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11992 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11993 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11994 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11995 instead of direction. Use targetm.calls.function_arg_padding.
11996 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11997 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11998 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11999 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12000 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12001 (function_arg_padding): Rename to...
12002 (rs6000_function_arg_padding): ...this. Make static. Return
12003 pad_direction instead of direction.
12004 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12005 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12006 * config/s390/s390.c (s390_function_arg_padding): New function.
12007 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12008 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12009 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12010 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12011 (function_arg_padding): Rename to...
12012 (sparc_function_arg_padding): ...this. Make static. Return
12013 pad_direction instead of direction.
12014 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12015 * config/spu/spu.c (spu_function_arg_padding): New function.
12016 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12017 * system.h (FUNCTION_ARG_PADDING): Poison.
12019 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12020 Alan Hayward <alan.hayward@arm.com>
12021 David Sherwood <david.sherwood@arm.com>
12023 * target.def (modes_tieable_p): New hook.
12024 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12025 (TARGET_MODES_TIEABLE_P): ...this.
12026 * doc/tm.texi.in: Regenerate.
12027 * hooks.h (hook_bool_mode_mode_true): Declare.
12028 * hooks.c (hook_bool_mode_mode_true): New function.
12029 * combine.c (subst): Use targetm.modes_tieable_p instead of
12031 * dse.c (find_shift_sequence): Likewise.
12032 * expmed.c (extract_low_bits): Likewise.
12033 * lower-subreg.c: Include target.h.
12034 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12036 * rtlanal.c (rtx_cost): Likewise.
12037 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12038 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12039 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12040 (TARGET_MODES_TIEABLE_P): Redefine.
12041 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12042 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12043 (TARGET_MODES_TIEABLE_P): Redefine.
12044 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12045 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12046 (arc_modes_tieable_p): New function.
12047 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12048 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12049 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12050 (arm_modes_tieable_p): Make static.
12051 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12052 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12053 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12054 (TARGET_MODES_TIEABLE_P): Redefine.
12055 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12056 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12057 (TARGET_MODES_TIEABLE_P): Redefine.
12058 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12059 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12060 (cr16_modes_tieable_p): New function.
12061 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12062 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12063 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12064 (TRULY_NOOP_TRUNCATION): Update comment.
12065 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12066 (TRULY_NOOP_TRUNCATION): Update comment.
12067 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12068 (frv_modes_tieable_p): New function.
12069 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12070 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12071 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12072 (TARGET_MODES_TIEABLE_P): Redefine.
12073 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12074 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12075 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12076 (TARGET_MODES_TIEABLE_P): Redefine.
12077 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12078 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12079 (ia64_modes_tieable_p): New function.
12080 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12081 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12082 (iq2000_modes_tieable_p): New function.
12083 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12084 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12085 (lm32_modes_tieable_p): New function.
12086 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12087 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12088 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12089 (TARGET_MODES_TIEABLE_P): Redefine.
12090 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12091 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12092 (m32r_modes_tieable_p): New function.
12093 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12094 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12095 (m68k_modes_tieable_p): New function.
12096 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12097 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12098 (mcore_modes_tieable_p): New function.
12099 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12100 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12102 (TARGET_MODES_TIEABLE_P): Redefine.
12103 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12104 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12105 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12106 (TARGET_MODES_TIEABLE_P): Redefine.
12107 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12108 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12109 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12110 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12111 (mn10300_modes_tieable_p): ...this and make static.
12112 (TARGET_MODES_TIEABLE_P): Redefine.
12113 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12114 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12115 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12116 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12117 (msp430_modes_tieable_p): Make static.
12118 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12119 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12120 (TARGET_MODES_TIEABLE_P): Redefine.
12121 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12122 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12123 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12124 (TARGET_MODES_TIEABLE_P): Redefine.
12125 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12126 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12127 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12128 (TARGET_MODES_TIEABLE_P): Redefine.
12129 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12130 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12131 (pdp11_modes_tieable_p): New function.
12132 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12133 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12134 (rs6000_modes_tieable_p): New function.
12135 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12136 * config/powerpcspe/powerpcspe.md: Update comment.
12137 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12138 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12139 (TARGET_MODES_TIEABLE_P): Redefine.
12140 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12141 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12142 (rl78_modes_tieable_p): New function.
12143 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12144 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12145 (rs6000_modes_tieable_p): New function.
12146 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12147 * config/rs6000/rs6000.md: Update comment.
12148 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12149 * config/rx/rx.c (rx_modes_tieable_p): New function.
12150 (TARGET_MODES_TIEABLE_P): Redefine.
12151 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12152 * config/s390/s390.c (s390_modes_tieable_p): New function.
12153 (TARGET_MODES_TIEABLE_P): Redefine.
12154 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12155 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12156 (sh_modes_tieable_p): New function.
12157 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12158 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12159 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12160 (sparc_modes_tieable_p): Make static.
12161 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12162 * config/spu/spu.c (spu_modes_tieable_p): New function.
12163 (TARGET_MODES_TIEABLE_P): Redefine.
12164 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12165 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12166 (TARGET_MODES_TIEABLE_P): Redefine.
12167 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12168 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12169 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12170 * config/v850/v850.c (v850_modes_tieable_p): New function.
12171 (TARGET_MODES_TIEABLE_P): Redefine.
12172 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12173 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12174 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12175 (visium_modes_tieable_p): New function.
12176 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12177 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12178 (xtensa_modes_tieable_p): New function.
12179 * system.h (MODES_TIEABLE_P): Poison.
12181 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12182 Alan Hayward <alan.hayward@arm.com>
12183 David Sherwood <david.sherwood@arm.com>
12185 * target.def (hard_regno_mode_ok): New hook.
12186 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12187 (TARGET_HARD_REGNO_MODE_OK): ...this.
12188 * doc/tm.texi.in: Regenerate.
12189 * hooks.h (hook_bool_uint_mode_true): Declare.
12190 * hooks.c (hook_bool_uint_mode_true): New function.
12191 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12192 HARD_REGNO_MODE_OK.
12193 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12194 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12195 instead of HARD_REGNO_MODE_OK.
12196 * caller-save.c: Include target.h.
12197 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12198 HARD_REGNO_MODE_OK.
12199 * combine.c (can_combine_p): Likewise.
12200 (combinable_i3pat): Likewise.
12201 (can_change_dest_mode): Likewise.
12202 * expr.c (init_expr_target): Likewise.
12203 (convert_move): Likewise.
12204 (convert_modes): Likewise.
12205 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12206 (setup_prohibited_mode_move_regs): Likewise.
12207 * ira.h (target_ira): Likewise.
12208 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12209 * lra-constraints.c (process_alt_operands): Likewise.
12210 (split_reg): Likewise.
12211 * recog.c (peep2_find_free_register): Likewise.
12212 * ree.c (combine_reaching_defs): Likewise.
12213 * regcprop.c (maybe_mode_change): Likewise.
12214 * reginfo.c (init_reg_sets_1): Likewise.
12215 (choose_hard_reg_mode): Likewise.
12216 (simplifiable_subregs): Likewise.
12217 * regrename.c (check_new_reg_p): Likewise.
12218 * reload.c (find_valid_class): Likewise.
12219 (find_valid_class_1): Likewise.
12220 (reload_inner_reg_of_subreg): Likewise.
12221 (push_reload): Likewise.
12222 (combine_reloads): Likewise.
12223 (find_dummy_reload): Likewise.
12224 (find_reloads): Likewise.
12225 * reload1.c (find_reg): Likewise.
12226 (set_reload_reg): Likewise.
12227 (allocate_reload_reg): Likewise.
12228 (choose_reload_regs): Likewise.
12229 (reload_adjust_reg_for_temp): Likewise.
12230 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12231 (simplify_subreg_regno): Likewise.
12232 * sel-sched.c (init_regs_for_mode): Likewise.
12233 * varasm.c (make_decl_rtl): Likewise.
12234 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12235 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12236 HARD_REGNO_MODE_OK.
12237 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12238 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12239 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12240 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12241 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12243 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12244 (arc_mode_class): Delete.
12245 (HARD_REGNO_MODE_OK): Delete.
12246 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12247 (arc_hard_regno_mode_ok): Rename old array to...
12248 (arc_hard_regno_mode_ok_modes): ...this.
12249 (arc_conditional_register_usage): Update accordingly.
12250 (arc_mode_class): Make static.
12251 (arc_hard_regno_mode_ok): New function.
12252 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12253 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12254 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12255 (arm_hard_regno_mode_ok): Make static.
12256 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12257 HARD_REGNO_MODE_OK.
12258 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12259 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12260 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12262 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12263 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12264 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12265 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12266 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12268 * config/bfin/predicates.md (valid_reg_operand): Use
12269 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12270 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12271 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12273 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12274 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12275 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12276 (cr16_hard_regno_mode_ok): Make static and return a bool.
12277 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12278 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12279 (cris_hard_regno_mode_ok): New function.
12280 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12281 (epiphany_mode_class): Delete.
12282 (HARD_REGNO_MODE_OK): Delete.
12283 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12284 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12285 (hard_regno_mode_ok): Rename to...
12286 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12287 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12288 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12289 HARD_REGNO_MODE_OK.
12290 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12291 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12292 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12293 (frv_hard_regno_mode_ok): Make static and return a bool.
12294 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12295 HARD_REGNO_MODE_OK.
12296 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12297 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12298 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12299 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12301 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12302 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12303 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12304 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12307 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12308 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12309 (ia64_hard_regno_mode_ok): New function.
12310 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12311 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12312 (iq2000_hard_regno_mode_ok): New function.
12313 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12314 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12315 (lm32_hard_regno_mode_ok): New function.
12316 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12317 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12318 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12319 instead of HARD_REGNO_MODE_OK.
12320 (m32c_hard_regno_ok): Rename to...
12321 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12322 (m32c_cannot_change_mode_class): Update accordingly.
12323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12324 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12325 (m32r_mode_class): Delete.
12326 (HARD_REGNO_MODE_OK): Delete.
12327 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12328 (m32r_hard_regno_mode_ok): Rename to...
12329 (m32r_hard_regno_modes): ...this.
12330 (m32r_mode_class): Make static.
12331 (m32r_hard_regno_mode_ok): New function.
12332 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12333 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12334 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12335 (m68k_hard_regno_mode_ok): Make static.
12336 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12337 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12338 (mcore_hard_regno_mode_ok): New function.
12339 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12340 (HARD_REGNO_MODE_OK): Delete.
12341 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12343 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12344 (microblaze_hard_regno_mode_ok): New function.
12345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12346 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12347 (mips_hard_regno_mode_ok): Delete.
12348 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12349 (mips_hard_regno_mode_ok_p): ...this and make static.
12350 (mips_hard_regno_mode_ok_p): Rename to...
12351 (mips_hard_regno_mode_ok_uncached): ...this.
12352 (mips_hard_regno_mode_ok): New function.
12353 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12354 of HARD_REGNO_MODE_OK.
12355 (mips_option_override): Update after above name changes.
12356 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12357 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12358 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12359 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12360 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12362 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12363 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12364 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12365 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12366 (msp430_hard_regno_mode_ok): Make static and return a bool.
12367 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12368 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12369 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12371 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12372 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12373 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12374 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12375 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12376 (PA_HARD_REGNO_MODE_OK): ...this
12377 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12378 (PA_HARD_REGNO_MODE_OK): ...this.
12379 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12380 (pa_hard_regno_mode_ok): New function.
12381 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12382 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12383 (pdp11_hard_regno_mode_ok): New function.
12384 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12385 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12387 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12389 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12390 (rs6000_hard_regno_mode_ok): Rename to...
12391 (rs6000_hard_regno_mode_ok_uncached): ...this.
12392 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12393 (rs6000_hard_regno_mode_ok): New function.
12394 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12395 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12396 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12397 (riscv_hard_regno_mode_ok): ...this and make static.
12398 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12399 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12400 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12401 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12402 (rl78_hard_regno_mode_ok): Make static and return bool.
12403 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12404 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12406 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12408 (rs6000_hard_regno_mode_ok): Rename to...
12409 (rs6000_hard_regno_mode_ok_uncached): ...this.
12410 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12411 (rs6000_hard_regno_mode_ok): New function.
12412 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12413 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12414 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12415 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12416 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12417 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12418 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12419 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12420 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12421 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12422 (sh_hard_regno_mode_ok): Make static.
12423 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12424 instead of HARD_REGNO_MODE_OK.
12425 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12426 (sparc_mode_class): Delete.
12427 (HARD_REGNO_MODE_OK): Delete.
12428 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12429 (hard_regno_mode_classes): Make static.
12430 (sparc_mode_class): Likewise.
12431 (sparc_hard_regno_mode_ok): New function.
12432 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12433 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12434 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12436 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12437 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12438 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12439 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12440 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12442 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12443 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12444 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12445 (visium_hard_regno_mode_ok): New function.
12446 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12447 instead of HARD_REGNO_MODE_OK.
12448 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12449 (HARD_REGNO_MODE_OK): Delete.
12450 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12451 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12452 (xtensa_option_override): Update accordingly.
12453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12454 (xtensa_hard_regno_mode_ok): New function.
12455 * system.h (HARD_REGNO_MODE_OK): Poison.
12457 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12458 Alan Hayward <alan.hayward@arm.com>
12459 David Sherwood <david.sherwood@arm.com>
12461 * target.def (hard_regno_call_part_clobbered): New hook.
12462 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12463 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12464 * doc/tm.texi: Regenerate.
12465 * hooks.h (hook_bool_uint_mode_false): Declare.
12466 * hooks.c (hook_bool_uint_mode_false): New function.
12467 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12468 * cselib.c (cselib_process_insn): Use
12469 targetm.hard_regno_call_part_clobbered instead of
12470 HARD_REGNO_CALL_PART_CLOBBERED.
12471 * ira-conflicts.c (ira_build_conflicts): Likewise.
12472 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12473 * lra-constraints.c (need_for_call_save_p): Likewise.
12474 * lra-lives.c: Include target.h.
12475 (check_pseudos_live_through_calls): Use
12476 targetm.hard_regno_call_part_clobbered instead of
12477 HARD_REGNO_CALL_PART_CLOBBERED.
12478 * regcprop.c: Include target.h.
12479 (copyprop_hardreg_forward_1): Use
12480 targetm.hard_regno_call_part_clobbered instead of
12481 HARD_REGNO_CALL_PART_CLOBBERED.
12482 * reginfo.c (choose_hard_reg_mode): Likewise.
12483 * regrename.c (check_new_reg_p): Likewise.
12484 * reload.c (find_equiv_reg): Likewise.
12485 * reload1.c (emit_reload_insns): Likewise.
12486 * sched-deps.c (deps_analyze_insn): Likewise.
12487 * sel-sched.c (init_regs_for_mode): Likewise.
12488 (mark_unavailable_hard_regs): Likewise.
12489 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12490 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12491 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12493 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12494 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12495 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12497 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12499 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12500 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12501 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12503 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12504 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12505 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12507 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12508 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12510 * config/powerpcspe/powerpcspe.c
12511 (rs6000_hard_regno_call_part_clobbered): New function.
12512 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12513 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12514 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12516 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12517 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12518 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12520 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12521 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12522 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12524 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12525 Alan Hayward <alan.hayward@arm.com>
12526 David Sherwood <david.sherwood@arm.com>
12528 * rtl.h (subreg_memory_offset): Declare.
12529 * emit-rtl.c (subreg_memory_offset): New function.
12530 * expmed.c (store_bit_field_1): Use it.
12531 * expr.c (undefined_operand_subword_p): Likewise.
12532 * simplify-rtx.c (simplify_subreg): Likewise.
12534 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12536 PR rtl-optimization/57448
12539 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12540 using atomic_load pattern.
12541 (expand_atomic_store): Likewise.
12543 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12546 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12547 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12548 handling. Use replace_call_with_value with NULL instead of
12549 gsi_replace, unlink_stmt_vdef and release_defs.
12551 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12554 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12556 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12559 * rtl.h (emit_library_call_value_1): Declare.
12560 (emit_library_call): Replace declaration with a series of overloads.
12561 Remove the parameter count argument.
12562 (emit_library_call_value): Likewise.
12563 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12564 with an "rtx_mode_t *".
12565 (emit_library_call_value): Delete.
12566 (emit_library_call): Likewise.
12567 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12568 (asan_emit_allocas_unpoison): Likewise.
12569 * builtins.c (expand_builtin_powi): Likewise.
12570 (expand_asan_emit_allocas_unpoison): Likewise.
12571 * cfgexpand.c (expand_main_function): Likewise.
12572 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12573 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12574 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12575 * config/arm/arm.c (arm_trampoline_init): Likewise.
12576 (arm_call_tls_get_addr): Likewise.
12577 (arm_expand_divmod_libfunc): Likewise.
12578 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12579 (smulsi3_highpart): Likewise.
12580 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12581 (c6x_expand_compare): Likewise.
12582 (c6x_expand_movmem): Likewise.
12583 * config/frv/frv.c (frv_trampoline_init): Likewise.
12584 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12585 (ix86_expand_divmod_libfunc): Likewise.
12586 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12587 (ia64_expand_compare): Likewise.
12588 (ia64_profile_hook): Likewise.
12589 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12590 (nonlocal_goto): Likewise.
12591 (restore_stack_nonlocal): Likewise.
12592 * config/m32r/m32r.c (block_move_call): Likewise.
12593 (m32r_trampoline_init): Likewise.
12594 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12595 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12596 (m68k_call_m68k_read_tp): Likewise.
12597 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12598 (microblaze_expand_divide): Likewise.
12599 * config/mips/mips.h (mips_args): Likewise.
12600 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12601 (MIPS_ICACHE_SYNC): Likewise.
12602 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12603 (nios2_trampoline_init): Likewise.
12604 * config/pa/pa.c (hppa_tls_call): Likewise.
12605 (pa_trampoline_init): Likewise.
12606 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12607 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12608 (expand_strn_compare): Likewise.
12609 (rs6000_generate_compare): Likewise.
12610 (rs6000_expand_float128_convert): Likewise.
12611 (output_profile_hook): Likewise.
12612 (rs6000_trampoline_init): Likewise.
12613 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12614 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12615 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12616 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12617 (rs6000_generate_compare): Likewise.
12618 (rs6000_expand_float128_convert): Likewise.
12619 (output_profile_hook): Likewise.
12620 (rs6000_trampoline_init): Likewise.
12621 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12622 * config/sh/sh.c (sh_trampoline_init): Likewise.
12623 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12624 (sparc_emit_float_lib_cmp): Likewise.
12625 (sparc32_initialize_trampoline): Likewise.
12626 (sparc64_initialize_trampoline): Likewise.
12627 (sparc_profile_hook): Likewise.
12628 * config/spu/spu.c (ea_load_store): Likewise.
12629 * config/spu/spu.md (floatunssidf2): Likewise.
12630 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12631 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12632 * config/visium/visium.c (expand_block_move_4): Likewise.
12633 (expand_block_move_2): Likewise.
12634 (expand_block_move_1): Likewise.
12635 (expand_block_set_4): Likewise.
12636 (expand_block_set_2): Likewise.
12637 (expand_block_set_1): Likewise.
12638 (visium_trampoline_init): Likewise.
12639 (visium_profile_hook): Likewise.
12640 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12641 (xtensa_setup_frame_addresses): Likewise.
12642 (xtensa_trampoline_init): Likewise.
12643 * except.c (sjlj_emit_function_enter): Likewise.
12644 (sjlj_emit_function_exit): Likewise.
12645 * explow.c (allocate_dynamic_stack_space): Likewise.
12646 (probe_stack_range): Likewise.
12647 * expr.c (convert_mode_scalar): Likewise.
12648 * optabs.c (expand_binop): Likewise.
12649 (expand_twoval_binop_libfunc): Likewise.
12650 (expand_unop): Likewise.
12651 (prepare_cmp_insn): Likewise.
12652 (prepare_float_lib_cmp): Likewise.
12653 (expand_float): Likewise.
12654 (expand_fix): Likewise.
12655 (expand_fixed_convert): Likewise.
12656 (maybe_emit_sync_lock_test_and_set): Likewise.
12657 (expand_atomic_compare_and_swap): Likewise.
12658 (expand_mem_thread_fence): Likewise.
12659 (expand_atomic_fetch_op): Likewise.
12661 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12663 * doc/generic.texi (OpenACC): Adjust URL.
12664 * doc/invoke.texi (C Dialect Options): Ditto.
12666 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12668 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12669 predicate for operand 1. Add (m,<S>) constraint.
12670 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12671 Prevent memory operand 1 with register operand 2.
12673 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12675 PR rtl-optimization/82024
12676 * combine.c (try_combine): If the combination result is a PARALLEL,
12677 and we only need to retain the SET in there that would be placed
12678 at I2, check that we can place that at I3 instead, before doing so.
12680 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12683 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12685 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12686 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12689 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12690 Jeff Law <law@redhat.com>
12692 * varasm.c (bss_initializer_p): Do not put constants into .bss
12693 (categorize_decl_for_section): Handle bss_initializer_p returning
12694 false when DECL_INITIAL is NULL.
12696 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12699 * config/s390/s390.c (s390_can_inline_p): New function.
12701 2017-09-01 Jeff Law <law@redhat.com>
12703 PR tree-optimization/82052
12704 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12705 Always initialize the returned slot after a hash table miss
12706 when INSERT is true.
12708 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
12710 * config/s390/s390.md (mem_signal_fence): Remove.
12711 * doc/md.texi (mem_signal_fence): Remove.
12712 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
12714 * target-insns.def (mem_signal_fence): Remove.
12716 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12719 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
12722 * asan.c (create_odr_indicator): Strip name encoding from assembler
12723 name before appending it after __odr_asan_.
12725 2017-09-01 Martin Liska <mliska@suse.cz>
12727 PR tree-optimization/82059
12728 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
12729 frequency only when an edge is redirected.
12731 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12733 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
12734 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
12735 (arc_conditional_register_usage): Remove ARC600 lp_count
12737 (arc_file_start): Emit Tag_ARC_CPU_variation.
12738 (arc_can_use_doloop_p): New conditions to use ZOLs.
12739 (hwloop_fail): New function.
12740 (hwloop_optimize): Likewise.
12741 (hwloop_pattern_reg): Likewise.
12742 (arc_doloop_hooks): New struct, to be used with reorg_loops.
12743 (arc_reorg_loops): New function, calls reorg_loops.
12744 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
12745 (arc600_corereg_hazard): Remove ZOL checking, case handled by
12747 (arc_loop_hazard): Remove function, functionality moved into
12749 (arc_hazard): Remove arc_loop_hazard call.
12750 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
12751 into hwloop_optimize.
12752 (arc_label_align): Remove ZOL handling.
12753 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
12754 * config/arc/arc.md (doloop_begin): Remove pattern.
12755 (doloop_begin_i): Likewise.
12756 (doloop_end_i): Likewise.
12757 (doloop_fallback): Likewise.
12758 (doloop_fallback_m): Likewise.
12759 (doloop_end): Reimplement expand.
12760 (arc_lp): New pattern for LP instruction.
12761 (loop_end): New pattern.
12762 (loop_fail): Likewise.
12763 (decrement_and_branch_until_zero): Likewise.
12764 * config/arc/arc.opt (mlpc-width): New option.
12765 * doc/invoke.texi (mlpc-width): Document option.
12767 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12769 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
12770 (arc_ccfsm_advance): Fix checking for delay slots.
12771 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
12773 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12775 * config/arc/arc.md (movqi_insn): Add stores to save constant long
12777 (movhi_insn): Update store instruction constraint which are saving
12778 6-bit short immediates.
12779 (movsi_insn): Consider also short scaled load operations.
12780 (zero_extendhisi2_i): Use Usd constraint instead of T.
12781 (extendhisi2_i): Add q constraint.
12782 (arc_clzsi2): Add type and length attributes.
12783 (arc_ctzsi2): Likewise.
12784 * config/arc/constraints.md (Usc): Update constraint, the
12785 assembler can parse two relocations for a single instruction.
12787 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
12789 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
12790 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
12792 2017-08-31 Olivier Hainque <hainque@adacore.com>
12794 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
12795 match as powerpc-wrs-vxworks*.
12797 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
12799 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
12800 register constraint for by-element operand.
12801 (aarch64_mls_elt_merge<mode>): Likewise.
12803 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12805 * config/arc/arc.c (arc_can_follow_jump): Check for short
12808 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12810 * config.gcc: Use g.opt for arc.
12811 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
12812 functionality moved to ...
12813 (legitimate_scaled_address_p): New function, ...here.
12814 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
12815 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
12816 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
12818 (arc_override_options): Handle G option.
12819 (arc_output_pic_addr_const): Correct function definition.
12820 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
12821 (arc_decl_anon_ns_mem_p): Delete.
12822 (arc_in_small_data_p): Overhaul this function to take into
12823 consideration the value given via G option.
12824 (arc_rewrite_small_data_1): Renamed and corrected old
12825 arc_rewrite_small_data function.
12826 (arc_rewrite_small_data): New function.
12827 (small_data_pattern): Don't use pic_offset_table_rtx.
12828 * config/arc/arc.h (CC1_SPEC): Recognize G option.
12829 * config/arc/simdext.md (movmisalignv2hi): Use
12830 prepare_move_operands function.
12832 (movmisalign*): Likewise.
12833 * doc/invoke.texi (ARC options): Document -G option.
12835 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
12837 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
12839 * config/arc/arc.c (arc_print_operand): Output scalled address for
12840 sdata whenever is possible.
12841 (arc_in_small_data_p): Allow sdata for 64bit datum when double
12842 load/stores are available.
12843 (compact_sda_memory_operand): Check for the alignment required by
12844 code density instructions.
12845 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
12847 * config/arc/constraints.md (Usd): Update constraint.
12848 (Us0): New constraint.
12849 (Usc): Update constraint.
12851 2017-08-31 Richard Biener <rguenther@suse.de>
12853 PR middle-end/82054
12854 * dwarf2out.c (dwarf2out_early_global_decl): Process each
12855 function only once.
12857 2017-08-31 Tamar Christina <tamar.christina@arm.com>
12859 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
12860 Resize type_signature.
12862 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12863 Alan Hayward <alan.hayward@arm.com>
12864 David Sherwood <david.sherwood@arm.com>
12866 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
12867 subregs whose inner modes can be stored in GPRs.
12868 (aarch64_classify_index): Likewise.
12870 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12871 Alan Hayward <alan.hayward@arm.com>
12872 David Sherwood <david.sherwood@arm.com>
12874 * config/aarch64/iterators.md (V_cmp_result): Rename to...
12875 (V_INT_EQUIV): ...this.
12876 (v_cmp_result): Rename to...
12877 (v_int_equiv): ...this.
12878 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
12879 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
12880 (copysign<mode>3): Likewise.
12881 (aarch64_simd_bsl<mode>_internal): Likewise.
12882 (aarch64_simd_bsl<mode>): Likewise.
12883 (vec_cmp<mode><mode>): Likewise.
12884 (vcond<mode><mode>): Likewise.
12885 (vcond<v_cmp_mixed><mode>): Likewise.
12886 (vcondu<mode><v_cmp_mixed>): Likewise.
12887 (aarch64_cm<optab><mode>): Likewise.
12888 (aarch64_cmtst<mode>): Likewise.
12889 (aarch64_fac<optab><mode>): Likewise.
12890 (vec_perm_const<mode>): Likewise.
12891 (vcond_mask_<mode><v_cmp_result>): Rename to...
12892 (vcond_mask_<mode><v_int_equiv>): ...this.
12893 (vec_cmp<mode><v_cmp_result>): Rename to...
12894 (vec_cmp<mode><v_int_equiv>): ...this.
12896 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12897 Alan Hayward <alan.hayward@arm.com>
12898 David Sherwood <david.sherwood@arm.com>
12900 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12902 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12903 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12904 (UNSPEC_LD4_DREG): New unspecs.
12905 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12906 (aarch64_ld2<mode>_dreg_be): Replace with...
12907 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12909 (aarch64_ld3<mode>_dreg_le)
12910 (aarch64_ld3<mode>_dreg_be): Replace with...
12911 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12913 (aarch64_ld4<mode>_dreg_le)
12914 (aarch64_ld4<mode>_dreg_be): Replace with...
12915 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12918 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12920 PR tree-optimization/81987
12921 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12922 insert an initializer in a location not dominated by the stride
12925 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12927 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12928 on the entire header of the finally block in the fallthru case.
12930 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12932 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12934 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12936 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12937 rs6000_emit_move_from_cr and call renamed function.
12938 (rs6000_emit_prologue): Call renamed functions.
12939 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12940 movesi_from_cr, remove volatile CRs.
12942 2017-08-30 Jon Beniston <jon@beniston.com>
12943 Richard Biener <rguenther@suse.de>
12945 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12946 of VECTOR_MODE_P check.
12947 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12948 element vector types.
12950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12952 * df.h (df_read_modify_subreg_p): Remove in favor of...
12953 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12954 const_rtx instead of an rtx.
12955 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12956 * df-problems.c (df_word_lr_mark_ref): Likewise.
12957 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12958 (mark_pseudo_reg_dead): Likewise.
12959 (mark_ref_dead): Likewise.
12960 * reginfo.c (init_subregs_of_mode): Likewise.
12961 * sched-deps.c (sched_analyze_1): Likewise.
12962 * df-scan.c (df_def_record_1): Likewise.
12963 (df_uses_record): Likewise.
12964 (df_read_modify_subreg_p): Remove in favor of...
12965 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12966 const_rtx instead of an rtx.
12968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12969 Alan Hayward <alan.hayward@arm.com>
12970 David Sherwood <david.sherwood@arm.com>
12972 * rtl.h (partial_subreg_p): New function.
12973 * caller-save.c (save_call_clobbered_regs): Use it.
12974 * calls.c (expand_call): Likewise.
12975 * combine.c (combinable_i3pat): Likewise.
12976 (simplify_set): Likewise.
12977 (make_extraction): Likewise.
12978 (make_compound_operation_int): Likewise.
12979 (gen_lowpart_or_truncate): Likewise.
12980 (force_to_mode): Likewise.
12981 (make_field_assignment): Likewise.
12982 (reg_truncated_to_mode): Likewise.
12983 (record_truncated_value): Likewise.
12984 (move_deaths): Likewise.
12985 * cse.c (record_jump_cond): Likewise.
12986 (cse_insn): Likewise.
12987 * cselib.c (cselib_lookup_1): Likewise.
12988 * expmed.c (extract_bit_field_using_extv): Likewise.
12989 * function.c (assign_parm_setup_reg): Likewise.
12990 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12991 * ira-build.c (create_insn_allocnos): Likewise.
12992 * lra-coalesce.c (merge_pseudos): Likewise.
12993 * lra-constraints.c (match_reload): Likewise.
12994 (simplify_operand_subreg): Likewise.
12995 (curr_insn_transform): Likewise.
12996 * lra-lives.c (process_bb_lives): Likewise.
12997 * lra.c (new_insn_reg): Likewise.
12998 (lra_substitute_pseudo): Likewise.
12999 * regcprop.c (mode_change_ok): Likewise.
13000 (maybe_mode_change): Likewise.
13001 (copyprop_hardreg_forward_1): Likewise.
13002 * reload.c (push_reload): Likewise.
13003 (find_reloads): Likewise.
13004 (find_reloads_subreg_address): Likewise.
13005 * reload1.c (alter_reg): Likewise.
13006 (eliminate_regs_1): Likewise.
13007 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13009 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13011 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13012 back to if statements, including unpack.
13014 2017-08-30 Martin Liska <mliska@suse.cz>
13016 PR inline-asm/82001
13017 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13019 (func_checker::compare_asm_inputs_outputs): ... this function.
13020 (func_checker::compare_gimple_asm): Use the function to compare
13021 also ASM constrains.
13022 * ipa-icf-gimple.h: Rename the function.
13024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13025 Alan Hayward <alan.hayward@arm.com>
13026 David Sherwood <david.sherwood@arm.com>
13028 * coretypes.h (complex_mode): New type.
13029 * gdbhooks.py (build_pretty_printer): Handle it.
13030 * machmode.h (complex_mode): New class.
13031 (complex_mode::includes_p): New function.
13032 (is_complex_int_mode): Likewise.
13033 (is_complex_float_mode): Likewise.
13034 * genmodes.c (get_mode_class): Handle complex mode classes.
13035 * function.c (expand_function_end): Use is_complex_int_mode.
13037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13038 Alan Hayward <alan.hayward@arm.com>
13039 David Sherwood <david.sherwood@arm.com>
13041 * coretypes.h (scalar_mode_pod): New typedef.
13042 * gdbhooks.py (build_pretty_printer): Handle it.
13043 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13044 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13045 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13046 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13047 as_a <scalar_mode>.
13049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13050 Alan Hayward <alan.hayward@arm.com>
13051 David Sherwood <david.sherwood@arm.com>
13053 * machmode.h (mode_for_vector): Take a scalar_mode instead
13055 * stor-layout.c (mode_for_vector): Likewise.
13056 * explow.c (promote_mode): Use as_a <scalar_mode>.
13057 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13060 Alan Hayward <alan.hayward@arm.com>
13061 David Sherwood <david.sherwood@arm.com>
13063 * target.def (preferred_simd_mode): Take a scalar_mode
13064 instead of a machine_mode.
13065 * targhooks.h (default_preferred_simd_mode): Likewise.
13066 * targhooks.c (default_preferred_simd_mode): Likewise.
13067 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13068 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13069 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13070 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13071 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13072 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13073 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13074 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13076 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13077 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13078 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13079 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13080 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13081 * doc/tm.texi: Regenerate.
13082 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13086 Alan Hayward <alan.hayward@arm.com>
13087 David Sherwood <david.sherwood@arm.com>
13089 * target.def (scalar_mode_supported_p): Take a scalar_mode
13090 instead of a machine_mode.
13091 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13092 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13093 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13094 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13095 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13096 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13097 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13098 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13099 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13100 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13101 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13102 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13103 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13104 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13106 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13107 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13108 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13109 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13110 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13112 * doc/tm.texi: Regenerate.
13114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13115 Alan Hayward <alan.hayward@arm.com>
13116 David Sherwood <david.sherwood@arm.com>
13118 * coretypes.h (opt_scalar_mode): New typedef.
13119 * gdbhooks.py (build_pretty_printers): Handle it.
13120 * machmode.h (mode_iterator::get_2xwider): Add overload for
13122 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13124 * expr.c (convert_mode_scalar): Likewise.
13125 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13126 * optabs.c (expand_float): Likewise.
13127 (expand_fix): Likewise.
13128 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13131 Alan Hayward <alan.hayward@arm.com>
13132 David Sherwood <david.sherwood@arm.com>
13134 * optabs.c (expand_float): Explicitly check for scalars before
13135 using a branching expansion.
13136 (expand_fix): Likewise.
13138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13139 Alan Hayward <alan.hayward@arm.com>
13140 David Sherwood <david.sherwood@arm.com>
13142 * expr.c (convert_mode): Split scalar handling out into...
13143 (convert_mode_scalar): ...this new function. Treat the modes
13146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13147 Alan Hayward <alan.hayward@arm.com>
13148 David Sherwood <david.sherwood@arm.com>
13150 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13152 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13155 Alan Hayward <alan.hayward@arm.com>
13156 David Sherwood <david.sherwood@arm.com>
13158 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13159 rather than a machine_mode.
13160 (fixed_from_string): Likewise.
13161 (fixed_convert): Likewise.
13162 (fixed_convert_from_int): Likewise.
13163 (fixed_convert_from_real): Likewise.
13164 (real_convert_from_fixed): Likewise.
13165 * fixed-value.c (fixed_from_double_int): Likewise.
13166 (fixed_from_string): Likewise.
13167 (fixed_convert): Likewise.
13168 (fixed_convert_from_int): Likewise.
13169 (fixed_convert_from_real): Likewise.
13170 (real_convert_from_fixed): Likewise.
13171 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13178 of separate mode class checks. Do not allow vector modes here.
13179 (immed_wide_int_const): Use as_a <scalar_mode>.
13180 * explow.c (trunc_int_for_mode): Likewise.
13181 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13182 (wi::shwi): Likewise.
13183 (wi::min_value): Likewise.
13184 (wi::max_value): Likewise.
13185 * dwarf2out.c (loc_descriptor): Likewise.
13186 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13187 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
13194 * expr.c (expand_expr_addr_expr_1): Use it.
13195 (expand_expr_real_2): Likewise.
13196 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13197 (fold_convert_const_fixed_from_int): Likewise.
13198 (fold_convert_const_fixed_from_real): Likewise.
13199 (native_encode_fixed): Likewise
13200 (native_encode_complex): Likewise
13201 (native_encode_vector): Likewise.
13202 (native_interpret_fixed): Likewise.
13203 (native_interpret_real): Likewise.
13204 (native_interpret_complex): Likewise.
13205 (native_interpret_vector): Likewise.
13206 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13207 (simd_clone_adjust_argument_types): Likewise.
13208 (simd_clone_init_simd_arrays): Likewise.
13209 (simd_clone_adjust): Likewise.
13210 * stor-layout.c (layout_type): Likewise.
13211 * tree.c (build_minus_one_cst): Likewise.
13212 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13213 * tree-inline.c (estimate_move_cost): Likewise.
13214 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13215 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13216 (vectorizable_reduction): Likewise.
13217 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13218 (vect_recog_mixed_size_cond_pattern): Likewise.
13219 (check_bool_pattern): Likewise.
13220 (adjust_bool_pattern): Likewise.
13221 (search_type_for_mask_1): Likewise.
13222 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13223 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13224 (vectorizable_load): Likewise.
13225 (vectorizable_store): Likewise.
13226 * ubsan.c (ubsan_encode_value): Likewise.
13227 * varasm.c (output_constant): Likewise.
13229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13230 Alan Hayward <alan.hayward@arm.com>
13231 David Sherwood <david.sherwood@arm.com>
13233 * coretypes.h (scalar_mode): New class.
13234 * machmode.h (scalar_mode): Likewise.
13235 (scalar_mode::includes_p): New function.
13236 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13237 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13238 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13239 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13240 * expmed.c (store_bit_field_1): Likewise.
13241 (extract_bit_field_1): Likewise.
13242 * expr.c (write_complex_part): Likewise.
13243 (read_complex_part): Likewise.
13244 (emit_move_complex_push): Likewise.
13245 (expand_expr_real_2): Likewise.
13246 * function.c (assign_parm_setup_reg): Likewise.
13247 (assign_parms_unsplit_complex): Likewise.
13248 * optabs.c (expand_binop): Likewise.
13249 * rtlanal.c (subreg_get_info): Likewise.
13250 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13251 * varasm.c (output_constant_pool_2): Likewise.
13253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13254 Alan Hayward <alan.hayward@arm.com>
13255 David Sherwood <david.sherwood@arm.com>
13257 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13259 (expmed_mult_highpart_optab): Likewise.
13260 (expmed_mult_highpart): Likewise.
13262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13263 Alan Hayward <alan.hayward@arm.com>
13264 David Sherwood <david.sherwood@arm.com>
13266 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13267 instead of a machine_mode.
13268 (builtin_memset_read_str): Likewise.
13269 * builtins.c (c_readstr): Likewise.
13270 (builtin_memcpy_read_str): Likewise.
13271 (builtin_strncpy_read_str): Likewise.
13272 (builtin_memset_read_str): Likewise.
13273 (builtin_memset_gen_str): Likewise.
13274 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13275 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13276 instead of a machine_mode.
13277 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13279 (make_extraction): Likewise.
13280 (try_widen_shift_mode): Take and return scalar_int_modes instead
13282 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13283 a scalar_int_mode instead of a machine_mode.
13284 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13285 (avr_addr_space_pointer_mode): Likewise.
13286 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13287 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13288 (msp430_unwind_word_mode): Likewise.
13289 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13290 (spu_addr_space_pointer_mode): Likewise.
13291 (spu_addr_space_address_mode): Likewise.
13292 (spu_libgcc_cmp_return_mode): Likewise.
13293 (spu_libgcc_shift_count_mode): Likewise.
13294 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13295 (rl78_addr_space_pointer_mode): Likewise.
13296 (fl78_unwind_word_mode): Likewise.
13297 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13299 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13300 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13301 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13302 (mips_valid_pointer_mode): Likewise.
13303 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13304 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13305 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13307 (ft32_addr_space_address_mode): Likewise.
13308 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13309 scalar_int_mode instead of a machine_mode.
13310 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13312 (m32c_addr_space_address_mode): Likewise.
13313 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13314 (rs6000_eh_return_filter_mode): Likewise.
13315 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13316 (rs6000_eh_return_filter_mode): Likewise.
13317 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13318 (s390_libgcc_shift_count_mode): Likewise.
13319 (s390_unwind_word_mode): Likewise.
13320 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13322 * target.def (mode_rep_extended): Likewise.
13323 (valid_pointer_mode): Likewise.
13324 (addr_space.valid_pointer_mode): Likewise.
13325 (eh_return_filter_mode): Return a scalar_int_mode rather than
13327 (libgcc_cmp_return_mode): Likewise.
13328 (libgcc_shift_count_mode): Likewise.
13329 (unwind_word_mode): Likewise.
13330 (addr_space.pointer_mode): Likewise.
13331 (addr_space.address_mode): Likewise.
13332 * doc/tm.texi: Regenerate.
13333 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13335 (do_jump): Use scalar_int_mode for local variables.
13336 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13337 rather than a machine_mode.
13338 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13339 (scompare_loc_descriptor_wide): Likewise.
13340 (scompare_loc_descriptor_narrow): Likewise.
13341 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13343 * except.c (sjlj_emit_dispatch_table): Likewise.
13344 (expand_builtin_eh_copy_values): Likewise.
13345 * explow.c (convert_memory_address_addr_space_1): Likewise.
13346 Take a scalar_int_mode rather than a machine_mode.
13347 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13348 than a machine_mode.
13349 (memory_address_addr_space): Use scalar_int_mode for local variables.
13350 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13351 rather than a machine_mode.
13352 * expmed.c (mask_rtx): Likewise.
13353 (init_expmed_one_conv): Likewise.
13354 (expand_mult_highpart_adjust): Likewise.
13355 (extract_high_half): Likewise.
13356 (expmed_mult_highpart_optab): Likewise.
13357 (expmed_mult_highpart): Likewise.
13358 (expand_smod_pow2): Likewise.
13359 (expand_sdiv_pow2): Likewise.
13360 (emit_store_flag_int): Likewise.
13361 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13363 (extract_low_bits): Likewise.
13364 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13366 * expr.c (pieces_addr::adjust): Likewise.
13367 (can_store_by_pieces): Likewise.
13368 (store_by_pieces): Likewise.
13369 (clear_by_pieces_1): Likewise.
13370 (expand_expr_addr_expr_1): Likewise.
13371 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13372 (expand_expr_real_1): Likewise.
13373 (try_casesi): Likewise.
13374 * final.c (shorten_branches): Likewise.
13375 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13376 type of "mode" to machine_mode.
13377 * internal-fn.c (expand_arith_overflow_result_store): Take a
13378 scalar_int_mode rather than a machine_mode.
13379 (expand_mul_overflow): Use scalar_int_mode for local variables.
13380 * loop-doloop.c (doloop_modify): Likewise.
13381 (doloop_optimize): Likewise.
13382 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13383 than a machine_mode.
13384 (expand_doubleword_shift_condmove): Likewise.
13385 (expand_doubleword_shift): Likewise.
13386 (expand_doubleword_clz): Likewise.
13387 (expand_doubleword_popcount): Likewise.
13388 (expand_doubleword_parity): Likewise.
13389 (expand_absneg_bit): Use scalar_int_mode for local variables.
13390 (prepare_float_lib_cmp): Likewise.
13391 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13392 rather than a machine_mode.
13393 (convert_memory_address_addr_space): Likewise.
13394 (get_mode_bounds): Likewise.
13395 (get_address_mode): Return a scalar_int_mode rather than a
13397 * rtlanal.c (get_address_mode): Likewise.
13398 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13399 than a machine_mode.
13400 * targhooks.c (default_mode_rep_extended): Likewise.
13401 (default_valid_pointer_mode): Likewise.
13402 (default_addr_space_valid_pointer_mode): Likewise.
13403 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13404 than a machine_mode.
13405 (default_libgcc_cmp_return_mode): Likewise.
13406 (default_libgcc_shift_count_mode): Likewise.
13407 (default_unwind_word_mode): Likewise.
13408 (default_addr_space_pointer_mode): Likewise.
13409 (default_addr_space_address_mode): Likewise.
13410 * targhooks.h (default_eh_return_filter_mode): Likewise.
13411 (default_libgcc_cmp_return_mode): Likewise.
13412 (default_libgcc_shift_count_mode): Likewise.
13413 (default_unwind_word_mode): Likewise.
13414 (default_addr_space_pointer_mode): Likewise.
13415 (default_addr_space_address_mode): Likewise.
13416 (default_mode_rep_extended): Take a scalar_int_mode rather than
13418 (default_valid_pointer_mode): Likewise.
13419 (default_addr_space_valid_pointer_mode): Likewise.
13420 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13422 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13423 rather than a machine_mode.
13424 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13425 for local variables.
13426 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13427 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13428 than a machine_mode.
13430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13431 Alan Hayward <alan.hayward@arm.com>
13432 David Sherwood <david.sherwood@arm.com>
13434 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13435 the mode argument to scalar_int_mode.
13436 (do_jump_by_parts_zero_rtx): Likewise.
13437 (do_jump_by_parts_equality_rtx): Likewise.
13438 (do_jump_by_parts_greater): Take a mode argument.
13439 (do_jump_by_parts_equality): Likewise.
13440 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
13447 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13448 (jump_table_for_label): Likewise.
13449 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13450 instead of an rtx_insn *.
13451 (shorten_branches): Use dyn_cast instead of LABEL_P and
13452 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13453 rtx_jump_table_data::get_data_mode.
13454 (final_scan_insn): Likewise.
13456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13457 Alan Hayward <alan.hayward@arm.com>
13458 David Sherwood <david.sherwood@arm.com>
13460 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13461 trying to combine a full-register integer set with a subreg
13464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13465 Alan Hayward <alan.hayward@arm.com>
13466 David Sherwood <david.sherwood@arm.com>
13468 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13469 that is always either address_mode or pointer_mode.
13471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13472 Alan Hayward <alan.hayward@arm.com>
13473 David Sherwood <david.sherwood@arm.com>
13475 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13476 when the two are known to be equal.
13478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13479 Alan Hayward <alan.hayward@arm.com>
13480 David Sherwood <david.sherwood@arm.com>
13482 * simplify-rtx.c (simplify_const_unary_operation): Use
13483 is_a <scalar_int_mode> instead of checking for a nonzero
13484 precision. Forcibly convert op_mode to a scalar_int_mode
13485 in that case. More clearly differentiate the operand and
13486 result modes and use the former when deciding what the value
13487 of a count-bits operation should be. Use is_int_mode instead
13488 of checking for a MODE_INT. Remove redundant check for whether
13489 this mode has a zero precision.
13491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13492 Alan Hayward <alan.hayward@arm.com>
13493 David Sherwood <david.sherwood@arm.com>
13495 * optabs.c (widen_leading): Change the type of the mode argument
13496 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13497 (widen_bswap): Likewise.
13498 (expand_parity): Likewise.
13499 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13500 (expand_ffs): Likewise.
13501 (epand_unop): Check for scalar integer modes before calling the
13504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13505 Alan Hayward <alan.hayward@arm.com>
13506 David Sherwood <david.sherwood@arm.com>
13508 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13510 (expand_expr_real_1): Update call accordingly.
13512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13513 Alan Hayward <alan.hayward@arm.com>
13514 David Sherwood <david.sherwood@arm.com>
13516 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13517 value_mode arguments. Use scalar_int_mode internally.
13518 (store_bit_field_1): Rename the new integer mode from imode
13519 to op0_mode and use it instead of GET_MODE (op0). Update calls
13520 to store_split_bit_field, store_bit_field_using_insv and
13521 store_fixed_bit_field.
13522 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13523 Use scalar_int_mode internally. Use a bit count rather than a mode
13524 when calculating the largest bit size for get_best_mode.
13525 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13526 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13527 Remove assertion that OP0 has a scalar integer mode.
13528 (store_split_bit_field): Add op0_mode and value_mode arguments.
13529 Update calls to extract_fixed_bit_field.
13530 (extract_bit_field_using_extv): Add an op0_mode argument.
13531 Use scalar_int_mode internally.
13532 (extract_bit_field_1): Rename the new integer mode from imode to
13533 op0_mode and use it instead of GET_MODE (op0). Update calls to
13534 extract_split_bit_field, extract_bit_field_using_extv and
13535 extract_fixed_bit_field.
13536 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13537 to extract_split_bit_field and extract_fixed_bit_field_1.
13538 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13539 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13540 on the target mode.
13541 (extract_split_bit_field): Add an op0_mode argument. Update call
13542 to extract_fixed_bit_field.
13544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13545 Alan Hayward <alan.hayward@arm.com>
13546 David Sherwood <david.sherwood@arm.com>
13548 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13549 * explow.c (hard_function_value): Likewise.
13550 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13551 convert_to_mode call outside the loop.
13552 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13553 for the mode iterator. Require the mode specified by max_pieces
13555 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13557 (copy_blkmode_to_reg): Likewise.
13558 (set_storage_via_setmem): Likewise.
13559 * optabs.c (prepare_cmp_insn): Likewise.
13560 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13561 * stor-layout.c (finish_bitfield_representative): Likewise.
13563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13564 Alan Hayward <alan.hayward@arm.com>
13565 David Sherwood <david.sherwood@arm.com>
13567 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13568 * expr.c (convert_move): Use them.
13569 (convert_modes): Likewise.
13570 (store_expr_with_bounds): Likewise.
13572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13573 Alan Hayward <alan.hayward@arm.com>
13574 David Sherwood <david.sherwood@arm.com>
13576 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13577 parameter for the mode of "x". Remove the "known_x", "known_mode"
13578 and "known_ret" arguments. Change the type of the mode argument
13579 to scalar_int_mode.
13580 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13581 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13582 (reg_num_sign_bit_copies_for_combine): Likewise.
13583 * rtlanal.c (nonzero_bits1): Likewise.
13584 (num_sign_bit_copies1): Likewise.
13585 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13586 (reg_num_sign_bit_copies_general): Likewise.
13587 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13588 (reg_nonzero_bits_general): Likewise.
13590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13591 Alan Hayward <alan.hayward@arm.com>
13592 David Sherwood <david.sherwood@arm.com>
13594 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13595 than in subroutines. Return 1 for non-integer modes.
13596 (cached_num_sign_bit_copies): Change the type of the mode parameter
13597 to scalar_int_mode.
13598 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13599 classes. Handle CONST_INT_P first and then check whether X also
13600 has a scalar integer mode. Check the same thing for inner registers
13601 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13608 in subroutines. Return the mode mask for non-integer modes.
13609 (cached_nonzero_bits): Change the type of the mode parameter
13610 to scalar_int_mode.
13611 (nonzero_bits1): Likewise. Remove early exit for other mode
13612 classes. Handle CONST_INT_P first and then check whether X
13613 also has a scalar integer mode.
13615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13616 Alan Hayward <alan.hayward@arm.com>
13617 David Sherwood <david.sherwood@arm.com>
13619 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13620 Return a scalar_int_mode and assert that the size is greater than
13622 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13623 (op_by_pieces_d::op_by_pieces_d): Likewise.
13624 (op_by_pieces_d::run): Likewise.
13625 (can_store_by_pieces): Likewise.
13627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13628 Alan Hayward <alan.hayward@arm.com>
13629 David Sherwood <david.sherwood@arm.com>
13631 * combine.c (extract_left_shift): Add a mode argument and update
13633 (make_compound_operation_int): Change the type of the mode parameter
13634 to scalar_int_mode and update the call to extract_left_shift.
13636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13637 Alan Hayward <alan.hayward@arm.com>
13638 David Sherwood <david.sherwood@arm.com>
13640 * combine.c (simplify_and_const_int): Change the type of the mode
13641 parameter to scalar_int_mode.
13642 (simplify_and_const_int_1): Likewise. Update recursive call.
13644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13645 Alan Hayward <alan.hayward@arm.com>
13646 David Sherwood <david.sherwood@arm.com>
13648 * combine.c (simplify_compare_const): Check that the mode is a
13649 scalar_int_mode (rather than VOIDmode) before testing its
13651 (simplify_comparison): Move COMPARISON_P handling out of the
13652 loop and restrict the latter part of the loop to scalar_int_modes.
13653 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13654 and when considering SUBREG_REGs. Use is_int_mode instead of
13655 checking GET_MODE_CLASS against MODE_INT.
13657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13658 Alan Hayward <alan.hayward@arm.com>
13659 David Sherwood <david.sherwood@arm.com>
13661 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13662 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13663 shift_unit_mode and for modes involved in scalar shifts.
13665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13666 Alan Hayward <alan.hayward@arm.com>
13667 David Sherwood <david.sherwood@arm.com>
13669 * combine.c (force_int_to_mode): New function, split out from...
13670 (force_to_mode): ...here. Keep xmode up-to-date and use it
13671 instead of GET_MODE (x).
13673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13674 Alan Hayward <alan.hayward@arm.com>
13675 David Sherwood <david.sherwood@arm.com>
13677 * optabs-query.h (extraction_insn::struct_mode): Change type to
13678 opt_scalar_int_mode and update comment.
13679 (extraction_insn::field_mode): Change type to scalar_int_mode.
13680 (extraction_insn::pos_mode): Likewise.
13681 * combine.c (make_extraction): Update accordingly.
13682 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13683 (get_optab_extraction_insn): Likewise.
13684 * recog.c (simplify_while_replacing): Likewise.
13685 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13686 parameter to opt_scalar_int_mode.
13688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13689 Alan Hayward <alan.hayward@arm.com>
13690 David Sherwood <david.sherwood@arm.com>
13692 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13693 to a scalar_int_mode instead of a machine_mode.
13694 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13695 (get_best_mode): Return a boolean and use a pointer argument to store
13696 the selected mode. Replace the limit mode parameter with a bit limit.
13697 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
13698 for the values returned by bit_field_mode_iterator::next_mode.
13699 (store_bit_field): Update call to get_best_mode.
13700 (store_fixed_bit_field): Likewise.
13701 (extract_fixed_bit_field): Likewise.
13702 * expr.c (optimize_bitfield_assignment_op): Likewise.
13703 * fold-const.c (optimize_bit_field_compare): Likewise.
13704 (fold_truth_andor_1): Likewise.
13705 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
13706 Update for new type of m_mode.
13707 (get_best_mode): As above.
13709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13710 Alan Hayward <alan.hayward@arm.com>
13711 David Sherwood <david.sherwood@arm.com>
13713 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
13714 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
13715 (store_bit_field): Check is_a <scalar_int_mode> before calling
13716 strict_volatile_bitfield_p.
13717 (extract_bit_field): Likewise.
13719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13720 Alan Hayward <alan.hayward@arm.com>
13721 David Sherwood <david.sherwood@arm.com>
13723 * target.def (cstore_mode): Return a scalar_int_mode.
13724 * doc/tm.texi: Regenerate.
13725 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
13726 * targhooks.h (default_cstore_mode): Likewise.
13727 * targhooks.c (default_cstore_mode): Likewise, using a forced
13729 * expmed.c (emit_cstore): Expect the target of the cstore to be
13732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13733 Alan Hayward <alan.hayward@arm.com>
13734 David Sherwood <david.sherwood@arm.com>
13736 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
13738 (niter_desc): Likewise mode.
13739 (iv_analyze): Add a mode parameter.
13741 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
13742 and change its type to scalar_int_mode.
13743 * loop-iv.c: Update commentary at head of file.
13744 (iv_constant): Pass the mode paraeter before the rtx it describes
13745 and change its type to scalar_int_mode. Remove VOIDmode handling.
13746 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
13747 (iv_extend): Likewise.
13748 (shorten_into_mode): Likewise.
13749 (iv_add): Use scalar_int_mode.
13750 (iv_mult): Likewise.
13751 (iv_shift): Likewise.
13752 (canonicalize_iv_subregs): Likewise.
13753 (get_biv_step_1): Pass the outer_mode parameter before the rtx
13754 it describes and change its mode to scalar_int_mode. Also change
13755 the type of the returned inner_mode to scalar_int_mode.
13756 (get_biv_step): Likewise, turning outer_mode from a pointer
13757 into a direct parameter. Update call to get_biv_step_1.
13758 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
13759 iv_constant and get_biv_step.
13760 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
13761 and change its type to scalar_int_mode. Don't initialise iv->mode
13762 to VOIDmode and remove later checks for its still being VOIDmode.
13763 Update calls to iv_analyze_op and iv_analyze_expr. Check
13764 is_a <scalar_int_mode> when changing the mode under consideration.
13765 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
13766 Update call to iv_analyze_expr.
13767 (iv_analyze_op): Add a mode parameter. Reject subregs whose
13768 inner register is not also a scalar_int_mode. Update call to
13770 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
13771 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
13772 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
13773 separate mode class checks. Update calls to iv_analyze. Remove
13774 fix-up of VOIDmodes after iv_analyze_biv.
13775 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
13776 don't have a scalar_int_mode. Update call to biv_p.
13778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13779 Alan Hayward <alan.hayward@arm.com>
13780 David Sherwood <david.sherwood@arm.com>
13782 * cfgexpand.c (convert_debug_memory_address): Use
13783 as_a <scalar_int_mode>.
13784 * combine.c (expand_compound_operation): Likewise.
13785 (make_extraction): Likewise.
13786 (change_zero_ext): Likewise.
13787 (simplify_comparison): Likewise.
13788 * cse.c (cse_insn): Likewise.
13789 * dwarf2out.c (minmax_loc_descriptor): Likewise.
13790 (mem_loc_descriptor): Likewise.
13791 (loc_descriptor): Likewise.
13792 * expmed.c (init_expmed_one_mode): Likewise.
13793 (synth_mult): Likewise.
13794 (emit_store_flag_1): Likewise.
13795 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
13796 of a comparison with size.
13797 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
13798 (reduce_to_bit_field_precision): Likewise.
13799 * function.c (expand_function_end): Likewise.
13800 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
13801 * loop-doloop.c (doloop_modify): Likewise.
13802 * optabs.c (expand_binop): Likewise.
13803 (expand_unop): Likewise.
13804 (expand_copysign_absneg): Likewise.
13805 (prepare_cmp_insn): Likewise.
13806 (maybe_legitimize_operand): Likewise.
13807 * recog.c (const_scalar_int_operand): Likewise.
13808 * rtlanal.c (get_address_mode): Likewise.
13809 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13810 (simplify_cond_clz_ctz): Likewise.
13811 * tree-nested.c (get_nl_goto_field): Likewise.
13812 * tree.c (build_vector_type_for_mode): Likewise.
13813 * var-tracking.c (use_narrower_mode): Likewise.
13815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13816 Alan Hayward <alan.hayward@arm.com>
13817 David Sherwood <david.sherwood@arm.com>
13819 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
13820 * builtins.c (expand_builtin_signbit): Use it.
13821 * cfgexpand.c (expand_debug_expr): Likewise.
13822 * dojump.c (do_jump): Likewise.
13823 (do_compare_and_jump): Likewise.
13824 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
13825 * expmed.c (make_tree): Likewise.
13826 * expr.c (expand_expr_real_2): Likewise.
13827 (expand_expr_real_1): Likewise.
13828 (try_casesi): Likewise.
13829 * fold-const-call.c (fold_const_call_ss): Likewise.
13830 * fold-const.c (unextend): Likewise.
13831 (extract_muldiv_1): Likewise.
13832 (fold_single_bit_test): Likewise.
13833 (native_encode_int): Likewise.
13834 (native_encode_string): Likewise.
13835 (native_interpret_int): Likewise.
13836 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
13837 * internal-fn.c (expand_addsub_overflow): Likewise.
13838 (expand_neg_overflow): Likewise.
13839 (expand_mul_overflow): Likewise.
13840 (expand_arith_overflow): Likewise.
13841 * match.pd: Likewise.
13842 * stor-layout.c (layout_type): Likewise.
13843 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13844 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
13845 * tree-ssanames.c (get_range_info): Likewise.
13846 * tree-switch-conversion.c (array_value_type) Likewise.
13847 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
13848 (vect_recog_divmod_pattern): Likewise.
13849 (vect_recog_mixed_size_cond_pattern): Likewise.
13850 * tree-vrp.c (extract_range_basic): Likewise.
13851 (simplify_float_conversion_using_ranges): Likewise.
13852 * tree.c (int_fits_type_p): Likewise.
13853 * ubsan.c (instrument_bool_enum_load): Likewise.
13854 * varasm.c (mergeable_string_section): Likewise.
13855 (narrowing_initializer_constant_valid_p): Likewise.
13856 (output_constant): Likewise.
13858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13859 Alan Hayward <alan.hayward@arm.com>
13860 David Sherwood <david.sherwood@arm.com>
13862 * machmode.h (NARROWEST_INT_MODE): New macro.
13863 * expr.c (alignment_for_piecewise_move): Use it instead of
13864 GET_CLASS_NARROWEST_MODE (MODE_INT).
13865 (push_block): Likewise.
13866 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
13868 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13871 Alan Hayward <alan.hayward@arm.com>
13872 David Sherwood <david.sherwood@arm.com>
13874 * postreload.c (move2add_valid_value_p): Change the type of the
13875 mode parameter to scalar_int_mode.
13876 (move2add_use_add2_insn): Add a mode parameter and use it instead
13878 (move2add_use_add3_insn): Likewise.
13879 (reload_cse_move2add): Update accordingly.
13881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13882 Alan Hayward <alan.hayward@arm.com>
13883 David Sherwood <david.sherwood@arm.com>
13885 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13887 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13888 * optabs.c (expand_unop): Likewise.
13890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13891 Alan Hayward <alan.hayward@arm.com>
13892 David Sherwood <david.sherwood@arm.com>
13894 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13895 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13896 (popcount_loc_descriptor): Likewise.
13897 (bswap_loc_descriptor): Likewise.
13898 (rotate_loc_descriptor): Likewise.
13899 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13900 calling the functions above.
13902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13903 Alan Hayward <alan.hayward@arm.com>
13904 David Sherwood <david.sherwood@arm.com>
13906 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13908 (try_combine): Likewise.
13909 (simplify_if_then_else): Likewise.
13910 * cse.c (cse_insn): Likewise.
13911 * dwarf2out.c (mem_loc_descriptor): Likewise.
13912 * emit-rtl.c (gen_lowpart_common): Likewise.
13913 * simplify-rtx.c (simplify_truncation): Likewise.
13914 (simplify_binary_operation_1): Likewise.
13915 (simplify_const_relational_operation): Likewise.
13916 (simplify_ternary_operation): Likewise.
13917 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13920 Alan Hayward <alan.hayward@arm.com>
13921 David Sherwood <david.sherwood@arm.com>
13923 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13924 * reload.c (push_reload): Likewise.
13925 (find_reloads): Likewise.
13927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13928 Alan Hayward <alan.hayward@arm.com>
13929 David Sherwood <david.sherwood@arm.com>
13931 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13932 (make_compound_operation_int): Likewise.
13933 (change_zero_ext): Likewise.
13934 * expr.c (convert_move): Likewise.
13935 (convert_modes): Likewise.
13936 * fwprop.c (forward_propagate_subreg): Likewise.
13937 * loop-iv.c (get_biv_step_1): Likewise.
13938 * optabs.c (widen_operand): Likewise.
13939 * postreload.c (move2add_valid_value_p): Likewise.
13940 * recog.c (simplify_while_replacing): Likewise.
13941 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13942 (simplify_binary_operation_1): Likewise. Remove redundant
13943 mode equality check.
13945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13946 Alan Hayward <alan.hayward@arm.com>
13947 David Sherwood <david.sherwood@arm.com>
13949 * combine.c (combine_simplify_rtx): Add checks for
13950 is_a <scalar_int_mode>.
13951 (simplify_if_then_else): Likewise.
13952 (make_field_assignment): Likewise.
13953 (simplify_comparison): Likewise.
13954 * ifcvt.c (noce_try_bitop): Likewise.
13955 * loop-invariant.c (canonicalize_address_mult): Likewise.
13956 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13959 Alan Hayward <alan.hayward@arm.com>
13960 David Sherwood <david.sherwood@arm.com>
13962 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13963 is_a <scalar_int_mode> instead of != BLKmode.
13965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13966 Alan Hayward <alan.hayward@arm.com>
13967 David Sherwood <david.sherwood@arm.com>
13969 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13970 instead of != VOIDmode.
13971 * combine.c (if_then_else_cond): Likewise.
13972 (change_zero_ext): Likewise.
13973 * dwarf2out.c (mem_loc_descriptor): Likewise.
13974 (loc_descriptor): Likewise.
13975 * rtlanal.c (canonicalize_condition): Likewise.
13976 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13979 Alan Hayward <alan.hayward@arm.com>
13980 David Sherwood <david.sherwood@arm.com>
13982 * simplify-rtx.c (simplify_binary_operation_1): Use
13983 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
13990 (int_traits<unsigned short>) Likewise.
13991 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13992 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13994 * combine.c (set_nonzero_bits_and_sign_copies): Use
13995 is_a <scalar_int_mode>.
13996 (find_split_point): Likewise.
13997 (combine_simplify_rtx): Likewise.
13998 (simplify_logical): Likewise.
13999 (expand_compound_operation): Likewise.
14000 (expand_field_assignment): Likewise.
14001 (make_compound_operation): Likewise.
14002 (extended_count): Likewise.
14003 (change_zero_ext): Likewise.
14004 (simplify_comparison): Likewise.
14005 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14006 (ucompare_loc_descriptor): Likewise.
14007 (minmax_loc_descriptor): Likewise.
14008 (mem_loc_descriptor): Likewise.
14009 (loc_descriptor): Likewise.
14010 * expmed.c (init_expmed_one_mode): Likewise.
14011 * lra-constraints.c (lra_constraint_offset): Likewise.
14012 * optabs.c (prepare_libcall_arg): Likewise.
14013 * postreload.c (move2add_note_store): Likewise.
14014 * reload.c (operands_match_p): Likewise.
14015 * rtl.h (load_extend_op): Likewise.
14016 * rtlhooks.c (gen_lowpart_general): Likewise.
14017 * simplify-rtx.c (simplify_truncation): Likewise.
14018 (simplify_unary_operation_1): Likewise.
14019 (simplify_binary_operation_1): Likewise.
14020 (simplify_const_binary_operation): Likewise.
14021 (simplify_const_relational_operation): Likewise.
14022 (simplify_subreg): Likewise.
14023 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14024 * var-tracking.c (adjust_mems): Likewise.
14025 (prepare_call_arguments): Likewise.
14027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14028 Alan Hayward <alan.hayward@arm.com>
14029 David Sherwood <david.sherwood@arm.com>
14031 * machmode.h (is_int_mode): New fuction.
14032 * combine.c (find_split_point): Use it.
14033 (combine_simplify_rtx): Likewise.
14034 (simplify_if_then_else): Likewise.
14035 (simplify_set): Likewise.
14036 (simplify_shift_const_1): Likewise.
14037 (simplify_comparison): Likewise.
14038 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14039 * cse.c (notreg_cost): Likewise.
14040 (cse_insn): Likewise.
14041 * cselib.c (cselib_lookup_1): Likewise.
14042 * dojump.c (do_jump_1): Likewise.
14043 (do_compare_rtx_and_jump): Likewise.
14044 * dse.c (get_call_args): Likewise.
14045 * dwarf2out.c (rtl_for_decl_init): Likewise.
14046 (native_encode_initializer): Likewise.
14047 * expmed.c (emit_store_flag_1): Likewise.
14048 (emit_store_flag): Likewise.
14049 * expr.c (convert_modes): Likewise.
14050 (store_field): Likewise.
14051 (expand_expr_real_1): Likewise.
14052 * fold-const.c (fold_read_from_constant_string): Likewise.
14053 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14054 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14055 * optabs.c (expand_binop): Likewise.
14056 (expand_unop): Likewise.
14057 (expand_abs_nojump): Likewise.
14058 (expand_one_cmpl_abs_nojump): Likewise.
14059 * simplify-rtx.c (mode_signbit_p): Likewise.
14060 (val_signbit_p): Likewise.
14061 (val_signbit_known_set_p): Likewise.
14062 (val_signbit_known_clear_p): Likewise.
14063 (simplify_relational_operation_1): Likewise.
14064 * tree.c (vector_type_mode): Likewise.
14066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14067 Alan Hayward <alan.hayward@arm.com>
14068 David Sherwood <david.sherwood@arm.com>
14070 * machmode.h (smallest_mode_for_size): Fix formatting.
14071 (smallest_int_mode_for_size): New function.
14072 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14073 instead of smallest_mode_for_size.
14074 * combine.c (make_extraction): Likewise.
14075 * config/arc/arc.c (arc_expand_movmem): Likewise.
14076 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14077 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14078 * config/s390/s390.c (s390_expand_insv): Likewise.
14079 * config/sparc/sparc.c (assign_int_registers): Likewise.
14080 * config/spu/spu.c (spu_function_value): Likewise.
14081 (spu_function_arg): Likewise.
14082 * coverage.c (get_gcov_type): Likewise.
14083 (get_gcov_unsigned_t): Likewise.
14084 * dse.c (find_shift_sequence): Likewise.
14085 * expmed.c (store_bit_field_1): Likewise.
14086 * expr.c (convert_move): Likewise.
14087 (store_field): Likewise.
14088 * internal-fn.c (expand_arith_overflow): Likewise.
14089 * optabs-query.c (get_best_extraction_insn): Likewise.
14090 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14091 * stor-layout.c (layout_type): Likewise.
14092 (initialize_sizetypes): Likewise.
14093 * targhooks.c (default_get_mask_mode): Likewise.
14094 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14097 Alan Hayward <alan.hayward@arm.com>
14098 David Sherwood <david.sherwood@arm.com>
14100 * machmode.h (opt_mode::else_blk): New function.
14101 (int_mode_for_mode): Declare.
14102 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14103 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14105 * cfgexpand.c (expand_debug_expr): Likewise.
14106 * combine.c (gen_lowpart_or_truncate): Likewise.
14107 (gen_lowpart_for_combine): Likewise.
14108 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14109 * config/avr/avr.c (avr_to_int_mode): Likewise.
14110 (avr_out_plus_1): Likewise.
14111 (avr_out_plus): Likewise.
14112 (avr_out_round): Likewise.
14113 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14114 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14115 (s390_expand_vcond): Likewise.
14116 * config/spu/spu.c (spu_split_immediate): Likewise.
14117 (spu_expand_mov): Likewise.
14118 * dse.c (get_stored_val): Likewise.
14119 * expmed.c (store_bit_field_1): Likewise.
14120 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14122 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14123 (extract_low_bits): Likewise.
14124 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14125 handling rather than repeating the check.
14126 (emit_group_store): Likewise.
14127 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14128 * optabs.c (expand_absneg_bit): Likewise.
14129 (expand_copysign_absneg): Likewise.
14130 (expand_copysign_bit): Likewise.
14131 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14132 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14133 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14134 * var-tracking.c (prepare_call_arguments): Likewise.
14135 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14136 int_mode_for_mode instead of mode_for_size.
14137 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14140 Alan Hayward <alan.hayward@arm.com>
14141 David Sherwood <david.sherwood@arm.com>
14143 * machmode.h (int_mode_for_size): New function.
14144 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14145 instead of mode_for_size.
14146 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14148 * combine.c (expand_field_assignment): Use int_mode_for_size
14149 instead of mode_for_size.
14150 (make_extraction): Likewise.
14151 (simplify_shift_const_1): Likewise.
14152 (simplify_comparison): Likewise.
14153 * dojump.c (do_jump): Likewise.
14154 * dwarf2out.c (mem_loc_descriptor): Likewise.
14155 * emit-rtl.c (init_derived_machine_modes): Likewise.
14156 * expmed.c (flip_storage_order): Likewise.
14157 (convert_extracted_bit_field): Likewise.
14158 * expr.c (copy_blkmode_from_reg): Likewise.
14159 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14160 * internal-fn.c (expand_mul_overflow): Likewise.
14161 * lower-subreg.c (simple_move): Likewise.
14162 * optabs-libfuncs.c (init_optabs): Likewise.
14163 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14164 * tree.c (vector_type_mode): Likewise.
14165 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14166 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14167 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14168 * tree-vect-stmts.c (vectorizable_load): Likewise.
14169 (vectorizable_store): Likewise.
14171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14172 Alan Hayward <alan.hayward@arm.com>
14173 David Sherwood <david.sherwood@arm.com>
14175 * coretypes.h (pod_mode): New type.
14176 (scalar_int_mode_pod): New typedef.
14177 * machmode.h (pod_mode): New class.
14178 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14179 * genmodes.c (emit_mode_int_n): Update accordingly.
14180 * lower-subreg.h (target_lower_subreg): Change type to
14181 scalar_int_mode_pod.
14182 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14183 scalar_int_mode_pod.
14185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14186 Alan Hayward <alan.hayward@arm.com>
14187 David Sherwood <david.sherwood@arm.com>
14189 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14190 machine_mode to scalar_int_mode.
14191 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14192 (rs6000_option_override_internal): Remove cast to int.
14193 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14194 machine_mode to scalar_int_mode.
14195 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14196 (rs6000_option_override_internal): Remove cast to int.
14197 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14198 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14200 * config/s390/s390.c (s390_expand_builtin): Likewise.
14201 * coretypes.h (scalar_int_mode): New type.
14202 (opt_scalar_int_mode): New typedef.
14203 * machmode.h (scalar_int_mode): New class.
14204 (scalar_int_mode::includes_p): New function.
14205 (byte_mode): Change type to scalar_int_mode.
14206 (word_mode): Likewise.
14207 (ptr_mode): Likewise.
14208 * emit-rtl.c (byte_mode): Likewise.
14209 (word_mode): Likewise.
14210 (ptr_mode): Likewise.
14211 (init_derived_machine_modes): Update accordingly.
14212 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14213 and MODE_PARTIAL_INT.
14214 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14215 opt_scalar_int_mode.
14217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14218 Alan Hayward <alan.hayward@arm.com>
14219 David Sherwood <david.sherwood@arm.com>
14221 * target.def (libgcc_floating_mode_supported_p): Take a
14223 * doc/tm.texi: Regenerate.
14224 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14226 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14227 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14235 * doc/tm.texi: Regenerate.
14236 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14237 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14238 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14239 * targhooks.h (default_floatn_mode): Likewise.
14240 * targhooks.c (default_floatn_mode): Likewise.
14241 * tree.c (build_common_tree_nodes): Update accordingly.
14243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14244 Alan Hayward <alan.hayward@arm.com>
14245 David Sherwood <david.sherwood@arm.com>
14247 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14248 (mode_iterator::iterate_p): Likewise.
14249 (mode_iterator::get_wider): Likewise.
14250 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14253 Alan Hayward <alan.hayward@arm.com>
14254 David Sherwood <david.sherwood@arm.com>
14256 * coretypes.h (opt_scalar_float_mode): New typedef.
14257 * machmode.h (float_mode_for_size): New function.
14258 * emit-rtl.c (double_mode): Delete.
14259 (init_emit_once): Use float_mode_for_size.
14260 * stor-layout.c (layout_type): Likewise.
14261 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14264 Alan Hayward <alan.hayward@arm.com>
14265 David Sherwood <david.sherwood@arm.com>
14267 * output.h (assemble_real): Take a scalar_float_mode.
14268 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14269 * config/arm/arm.md (consttable_4): Likewise.
14270 (consttable_8): Likewise.
14271 (consttable_16): Likewise.
14272 * config/mips/mips.md (consttable_float): Likewise.
14273 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14274 * varasm.c (assemble_real): Take a scalar_float_mode.
14275 (output_constant_pool_2): Update accordingly.
14276 (output_constant): Likewise.
14278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14279 Alan Hayward <alan.hayward@arm.com>
14280 David Sherwood <david.sherwood@arm.com>
14282 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14283 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14284 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14285 (native_encode_real): Likewise.
14286 (native_interpret_real): Likewise.
14287 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14288 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14291 Alan Hayward <alan.hayward@arm.com>
14292 David Sherwood <david.sherwood@arm.com>
14294 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14295 <scalar_float_mode>. Simplify.
14296 (gen_extend_conv_libfunc): Likewise.
14298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14299 Alan Hayward <alan.hayward@arm.com>
14300 David Sherwood <david.sherwood@arm.com>
14302 * coretypes.h (scalar_float_mode): New type.
14303 * machmode.h (mode_traits::from_int): Use machine_mode if
14304 USE_ENUM_MODES is defined.
14305 (is_a): New function.
14307 (dyn_cast): Likewise.
14308 (scalar_float_mode): New class.
14309 (scalar_float_mode::includes_p): New function.
14310 (is_float_mode): Likewise.
14311 * gdbhooks.py (MachineModePrinter): New class.
14312 (build_pretty_printer): Use it for scalar_float_mode.
14313 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14314 (format_helper::format_helper): Turn into a template.
14315 * genmodes.c (get_mode_class): New function.
14316 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14317 or machine_mode if none.
14318 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14319 as_a <scalar_float_mode>.
14320 * dwarf2out.c (mem_loc_descriptor): Likewise.
14321 (insert_float): Likewise.
14322 (add_const_value_attribute): Likewise.
14323 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14324 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14325 (expand_unop): Update accordingly.
14326 (expand_abs_nojump): Likewise.
14327 (expand_copysign_absneg): Take a scalar_float_mode.
14328 (expand_copysign_bit): Likewise.
14329 (expand_copysign): Update accordingly.
14331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14332 Alan Hayward <alan.hayward@arm.com>
14333 David Sherwood <david.sherwood@arm.com>
14335 * coretypes.h (opt_mode): New class.
14336 * machmode.h (opt_mode): Likewise.
14337 (opt_mode::else_void): New function.
14338 (opt_mode::require): Likewise.
14339 (opt_mode::exists): Likewise.
14340 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14341 (GET_MODE_2XWIDER_MODE): Likewise.
14342 (mode_iterator::get_wider): Update accordingly.
14343 (mode_iterator::get_2xwider): Likewise.
14344 (mode_iterator::get_known_wider): Likewise, turning into a template.
14345 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14346 forcing a wider mode to exist.
14347 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14348 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14349 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14350 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14351 * lower-subreg.c (init_lower_subreg): Likewise.
14352 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14353 on the final iteration.
14354 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14355 a wider mode exists before asking for a move pattern.
14356 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14357 forcing a wider mode to exist.
14358 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14359 returning false if no such mode exists.
14360 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14361 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14362 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14363 Avoid checking for a MODE_INT if we already know the mode is not a
14365 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14366 forcing a wider mode to exist.
14367 (expmed_mult_highpart_optab): Likewise.
14368 (expmed_mult_highpart): Likewise.
14369 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14371 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14372 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14373 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14374 * internal-fn.c (expand_mul_overflow): Update use of
14375 GET_MODE_2XWIDER_MODE.
14376 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14377 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14378 GET_MODE_WIDER_MODE.
14379 (convert_plusminus_to_widen): Likewise.
14380 * tree-switch-conversion.c (array_value_type): Likewise.
14381 * var-tracking.c (emit_note_insn_var_location): Likewise.
14382 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14383 Return false inside rather than outside the loop if no wider mode
14385 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14386 and GET_MODE_2XWIDER_MODE
14387 (can_compare_p): Use else_void.
14388 * gdbhooks.py (OptMachineModePrinter): New class.
14389 (build_pretty_printer): Use it for opt_mode.
14391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14392 Alan Hayward <alan.hayward@arm.com>
14393 David Sherwood <david.sherwood@arm.com>
14395 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14396 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14399 Alan Hayward <alan.hayward@arm.com>
14400 David Sherwood <david.sherwood@arm.com>
14402 * machmode.h (mode_traits): New structure.
14403 (get_narrowest_mode): New function.
14404 (mode_iterator::start): Likewise.
14405 (mode_iterator::iterate_p): Likewise.
14406 (mode_iterator::get_wider): Likewise.
14407 (mode_iterator::get_known_wider): Likewise.
14408 (mode_iterator::get_2xwider): Likewise.
14409 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14410 (FOR_EACH_MODE): Likewise.
14411 (FOR_EACH_MODE_FROM): Likewise.
14412 (FOR_EACH_MODE_UNTIL): Likewise.
14413 (FOR_EACH_WIDER_MODE): Likewise.
14414 (FOR_EACH_2XWIDER_MODE): Likewise.
14415 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14416 * combine.c (simplify_comparison): Likewise
14417 * config/i386/i386.c (type_natural_mode): Likewise.
14418 * cse.c (cse_insn): Likewise.
14419 * dse.c (find_shift_sequence): Likewise.
14420 * emit-rtl.c (init_derived_machine_modes): Likewise.
14421 (init_emit_once): Likewise.
14422 * explow.c (hard_function_value): Likewise.
14423 * expmed.c (extract_fixed_bit_field_1): Likewise.
14424 (extract_bit_field_1): Likewise.
14425 (expand_divmod): Likewise.
14426 (emit_store_flag_1): Likewise.
14427 * expr.c (init_expr_target): Likewise.
14428 (convert_move): Likewise.
14429 (alignment_for_piecewise_move): Likewise.
14430 (widest_int_mode_for_size): Likewise.
14431 (emit_block_move_via_movmem): Likewise.
14432 (copy_blkmode_to_reg): Likewise.
14433 (set_storage_via_setmem): Likewise.
14434 (compress_float_constant): Likewise.
14435 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14436 * optabs-query.c (get_best_extraction_insn): Likewise.
14437 * optabs.c (expand_binop): Likewise.
14438 (expand_twoval_unop): Likewise.
14439 (expand_twoval_binop): Likewise.
14440 (widen_leading): Likewise.
14441 (widen_bswap): Likewise.
14442 (expand_parity): Likewise.
14443 (expand_unop): Likewise.
14444 (prepare_cmp_insn): Likewise.
14445 (prepare_float_lib_cmp): Likewise.
14446 (expand_float): Likewise.
14447 (expand_fix): Likewise.
14448 (expand_sfix_optab): Likewise.
14449 * postreload.c (move2add_use_add2_insn): Likewise.
14450 * reg-stack.c (reg_to_stack): Likewise.
14451 * reginfo.c (choose_hard_reg_mode): Likewise.
14452 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14453 * stor-layout.c (mode_for_size): Likewise.
14454 (smallest_mode_for_size): Likewise.
14455 (mode_for_vector): Likewise.
14456 (finish_bitfield_representative): Likewise.
14457 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14458 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14459 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14460 * var-tracking.c (prepare_call_arguments): Likewise.
14462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14463 Alan Hayward <alan.hayward@arm.com>
14464 David Sherwood <david.sherwood@arm.com>
14466 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14467 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14468 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14469 * machmode.h (mode_size): Move earlier in file.
14470 (mode_precision): Likewise.
14471 (mode_inner): Likewise.
14472 (mode_nunits): Likewise.
14473 (mode_unit_size): Likewise.
14474 (unit_unit_precision): Likewise.
14475 (mode_wider): Likewise.
14476 (mode_2xwider): Likewise.
14477 (machine_mode): New class.
14478 (mode_to_bytes): New function.
14479 (mode_to_bits): Likewise.
14480 (mode_to_precision): Likewise.
14481 (mode_to_inner): Likewise.
14482 (mode_to_unit_size): Likewise.
14483 (mode_to_unit_precision): Likewise.
14484 (mode_to_nunits): Likewise.
14485 (GET_MODE_SIZE): Use mode_to_bytes.
14486 (GET_MODE_BITSIZE): Use mode_to_bits.
14487 (GET_MODE_PRECISION): Use mode_to_precision.
14488 (GET_MODE_INNER): Use mode_to_inner.
14489 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14490 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14491 (GET_MODE_NUNITS): Use mode_to_nunits.
14492 * system.h (ALWAYS_INLINE): New macro.
14493 * config/powerpcspe/powerpcspe-c.c
14494 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14495 int for arg1_mode and arg2_mode.
14497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14498 Alan Hayward <alan.hayward@arm.com>
14499 David Sherwood <david.sherwood@arm.com>
14501 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14502 Prefix mode names with E_ in case statements.
14503 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14504 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14505 (aarch64_split_simd_move): Likewise.
14506 (aarch64_gen_storewb_pair): Likewise.
14507 (aarch64_gen_loadwb_pair): Likewise.
14508 (aarch64_gen_store_pair): Likewise.
14509 (aarch64_gen_load_pair): Likewise.
14510 (aarch64_get_condition_code_1): Likewise.
14511 (aarch64_constant_pool_reload_icode): Likewise.
14512 (get_rsqrte_type): Likewise.
14513 (get_rsqrts_type): Likewise.
14514 (get_recpe_type): Likewise.
14515 (get_recps_type): Likewise.
14516 (aarch64_gimplify_va_arg_expr): Likewise.
14517 (aarch64_simd_container_mode): Likewise.
14518 (aarch64_emit_load_exclusive): Likewise.
14519 (aarch64_emit_store_exclusive): Likewise.
14520 (aarch64_expand_compare_and_swap): Likewise.
14521 (aarch64_gen_atomic_cas): Likewise.
14522 (aarch64_emit_bic): Likewise.
14523 (aarch64_emit_atomic_swap): Likewise.
14524 (aarch64_emit_atomic_load_op): Likewise.
14525 (aarch64_evpc_trn): Likewise.
14526 (aarch64_evpc_uzp): Likewise.
14527 (aarch64_evpc_zip): Likewise.
14528 (aarch64_evpc_ext): Likewise.
14529 (aarch64_evpc_rev): Likewise.
14530 (aarch64_evpc_dup): Likewise.
14531 (aarch64_gen_ccmp_first): Likewise.
14532 (aarch64_gen_ccmp_next): Likewise.
14533 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14534 (alpha_emit_xfloating_libcall): Likewise.
14535 (emit_insxl): Likewise.
14536 (alpha_arg_type): Likewise.
14537 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14538 (arc_preferred_simd_mode): Likewise.
14539 (arc_secondary_reload): Likewise.
14540 (get_arc_condition_code): Likewise.
14541 (arc_print_operand): Likewise.
14542 (arc_legitimate_constant_p): Likewise.
14543 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14544 * config/arc/arc.md (casesi_load): Likewise.
14545 (casesi_compact_jump): Likewise.
14546 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14547 (cc_use_register): Likewise.
14548 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14549 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14550 (arm_init_iwmmxt_builtins): Likewise.
14551 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14552 (neon_expand_vector_init): Likewise.
14553 (arm_attr_length_move_neon): Likewise.
14554 (maybe_get_arm_condition_code): Likewise.
14555 (arm_emit_vector_const): Likewise.
14556 (arm_preferred_simd_mode): Likewise.
14557 (arm_output_iwmmxt_tinsr): Likewise.
14558 (thumb1_output_casesi): Likewise.
14559 (thumb2_output_casesi): Likewise.
14560 (arm_emit_load_exclusive): Likewise.
14561 (arm_emit_store_exclusive): Likewise.
14562 (arm_expand_compare_and_swap): Likewise.
14563 (arm_evpc_neon_vuzp): Likewise.
14564 (arm_evpc_neon_vzip): Likewise.
14565 (arm_evpc_neon_vrev): Likewise.
14566 (arm_evpc_neon_vtrn): Likewise.
14567 (arm_evpc_neon_vext): Likewise.
14568 (arm_validize_comparison): Likewise.
14569 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14570 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14571 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14572 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14573 (c6x_preferred_simd_mode): Likewise.
14574 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14575 (epiphany_rtx_costs): Likewise.
14576 * config/epiphany/predicates.md (proper_comparison_operator):
14578 * config/frv/frv.c (condexec_memory_operand): Likewise.
14579 (frv_emit_move): Likewise.
14580 (output_move_single): Likewise.
14581 (output_condmove_single): Likewise.
14582 (frv_hard_regno_mode_ok): Likewise.
14583 (frv_matching_accg_mode): Likewise.
14584 * config/h8300/h8300.c (split_adds_subs): Likewise.
14585 (h8300_rtx_costs): Likewise.
14586 (h8300_print_operand): Likewise.
14587 (compute_mov_length): Likewise.
14588 (output_logical_op): Likewise.
14589 (compute_logical_op_length): Likewise.
14590 (compute_logical_op_cc): Likewise.
14591 (h8300_shift_needs_scratch_p): Likewise.
14592 (output_a_shift): Likewise.
14593 (compute_a_shift_length): Likewise.
14594 (compute_a_shift_cc): Likewise.
14595 (expand_a_rotate): Likewise.
14596 (output_a_rotate): Likewise.
14597 * config/i386/i386.c (classify_argument): Likewise.
14598 (function_arg_advance_32): Likewise.
14599 (function_arg_32): Likewise.
14600 (function_arg_64): Likewise.
14601 (function_value_64): Likewise.
14602 (ix86_gimplify_va_arg): Likewise.
14603 (ix86_legitimate_constant_p): Likewise.
14604 (put_condition_code): Likewise.
14605 (split_double_mode): Likewise.
14606 (ix86_avx256_split_vector_move_misalign): Likewise.
14607 (ix86_expand_vector_logical_operator): Likewise.
14608 (ix86_split_idivmod): Likewise.
14609 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14610 (ix86_build_const_vector): Likewise.
14611 (ix86_build_signbit_mask): Likewise.
14612 (ix86_match_ccmode): Likewise.
14613 (ix86_cc_modes_compatible): Likewise.
14614 (ix86_expand_branch): Likewise.
14615 (ix86_expand_sse_cmp): Likewise.
14616 (ix86_expand_sse_movcc): Likewise.
14617 (ix86_expand_int_sse_cmp): Likewise.
14618 (ix86_expand_vec_perm_vpermi2): Likewise.
14619 (ix86_expand_vec_perm): Likewise.
14620 (ix86_expand_sse_unpack): Likewise.
14621 (ix86_expand_int_addcc): Likewise.
14622 (ix86_split_to_parts): Likewise.
14623 (ix86_vectorize_builtin_gather): Likewise.
14624 (ix86_vectorize_builtin_scatter): Likewise.
14625 (avx_vpermilp_parallel): Likewise.
14626 (inline_memory_move_cost): Likewise.
14627 (ix86_tieable_integer_mode_p): Likewise.
14628 (x86_maybe_negate_const_int): Likewise.
14629 (ix86_expand_vector_init_duplicate): Likewise.
14630 (ix86_expand_vector_init_one_nonzero): Likewise.
14631 (ix86_expand_vector_init_one_var): Likewise.
14632 (ix86_expand_vector_init_concat): Likewise.
14633 (ix86_expand_vector_init_interleave): Likewise.
14634 (ix86_expand_vector_init_general): Likewise.
14635 (ix86_expand_vector_set): Likewise.
14636 (ix86_expand_vector_extract): Likewise.
14637 (emit_reduc_half): Likewise.
14638 (ix86_emit_i387_round): Likewise.
14639 (ix86_mangle_type): Likewise.
14640 (ix86_expand_round_sse4): Likewise.
14641 (expand_vec_perm_blend): Likewise.
14642 (canonicalize_vector_int_perm): Likewise.
14643 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14644 (expand_vec_perm_1): Likewise.
14645 (expand_vec_perm_interleave3): Likewise.
14646 (expand_vec_perm_even_odd_pack): Likewise.
14647 (expand_vec_perm_even_odd_1): Likewise.
14648 (expand_vec_perm_broadcast_1): Likewise.
14649 (ix86_vectorize_vec_perm_const_ok): Likewise.
14650 (ix86_expand_vecop_qihi): Likewise.
14651 (ix86_expand_mul_widen_hilo): Likewise.
14652 (ix86_expand_sse2_abs): Likewise.
14653 (ix86_expand_pextr): Likewise.
14654 (ix86_expand_pinsr): Likewise.
14655 (ix86_preferred_simd_mode): Likewise.
14656 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14657 * config/i386/sse.md (*andnot<mode>3): Likewise.
14658 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14659 (*<code><mode>3): Likewise.
14660 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14661 (ia64_expand_atomic_op): Likewise.
14662 (ia64_arg_type): Likewise.
14663 (ia64_mode_to_int): Likewise.
14664 (ia64_scalar_mode_supported_p): Likewise.
14665 (ia64_vector_mode_supported_p): Likewise.
14666 (expand_vec_perm_broadcast): Likewise.
14667 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14668 (iq2000_function_arg_advance): Likewise.
14669 (iq2000_function_arg): Likewise.
14670 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14671 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14672 (m68k_libcall_value): Likewise.
14673 (m68k_function_value): Likewise.
14674 (sched_attr_op_type): Likewise.
14675 * config/mcore/mcore.c (mcore_output_move): Likewise.
14676 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14678 (microblaze_function_arg): Likewise.
14679 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14680 (mips_print_operand): Likewise.
14681 (mips_mode_ok_for_mov_fmt_p): Likewise.
14682 (mips_vector_mode_supported_p): Likewise.
14683 (mips_preferred_simd_mode): Likewise.
14684 (mips_expand_vpc_loongson_even_odd): Likewise.
14685 (mips_expand_vec_unpack): Likewise.
14686 (mips_expand_vi_broadcast): Likewise.
14687 (mips_expand_vector_init): Likewise.
14688 (mips_expand_vec_reduc): Likewise.
14689 (mips_expand_msa_cmp): Likewise.
14690 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14691 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14692 (cc_flags_for_mode): Likewise.
14693 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14694 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14695 (nds32_output_casesi_pc_relative): Likewise.
14696 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14697 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
14698 (nvptx_gen_unpack): Likewise.
14699 (nvptx_gen_pack): Likewise.
14700 (nvptx_gen_shuffle): Likewise.
14701 (nvptx_gen_wcast): Likewise.
14702 (nvptx_preferred_simd_mode): Likewise.
14703 * config/pa/pa.c (pa_secondary_reload): Likewise.
14704 * config/pa/predicates.md (base14_operand): Likewise.
14705 * config/powerpcspe/powerpcspe-c.c
14706 (altivec_resolve_overloaded_builtin): Likewise.
14707 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
14709 (rs6000_preferred_simd_mode): Likewise.
14710 (output_vec_const_move): Likewise.
14711 (rs6000_expand_vector_extract): Likewise.
14712 (rs6000_split_vec_extract_var): Likewise.
14713 (reg_offset_addressing_ok_p): Likewise.
14714 (rs6000_legitimate_offset_address_p): Likewise.
14715 (rs6000_legitimize_address): Likewise.
14716 (rs6000_emit_set_const): Likewise.
14717 (rs6000_const_vec): Likewise.
14718 (rs6000_emit_move): Likewise.
14719 (spe_build_register_parallel): Likewise.
14720 (rs6000_darwin64_record_arg_recurse): Likewise.
14721 (swap_selector_for_mode): Likewise.
14722 (spe_init_builtins): Likewise.
14723 (paired_init_builtins): Likewise.
14724 (altivec_init_builtins): Likewise.
14725 (do_load_for_compare): Likewise.
14726 (rs6000_generate_compare): Likewise.
14727 (rs6000_expand_float128_convert): Likewise.
14728 (emit_load_locked): Likewise.
14729 (emit_store_conditional): Likewise.
14730 (rs6000_output_function_epilogue): Likewise.
14731 (rs6000_handle_altivec_attribute): Likewise.
14732 (rs6000_function_value): Likewise.
14733 (emit_fusion_gpr_load): Likewise.
14734 (emit_fusion_p9_load): Likewise.
14735 (emit_fusion_p9_store): Likewise.
14736 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
14737 (fusion_gpr_mem_load): Likewise.
14738 (fusion_addis_mem_combo_load): Likewise.
14739 (fusion_addis_mem_combo_store): Likewise.
14740 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
14741 (fusion_gpr_mem_load): Likewise.
14742 (fusion_addis_mem_combo_load): Likewise.
14743 (fusion_addis_mem_combo_store): Likewise.
14744 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14746 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
14747 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14748 (rs6000_preferred_simd_mode): Likewise.
14749 (output_vec_const_move): Likewise.
14750 (rs6000_expand_vector_extract): Likewise.
14751 (rs6000_split_vec_extract_var): Likewise.
14752 (reg_offset_addressing_ok_p): Likewise.
14753 (rs6000_legitimate_offset_address_p): Likewise.
14754 (rs6000_legitimize_address): Likewise.
14755 (rs6000_emit_set_const): Likewise.
14756 (rs6000_const_vec): Likewise.
14757 (rs6000_emit_move): Likewise.
14758 (rs6000_darwin64_record_arg_recurse): Likewise.
14759 (swap_selector_for_mode): Likewise.
14760 (paired_init_builtins): Likewise.
14761 (altivec_init_builtins): Likewise.
14762 (rs6000_expand_float128_convert): Likewise.
14763 (emit_load_locked): Likewise.
14764 (emit_store_conditional): Likewise.
14765 (rs6000_output_function_epilogue): Likewise.
14766 (rs6000_handle_altivec_attribute): Likewise.
14767 (rs6000_function_value): Likewise.
14768 (emit_fusion_gpr_load): Likewise.
14769 (emit_fusion_p9_load): Likewise.
14770 (emit_fusion_p9_store): Likewise.
14771 * config/rx/rx.c (rx_gen_move_template): Likewise.
14772 (flags_from_mode): Likewise.
14773 * config/s390/predicates.md (s390_alc_comparison): Likewise.
14774 (s390_slb_comparison): Likewise.
14775 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
14776 (s390_vector_mode_supported_p): Likewise.
14777 (s390_cc_modes_compatible): Likewise.
14778 (s390_match_ccmode_set): Likewise.
14779 (s390_canonicalize_comparison): Likewise.
14780 (s390_emit_compare_and_swap): Likewise.
14781 (s390_branch_condition_mask): Likewise.
14782 (s390_rtx_costs): Likewise.
14783 (s390_secondary_reload): Likewise.
14784 (__SECONDARY_RELOAD_CASE): Likewise.
14785 (s390_expand_cs): Likewise.
14786 (s390_preferred_simd_mode): Likewise.
14787 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
14788 * config/sh/sh.c (sh_print_operand): Likewise.
14789 (dump_table): Likewise.
14790 (sh_secondary_reload): Likewise.
14791 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14792 * config/sh/sh.md (casesi_worker_1): Likewise.
14793 (casesi_worker_2): Likewise.
14794 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
14795 (fcc_comparison_operator): Likewise.
14796 * config/sparc/sparc.c (sparc_expand_move): Likewise.
14797 (emit_soft_tfmode_cvt): Likewise.
14798 (sparc_preferred_simd_mode): Likewise.
14799 (output_cbranch): Likewise.
14800 (sparc_print_operand): Likewise.
14801 (sparc_expand_vec_perm_bmask): Likewise.
14802 (vector_init_bshuffle): Likewise.
14803 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14804 (spu_vector_mode_supported_p): Likewise.
14805 (spu_expand_insv): Likewise.
14806 (spu_emit_branch_or_set): Likewise.
14807 (spu_handle_vector_attribute): Likewise.
14808 (spu_builtin_splats): Likewise.
14809 (spu_builtin_extract): Likewise.
14810 (spu_builtin_promote): Likewise.
14811 (spu_expand_sign_extend): Likewise.
14812 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14813 (tilegx_simd_int): Likewise.
14814 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
14815 (tilepro_simd_int): Likewise.
14816 * config/v850/v850.c (const_double_split): Likewise.
14817 (v850_print_operand): Likewise.
14818 (ep_memory_offset): Likewise.
14819 * config/vax/vax.c (vax_rtx_costs): Likewise.
14820 (vax_output_int_move): Likewise.
14821 (vax_output_int_add): Likewise.
14822 (vax_output_int_subtract): Likewise.
14823 * config/visium/predicates.md (visium_branch_operator): Likewise.
14824 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
14825 (visium_print_operand_address): Likewise.
14826 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14827 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
14828 (xtensa_expand_conditional_branch): Likewise.
14829 (xtensa_copy_incoming_a7): Likewise.
14830 (xtensa_output_literal): Likewise.
14831 * dfp.c (decimal_real_maxval): Likewise.
14832 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14835 Alan Hayward <alan.hayward@arm.com>
14836 David Sherwood <david.sherwood@arm.com>
14838 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
14839 (mode_nunits_inline): Likewise.
14840 (mode_inner_inline): Likewise.
14841 (mode_unit_size_inline): Likewise.
14842 (mode_unit_precision_inline): Likewise.
14843 (emit_insn_modes_h): Likewise. Also emit a #define of the
14845 (emit_mode_wider): Add an E_ prefix to mode names.
14846 (emit_mode_complex): Likewise.
14847 (emit_mode_inner): Likewise.
14848 (emit_mode_adjustments): Likewise.
14849 (emit_mode_int_n): Likewise.
14850 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
14851 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
14852 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
14853 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
14854 (CRC32_BUILTIN, ENTRY): Likewise.
14855 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
14856 (aarch64_pop_regs): Likewise.
14857 (aarch64_process_components): Likewise.
14858 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
14859 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
14860 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
14861 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
14862 * config/arm/arm.c (arm_init_libfuncs): Likewise.
14863 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
14865 * config/i386/i386-builtin.def (pcmpestr): Likewise.
14866 (pcmpistr): Likewise.
14867 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
14868 * config/mmix/mmix.c (mmix_output_condition): Likewise.
14869 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14871 * config/rl78/rl78.c (mduc_regs): Likewise.
14872 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14873 (htm_expand_builtin): Likewise.
14874 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
14875 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
14876 * config/xtensa/xtensa.c (print_operand): Likewise.
14877 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
14878 (NUM_MODE_VECTOR_INT): Likewise.
14879 * genoutput.c (null_operand): Likewise.
14880 (output_operand_data): Likewise.
14881 * genrecog.c (print_parameter_value): Likewise.
14882 * lra.c (debug_operand_data): Likewise.
14884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14886 * dwarf2out.c (scompare_loc_descriptor_wide)
14887 (scompare_loc_descriptor_narrow): New functions, split out from...
14888 (scompare_loc_descriptor): ...here.
14889 * expmed.c (emit_store_flag_int): New function, split out from...
14890 (emit_store_flag): ...here.
14892 2017-08-30 Richard Biener <rguenther@suse.de>
14894 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14895 (dwarf2out_early_finish): Move setting of AT_pubnames from
14896 early debug output to early finish.
14898 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14900 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14901 and -mdata-region to the assembler.
14903 2017-08-30 Richard Biener <rguenther@suse.de>
14905 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14907 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14908 (dwarf2out_early_global_decl): Only generate a DIE for the
14909 abstract origin if it doesn't already exist or is a declaration DIE.
14910 (resolve_addr): Do not add the linkage name twice when
14911 generating a stub DIE for the DW_TAG_GNU_call_site target.
14913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14915 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14916 Use machine_mode rather than int for arg1_mode.
14918 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14921 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14922 that the second argument of the built-in functions to unpack
14923 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14924 switch statement instead a lot of if statements.
14925 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14926 Allow 64-bit values to be in Altivec registers as well as
14927 traditional floating point registers.
14928 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14930 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14932 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14933 MAX_REGS_PER_ADDRESS == 1.
14935 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14937 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14938 * config/i386/i386.c (ix86_option_override_internal): Simplify
14939 setting of opts->x_flag_entry.
14941 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14942 Jakub Jelinek <jakub@redhat.com>
14943 Richard Biener <rguenther@suse.de>
14945 PR tree-optimization/81503
14946 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14947 folded constant fits in the target type; reorder tests for clarity.
14949 2017-08-29 Martin Liska <mliska@suse.cz>
14951 * passes.def: Include pass_lower_switch.
14952 * stmt.c (dump_case_nodes): Remove and move to
14953 tree-switch-conversion.
14954 (case_values_threshold): Likewise.
14955 (expand_switch_as_decision_tree_p): Likewise.
14956 (emit_case_decision_tree): Likewise.
14957 (expand_case): Likewise.
14958 (balance_case_nodes): Likewise.
14959 (node_has_low_bound): Likewise.
14960 (node_has_high_bound): Likewise.
14961 (node_is_bounded): Likewise.
14962 (emit_case_nodes): Likewise.
14963 (struct simple_case_node): New struct.
14964 (add_case_node): Remove.
14965 (emit_case_dispatch_table): Use vector instead of case_list.
14966 (reset_out_edges_aux): Remove.
14967 (compute_cases_per_edge): Likewise.
14968 (expand_case): Build list of simple_case_node.
14969 (expand_sjlj_dispatch_table): Use it.
14970 * tree-switch-conversion.c (struct case_node): Moved from
14971 stmt.c and adjusted.
14972 (emit_case_nodes): Likewise.
14973 (node_has_low_bound): Likewise.
14974 (node_has_high_bound): Likewise.
14975 (node_is_bounded): Likewise.
14976 (case_values_threshold): Likewise.
14977 (reset_out_edges_aux): Likewise.
14978 (compute_cases_per_edge): Likewise.
14979 (add_case_node): Likewise.
14980 (dump_case_nodes): Likewise.
14981 (balance_case_nodes): Likewise.
14982 (expand_switch_as_decision_tree_p): Likewise.
14983 (emit_jump): Likewise.
14984 (emit_case_decision_tree): Likewise.
14985 (try_switch_expansion): Likewise.
14986 (do_jump_if_equal): Likewise.
14987 (emit_cmp_and_jump_insns): Likewise.
14988 (fix_phi_operands_for_edge): New function.
14989 (record_phi_operand_mapping): Likewise.
14990 (class pass_lower_switch): New pass.
14991 (pass_lower_switch::execute): New function.
14992 (make_pass_lower_switch): Likewise.
14993 (conditional_probability):
14994 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14995 * tree-pass.h: Add make_pass_lower_switch.
14997 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15000 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15003 2017-08-29 Richard Biener <rguenther@suse.de>
15005 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15006 we do not add a DW_AT_inline attribute twice.
15007 (gen_subprogram_die): Remove code setting DW_AT_inline on
15008 DECL_ABSTRACT_P nodes.
15010 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15012 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15013 calls to internal functions.
15014 (gimplify_modify_expr): Likewise.
15015 * tree-call-cdce.c (use_internal_fn): Likewise.
15016 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15017 (convert_to_divmod): Set the nothrow flag.
15018 * tree-if-conv.c (predicate_mem_writes): Likewise.
15019 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15020 (vectorizable_call): Likewise.
15021 (vectorizable_store): Likewise.
15022 (vectorizable_load): Likewise.
15023 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15024 (vect_recog_mask_conversion_pattern): Likewise.
15026 2017-08-29 Martin Liska <mliska@suse.cz>
15029 * gcc.c (driver_handle_option): Add new argument.
15030 * opts-common.c (handle_option): Pass
15031 target_option_override_hook.
15032 * opts-global.c (lang_handle_option): Add new option.
15033 (set_default_handlers): Add new argument.
15034 (decode_options): Likewise.
15035 * opts.c (target_handle_option): Likewise.
15036 (common_handle_option): Call target_option_override_hook.
15037 * opts.h (struct cl_option_handler_func): Add hook for
15038 target option override.
15039 (struct cl_option_handlers): Likewise.
15040 (set_default_handlers): Add new argument.
15041 (decode_options): Likewise.
15042 (common_handle_option): Likewise.
15043 (target_handle_option): Likewise.
15044 * toplev.c (toplev::main): Pass targetm.target_option.override
15047 2017-08-29 Richard Biener <rguenther@suse.de>
15048 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15050 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15051 life to the active subtree.
15053 2017-08-28 Jeff Law <law@redhat.com>
15055 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15056 derive_equivalences.
15057 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15058 Code moved into....
15059 (edge_info::derive_equivalences): New private member function
15061 * tree-ssa-dom.c (class edge_info): Changed from a struct
15062 to a class. Add ctor/dtor, methods and data members.
15063 (edge_info::edge_info): Renamed from allocate_edge_info.
15064 Initialize additional members.
15065 (edge_info::~edge_info): New.
15066 (free_dom_edge_info): Delete the edge info.
15067 (record_edge_info): Use new class & associated member functions.
15068 Tighten forms for testing for edge equivalences.
15069 (record_temporary_equivalences): Iterate over the simple
15070 equivalences rather than assuming there's only one per edge.
15071 (cprop_into_successor_phis): Iterate over the simple
15072 equivalences rather than assuming there's only one per edge.
15073 (optimize_stmt): Use operand_equal_p rather than pointer
15074 equality for mini-DSE code.
15076 2017-08-28 Nathan Sidwell <nathan@acm.org>
15078 * gcc.c (execute): Fold SIGPIPE handling into switch
15079 statement. Adjust internal error message.
15081 2017-08-28 Richard Biener <rguenther@suse.de>
15084 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15085 Do nothing for removed DIEs.
15087 2017-08-28 Richard Biener <rguenther@suse.de>
15089 PR tree-optimization/81977
15090 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15093 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15096 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15097 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15098 using targetm.gen_mem_thread_fence.
15100 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15103 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15104 predicate to register_operand. Reorder operands.
15105 (*btr<mode>): Ditto.
15106 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15107 (*btr<mode>_mask): Ditto.
15109 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15111 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15112 * config/rs6000/xmmintrin.h: New file.
15113 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15115 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15118 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15119 parameter and_insn and return it.
15120 (recombine_lvx_pattern): Insert a copy to ensure availability of
15121 the base register of the copied masking operation at the point of
15122 the instruction replacement.
15123 (recombine_stvx_pattern): Likewise.
15125 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15127 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15128 undocumented switches.
15129 (-mpower9-dform-vector): Likewise.
15130 (-mpower9-dform): Likewise.
15131 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15132 comments to delete references to -mpower9-dform* switches.
15133 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15134 Delete reference to -mpower9-dform* switches, test for
15135 -mpower9-vector instead.
15136 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15137 (OTHER_P9_VECTOR_MASKS): Likewise.
15138 (POWERPC_MASKS): Likewise.
15139 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15140 tests against -mpower9-dform* to -mpower9-vector. Delete code
15141 that checked for -mpower9-dform* consistancy with other options.
15142 Add test for -mpower9-misc to enable other power9 switches.
15143 (rs6000_init_hard_regno_mode_ok): Likewise.
15144 (rs6000_option_override_internal): Likewise.
15145 (rs6000_emit_prologue): Likewise.
15146 (rs6000_emit_epilogue): Likewise.
15147 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15148 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15149 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15151 (emit_fusion_p9_store): Likewise.
15152 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15153 resetting these macros if the assembler does not support ISA 3.0
15155 (TARGET_P9_DFORM_VECTOR): Likewise.
15156 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15157 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15159 2017-08-25 Alan Modra <amodra@gmail.com>
15161 PR rtl-optimization/81747
15162 * cse.c (cse_extended_basic_block): Don't attempt to record
15163 equivalences for degenerate conditional branches that branch
15164 to their fall-through.
15166 2017-08-24 Martin Sebor <msebor@redhat.com>
15168 PR middle-end/81908
15169 * gimple-fold.c (size_must_be_zero_p): New function.
15170 (gimple_fold_builtin_memory_op): Call it.
15172 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15174 * config/rs6000/mm_malloc.h: New file.
15176 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15178 PR tree-optimization/81913
15179 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15180 analysis when either IVs in condition can wrap.
15182 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15184 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15185 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15187 2017-08-24 Richard Biener <rguenther@suse.de>
15190 * targhooks.c (default_target_can_inline_p): Properly
15191 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15192 is present and always compare.
15193 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15194 infer -mfpmath=sse from TARGET_SSE_P.
15195 (ix86_can_inline_p): Properly use target_option_default_node when
15196 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15198 2017-08-24 Richard Biener <rguenther@suse.de>
15201 * dwarf2out.c (output_die): Handle flag_generate_offload like
15203 (output_comp_unit): Likewise.
15204 (gen_array_type_die): Likewise.
15205 (dwarf2out_early_finish): Likewise.
15206 (note_variable_value_in_expr): Likewise.
15207 (dwarf2out_finish): Likewise. Adjust assert.
15208 * cgraphunit.c (symbol_table::compile): Move setting of
15209 flag_generate_offload earlier ...
15210 (symbol_table::finalize_compilation_unit): ... here, before
15211 early debug finalization.
15213 2017-08-24 Richard Biener <rguenther@suse.de>
15215 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15216 and ipa-fnsummary.h.
15217 (ix86_can_inline_p): When ix86_fpmath flags do not match
15218 check whether the callee uses FP math at all.
15220 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15222 PR middle-end/81931
15223 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15224 instead of TYPE_PRECISION.
15226 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15227 Alan Hayward <alan.hayward@arm.com>
15228 David Sherwood <david.sherwood@arm.com>
15230 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15232 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15234 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15236 (ix86_frame::stack_realign_allocate): New field.
15237 (struct machine_frame_state): Modify comments.
15238 (machine_frame_state::sp_realigned_fp_end): New field.
15239 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15240 layout calculation.
15241 (sp_valid_at): Add assertion to assure no attempt to access invalid
15242 offset of a realigned stack.
15243 (fp_valid_at): Likewise.
15244 (choose_baseaddr): Modify comments.
15245 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15246 ix86_expand_prologue.
15247 (ix86_expand_prologue): Modify stack realignment and allocation.
15248 (ix86_expand_epilogue): Modify comments.
15249 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15250 avx2_runtime, avx512f, and avx512f_runtime.
15252 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15254 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15255 (mstackrealign): Do not init to -1.
15256 * config/i386/i386.c (ix86_option_override_internal):
15257 Check opts_set, not opts when setting default value of
15258 opts->x_ix86_force_align_arg_pointer.
15260 2017-08-23 Richard Biener <rguenther@suse.de>
15262 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15263 lang_hooks.decl_printable_name.
15264 * print-rtl-function.c (print_rtx_function): Likewise.
15265 * tree-pretty-print.c (dump_function_header): Likewise.
15267 2017-08-23 Richard Biener <rguenther@suse.de>
15270 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15271 -g0 at compile-time.
15273 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15275 PR middle-end/19706
15276 * doc/sourcebuild.texi (Other hardware attributes):
15279 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15281 PR middle-end/19706
15282 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15283 Fix single-use check.
15285 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15287 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15289 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15291 * doc/install.texi: Modify to add more details on running selected
15294 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15296 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15297 is combined with -mabi=ms.
15298 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15301 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15303 PR tree-optimization/81488
15304 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15305 and cached_basis fields.
15306 (MAX_SPREAD): New constant.
15307 (alloc_cand_and_find_basis): Initialize new fields.
15308 (clear_visited): New function.
15309 (create_phi_basis_1): Rename from create_phi_basis, set visited
15310 and cached_basis fields.
15311 (create_phi_basis): New wrapper function.
15312 (phi_add_costs_1): Rename from phi_add_costs, add spread
15313 parameter, set visited field, short-circuit when limits reached.
15314 (phi_add_costs): New wrapper function.
15315 (record_phi_increments_1): Rename from record_phi_increments, set
15317 (record_phi_increments): New wrapper function.
15318 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15319 (phi_incr_cost): New wrapper function.
15320 (all_phi_incrs_profitable_1): Rename from
15321 all_phi_incrs_profitable, set visited field.
15322 (all_phi_incrs_profitable): New wrapper function.
15324 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15325 Alan Hayward <alan.hayward@arm.com>
15326 David Sherwood <david.sherwood@arm.com>
15328 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15329 that takes the outer and inner modes.
15330 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15331 comparison as the canonical test for a paradoxical subreg.
15332 * combine.c (simplify_set): Use paradoxical_subreg_p.
15333 (make_extraction): Likewise.
15334 (force_to_mode): Likewise.
15335 (rtx_equal_for_field_assignment_p): Likewise.
15336 (gen_lowpart_for_combine): Likewise.
15337 (simplify_comparison): Likewise.
15338 * cse.c (equiv_constant): Likewise.
15339 * expmed.c (store_bit_field_1): Likewise.
15340 * final.c (alter_subreg): Likewise.
15341 * fwprop.c (propagate_rtx): Likewise.
15342 (forward_propagate_subreg): Likewise.
15343 * ira-conflicts.c (ira_build_conflicts): Likewise.
15344 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15345 * lra-constraints.c (curr_insn_transform): Likewise.
15346 (split_reg): Likewise.
15347 * lra-eliminations.c (move_plus_up): Likewise.
15348 (lra_eliminate_regs_1): Likewise.
15349 * recog.c (general_operand): Likewise.
15350 * ree.c (combine_reaching_defs): Likewise.
15351 * reload.c (push_reload): Likewise.
15352 (find_reloads): Likewise.
15353 * reload1.c (elimination_effects): Likewise.
15354 (compute_reload_subreg_offset): Likewise.
15355 (choose_reload_regs): Likewise.
15356 * rtlanal.c (subreg_lsb_1): Likewise.
15357 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15358 (simplify_subreg): Likewise.
15359 * var-tracking.c (track_loc_p): Likewise.
15360 * emit-rtl.c (byte_lowpart_offset): Likewise.
15361 (paradoxical_subreg_p): Delete out-of-line definition.
15363 2017-08-22 Jeff Law <law@redhat.com>
15365 PR tree-optimization/81741
15366 PR tree-optimization/71947
15367 * tree-ssa-dom.c: Include tree-inline.h.
15368 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15369 equivalences if one is more expensive to compute than the other.
15370 * tree-ssa-scopedtables.h (class const_or_copies): Make
15371 record_const_or_copy_raw method private.
15372 (class avail_exprs_stack): New method simplify_binary_operation.
15373 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15374 avail_exprs_stack::simplify_binary_operation as needed.
15375 (avail_exprs_stack::simplify_binary_operation): New function.
15377 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15379 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15380 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15381 (DOT_SYMBOLS): Likewise.
15382 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15383 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15384 (RS6000_ABI_NAME): Likewise.
15385 (TARGET_CMODEL): Likewise.
15386 (TOC_SECTION_ASM_OP): Likewise.
15387 (SET_CMODEL): New macro.
15388 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15390 2017-08-22 Richard Biener <rguenther@suse.de>
15392 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15393 to allow for free-lang-data replacements similar to verify_type_variant.
15395 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15396 Alan Hayward <alan.hayward@arm.com>
15397 David Sherwood <david.sherwood@arm.com>
15399 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15400 VOIDmode for the LABEL_REF.
15402 2017-08-22 Richard Biener <rguenther@suse.de>
15404 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15406 2017-08-22 Richard Biener <rguenther@suse.de>
15408 * common.opt (feliminate-dwarf2-dups): Ignore.
15409 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15410 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15411 same_die_p_wrap, compute_section_prefix,
15412 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15413 (comdat_symbol_id, comdat_symbol_number): Likewise.
15414 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15416 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15417 (output_die): Mark unreachable path unreachable.
15418 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15419 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15420 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15421 (dwarf2out_early_finish): Likewise.
15423 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15425 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15427 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15430 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15431 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15432 (avr_attribute_table) <io, io_low, address>: Initialize
15433 .decl_required with true.
15435 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15437 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15438 undocumented debugging options.
15439 (-mvsx-scalar-double): Likewise.
15440 (-mallow-df-permute): Likewise.
15441 (-mvectorize-builtins): Likewise.
15442 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15443 (rs6000_builtin_vectorized_function): Likewise.
15444 (rs6000_builtin_md_vectorized_function): Likewise.
15445 (rs6000_opt_vars): Likewise.
15447 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15450 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15451 (*btrq_imm): Rename from *btrq.
15452 (*btcq_imm): Rename from *btcq.
15453 (btsc): New code attribute.
15454 (*<btsc><mode>): New insn pattern.
15455 (*btr<mode>): Ditto.
15456 (*<btsc><mode>_mask): New insn_and_split pattern.
15457 (*btr<mode>_mask): Ditto.
15459 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15460 Alan Hayward <alan.hayward@arm.com>
15461 David Sherwood <david.sherwood@arm.com>
15463 * function.c (pad_below): Simplify padding calculation.
15465 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15466 Alan Hayward <alan.hayward@arm.com>
15467 David Sherwood <david.sherwood@arm.com>
15469 * target.def (function_prologue): Remove frame size argument.
15470 (function_epilogue): Likewise.
15471 * doc/tm.texi: Regenerate.
15472 * final.c (final_start_function): Update call to function_prologue.
15473 (final_end_function): Update call to function_epilogue.
15474 (default_function_pro_epilogue): Remove frame size argument.
15475 * output.h (default_function_pro_epilogue): Likewise.
15476 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15477 (arm_output_function_prologue): Likewise.
15478 * config/frv/frv.c (frv_function_prologue): Likewise.
15479 (frv_function_epilogue): Likewise.
15480 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15481 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15482 (ia64_output_function_epilogue): Likewise.
15483 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15484 (m32r_output_function_epilogue): Likewise.
15485 * config/microblaze/microblaze.c (microblaze_function_prologue)
15486 (microblaze_function_epilogue): Likewise.
15487 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15488 (mips_output_function_epilogue): Likewise.
15489 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15490 (mmix_target_asm_function_epilogue): Likewise.
15491 * config/msp430/msp430.c (msp430_start_function): Likewise.
15492 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15493 (nds32_asm_function_epilogue): Likewise.
15494 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15495 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15496 (pa_output_function_epilogue): Likewise.
15497 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15498 (rs6000_output_function_epilogue): Likewise.
15499 * config/rl78/rl78.c (rl78_start_function): Likewise.
15500 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15501 (rs6000_output_function_epilogue): Likewise.
15502 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15503 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15504 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15505 (sparc_asm_function_epilogue): Likewise.
15507 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15509 * tree.h (type_has_mode_precision_p): New function.
15510 * convert.c (convert_to_integer_1): Use it.
15511 * expr.c (expand_expr_real_2): Likewise.
15512 (expand_expr_real_1): Likewise.
15513 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15514 * match.pd: Likewise.
15515 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15516 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15517 * tree-tailcall.c (process_assignment): Likewise.
15518 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15519 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15520 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15521 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15522 (vectorizable_assignment): Likewise.
15523 (vectorizable_shift): Likewise.
15524 (vectorizable_operation): Likewise.
15525 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15527 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15529 * match.pd: Add pow (C, x) simplification.
15531 2017-08-21 Richard Biener <rguenther@suse.de>
15533 PR tree-optimization/81900
15534 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15535 for blocks with abnormal predecessors.
15536 (compute_antic): Do not set visited flag prematurely.
15538 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15541 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15543 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15545 * stor-layout.h (vector_type_mode): Move to...
15546 * tree.h (vector_type_mode): ...here.
15547 * stor-layout.c (vector_type_mode): Move to...
15548 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15550 2017-08-21 Richard Biener <rguenther@suse.de>
15552 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15553 register_external_die hooks.
15554 (debug_false_tree_charstarstar_uhwistar): Declare.
15555 (debug_nothing_tree_charstar_uhwi): Likewise.
15556 * debug.c (do_nothing_debug_hooks): Adjust.
15557 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15558 (debug_nothing_tree_charstar_uhwi): Likewise.
15559 * dbxout.c (dbx_debug_hooks): Adjust.
15560 (xcoff_debug_hooks): Likewise.
15561 * sdbout.c (sdb_debug_hooks): Likewise.
15562 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15563 * dwarf2out.c (macinfo_label_base): New global.
15564 (dwarf2out_register_external_die): New function for the
15565 register_external_die hook.
15566 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15567 (dwarf2_debug_hooks): Use them.
15568 (dwarf2_lineno_debug_hooks): Adjust.
15569 (struct die_struct): Add with_offset flag.
15570 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15571 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15572 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15573 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15574 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15575 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15576 defining section names for the early LTO debug variants.
15577 (reset_indirect_string): New helper.
15578 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15579 (print_dw_val): Add support for offsetted symbol references.
15580 (get_ultimate_context): Split out from is_cxx.
15581 (is_cxx): Use get_ultimate_context.
15582 (is_fortran): Add decl overload.
15583 (compute_comp_unit_symbol): Split out worker from
15584 compute_section_prefix.
15585 (compute_section_prefix): Call compute_comp_unit_symbol and
15586 set comdat_type_p here.
15587 (output_die): Skip DIE symbol output for the LTO added one.
15588 Handle DIE symbol references with offset.
15589 (output_comp_unit): Guard section name mangling properly.
15590 For LTO debug sections emit a symbol at the section beginning
15591 which we use to refer to its DIEs.
15592 (add_abstract_origin_attribute): For DIEs registered via
15593 dwarf2out_register_external_die directly refer to the early
15594 DIE rather than indirectly through the shadow one we created.
15595 Remove obsolete call to dwarf2out_abstract_function for
15596 non-function/block origins.
15597 (gen_array_type_die): When generating early LTO debug do
15598 not emit DW_AT_string_length.
15599 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15600 late when in LTO. As suggested place a gcc_unreachable for
15601 the DECL_ABSTRACT_P case.
15602 (gen_subprogram_die): Avoid another specification DIE
15603 for early built declarations/definitions for the late LTO case.
15604 (gen_variable_die): Add type references for late duplicated VLA dies
15606 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15607 we have the abstract instance already.
15608 (process_scope_var): Adjust decl DIE contexts in LTO which
15609 first puts them in limbo.
15610 (gen_decl_die): Do not generate type DIEs late apart from
15611 types for VLAs or for decls we do not yet have a DIE. Do not
15612 call dwarf2out_abstract_function late.
15613 (dwarf2out_early_global_decl): Make sure to create DIEs
15614 for abstract instances of a decl first.
15615 (dwarf2out_late_global_decl): Adjust comment.
15616 (output_macinfo_op): With multiple macro sections use
15617 macinfo_label_base to distinguish labels.
15618 (output_macinfo): Likewise. Update macinfo_label_base.
15619 Pass in the line info label.
15620 (note_variable_value_in_expr): When generating LTO resolve
15621 all variable values here by generating DIEs as needed.
15622 (init_sections_and_labels): Add early LTO debug flag parameter
15623 and generate different sections and names if set. Add generation
15624 counter for the labels so we can have multiple of them.
15625 (reset_dies): Helper to allow DIEs to be output multiple times.
15626 (dwarf2out_finish): When outputting DIEs to the fat part of an
15627 LTO object first reset DIEs.
15628 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15629 (modified_type_die): Check for decl_ultimate_origin being self
15631 (gen_type_die_with_usage): Likewise.
15632 (gen_typedef_die): Allow decl_ultimate_origin being self.
15633 (set_decl_abstract_flags): Remove.
15634 (set_block_abstract_flags): Likewise.
15635 (dwarf2out_abstract_function): Treat the early generated DIEs
15636 as the abstract copy and only add DW_AT_inline and
15637 DW_AT_artificial here and call set_decl_origin_self.
15638 If the DIE has an abstract origin don't do anything.
15639 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15640 if we have none yet (Go fails to build one, PR78628).
15641 (variably_modified_type_p): Prevent endless recursion for Ada
15642 cyclic pointer types.
15643 * lto-streamer-in.c: Include debug.h.
15644 (dref_queue): New global.
15645 (lto_read_tree_1): Stream in DIE references.
15646 (lto_input_tree): Register DIE references.
15647 (input_function): Stream DECL_DEBUG_ARGS.
15648 * lto-streamer-out.c: Include debug.h.
15649 (lto_write_tree_1): Output DIE references.
15650 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15651 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15652 (output_function): Stream DECL_DEBUG_ARGS.
15653 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15654 Stream DECL_ABSTRACT_ORIGIN.
15655 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15656 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15657 DECL_CONTEXT for file-scope decls.
15658 * lto-streamer.h (struct dref_entry): Declare.
15659 (dref_queue): Likewise.
15660 * cfgexpand.c (pass_expand::execute): Do not call the
15661 outlining_inline_function hook here.
15662 * lto-wrapper.c (debug_obj): New global.
15663 (tool_cleanup): Unlink it if required.
15664 (debug_objcopy): New function.
15665 (run_gcc): Handle early debug sections in the IL files by
15666 extracting them to separate files, partially linkin them and
15667 feeding the result back as result to the linker.
15668 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15669 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15670 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15671 sections into a separate segment.
15672 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15674 (darwin_asm_dwarf_section): Likewise.
15675 (darwin_asm_output_dwarf_offset): Likewise.
15676 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15678 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15679 Alan Hayward <alan.hayward@arm.com>
15680 David Sherwood <david.sherwood@arm.com>
15682 * read-md.h (md_reader::record_potential_iterator_use): Replace
15683 pointer argument with an rtx and an index.
15684 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15685 (apply_mode_iterator): Likewise.
15686 (apply_code_iterator): Likewise.
15687 (apply_int_iterator): Likewise.
15688 (apply_subst_iterator): Likewise.
15689 (record_iterator_use): Likewise.
15690 (record_attribute_use): Likewise.
15691 (md_reader::record_potential_iterator_use): Likewise. Update calls
15692 to record_iterator_use and apply_iterator.
15693 (iterator_use): Replace ptr with x and index.
15694 (attribute_use): Likewise.
15695 (apply_attribute_uses): Update calls to apply_iterator.
15696 (apply_iterators): Likewise. Update initialization of iterator_use.
15697 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
15698 and record_potential_iterator_use.
15699 (rtx_reader::read_rtx_operand): Likewise.
15701 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15702 Alan Hayward <alan.hayward@arm.com>
15703 David Sherwood <david.sherwood@arm.com>
15705 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
15708 2017-08-21 Richard Biener <rguenther@suse.de>
15710 PR middle-end/81884
15711 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
15712 at struct end conservatively when comparing common bases.
15714 2017-08-21 Richard Biener <rguenther@suse.de>
15716 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
15717 (mem_ref_in_stmt): Remove.
15718 (determine_max_movement): Use ref index to get at the reference.
15719 (invariantness_dom_walker::before_dom_children): Deal with
15720 lim data already initialized.
15721 (gather_mem_refs_stmt): Initialize lim data and record ref index.
15723 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15725 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
15726 (TARGET_ISA_ROUND): Ditto.
15727 (TARGET_ROUND): Ditto.
15728 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
15729 * config/i386/i386.md: Ditto.
15730 * config/i386/sse.md: Ditto.
15731 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
15732 with OPTION_MASK_ISA_SSE4_1.
15734 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
15737 * doc/extend.texi (x86 Built-in Functions): Correct the name of
15738 __builtin_ia32_lzcnt_u16.
15740 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
15743 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
15744 (rs6000_set_current_function): Rewrite function to use it.
15746 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
15749 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
15750 and TYPE_WARN_IF_NOT_ALIGN.
15751 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
15752 (handle_warn_if_not_align): New.
15753 (place_union_field): Call handle_warn_if_not_align.
15754 (place_field): Call handle_warn_if_not_align.
15755 Copy TYPE_WARN_IF_NOT_ALIGN.
15756 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
15757 (layout_type): Likewise.
15758 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
15760 (tree_decl_common): Add warn_if_not_align.
15761 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
15762 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
15763 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
15764 (DECL_WARN_IF_NOT_ALIGN): Likewise.
15765 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
15766 * doc/extend.texi: Document warn_if_not_aligned attribute.
15767 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
15769 2017-08-17 Martin Liska <mliska@suse.cz>
15772 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
15773 (get_data_dependence): Use it as pointer type.
15774 (distribute_loop): Likewise.
15776 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15778 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
15779 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
15780 (p8_vmrgow_<mode>_direct): New define_insn.
15781 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
15782 handle endianness for vmrgew and vmrgow permute patterns.
15784 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15786 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
15787 * config/rs6000/rs6000-cpus.def: Remove comment.
15788 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
15789 (POWERPC_MASKS): Likewise.
15790 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
15791 use of TARGET_VSX_TIMODE.
15792 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
15793 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
15794 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
15795 (rs6000_option_override_internal): Remove dead code.
15796 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
15797 (rs6000_legitimize_reload_address): Likewise.
15798 (rs6000_legitimate_address_p): Likewise.
15799 (rs6000_opt_masks): Delete "vsx-timode".
15800 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
15801 from function comment.
15802 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
15803 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
15804 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
15806 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
15807 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
15808 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
15809 (VSX_TI): Likewise.
15811 (define_peephole2): Likewise.
15813 2017-08-17 Martin Sebor <msebor@redhat.com>
15816 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
15817 past the end of an array.
15818 (test_pp_format): Add test cases.
15820 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
15822 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
15823 missing ECF_NOTHROW flags.
15825 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
15828 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
15829 operands residing in integer registers.
15830 (*vsx_le_perm_load_<mode>): Likewise.
15831 (*vsx_le_perm_store_<mode>): Likewise.
15832 (define_peephole2): Add peepholes to optimize the above.
15834 2017-08-17 Marek Polacek <polacek@redhat.com>
15836 PR middle-end/81814
15837 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
15838 to mimic what shorten_compare did. Change the return type to bool.
15839 (fold_cond_expr_with_comparison): Update call to
15840 operand_equal_for_comparison_p.
15841 (fold_ternary_loc): Likewise.
15843 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
15845 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
15847 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
15849 2017-08-17 Richard Biener <rguenther@suse.de>
15851 * tree-ssa-structalias.c (solve_graph): When propagating
15852 to successors update the graphs succ edges and avoid duplicate work.
15854 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
15857 * config/i386/i386.c (ix86_option_override_internal): Save target
15858 specific options after ix86_stack_protector_guard_reg was changed.
15860 2017-08-17 Richard Biener <rguenther@suse.de>
15862 PR tree-optimization/81827
15863 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
15864 (new_var_info): Initialize it conservatively.
15865 (get_call_vi): Mark register vars.
15866 (new_scalar_tmp_constraint_exp): Likewise.
15867 (handle_rhs_call): Likewise.
15868 (handle_const_call): Likewise.
15869 (create_function_info_for): Likewise.
15870 (solve_constraints): Sort varinfos to separate register from
15871 non-register vars to pack points-to solution bitmaps during
15874 2017-08-17 Marek Polacek <polacek@redhat.com>
15876 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
15878 2017-08-17 Richard Biener <rguenther@suse.de>
15880 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
15881 to true when overflow is undefined and we saturated the result.
15883 2017-08-17 Alan Modra <amodra@gmail.com>
15886 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15887 Don't use store multiple if only one reg needs saving.
15888 (interesting_frame_related_regno): New function.
15889 (rs6000_frame_related): Don't emit frame info for regs that
15891 (rs6000_emit_epilogue): Likewise.
15893 2017-08-16 Nathan Sidwell <nathan@acm.org>
15895 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15896 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15897 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15898 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15899 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15900 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15901 (verify_type): Adjust for TYPE_BINFO move.
15902 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15903 process TYPE_BINFO directly.
15904 (hash_tree): Likewise.
15905 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15907 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15910 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15912 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15914 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15917 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15918 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15919 Add ix86_binary_operator_ok to insn constraint.
15920 (*iordi_1_bts): Ditto.
15921 (*xordi_1_btc): Ditto.
15922 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15923 Update corresponding peephole2 pattern.
15927 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15929 PR tree-optimization/81832
15930 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15931 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15933 2017-08-16 Marek Polacek <polacek@redhat.com>
15936 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15937 regarding size_zero_node.
15939 2017-08-16 Martin Liska <mliska@suse.cz>
15942 * config.gcc: Respect previously set extra_objs in case
15945 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15947 PR tree-optimization/81835
15948 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15949 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15950 not depend on the phi.
15952 2017-08-16 Alan Modra <amodra@gmail.com>
15954 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15957 2017-08-16 Alan Modra <amodra@gmail.com>
15959 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15960 (save_reg_p): ..into this. Update all callers.
15961 (first_reg_to_save): Simplify.
15963 2017-08-16 Alan Modra <amodra@gmail.com>
15965 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15968 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15972 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15973 constant count if that count is less than 32.
15975 2017-08-15 Nathan Sidwell <nathan@acm.org>
15977 * gcc.c (execute): Emit friendlier message if inferior is killed
15978 by an external cause.
15980 2017-08-15 Richard Biener <rguenther@suse.de>
15982 PR tree-optimization/81790
15983 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15984 CONSTRUCTORs from simplifying and VN.
15986 2017-08-14 Martin Sebor <msebor@redhat.com>
15988 * builtin-attrs.def: Add comments.
15990 2017-08-14 Martin Sebor <msebor@redhat.com>
15993 * doc/extend.texi (attribute nonstring): Document new attribute.
15995 2017-08-14 Martin Sebor <msebor@redhat.com>
15998 * tree-diagnostic.c (default_tree_printer): Handle %G.
15999 * gimple-pretty-print.h (percent_G_format): Declare new function.
16000 * gimple-pretty-print.c (percent_G_format): Define.
16001 * tree-pretty-print.c (percent_K_format): Add argument.
16003 2017-08-14 Martin Sebor <msebor@redhat.com>
16005 PR translation/79998
16006 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16007 Remove a stray space.
16009 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16012 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16013 (*iordi_1_bts): Ditto.
16014 (*xordi_1_btc): Ditto.
16016 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16019 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16020 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16022 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16023 quoted strings, and make more translator-friendly.
16024 (darwin_rs6000_override_options): Likewise.
16025 (rs6000_option_override_internal): Likewise.
16026 (rs6000_return_in_memory): Fix overlong line.
16027 (init_cmulative_args): Use quoted strings, and make more
16028 translator-friendly.
16029 (rs6000_pass_by_reference): Fix overlong line.
16030 (def_builtin): Use quoted strings.
16031 (altivec_expand_predicate_builtin): Use quoted strings, and make
16032 more translator-friendly.
16033 (htm_expand_builtin): Use quoted strings.
16034 (cpu_expand_builtin): Use quoted strings, and make more
16035 translator-friendly.
16036 (altivec_expand_builtin): Likewise.
16037 (paired_expand_predicate_builtin): Likewise.
16038 (rs6000_invalid_builtin): Likewise.
16039 (builtin_function_type): Use quoted strings.
16040 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16041 more translator-friendly.
16042 (rs6000_trampoline_init): Likewise.
16043 (rs6000_handle_altivec_attribute): Likewise.
16044 (rs6000_inner_target_options): Use quoted strings.
16045 (rs6000_disable_incompatible_switches): Likewise.
16046 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16047 strings, and make more translator-friendly.
16048 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16050 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16052 PR tree-optimization/81799
16053 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16054 cond_expr to simple gimple operand.
16056 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16058 PR middle-end/46932
16059 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16061 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16065 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16066 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16067 TARGET_GASISR_PROLOGUES.
16068 * config/avr/avr.c (avr_option_override): Same.
16069 (avr_pass_pre_proep::execute): Same.
16071 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16074 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16075 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16077 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16079 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16080 define_insn_and_split. Split to a memory load from 0 in
16081 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16082 using PTR mode iterator.
16083 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16084 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16085 (*add_tp_<mode>): Redefine as define_insn_and_split.
16086 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16087 address space. Merge with *add_tp_x32 using PTR mode iterator.
16088 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16089 Split to an add with a memory load from 0 in
16090 DEFAULT_TLS_SEG_REG address space.
16092 2017-08-12 Andrew Pinski <apinski@cavium.com>
16094 * config/aarch64/aarch64-option-extensions.def (rdma):
16095 Fix feature string to what Linux prints out in /proc/cpuinfo.
16097 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16100 * dwarf2out.c (modified_type_die): For C typedef types that have
16101 an ultimate origin, process the ultimate origin instead of the
16103 (gen_typedef_die): Assert that input DECLs have no ultimate
16105 (gen_type_die_with_usage): For typedef variants that have an
16106 ultimate origin, just call gen_decl_die on the original DECL.
16107 (process_scope_var): Avoid creating DIEs for local typedefs and
16108 concrete static variables.
16110 2017-08-12 Alan Modra <amodra@gmail.com>
16114 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16115 match gnu-user.h startfile.
16116 (ENDFILE_LINUX_SPEC): Similarly.
16118 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16121 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16123 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16125 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16126 * config/aarch64/aarch64.md (mov<mode>): Change.
16127 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16128 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16129 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16131 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16133 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16134 for storage order barriers.
16136 2017-08-11 Martin Liska <mliska@suse.cz>
16139 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16140 variables of void type.
16142 2017-08-11 Martin Liska <mliska@suse.cz>
16144 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16145 TARGET_SUPPORTS_ALIASES.
16146 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16147 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16148 (optimize_weakref): Likewise.
16149 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16150 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16151 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16153 2017-08-11 Martin Liska <mliska@suse.cz>
16156 * config/i386/i386.c (make_resolver_func): Do complete
16157 refactoring of the function.
16159 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16162 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16163 * config/i386/i386.c (ix86_stack_protect_guard): Use
16164 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16165 * doc/invoke.texi (x86 Options): Document
16166 -mstack-protector-guard-symbol= option.
16168 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16170 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16171 * config/i386/i386.c (ix86_split_stack_guard): New function.
16172 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16173 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16174 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16175 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16176 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16177 (split_stack_space_check): Call ix86_split_stack_guard.
16179 2017-08-10 Martin Sebor <msebor@redhat.com>
16181 * print-tree.c (print_node): Print location using the established
16183 Replace spaces with colons.
16184 (debug_raw, debug): Ditto.
16186 2017-08-10 Martin Sebor <msebor@redhat.com>
16189 * pretty-print.c (pp_format): Correct the handling of %s precision.
16191 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16194 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16196 (ix86_finalize_stack_frame_flags): This. Also clear
16197 frame_pointer_needed if -fno-omit-frame-pointer is used without
16199 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16200 with ix86_finalize_stack_frame_flags.
16201 (ix86_expand_epilogue): Likewise.
16202 (ix86_expand_split_stack_prologue): Likewise.
16203 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16205 2017-08-10 Martin Liska <mliska@suse.cz>
16208 * c-attribs.c (handle_target_attribute):
16209 Report warning for an empty string argument of target attribute.
16211 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16214 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16216 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16217 or DECL_NONLOCAL labels.
16218 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16219 or DECL_NONLOCAL labels here.
16221 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16224 to indicate when early gimple folding has been disabled.
16225 (rs6000_gimple_fold_builtin): Add debug content.
16226 (rs6000_invalid_builtin): Fix whitespace.
16227 (rs6000_expand_builtin): Fix whitespace.
16228 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16230 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16233 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16234 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16236 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16238 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16239 (qdf24xx): Likewise.
16240 * config/aarch64/aarch64.md: Include falkor.md.
16241 * config/aarch64/falkor.md: New.
16243 2017-08-09 Marek Polacek <polacek@redhat.com>
16246 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16247 * diagnostic.c (emit_diagnostic): Add a comment.
16248 (emit_diagnostic_valist): New function.
16250 2017-08-09 Marek Polacek <polacek@redhat.com>
16253 * input.c (make_location): New overload.
16254 * input.h (make_location): Declare.
16256 2017-08-08 Alan Modra <amodra@gmail.com>
16257 H.J. Lu <hongjiu.lu@intel.com>
16260 * gcc.c (NO_PIE_SPEC): Delete.
16261 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16263 (LINK_PIE_SPEC): ..to here.
16264 (LINK_COMMAND_SPEC): Support -no-pie.
16265 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16266 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16268 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16269 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16270 (ENDFILE_CRTEND_SPEC): Similarly.
16272 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16275 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16276 (mstack-protector-guard-offset=): Ditto.
16277 * config/i386/i386.c (ix86_option_override): Handle
16278 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16280 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16281 ix86_stack_protect_guard_offset variables.
16282 (TARGET_STACK_PROTECT_GUARD): Always define.
16283 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16284 and -mstack-protector-guard-offset= options.
16286 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16288 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16289 boundary case for the last candidate.
16291 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16293 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16294 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16296 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16298 PR middle-end/19706
16299 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16300 * config/aarch64/aarch64-builtins.c
16301 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16302 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16303 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16305 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16306 Andrew Pinski <pinskia@gmail.com>
16308 PR middle-end/19706
16309 * internal-fn.def (XORSIGN): New.
16310 * optabs.def (xorsign_optab): New.
16311 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16312 (convert_expand_mult_copysign): New.
16313 (pass_optimize_widening_mul::execute): Call
16314 convert_expand_mult_copysign.
16316 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16318 PR tree-optimization/81354
16319 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16320 Insert on edges rather than explicitly creating landing pads.
16321 (analyze_candidates_and_replace): Commit edge inserts.
16323 2017-08-08 Richard Biener <rguenther@suse.de>
16325 PR middle-end/81719
16326 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16327 (expand_simple_operations): Also look through ADDR_EXPRs with
16328 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16330 2017-08-08 Richard Biener <rguenther@suse.de>
16332 PR tree-optimization/81723
16333 * tree-vect-slp.c (struct bst_traits): New hash traits.
16334 (bst_fail): New global.
16335 (vect_build_slp_tree_2): New worker, split out from ...
16336 (vect_build_slp_tree): ... this now wrapping it with using
16337 bst_fail set to cache SLP tree build fails. Properly handle
16339 (vect_analyze_slp_instance): Allocate and free bst_fail.
16341 2017-08-08 Martin Liska <mliska@suse.cz>
16344 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16345 LABEL_DECLs that can be from a different function.
16347 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16349 PR tree-optimization/81744
16350 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16351 loop's number of iterations.
16353 2017-08-08 Martin Liska <mliska@suse.cz>
16355 * asan.c: Include header files.
16356 * attribs.c (build_decl_attribute_variant): New function moved
16358 (build_type_attribute_qual_variant): Likewise.
16359 (cmp_attrib_identifiers): Likewise.
16360 (simple_cst_list_equal): Likewise.
16361 (omp_declare_simd_clauses_equal): Likewise.
16362 (attribute_value_equal): Likewise.
16363 (comp_type_attributes): Likewise.
16364 (build_type_attribute_variant): Likewise.
16365 (lookup_ident_attribute): Likewise.
16366 (remove_attribute): Likewise.
16367 (merge_attributes): Likewise.
16368 (merge_type_attributes): Likewise.
16369 (merge_decl_attributes): Likewise.
16370 (merge_dllimport_decl_attributes): Likewise.
16371 (handle_dll_attribute): Likewise.
16372 (attribute_list_equal): Likewise.
16373 (attribute_list_contained): Likewise.
16374 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16375 (lookup_attribute_by_prefix): Likewise.
16376 * bb-reorder.c: Include header files.
16377 * builtins.c: Likewise.
16378 * calls.c: Likewise.
16379 * cfgexpand.c: Likewise.
16380 * cgraph.c: Likewise.
16381 * cgraphunit.c: Likewise.
16382 * convert.c: Likewise.
16383 * dwarf2out.c: Likewise.
16384 * final.c: Likewise.
16385 * fold-const.c: Likewise.
16386 * function.c: Likewise.
16387 * gimple-expr.c: Likewise.
16388 * gimple-fold.c: Likewise.
16389 * gimple-pretty-print.c: Likewise.
16390 * gimple.c: Likewise.
16391 * gimplify.c: Likewise.
16392 * hsa-common.c: Likewise.
16393 * hsa-gen.c: Likewise.
16394 * internal-fn.c: Likewise.
16395 * ipa-chkp.c: Likewise.
16396 * ipa-cp.c: Likewise.
16397 * ipa-devirt.c: Likewise.
16398 * ipa-fnsummary.c: Likewise.
16399 * ipa-inline.c: Likewise.
16400 * ipa-visibility.c: Likewise.
16402 * lto-cgraph.c: Likewise.
16403 * omp-expand.c: Likewise.
16404 * omp-general.c: Likewise.
16405 * omp-low.c: Likewise.
16406 * omp-offload.c: Likewise.
16407 * omp-simd-clone.c: Likewise.
16408 * opts-global.c: Likewise.
16409 * passes.c: Likewise.
16410 * predict.c: Likewise.
16411 * sancov.c: Likewise.
16412 * sanopt.c: Likewise.
16413 * symtab.c: Likewise.
16414 * toplev.c: Likewise.
16415 * trans-mem.c: Likewise.
16416 * tree-chkp.c: Likewise.
16417 * tree-eh.c: Likewise.
16418 * tree-into-ssa.c: Likewise.
16419 * tree-object-size.c: Likewise.
16420 * tree-parloops.c: Likewise.
16421 * tree-profile.c: Likewise.
16422 * tree-ssa-ccp.c: Likewise.
16423 * tree-ssa-live.c: Likewise.
16424 * tree-ssa-loop.c: Likewise.
16425 * tree-ssa-sccvn.c: Likewise.
16426 * tree-ssa-structalias.c: Likewise.
16427 * tree-ssa.c: Likewise.
16428 * tree-streamer-in.c: Likewise.
16429 * tree-vectorizer.c: Likewise.
16430 * tree-vrp.c: Likewise.
16431 * tsan.c: Likewise.
16432 * ubsan.c: Likewise.
16433 * varasm.c: Likewise.
16434 * varpool.c: Likewise.
16435 * tree.c: Remove functions moved to attribs.[ch].
16436 * tree.h: Likewise.
16437 * config/aarch64/aarch64.c: Add attrs.h header file.
16438 * config/alpha/alpha.c: Likewise.
16439 * config/arc/arc.c: Likewise.
16440 * config/arm/arm.c: Likewise.
16441 * config/avr/avr.c: Likewise.
16442 * config/bfin/bfin.c: Likewise.
16443 * config/c6x/c6x.c: Likewise.
16444 * config/cr16/cr16.c: Likewise.
16445 * config/cris/cris.c: Likewise.
16446 * config/darwin.c: Likewise.
16447 * config/epiphany/epiphany.c: Likewise.
16448 * config/fr30/fr30.c: Likewise.
16449 * config/frv/frv.c: Likewise.
16450 * config/ft32/ft32.c: Likewise.
16451 * config/h8300/h8300.c: Likewise.
16452 * config/i386/winnt.c: Likewise.
16453 * config/ia64/ia64.c: Likewise.
16454 * config/iq2000/iq2000.c: Likewise.
16455 * config/lm32/lm32.c: Likewise.
16456 * config/m32c/m32c.c: Likewise.
16457 * config/m32r/m32r.c: Likewise.
16458 * config/m68k/m68k.c: Likewise.
16459 * config/mcore/mcore.c: Likewise.
16460 * config/microblaze/microblaze.c: Likewise.
16461 * config/mips/mips.c: Likewise.
16462 * config/mmix/mmix.c: Likewise.
16463 * config/mn10300/mn10300.c: Likewise.
16464 * config/moxie/moxie.c: Likewise.
16465 * config/msp430/msp430.c: Likewise.
16466 * config/nds32/nds32-isr.c: Likewise.
16467 * config/nds32/nds32.c: Likewise.
16468 * config/nios2/nios2.c: Likewise.
16469 * config/nvptx/nvptx.c: Likewise.
16470 * config/pa/pa.c: Likewise.
16471 * config/pdp11/pdp11.c: Likewise.
16472 * config/powerpcspe/powerpcspe.c: Likewise.
16473 * config/riscv/riscv.c: Likewise.
16474 * config/rl78/rl78.c: Likewise.
16475 * config/rx/rx.c: Likewise.
16476 * config/s390/s390.c: Likewise.
16477 * config/sh/sh.c: Likewise.
16478 * config/sol2.c: Likewise.
16479 * config/sparc/sparc.c: Likewise.
16480 * config/spu/spu.c: Likewise.
16481 * config/stormy16/stormy16.c: Likewise.
16482 * config/tilegx/tilegx.c: Likewise.
16483 * config/tilepro/tilepro.c: Likewise.
16484 * config/v850/v850.c: Likewise.
16485 * config/vax/vax.c: Likewise.
16486 * config/visium/visium.c: Likewise.
16487 * config/xtensa/xtensa.c: Likewise.
16489 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16492 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16493 constraints since the -mupper-regs-* switches have been
16495 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16496 into a vector from a double word element that was extracted from
16497 another vector, and eliminate extra XXPERMDI instructions.
16498 (vsx_concat_<mode>_2): Likewise.
16499 (vsx_concat_<mode>_3): Likewise.
16500 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16501 concat to allow optimizing inserts from previous extracts.
16503 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16505 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16506 memory reference to a SSP offset in TLS address space.
16507 (ix86_print_operand) <case '@'>: Remove.
16508 (ix86_print_operand_punct_valid_p): Remove '@' code.
16509 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16510 UNSPEC_SP_TLS_TEST.
16511 (stack_tls_protect_set_<mode>): Remove.
16512 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16513 (stack_tls_protect_test_<mode>): Remove.
16514 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16516 2017-08-07 Olivier Hainque <hainque@adacore.com>
16519 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16521 2017-08-07 Douglas Rupp <rupp@adacore.com>
16523 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16524 variable was referenced as multidir in command.
16526 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16529 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16532 2017-08-07 Martin Liska <mliska@suse.cz>
16534 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16535 * config/rl78/rl78.c: Add include of attribs.h.
16536 * config/sh/sh.c: Likewise.
16537 * config/v850/v850.c: Likewise.
16539 2017-08-07 Tom de Vries <tom@codesourcery.com>
16541 PR middle-end/78266
16542 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16544 2017-08-07 Martin Liska <mliska@suse.cz>
16546 * config/mips/mips.c: Include attribs.h.
16548 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16551 * doc/invoke.texi: Document change in behvaior for -Ofast for
16554 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16556 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16558 (aarch64_pop_regs): Likewise.
16559 (aarch64_gen_load_pair): Likewise.
16560 (aarch64_save_callee_saves): Likewise.
16561 (aarch64_restore_callee_saves): Likewise.
16563 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16565 * config/i386/i386.c: Revert the last change.
16567 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16570 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16572 (ix86_finalize_stack_frame_flags): This. Also clear
16573 frame_pointer_needed if -fno-omit-frame-pointer is used without
16575 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16576 with ix86_finalize_stack_frame_flags.
16577 (ix86_expand_epilogue): Likewise.
16578 (ix86_expand_split_stack_prologue): Likewise.
16580 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16583 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16584 to P_AES for Westmere.
16586 2017-08-07 Jonathan Yong <10walls@gmail.com>
16588 * config/i386/mingw.opt (fset-stack-executable): Removed.
16589 * config/i386/cygming.opt (fset-stack-executable): Moved
16591 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16593 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16595 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16597 2017-08-07 Marek Polacek <polacek@redhat.com>
16599 PR middle-end/81737
16600 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16602 2017-08-07 Martin Liska <mliska@suse.cz>
16604 * attribs.h (canonicalize_attr_name): New function.
16605 (cmp_attribs): Move from c-format.c and adjusted.
16606 (is_attribute_p): Moved from tree.h.
16607 * tree-inline.c: Add new includes.
16608 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16609 (private_is_attribute_p): Remove.
16610 (private_lookup_attribute): Likewise.
16611 (private_lookup_attribute_by_prefix): Simplify.
16612 (remove_attribute): Use is_attribute_p.
16613 * tree.h: Remove removed declarations.
16615 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16617 PR middle-end/81698
16618 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16619 instead of computing it in the function. Formatting fix.
16620 (expand_case): Don't rely on default_edge being the first edge,
16621 clear it if removing it, pass default_edge to
16622 emit_case_dispatch_table.
16623 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16626 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16628 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16629 insn in the function, emit NOP after the insn.
16631 2017-08-06 Tom de Vries <tom@codesourcery.com>
16633 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16636 2017-08-06 Tom de Vries <tom@codesourcery.com>
16638 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16641 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16643 PR tree-optimization/57371
16644 * match.pd: New pattern.
16646 2017-08-04 Marek Polacek <polacek@redhat.com>
16648 PR middle-end/81695
16649 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16650 perform the computation in offset_int.
16652 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16654 PR tree-optimization/81136
16655 * tree-vectorizer.h: Include tree-hash-traits.h.
16656 (vec_base_alignments): New typedef.
16657 (vec_info): Add a base_alignments field.
16658 (vect_record_base_alignments): Declare.
16659 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16661 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16662 (create_data_ref): Add an is_conditional_in_stmt argument.
16663 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16664 the is_conditional_in_stmt field.
16665 (data_ref_loc): Add an is_conditional_in_stmt field.
16666 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16667 (find_data_references_in_stmt): Update call to create_data_ref.
16668 (graphite_find_data_references_in_stmt): Likewise.
16669 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16670 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16671 (vect_record_base_alignment): New function.
16672 (vect_record_base_alignments): Likewise.
16673 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16674 for nested statements even if we fail to compute a misalignment.
16675 Use pooled base alignments for unconditional references.
16676 (vect_find_same_alignment_drs): Compare base addresses instead
16678 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16679 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16681 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16683 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16684 Add an explicit name for the enum. Use auto_vec for slp_instances
16685 and grouped_stores.
16686 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16688 (_bb_vec_info): Add a constructor and destructor.
16689 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16690 (destroy_loop_vec_info): Delete.
16691 (vect_destroy_datarefs): Likewise.
16692 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16693 (vec_info::vec_info): New function.
16694 (vec_info::~vec_info): Likewise.
16695 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16696 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
16697 destroy_loop_vec_info.
16698 * tree-vect-loop.c (new_loop_vec_info): Replace with...
16699 (_loop_vec_info::_loop_vec_info): ...this.
16700 (destroy_loop_vec_info): Replace with...
16701 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
16702 the stmt_vec_infos. Leave handling of vec_info information to its
16703 destructor. Remove explicit vector releases.
16704 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
16705 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
16706 * tree-vect-slp.c (new_bb_vec_info): Replace with...
16707 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
16708 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
16709 (destroy_bb_vec_info): Replace with...
16710 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
16711 information to its destructor.
16712 (vect_slp_analyze_bb_1): Use new and delete instead of
16713 new_bb_vec_info and destroy_bb_vec_info.
16714 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
16717 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16719 * tree-data-ref.h (subscript): Add access_fn field.
16720 (data_dependence_relation): Add could_be_independent_p.
16721 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
16722 (same_access_functions): Move to tree-data-ref.c.
16723 * tree-data-ref.c (ref_contains_union_access_p): New function.
16724 (access_fn_component_p): Likewise.
16725 (access_fn_components_comparable_p): Likewise.
16726 (dr_analyze_indices): Add a reference to access_fn_component_p.
16727 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
16729 (constant_access_functions): Likewise.
16730 (add_other_self_distances): Likewise.
16731 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
16732 (initialize_data_dependence_relation): Use XCNEW and remove
16733 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
16734 of access functions that have the same type. Allow the
16735 subsequence to end with different bases in some circumstances.
16736 Record the chosen access functions in SUB_ACCESS_FN.
16737 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
16738 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
16739 (subscript_dependence_tester_1): Likewise dra and drb.
16740 (build_classic_dist_vector): Update calls accordingly.
16741 (subscript_dependence_tester): Likewise.
16742 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
16743 DDR_COULD_BE_INDEPENDENT_P.
16744 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
16745 comp_alias_ddrs instead of may_alias_ddrs.
16746 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16748 (vect_analyze_data_ref_dependence): Use it if
16749 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
16750 distance vectors if that fails.
16751 (dependence_distance_ge_vf): New function.
16752 (vect_prune_runtime_alias_test_list): Use it. Don't clear
16753 LOOP_VINFO_MAY_ALIAS_DDRS.
16755 2017-08-04 Richard Biener <rguenther@suse.de>
16757 PR middle-end/81705
16758 * fold-const.c (fold_binary_loc): Properly restrict
16759 minus_var0 && minus_var1 case when associating undefined overflow
16762 2017-08-04 Tom de Vries <tom@codesourcery.com>
16764 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
16766 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16769 Don't start diagnostic messages with a capital letter.
16770 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16772 (rs6000_invalid_builtin): Likewise.
16773 (rs6000_trampoline_init): Likewise.
16775 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16778 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
16779 after setting changeable df flags.
16781 2017-08-03 Richard Biener <rguenther@suse.de>
16783 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
16784 up if the use is in USE - X.
16786 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
16788 * toplev.c (dumpfile.h): New include.
16789 (internal_error_reentered): New static function. Use it...
16790 (internal_error_function): ...here to handle reentered internal_error.
16792 2017-08-03 Richard Biener <rguenther@suse.de>
16794 PR middle-end/81148
16795 * fold-const.c (split_tree): Add minus_var and minus_con
16796 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
16797 here but always use minus_*.
16798 (associate_trees): Assert we never associate with MINUS_EXPR
16799 and NULL first operand. Do not recurse for PLUS_EXPR operands
16800 when associating as MINUS_EXPR either.
16801 (fold_binary_loc): Track minus_var and minus_con.
16803 2017-08-03 Tom de Vries <tom@codesourcery.com>
16806 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
16807 ACCEL_COMPILER, apply finish_options on
16808 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
16810 2017-08-03 Tom de Vries <tom@codesourcery.com>
16813 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
16814 function_entry_patch_area_size > 0.
16816 2017-08-03 Jakub Jelinek <jakub@redhat.com>
16819 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
16820 instead of 10??LU, perform unit multiplication in wide_int,
16821 don't change alloc_object_size_limit if the limit is larger
16824 PR tree-optimization/81655
16825 PR tree-optimization/81588
16826 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
16827 the case when ranges[i].low and high are 1 for unsigned type with
16830 PR middle-end/81052
16831 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
16832 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
16834 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16836 * tree-vrp.h: Add include guard.
16838 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
16841 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
16842 (ud2): New insn pattern.
16843 * config/i386/i386.c (ix86_expand_epilogue):
16844 For naked functions, generate ud2 instead of trap insn.
16846 2017-08-02 Marek Polacek <polacek@redhat.com>
16849 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
16851 2017-08-02 Tom de Vries <tom@codesourcery.com>
16852 Cesar Philippidis <cesar@codesourcery.com>
16854 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
16855 Add missing edge probabilities.
16857 2017-08-02 Tamar Christina <tamar.christina@arm.com>
16859 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16860 Correct endianness.
16862 2017-08-02 Jakub Jelinek <jakub@redhat.com>
16864 PR middle-end/79499
16865 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
16866 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
16867 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
16868 prologue_seq sequences - if any.
16870 2017-08-02 Richard Biener <rguenther@suse.de>
16872 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
16873 via vectors if supported, integer extracts via punning if supported
16874 or otherwise vector extracts.
16876 2017-08-02 Richard Biener <rguenther@suse.de>
16878 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
16880 (bitmap_insert_into_set): ... this.
16882 2017-08-02 Richard Biener <rguenther@suse.de>
16884 PR tree-optimization/81633
16886 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16888 PR tree-optimization/71752
16889 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16891 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16893 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16894 (machine_function::call_ms2sysv_pad_out): Remove field.
16895 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16896 (ix86_compute_frame_layout): Likewise.
16898 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16901 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16902 attribute with interrupt attribute.
16904 2017-08-01 Andrew Pinski <apinski@cavium.com>
16906 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16907 BIT_INSERT_EXPR's operand 1
16908 to see if the types precision matches.
16910 2017-08-01 Martin Liska <mliska@suse.cz>
16912 PR middle-end/70140
16913 * builtins.c (expand_builtin_memcpy_args): Remove.
16914 (expand_builtin_memcpy): Call newly added function
16915 expand_builtin_memory_copy_args.
16916 (expand_builtin_memcpy_with_bounds): Likewise.
16917 (expand_builtin_mempcpy): Remove last argument.
16918 (expand_builtin_mempcpy_with_bounds): Likewise.
16919 (expand_builtin_memory_copy_args): New function created from
16920 expand_builtin_mempcpy_args with small modifications.
16921 (expand_builtin_mempcpy_args): Remove.
16922 (expand_builtin_stpcpy): Remove unused argument.
16923 (expand_builtin): Likewise.
16924 (expand_builtin_with_bounds): Likewise.
16926 2017-08-01 Martin Liska <mliska@suse.cz>
16929 Make mempcpy more optimal (PR middle-end/70140).
16931 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16934 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16935 __builtin_vec_cmpne verify both arguments are compatible vectors
16936 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16937 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16938 move computation of aligned to after checking the argument types.
16942 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16943 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16946 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16947 Doug Rupp <rupp@adacore.com>
16948 Olivier Hainque <hainque@adacore.com>
16950 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16951 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16953 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16954 for TARGET_OS_CPP_BUILTIN.
16955 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16956 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16958 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16959 passing required abi options to the assembler for EABI configurations.
16960 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16961 of .text.hot and .text.unlikely sections for kernel modules when
16962 using ARM style exceptions.
16963 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16964 options. Add EXTRA_CC1_SPEC.
16965 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16967 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16969 (ARM_TARGET2_DWARF_FORMAT): Define.
16970 * config/arm/t-vxworks: Adjust multilib control to removal of the
16971 Diab command line options.
16973 2017-08-01 Martin Liska <mliska@suse.cz>
16975 PR gcov-profile/81561
16976 * gcov.c (unblock): Make unblocking safe as we need to preserve
16977 index correspondence of blocks and block_lists.
16979 2017-08-01 Richard Biener <rguenther@suse.de>
16981 PR tree-optimization/81181
16982 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16983 (compute_antic): ... end of iteration here.
16985 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16987 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16988 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16989 (ftree-slp-vectorize): Likewise.
16990 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16991 can no longer be set independent of flag_tree_loop_vectorize.
16992 * omp-general.c (emp_max_vf): Likewise.
16993 * opts.c (enable_fdo_optimizations): Remove references to
16994 flag_tree_vectorize, these are now implicit.
16995 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16996 and leave it for the options machinery.
16998 2017-08-01 Martin Liska <mliska@suse.cz>
17000 PR middle-end/70140
17001 * builtins.c (expand_builtin_memcpy_args): Remove.
17002 (expand_builtin_memcpy): Call newly added function
17003 expand_builtin_memory_copy_args.
17004 (expand_builtin_memcpy_with_bounds): Likewise.
17005 (expand_builtin_mempcpy): Remove last argument.
17006 (expand_builtin_mempcpy_with_bounds): Likewise.
17007 (expand_builtin_memory_copy_args): New function created from
17008 expand_builtin_mempcpy_args with small modifications.
17009 (expand_builtin_mempcpy_args): Remove.
17010 (expand_builtin_stpcpy): Remove unused argument.
17011 (expand_builtin): Likewise.
17012 (expand_builtin_with_bounds): Likewise.
17014 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17017 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17018 print "ds:" only for immediates in generic address space.
17020 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17023 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17024 (ix86_function_ok_for_sibcall): Return false for naked functions.
17026 2017-08-01 Richard Biener <rguenther@suse.de>
17028 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17029 (compute_antic): Seed worklist with exit block predecessors.
17030 * cfganal.c (dfs_find_deadend): For a cycle return the source
17031 of the edge closing it.
17033 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17035 * config/aarch64/aarch64.c
17036 (aarch64_can_const_movi_rtx_p): Move 0 check.
17038 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17040 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17041 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17043 * match.pd: Ditto in address comparison pattern.
17045 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17047 PR tree-optimization/81627
17048 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17049 closed ssa form for store-store chain.
17051 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17053 PR tree-optimization/81620
17054 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17055 for store-store chain.
17057 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17059 PR tree-optimization/81588
17060 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17061 ranges[i].in_p, invert comparison code ccode. For >/>=,
17062 swap rhs1 and rhs2 and comparison code unconditionally,
17063 for </<= don't do that. Don't swap rhs1/rhs2 again if
17064 ranges[i].in_p, instead invert comparison code ccode if
17065 opcode or oe->rank is BIT_IOR_EXPR.
17068 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17069 a direct optab to conversion optab.
17070 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17071 with GET_MODE_INNER as last argument instead of optab_handler.
17072 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17073 vector extraction if possible and optab is available.
17074 * expr.c (store_constructor): Use convert_optab_handler instead
17075 of optab_handler. Use vector initialization from smaller
17076 vectors if possible and optab is available.
17077 * tree-vect-stmts.c (vectorizable_load): Likewise.
17078 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17079 now have two modes.
17080 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17081 of vec_init from half-sized vectors with the same element mode.
17082 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17083 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17084 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17085 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17086 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17087 after mode in gen_vec_extract* calls.
17088 (vec_extract<mode>): Renamed to ...
17089 (vec_extract<mode><ssescalarmodelower>): ... this.
17090 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17091 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17092 element mode after mode in gen_vec_init* calls.
17093 (VEC_INIT_HALF_MODE): New mode iterator.
17094 (vec_init<mode>): Renamed to ...
17095 (vec_init<mode><ssescalarmodelower>): ... this.
17096 (vec_init<mode><ssehalfvecmodelower>): New expander.
17097 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17098 (vec_extractv2sfsf): ... this.
17099 (vec_initv2sf): Renamed to ...
17100 (vec_initv2sfsf): ... this.
17101 (vec_extractv2si): Renamed to ...
17102 (vec_extractv2sisi): ... this.
17103 (vec_initv2si): Renamed to ...
17104 (vec_initv2sisi): ... this.
17105 (vec_extractv4hi): Renamed to ...
17106 (vec_extractv4hihi): ... this.
17107 (vec_initv4hi): Renamed to ...
17108 (vec_initv4hihi): ... this.
17109 (vec_extractv8qi): Renamed to ...
17110 (vec_extractv8qiqi): ... this.
17111 (vec_initv8qi): Renamed to ...
17112 (vec_initv8qiqi): ... this.
17113 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17114 (vec_init<mode>): Renamed to ...
17115 (vec_init<mode><VEC_base_l>): ... this.
17116 (vec_extract<mode>): Renamed to ...
17117 (vec_extract<mode><VEC_base_l>): ... this.
17118 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17119 (vec_initv2sfsf): ... this.
17120 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17121 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17122 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17123 element mode after mode in gen_vec_init* calls.
17124 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17125 (vec_init<mode><Vel>): ... this.
17126 (vec_extract<mode>): Renamed to ...
17127 (vec_extract<mode><Vel>): ... this.
17128 * config/aarch64/iterators.md (Vel): New mode attribute.
17129 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17130 Add element mode after mode in gen_vec_extract* calls.
17131 * config/s390/vector.md (non_vec_l): New mode attribute.
17132 (vec_extract<mode>): Renamed to ...
17133 (vec_extract<mode><non_vec_l>): ... this.
17134 (vec_init<mode>): Renamed to ...
17135 (vec_init<mode><non_vec_l>): ... this.
17136 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17137 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17139 * config/arm/iterators.md (V_elem_l): New mode attribute.
17140 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17141 (vec_extract<mode><V_elem_l>): ... this.
17142 (vec_extractv2di): Renamed to ...
17143 (vec_extractv2didi): ... this.
17144 (vec_init<mode>): Renamed to ...
17145 (vec_init<mode><V_elem_l>): ... this.
17146 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17147 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17148 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17149 Add element mode after gen_vec_extract* calls.
17150 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17151 (vec_init<mode><unitmode>): ... this.
17152 (vec_extract<mode>): Renamed to ...
17153 (vec_extract<mode><unitmode>): ... this.
17154 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17155 (vec_init<mode><unitmode>): ... this.
17156 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17157 (vec_initv2sfsf): ... this.
17158 (vec_extractv2sf): Renamed to ...
17159 (vec_extractv2sfsf): ... this.
17160 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17161 Add element mode after gen_vec_extract* calls.
17162 * config/mips/mips.md (unitmode): New mode iterator.
17163 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17164 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17165 * config/spu/spu.md (inner_l): New mode attribute.
17166 (vec_init<mode>): Renamed to ...
17167 (vec_init<mode><inner_l>): ... this.
17168 (vec_extract<mode>): Renamed to ...
17169 (vec_extract<mode><inner_l>): ... this.
17170 * config/sparc/sparc.md (veltmode): New mode iterator.
17171 (vec_init<VMALL:mode>): Renamed to ...
17172 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17173 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17174 (vec_initv2sisi): ... this.
17175 (vec_initv2sf): Renamed to ...
17176 (vec_initv2sfsf): ... this.
17177 (vec_extractv2sf): Renamed to ...
17178 (vec_extractv2sfsf): ... this.
17179 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17180 (vec_init<mode>): Renamed to ...
17181 (vec_init<mode><VEC_base_l>): ... this.
17182 (vec_extract<mode>): Renamed to ...
17183 (vec_extract<mode><VEC_base_l>): ... this.
17184 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17185 (vec_initv2sfsf): ... this.
17186 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17187 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17188 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17189 gen_vec_init* calls.
17191 2017-08-01 Richard Biener <rguenther@suse.de>
17193 PR tree-optimization/81297
17194 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17195 TREE_OVERFLOW from INTEGER_CSTs.
17197 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17199 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17201 2017-07-31 Carl Love <cel@us.ibm.com>
17203 * config/rs6000/rs6000-c: Add support for built-in functions
17204 vector signed char vec_xl_be (signed long long, signed char *);
17205 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17206 vector signed int vec_xl_be (signed long long, signed int *);
17207 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17208 vector signed long long vec_xl_be (signed long long, signed long long *);
17209 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17210 vector signed short vec_xl_be (signed long long, signed short *);
17211 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17212 vector double vec_xl_be (signed long long, double *);
17213 vector float vec_xl_be (signed long long, float *);
17214 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17215 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17216 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17218 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17219 (altivec_expand_builtin): Add switch statement to call
17220 altivec_expand_xl_be for each builtin.
17221 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17222 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17223 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17224 __builtin_vsx_le_be_v16qi.
17225 * doc/extend.texi: Update the built-in documentation file for the
17226 new built-in functions.
17228 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17231 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17233 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17235 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17237 * config.gcc: Add z14.
17238 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17239 CPU model numbers for z13s and z14.
17240 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17242 * config/s390/s390-opts.h (enum processor_type): Rename
17243 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17244 * config/s390/s390.c (processor_table): Add field for CPU name to
17245 be passed to Binutils.
17246 (s390_asm_output_machine_for_arch): Use the new field in
17247 processor_table for Binutils.
17248 (s390_expand_builtin): Replace arch12 with z14.
17249 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17250 (s390_get_sched_attrmask): Likewise.
17251 (s390_get_unit_mask): Likewise.
17252 * config/s390/s390.opt: Add z14 to processor_type enum.
17254 2017-07-31 Martin Jambor <mjambor@suse.cz>
17257 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17258 regardless of optimization level.
17260 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17261 Martin Liska <mliska@suse.cz>
17263 * predict.def: Remove old comment and adjust probability.
17264 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17265 PREDICT statements.
17267 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17270 * config/i386/i386.c (ix86_function_naked): New function.
17271 (ix86_can_use_return_insn_p): Return false for naked functions.
17272 (ix86_expand_prologue): Skip prologue for naked functions.
17273 (ix86_expand_epilogue): Skip epilogue for naked functions
17274 and emit trap instruction.
17275 (ix86_warn_func_return): New function.
17276 (ix86_attribute_table): Add "naked" attribute specification.
17277 (TARGET_WARN_FUNC_RETURN): Define.
17278 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17280 2017-07-31 Martin Liska <mliska@suse.cz>
17282 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17283 (dump_gimple_bb_header): Always dump BB info.
17284 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17286 2017-07-31 Martin Liska <mliska@suse.cz>
17289 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17290 also with current_function_decl non-null equality.
17292 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17295 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17296 change type to the element type, instead add eltype variable and
17297 use it where we are interested in the element type.
17299 PR tree-optimization/81603
17300 * ipa-polymorphic-call.c
17301 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17302 offset arithmetic in offset_int, bail out if the resulting bit offset
17303 doesn't fit into shwi.
17305 2017-07-31 Martin Liska <mliska@suse.cz>
17307 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17308 (gimplify_save_expr): Fix comment.
17310 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17313 * config/i386/i386.c (ix86_function_arg): Update arguments for
17315 (ix86_compute_frame_layout): Set the initial stack offset to
17316 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17317 INCOMING_FRAME_SP_OFFSET.
17318 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17319 stack before exception handler returns.
17320 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17321 the 'ERROR_CODE' for exception handler.
17323 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17325 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17326 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17327 (ASM_OUTPUT_REG_POP): Ditto.
17328 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17329 instead of asm_fprintf to output pure string.
17331 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17333 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17334 to imported_module_or_decl hook.
17335 (debug_nothing_tree_tree_tree_bool): Remove.
17336 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17337 * debug.c (do_nothing_debug_hooks): Use
17338 debug_nothing_tree_tree_tree_bool_bool instead of
17339 debug_nothing_tree_tree_tree_bool.
17340 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17341 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17342 * sdbout.c (sdb_debug_hooks): Likewise.
17343 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17344 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17346 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17347 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17348 attribute, don't add anything.
17350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17352 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17353 (fold_build2_stat_loc): Likewise.
17354 (fold_build3_stat_loc): Likewise.
17355 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17356 (fold_build1_loc): Remove macro.
17357 (fold_build2_loc): Likewise.
17358 (fold_build3_loc): Likewise.
17360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17362 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17363 (gimple_build_debug_bind_source_stat): Likewise.
17364 * gimple.h (gimple_build_debug_bind): Remove macro.
17365 (gimple_build_debug_bind_source): Likewise.
17367 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17369 * bitmap.c (bitmap_alloc): Adjust.
17370 (bitmap_gc_alloc): Likewise.
17371 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17373 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17375 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17376 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17377 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17378 (bitmap_gc_alloc_stat): Likewise.
17379 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17381 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17383 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17384 * rtl.h (shallow_copy_rtx): Remove macro.
17386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17388 * emit-rtl.c (gen_raw_REG): Adjust.
17389 * gengenrtl.c (gendef): Likewise.
17390 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17391 * rtl.h (rtx_alloc): Remove macro.
17393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17395 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17396 (build_tree_list_stat): Likewise.
17397 * tree.h (build_tree_list): Remove macro.
17398 (build_tree_list_vec): Likewise.
17400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17402 * tree.c (make_vector_stat): Remove _stat from name.
17403 (build_vector_stat): Likewise.
17404 * tree.h (make_vector_stat): Remove macro.
17405 (build_vector_stat): Likewise.
17407 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17409 * tree.h (build_var_debug_value): Remove prototype.
17411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17413 * tree.c (tree_cons_stat): Remove _stat from name.
17414 * tree.h (tree_cons): Remove macro.
17416 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17418 * tree.c (build_vl_exp_stat): Remove _stat from name.
17419 * tree.h (build_vl_exp): Remove macro.
17421 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17423 * tree.c (build_decl_stat): Remove _stat from name.
17424 * tree.h (build_decl): Remove macro.
17426 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17428 * gimple.c (gimple_build_with_ops_stat): Adjust.
17429 (gimple_alloc_stat): Remove _stat from name.
17430 * gimple.h (gimple_alloc): Remove macro.
17432 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17434 * tree.c (make_tree_vec_stat): Remove _stat from name.
17435 (grow_tree_vec_stat): Likewise.
17436 * tree.h (make_tree_vec_stat): Adjust prototype.
17437 (grow_tree_vec_stat): Likewise.
17438 (make_tree_vec): Remove macro.
17439 (grow_tree_vec): Likewise.
17441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17443 * fold-const.c (fold_build1_stat_loc): Adjust.
17444 (fold_build2_stat_loc): Likewise.
17445 (fold_build3_stat_loc): Likewise.
17446 * tree.c (build0_stat): Remove _stat from name.
17447 (build1_stat): Likewise.
17448 (build2_stat): Likewise.
17449 (build3_stat): Likewise.
17450 (build4_stat): Likewise.
17451 (build5_stat): Likewise.
17452 * tree.h (build1_loc): Remove macro, and rename _stat function
17454 (build2_loc): Likewise.
17455 (build3_loc): Likewise.
17456 (build4_loc): Likewise.
17457 (build5_loc): Likewise.
17459 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17461 * tree.c (make_int_cst_stat): Remove _stat from name.
17462 * tree.h (make_int_cst_stat): Adjust prototype.
17463 (make_int_cst): Remove macro.
17465 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17467 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17468 * tree.h (make_tree_binfo_stat): Adjust prototype.
17469 (make_tree_binfo): Remove.
17471 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17473 * tree.c (copy_node_stat): Rename to copy_node.
17474 (build_distinct_type_copy): Adjust.
17475 * tree.h (copy_node_stat): Adjust prototype.
17476 (copy_node): Remove macro.
17478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17480 * tree.c (make_node_stat): rename to make_node.
17481 (build_tree_list_stat): Adjust.
17482 (build0_stat): Likewise.
17483 (build2_stat): Likewise.
17484 (build3_stat): Likewise.
17485 (build4_stat): Likewise.
17486 (build5_stat): Likewise.
17487 (build_decl_stat): Likewise.
17488 * tree.h (make_node_stat): Adjust prototype.
17489 (make_node): remove macro.
17491 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17493 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17494 (PPC_FEATURE2_SCV): Likewise.
17495 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17497 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17499 * config/aarch64/aarch64.c
17500 (aarch64_internal_mov_immediate): Add new special pattern.
17501 * config/aarch64/aarch64.md (*movdi_aarch64):
17502 Add reg/32bit const mov case.
17504 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17505 Richard Sandiford <richard.sandiford@linaro.org>
17507 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17508 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17509 Add integer and movi cases.
17510 (movi-split-hf-df-sf split, fp16): New.
17511 (enabled): Added TARGET_FP_F16INST.
17512 * config/aarch64/iterators.md (GPF_HF): New.
17513 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17515 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17517 * config/aarch64/aarch64.c
17518 (aarch64_simd_container_mode): Add prototype.
17519 (aarch64_expand_mov_immediate): Add HI support.
17520 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17521 (aarch64_can_const_movi_rtx_p): New.
17522 (aarch64_preferred_reload_class):
17523 Remove restrictions of using FP registers for certain SIMD operations.
17524 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17525 (aarch64_valid_floating_const): Add integer move validation.
17526 (aarch64_simd_imm_scalar_p): Remove.
17527 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17528 (aarch64_legitimate_constant_p): Expand list of supported cases.
17529 * config/aarch64/aarch64-protos.h
17530 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17531 (aarch64_reinterpret_float_as_int): New.
17532 (aarch64_simd_imm_scalar_p): Remove.
17533 * config/aarch64/constraints.md (Uvi): New.
17534 (Dd): Split into Ds and new Dd.
17535 * config/aarch64/aarch64.md (*movsi_aarch64):
17537 (*movdi_aarch64): Add SIMD mov case.
17539 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17541 * tree-predcom.c: (struct chain): Handle store-store chain in which
17542 stores for elimination only store loop invariant values.
17543 (execute_pred_commoning_chain): Ditto.
17544 (prepare_initializers_chain_store_elim): Ditto.
17545 (prepare_finalizers): Ditto.
17546 (is_inv_store_elimination_chain): New function.
17547 (initialize_root_vars_store_elim_1): New function.
17549 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17551 * tree-predcom.c: Revise general description of the pass.
17552 (enum chain_type): New enum type for store elimination.
17553 (struct chain): New field supporting store elimination.
17554 (struct component): Ditto.
17555 (dump_chain): Dump store-stores chain.
17556 (release_chain): Release resources.
17557 (split_data_refs_to_components): Compute and create component
17558 contains only stores for elimination.
17559 (get_chain_last_ref_at): New function.
17560 (make_invariant_chain): Initialization.
17561 (make_rooted_chain): Specify chain type in parameter and record it.
17562 (add_looparound_copies): Skip for store-stores chain.
17563 (determine_roots_comp): Compute type of chain and pass it to
17565 (initialize_root_vars_store_elim_2): New function.
17566 (finalize_eliminated_stores): New function.
17567 (remove_stmt): Handle store for elimination.
17568 (execute_pred_commoning_chain): Execute predictive commoning on
17569 store-store chains.
17570 (determine_unroll_factor): Skip unroll for store-stores chain.
17571 (prepare_initializers_chain_store_elim): New function.
17572 (prepare_initializers_chain): Hanlde store-store chain.
17573 (prepare_finalizers_chain, prepare_finalizers): New function.
17574 (tree_predictive_commoning_loop): Return integer value indicating
17575 if loop is unrolled or lcssa form is corrupted.
17576 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17578 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17580 * tree-predcom.c (initialize_root): Delete.
17581 (execute_pred_commoning_chain): Initialize root vars and replace
17582 reference of non-combined chain directly, rather than call above
17585 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17587 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17588 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17590 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17592 * tree-predcom.c (struct chain): New field init_seq.
17593 (release_chain): Release init_seq.
17594 (prepare_initializers_chain): Record intialization stmts in above
17596 (insert_init_seqs): New function.
17597 (tree_predictive_commoning_loop): Call insert_init_seqs.
17599 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17601 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17603 2017-07-28 Richard Biener <rguenther@suse.de>
17605 * match.pd: Remove superfluous :c.
17606 * genmatch.c (simplify::id): Add member.
17607 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17609 (current_id): New global.
17610 (dt_node::parent): Move from ...
17611 (dt_operand::parent): ... here. Add for_id member.
17612 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17613 (decision_tree::find_node): Relax order requirement when
17614 merging DT_TRUE nodes to ones inbetween the current simplify
17615 and the one we try to merge with. Add diagnostic whenever
17616 we need to enforce pattern order by not merging.
17617 (decision_tree::insert): Set current_id.
17618 (decision_tree::print_node): Dump parent node and for_id.
17619 (parser::last_id): Add member.
17620 (parser::push_simplify): Assign unique id.
17621 (parser::parser): Initialize last_id.
17623 2017-07-28 Martin Liska <mliska@suse.cz>
17626 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17627 gimple_build_debug_bind.
17629 2017-07-28 Richard Biener <rguenther@suse.de>
17631 PR tree-optimization/81502
17632 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17634 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17636 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17637 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17638 for BIT_FIELD_REF args.
17639 * fold-const.c (make_bit_field_ref): Likewise.
17640 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17642 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17645 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17646 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17647 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17648 Or it into SANITIZER_UNDEFINED.
17649 * ubsan.c: Include gimple-fold.h and varasm.h.
17650 (ubsan_expand_ptr_ifn): New function.
17651 (instrument_pointer_overflow): New function.
17652 (maybe_instrument_pointer_overflow): New function.
17653 (instrument_object_size): Formatting fix.
17654 (pass_ubsan::execute): Call instrument_pointer_overflow
17655 and maybe_instrument_pointer_overflow.
17656 * internal-fn.c (expand_UBSAN_PTR): New function.
17657 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17658 * sanitizer.def (__ubsan_handle_pointer_overflow,
17659 __ubsan_handle_pointer_overflow_abort): New builtins.
17660 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17661 * internal-fn.def (UBSAN_PTR): New internal function.
17662 * opts.c (sanitizer_opts): Add pointer-overflow.
17663 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17664 * fold-const.c (build_range_check): Compute pointer range check in
17665 integral type if pointer arithmetics would be needed. Formatting
17668 2017-07-28 Martin Liska <mliska@suse.cz>
17671 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17672 parameters that are of a variable-length.
17674 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17676 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17678 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17679 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17680 (CRT_CALL_STATIC_FUNCTION): Likewise.
17681 (ASM_DEFAULT_SPEC): New define.
17682 (ASM_SPEC32): Likewise.
17683 (ASM_SPEC64): Likewise.
17684 (ASM_SPEC_COMMON): Likewise.
17685 (ASM_SPEC): Likewise.
17686 (INVALID_64BIT): Likewise.
17687 (LINK_OS_DEFAULT_SPEC): Likewise.
17688 (LINK_OS_SPEC32): Likewise.
17689 (LINK_OS_SPEC64): Likewise.
17690 (POWERPC_LINUX): Likewise.
17691 (PTRDIFF_TYPE): Likewise.
17692 (RESTORE_FP_PREFIX): Likewise.
17693 (RESTORE_FP_SUFFIX): Likewise.
17694 (SAVE_FP_PREFIX): Likewise.
17695 (SAVE_FP_SUFFIX): Likewise.
17696 (SIZE_TYPE): Likewise.
17697 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
17698 (TARGET_64BIT): Likewise.
17699 (TARGET_64BIT): Likewise.
17700 (TARGET_AIX): Likewise.
17701 (WCHAR_TYPE_SIZE): Likewise.
17702 (WCHAR_TYPE): Undefine.
17703 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
17704 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
17705 (CPP_OS_RTEMS_SPEC): Delete.
17706 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
17707 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
17709 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
17711 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17713 PR tree-optimization/81578
17714 * tree-parloops.c (build_new_reduction): Bail out if
17715 reduction_code isn't one of the standard OpenMP reductions.
17716 Move the details printing after that decision.
17718 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17720 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
17721 related to reload_in_progress.
17722 (splat_input_operand): Likewise.
17723 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
17725 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
17727 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
17728 (TARGET_INSTANTIATE_DECLS): Likewise.
17729 (legitimate_indexed_address_p): Delete reload_in_progress code.
17730 (rs6000_debug_legitimate_address_p): Likewise.
17731 (rs6000_eliminate_indexed_memrefs): Likewise.
17732 (rs6000_emit_le_vsx_store): Likewise.
17733 (rs6000_emit_move_si_sf_subreg): Likewise.
17734 (rs6000_emit_move): Likewise.
17735 (register_to_reg_type): Likewise.
17736 (rs6000_pre_atomic_barrier): Likewise.
17737 (rs6000_machopic_legitimize_pic_address): Likewise.
17738 (rs6000_allocate_stack_temp): Likewise.
17739 (rs6000_address_for_fpconvert): Likewise.
17740 (rs6000_address_for_altivec): Likewise.
17741 (rs6000_secondary_memory_needed_rtx): Delete function.
17742 (rs6000_check_sdmode): Likewise.
17743 (rs6000_alloc_sdmode_stack_slot): Likewise.
17744 (rs6000_instantiate_decls): Likewise.
17745 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
17746 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
17747 Delete reload_in_progress.
17748 (*vec_reload_and_plus_<mptrsize>): Likewise.
17749 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
17750 (vsx_div_v2di): Likewise.
17751 (vsx_udiv_v2di): Likewise.
17753 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17755 * config/rs6000/rs6000.opt (mlra): Replace with stub.
17756 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
17757 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
17758 (rs6000_debug_reg_global): Delete print of LRA status.
17759 (rs6000_option_override_internal): Delete dead LRA related code.
17760 (rs6000_lra_p): Delete function.
17761 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
17763 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
17765 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
17766 * config/riscv/rtems.h: New file.
17768 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17769 Sudakshina Das <sudi.das@arm.com>
17771 * config/aarch64/aarch64.md
17772 (define_split for and<mode>3nr_compare): Move
17773 non aarch64_logical_operand to a register.
17774 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
17775 register immediate operand to a register.
17776 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
17778 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
17780 PR middle-end/81564
17781 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
17783 2017-07-27 Richard Biener <rguenther@suse.de>
17785 PR tree-optimization/81573
17786 PR tree-optimization/81494
17787 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
17788 multi defuse cycle case.
17790 2017-07-27 Richard Biener <rguenther@suse.de>
17792 PR tree-optimization/81571
17793 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
17796 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
17798 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
17799 earlier and only if MASK_FPU is set. Adjust formatting.
17801 2017-07-27 Martin Liska <mliska@suse.cz>
17803 * opt-functions.awk: Add validation of value of Init.
17804 * optc-gen.awk: Pass new argument.
17806 2017-07-27 Martin Liska <mliska@suse.cz>
17808 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
17809 Fix wrong condition.
17811 2017-07-27 Martin Liska <mliska@suse.cz>
17813 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
17814 BBs and edges seen by autoFDO.
17816 2017-07-27 Richard Biener <rguenther@suse.de>
17818 PR tree-optimization/81502
17819 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
17820 with incompatible but same sized type.
17821 (execute_update_addresses_taken): Likewise.
17823 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
17825 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
17826 flag_tree_loop_vectorize rather than flag_tree_vectorize.
17828 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17831 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
17832 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
17833 Change s_operand to memory_operand.
17835 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
17837 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
17838 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
17839 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
17840 Emit instructions rather than returning an expression. Handle TFmode
17841 and KFmode by casting to TImode.
17842 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
17843 (rs6000_emit_le_vsx_store): Likewise.
17844 * config/rs6000/vsx.md (VSX_TI): New iterator.
17845 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
17846 (*vsx_le_undo_permute_<mode>): Likewise.
17847 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
17848 emit the split sequence.
17849 (*vsx_le_perm_store_<mode>): Likewise.
17851 2017-07-27 Jakub Jelinek <jakub@redhat.com>
17853 PR tree-optimization/81555
17854 PR tree-optimization/81556
17855 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
17856 if true, force CHANGED for the recursive invocation.
17857 (reassociate_bb): Remember original length of ops array, pass
17858 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
17860 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
17861 attributes for noipa attribute. For naked attribute use
17862 lookup_attribute first before lookup_attribute_spec.
17863 * final.c (rest_of_handle_final): Disable IPA RA for functions with
17865 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
17866 for functions with noipa attribute.
17867 (cgraph_externally_visible_p): Return true for functions with noipa
17869 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
17870 for functions with noipa attribute.
17871 * doc/extend.texi: Document noipa function attribute.
17872 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
17873 also for functions with noipa attribute.
17874 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
17876 2017-07-26 Andrew Pinski <apinski@cavium.com>
17878 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
17879 vec_unalign_load_cost and vec_unalign_store_cost.
17881 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17883 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17884 -mvsx-small-integer option.
17885 (ISA_3_0_MASKS_IEEE): Likewise.
17886 (OTHER_VSX_VECTOR_MASKS): Likewise.
17887 (POWERPC_MASKS): Likewise.
17888 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17889 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17890 code, only testing for DImode being allowed in non-VSX floating
17892 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17893 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17895 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17896 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17897 TARGET_P8_VECTOR test.
17898 (rs6000_secondary_reload_simple_move): Likewise.
17899 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17900 since TARGET_P9_VECTOR was already tested.
17901 (rs6000_opt_masks): Remove -mvsx-small-integer.
17902 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17903 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17905 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17906 test for TARGET_VEXTRACTUB was used, and that uses
17908 (p9 extract splitter): Likewise.
17909 (vsx_extract_<mode>_di_p9): Likewise.
17910 (vsx_extract_<mode>_store_p9): Likewise.
17911 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17912 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17913 the elimination of TARGET_VSX_SMALL_INTEGER.
17914 (vsx_extract_<mode>_p8): Likewise.
17915 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17916 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17917 (vsx_set_<mode>_p9): Likewise.
17918 (vsx_set_v4sf_p9): Likewise.
17919 (vsx_set_v4sf_p9_zero): Likewise.
17920 (vsx_insert_extract_v4sf_p9): Likewise.
17921 (vsx_insert_extract_v4sf_p9_2): Likewise.
17922 * config/rs6000/rs6000.md (sign extend splitter): Change
17923 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17924 (floatsi<mode>2_lfiwax_mem): Likewise.
17925 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17926 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17927 since a test for TARGET_P9_VECTOR was used.
17928 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17929 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17930 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17931 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17932 TARGET_P8_VECTOR test.
17933 (fix_trunc<mode>si2_stfiwx): Likewise.
17934 (fix_trunc<mode>si2_internal): Likewise.
17935 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17936 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17938 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17939 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17940 TARGET_P8_VECTOR test.
17941 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17942 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17943 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17945 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17946 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17947 since a test for TARGET_P9_VECTOR was used.
17948 (splitter for loading small constants): Likewise.
17950 2017-07-26 Andrew Pinski <apinski@cavium.com>
17952 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17955 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17958 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17959 (fp_valid_at): Likewise.
17961 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17963 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17964 (qdf24xx_addrcost_table): Likewise.
17965 (cortexa57_tunings): Update to use generic_branch_cost.
17966 (cortexa72_tunings): Likewise.
17967 (cortexa73_tunings): Likewise.
17968 (qdf24xx_tunings): Likewise.
17970 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17972 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17973 (thunderx2t99_branch_cost): Likewise.
17974 (cortexa35_tunings): Update to use generic_branch_cost.
17975 (cortexa53_tunings): Likewise.
17976 (cortexa57_tunings): Likewise.
17977 (cortexa72_tunings): Likewise.
17978 (cortexa73_tunings): Likewise.
17979 (thunderx2t99_tunings): Likewise.
17981 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17983 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17984 (sparc_option_override): Honour MASK_FSMULD.
17985 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17986 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17987 * config/sparc/sparc.opt (mfsmuld): New option.
17988 * doc/invoke.texi (mfsmuld): Document option.
17990 2017-07-26 Marek Polacek <polacek@redhat.com>
17992 PR middle-end/70992
17993 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17995 2017-07-26 Richard Biener <rguenther@suse.de>
17997 * gimple-match-head.c (do_valueize): Return OP if valueize
17999 (get_def): New helper to get at the def stmt of a SSA name
18000 if valueize allows.
18001 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18002 do_valueize to get at the def stmt.
18003 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18005 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18007 PR middle-end/46932
18008 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18010 2017-07-26 Martin Liska <mliska@suse.cz>
18013 * function.c (expand_function_start): Make expansion of
18014 nonlocal_goto_save_area after parm_birth_insn.
18016 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18018 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18019 from all CPU target flags enable members.
18021 2017-07-26 Richard Biener <rguenther@suse.de>
18023 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18024 (decision_tree::gen): Make 'type' const.
18025 (write_predicate): Likewise.
18027 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18029 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18030 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18031 (rs6000_option_override_internal): Likewise.
18032 (rs6000_expand_vector_set): Likewise.
18033 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18034 (TARGET_UPPER_REGS_SF): Likewise.
18035 (TARGET_UPPER_REGS_DI): Likewise.
18036 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18037 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18038 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18039 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18040 (Splitters for DI constants in Altivec registers): Likewise.
18041 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18042 (vsx_set_v4sf_p9): Likewise.
18043 (vsx_set_v4sf_p9_zero): Likewise.
18044 (vsx_insert_extract_v4sf_p9): Likewise.
18045 (vsx_insert_extract_v4sf_p9_2): Likewise.
18047 2017-07-25 Carl Love <cel@us.ibm.com>
18049 * doc/extend.texi: Update the built-in documentation file for the
18050 existing built-in functions
18051 vector signed char vec_cnttz (vector signed char);
18052 vector unsigned char vec_cnttz (vector unsigned char);
18053 vector signed short vec_cnttz (vector signed short);
18054 vector unsigned short vec_cnttz (vector unsigned short);
18055 vector signed int vec_cnttz (vector signed int);
18056 vector unsigned int vec_cnttz (vector unsigned int);
18057 vector signed long long vec_cnttz (vector signed long long);
18058 vector unsigned long long vec_cnttz (vector unsigned long long);
18060 2017-07-25 Andrew Pinski <apinski@cavium.com>
18062 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18063 accesses where the use is for the first operand of a BIT_INSERT.
18065 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18068 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18069 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18071 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18073 * config/i386/gstabs.h: Delete.
18074 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18076 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18078 * config/i386/i386.c (ix86_decompose_address): Do not check for
18079 register RTX when looking at index_reg or base_reg.
18080 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18082 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18084 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18085 to update EH info here.
18087 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18089 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18091 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18093 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18095 2017-07-25 Torsten Duwe <duwe@suse.de>
18097 * common.opt: Introduce -fpatchable-function-entry
18098 command line option, and its variables function_entry_patch_area_size
18099 and function_entry_patch_area_start.
18100 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18101 including a two-value parser.
18102 * target.def (print_patchable_function_entry): New target hook.
18103 * targhooks.h (default_print_patchable_function_entry): New function.
18104 * targhooks.c (default_print_patchable_function_entry): Likewise.
18105 * toplev.c (process_options): Switch off IPA-RA if
18106 patchable function entries are being generated.
18107 * varasm.c (assemble_start_function): Look at the
18108 patchable-function-entry command line switch and current
18109 function attributes and maybe generate NOP instructions by
18110 calling the print_patchable_function_entry hook.
18111 * doc/extend.texi: Document patchable_function_entry attribute.
18112 * doc/invoke.texi: Document -fpatchable_function_entry
18113 command line option.
18114 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18116 * doc/tm.texi: Re-generate.
18118 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18121 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18122 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18124 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18126 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18128 2017-07-25 Richard Biener <rguenther@suse.de>
18130 PR tree-optimization/81455
18131 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18132 not walk in cycles when looking for guards.
18134 2017-07-25 Richard Biener <rguenther@suse.de>
18136 PR tree-optimization/81529
18137 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18138 when optimizing backedge uses.
18140 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18142 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18144 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18145 to dl_section_ref. On AIX, append an expression to subtract
18146 the size of the section length to dl_section_ref.
18148 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18150 * configure.ac: If any of the config.* scripts fail, exit 1.
18151 * configure: Regenerate.
18153 2017-07-25 Richard Biener <rguenther@suse.de>
18155 PR middle-end/81546
18156 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18157 of immediate uses, be more verbose on errors.
18159 2017-07-25 Richard Biener <rguenther@suse.de>
18161 PR tree-optimization/81510
18162 * tree-vect-loop.c (vect_is_simple_reduction): When the
18163 reduction stmt is not inside the loop bail out.
18165 2017-07-25 Richard Biener <rguenther@suse.de>
18167 PR tree-optimization/81303
18168 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18169 profitability check against LOOP_VINFO_NITERSM1.
18171 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18173 * domwalk.c (cmp_bb_postorder): Simplify.
18174 (sort_bbs_postorder): New function. Use it...
18175 (dom_walker::walk): ...here to optimize common cases.
18177 2017-07-25 Martin Liska <mliska@suse.cz>
18180 * ipa-visibility.c (function_and_variable_visibility): Make the
18181 redirection just on target that supports aliasing.
18182 Fix GNU coding style.
18184 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18187 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18189 * config/arm/rtems.h: Likewise.
18190 * config/bfin/rtems.h: Likewise.
18191 * config/i386/rtemself.h: Likewise.
18192 * config/lm32/rtems.h: Likewise.
18193 * config/m32c/rtems.h: Likewise.
18194 * config/m68k/rtemself.h: Likewise.
18195 * config/microblaze/rtems.h: Likewise.
18196 * config/mips/rtems.h: Likewise.
18197 * config/moxie/rtems.h: Likewise.
18198 * config/nios2/rtems.h: Likewise.
18199 * config/powerpcspe/rtems.h: Likewise.
18200 * config/rs6000/rtems.h: Likewise.
18201 * config/rtems.h: Likewise.
18202 * config/sh/rtems.h: Likewise.
18203 * config/sh/rtemself.h: Likewise.
18204 * config/sparc/rtemself.h: Likewise.
18206 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18209 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18210 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18211 * tree-ssa-structalias.c (alias_get_name): Same.
18213 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18216 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18217 instructions if no du chain is found.
18219 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18221 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18223 2017-07-25 Richard Biener <rguenther@suse.de>
18225 PR middle-end/81505
18226 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18229 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18231 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18232 upper-regs options.
18233 (ISA_2_7_MASKS_SERVER): Likewise.
18234 (ISA_3_0_MASKS_IEEE): Likewise.
18235 (OTHER_P8_VECTOR_MASKS): Likewise.
18236 (OTHER_VSX_VECTOR_MASKS): Likewise.
18237 (POWERPC_MASKS): Likewise.
18238 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18239 duplicate list of options.
18240 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18241 explicit -mupper-regs options.
18242 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18243 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18244 alias for -mupper-regs-df.
18245 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18246 (rs6000_init_hard_regno_mode_ok): Likewise.
18247 (rs6000_option_override_internal): Likewise.
18248 (rs6000_opt_masks): Likewise.
18249 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18250 options in terms of whether -mvsx or -mpower8-vector was used.
18251 (TARGET_UPPER_REGS_DI): Likewise.
18252 (TARGET_UPPER_REGS_SF): Likewise.
18253 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18254 -mupper-regs-* options.
18256 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18258 * passes.c (emergency_dump_function): Print some empty lines and a
18259 header before the RTL dump.
18261 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18263 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18265 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18268 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18269 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18271 2017-07-24 Carl Love <cel@us.ibm.com>
18273 * config/rs6000/rs6000-c.c: Add support for built-in functions
18274 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18275 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18276 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18277 vec_extract_fp_from_shortl): Add defines for the two builtins.
18278 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18279 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18281 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18282 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18283 * doc/extend.texi: Update the built-in documentation file for the
18284 new built-in function.
18286 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18289 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18291 * doc/generic.texi: Likewise.
18292 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18293 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18295 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18297 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18298 (aarch64_mls_elt_merge<mode>): Likewise.
18300 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18302 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18303 having __cxa_atexit.
18305 2017-07-23 Michael Collison <michael.collison@arm.com>
18307 * config/arm/arm.c (arm_option_override): Deprecate
18308 use of -mstructure-size-boundary.
18309 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18310 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18312 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18315 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18316 Reduce cost estimate for direct moves.
18318 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18321 * config/i386/i386.c (ix86_option_override_internal): Disable
18322 BMI, BMI2 and TBM instructions for -m16.
18324 2017-07-21 Carl Love <cel@us.ibm.com>
18326 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18327 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18328 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18329 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18330 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18331 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18332 VMULOSW): New enum "unspec" values.
18333 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18334 altivec_vmulosw): New patterns.
18335 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18336 VMULOSW): Add definitions.
18338 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18340 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18341 (qdf24xx): Likewise.
18342 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18343 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18344 (AARCH64_FL_V8_1): Renumber.
18345 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18346 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18347 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18348 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18349 rdma to feature modifiers list.
18351 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18353 PR middle-end/56727
18354 * ipa-visibility (function_and_variable_visibility): Convert
18355 recursive PLT call to direct call if appropriate.
18357 2017-07-21 Andrew Pinski <apinski@cavium.com>
18359 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18360 operand 1 to see if the types precision matches.
18361 * fold-const.c (operand_equal_p): Likewise.
18363 2017-07-21 Richard Biener <rguenther@suse.de>
18365 PR tree-optimization/81303
18366 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18367 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18368 (vect_peeling_hash_get_lowest_cost): Adjust.
18369 (vect_enhance_data_refs_alignment): Likewise. Use
18370 vect_get_peeling_costs_all_drs to compute the penalty for no
18371 peeling to match up costs.
18373 2017-07-21 Richard Biener <rguenther@suse.de>
18375 PR tree-optimization/81500
18376 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18377 we didn't identify a reduction path.
18379 2017-07-21 Tom de Vries <tom@codesourcery.com>
18380 Cesar Philippidis <cesar@codesourcery.com>
18382 PR gcov-profile/81442
18383 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18386 2017-07-21 Tom de Vries <tom@codesourcery.com>
18389 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18391 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18392 nvptx_override_options_after_change.
18394 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18396 * dwarf2out.c (output_file_names): Avoid double testing for
18397 dwarf_version >= 5.
18399 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18401 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18403 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18405 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18407 (try_crossjump_to_edge): Do not punt on partitioned functions.
18409 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18411 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18412 Put all BBs reachable only via paths crossing cold region to cold
18414 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18416 2016-07-21 Richard Biener <rguenther@suse.de>
18418 PR tree-optimization/81303
18419 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18420 into account prologue and epilogue iterations when raising
18421 min_profitable_iters to sth at least covering one vector iteration.
18423 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18425 * config/arm/arm.c (arm_test_cpu_arch_dat):
18428 2017-07-20 Nathan Sidwell <nathan@acm.org>
18430 Remove TYPE_METHODS.
18431 * tree.h (TYPE_METHODS): Delete.
18432 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18433 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18434 (dbxout_type_methods): Scan TYPE_FIELDS.
18435 (dbxout_type): Don't check TYPE_METHODS here.
18436 * function.c (use_register_for_decl): Always ignore register for
18437 class types when not optimizing.
18438 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18439 * tree.c (free_lang_data_in_type): Stitch out member functions and
18440 templates from TYPE_FIELDS.
18441 (build_distinct_type_copy, verify_type_variant,
18442 verify_type): Member fns are on TYPE_FIELDS.
18443 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18444 * tree-pretty-print.c (dump_generic_node): Likewise.
18446 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18449 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18450 V2TImode and V4TImode.
18451 (ix86_expand_vector_extract): Likewise.
18452 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18453 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18454 (ssescalarmode): Handle V4TImode and V2TImode.
18455 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18456 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18457 (VEXTRACTI128_MODE): New mode iterator.
18458 (splitter for *vec_extractv?ti first element): New.
18459 (VEC_INIT_MODE): New mode iterator.
18460 (vec_init<mode>): Consolidate 3 expanders into one using
18461 VEC_INIT_MODE mode iterator.
18463 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18465 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18466 non_spilled_static_chain_regno_p.
18468 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18470 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18472 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18474 * bb-reorder.c (connect_traces): Allow copying of blocks within
18477 2017-07-20 Richard Biener <rguenther@suse.de>
18479 * gimple.h (gimple_phi_result): Add gphi * overload.
18480 (gimple_phi_result_ptr): Likewise.
18481 (gimple_phi_arg): Likewise. Adjust index assert to only
18482 allow actual argument accesses rather than all slots available
18484 (gimple_phi_arg_def): Add gphi * overload.
18485 * tree-phinodes.c (make_phi_node): Initialize only actual
18487 (resize_phi_node): Clear memory not covered by old node,
18488 do not initialize excess argument slots.
18489 (reserve_phi_args_for_new_edge): Initialize new argument slot
18492 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18494 PR tree-optimization/81388
18496 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18498 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18499 by removing computation of may_be_zero.
18501 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18502 Tom de Vries <tom@codesourcery.com>
18504 PR middle-end/81030
18505 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18506 when gimple level profile disagrees with what RTL expander did.
18508 2017-07-20 Richard Biener <rguenther@suse.de>
18510 PR tree-optimization/61171
18511 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18512 (vect_analyze_stmt): Add slp instance parameter.
18513 (vectorizable_reduction): Likewise.
18514 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18515 (vect_is_simple_reduction): Deal with chains not detected
18516 as SLP reduction chain, specifically not properly associated
18517 chains containing a mix of plus/minus.
18518 (get_reduction_op): Remove.
18519 (get_initial_defs_for_reduction): Simplify, pass in whether
18520 this is a reduction chain, pass in the SLP node for the PHIs.
18521 (vect_create_epilog_for_reduction): Get the SLP instance as
18523 (vectorizable_reduction): Get the SLP instance as arg.
18524 During analysis remember the SLP node with the PHIs in the
18525 instance. Simplify getting at the vectorized reduction PHIs.
18526 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18527 through SLP instance.
18528 (vect_slp_analyze_operations): Likewise.
18529 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18530 (vect_transform_stmt): Likewise.
18532 2017-07-20 Tom de Vries <tom@codesourcery.com>
18534 PR tree-optimization/81489
18535 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18536 read of phi arg location to before loop that modifies phi.
18538 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18540 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18543 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18545 PR middle-end/81331
18546 * except.c (execute): Fix ordering issue.
18548 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18550 PR rtl-optimization/81423
18551 * combine.c (make_compound_operation_int): Don't try to optimize
18552 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18554 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18556 PR rtl-optimization/81423
18557 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18558 with a constant that is -1 in the truncated to mode.
18560 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18562 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18563 (determine_unlikely_bbs): ... here.
18564 * predict.h (propagate_unlikely_bbs_forward): Declare.
18565 * cfgexpand.c (pass_expand::execute): Use it.
18566 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18568 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18569 propagate_unlikely_bbs_forward.
18571 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18573 PR middle-end/81331
18574 * except.c (maybe_add_nop_after_section_switch): New function.
18577 2017-07-19 Tom de Vries <tom@codesourcery.com>
18579 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18581 2017-07-19 Tom de Vries <tom@codesourcery.com>
18583 * gimple.h (gimple_phi_arg): Make assert more strict.
18585 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18587 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18588 * config/rs6000/mmintrin.h: New file.
18589 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18591 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18593 PR tree-optimization/81346
18594 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18596 2017-07-19 Tom de Vries <tom@codesourcery.com>
18598 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18600 2017-07-19 Tom de Vries <tom@codesourcery.com>
18602 * config/nvptx/nvptx-modes.def: Add V2DImode.
18603 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18604 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18605 (nvptx_output_mov_insn): Handle lack of mov.b128.
18606 (nvptx_print_operand): Handle 'H' and 'L' codes.
18607 (nvptx_vector_mode_supported): Allow V2DImode.
18608 (nvptx_preferred_simd_mode): New function.
18609 (nvptx_data_alignment): New function.
18610 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18611 nvptx_preferred_simd_mode.
18612 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18614 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18616 2017-07-19 Tom de Vries <tom@codesourcery.com>
18618 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18619 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18620 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18621 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18622 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18623 (mov<VECIM>_insn): New define_insn.
18624 (define_expand "mov<VECIM>): New define_expand.
18626 2017-07-19 Tom de Vries <tom@codesourcery.com>
18628 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18630 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18632 PR tree-optimization/81346
18633 * fold-const.h (fold_div_compare, range_check_type): Declare.
18634 * fold-const.c (range_check_type): New function.
18635 (build_range_check): Use range_check_type.
18636 (fold_div_compare): No longer static, rewritten into
18637 a match.pd helper function.
18638 (fold_comparison): Don't call fold_div_compare here.
18639 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18640 as helper function.
18642 2017-07-19 Nathan Sidwell <nathan@acm.org>
18644 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18645 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18646 * tree.c (find_decls_types_r, verify_type): Use
18647 TYPE_{MIN,MAX}_VALUE_RAW.
18648 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18649 (hash_tree): Likewise.
18650 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18652 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18655 2017-07-18 Tom de Vries <tom@codesourcery.com>
18657 PR middle-end/81464
18658 * omp-expand.c (expand_omp_for_static_chunk): Handle
18659 equal-argument loop exit phi.
18661 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18664 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18665 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18666 operand 2 predicate.
18667 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18668 operand 2 predicate.
18669 (ror,rol -> rorx splitters): Use const_int_operand as
18670 operand 2 predicate.
18672 2017-06-18 Richard Biener <rguenther@suse.de>
18674 PR tree-optimization/81410
18675 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18676 the gap in the ! slp_perm SLP case after each group.
18678 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18680 PR middle-end/81463
18681 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18684 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18686 PR middle-end/81462
18687 * predict.c (set_even_probabilities): Cleanup; do not affect
18688 probabilities that are already known.
18689 (combine_predictions_for_bb): Call even when count is set.
18691 2017-07-18 Nathan Sidwell <nathan@acm.org>
18693 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18696 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18699 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
18700 optimization for loop niter analysis.
18702 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
18705 * config/avr/avr.c (avr_optimize_casesi): Don't use
18706 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
18708 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
18710 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
18711 body_cost_vec from _vect_peel_extended_info.
18712 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
18713 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
18716 2017-07-18 Bin Cheng <bin.cheng@arm.com>
18718 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
18720 2017-07-18 Richard Biener <rguenther@suse.de>
18722 PR tree-optimization/80620
18723 PR tree-optimization/81403
18724 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
18725 info when re-using a VN table entry.
18727 2017-07-18 Richard Biener <rguenther@suse.de>
18729 PR tree-optimization/81418
18730 * tree-vect-loop.c (vectorizable_reduction): Properly compute
18731 vectype_in. Verify that with lane-reducing reduction operations
18732 we have a single def-use cycle.
18734 2017-07-17 Carl Love <cel@us.ibm.com>
18736 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
18738 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18739 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18740 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18741 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18742 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18743 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18744 VMULOSW): New enum "unspec" values.
18745 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18746 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18747 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18748 altivec_vmulosw): New patterns.
18749 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18750 VMULOSW): Add definitions.
18751 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
18753 * config/alpha/alpha.c: Include predict.h.
18755 2017-07-17 Yury Gribov <tetra2005@gmail.com>
18757 * tree-vrp.c (compare_assert_loc): Fix comparison function
18758 to return predictable results.
18760 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18762 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
18764 (subdi3): Likewise.
18765 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
18766 * doc/invoke.texi (mexpand-adddi): Update text.
18768 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18770 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
18771 that also clobbers the CC register. The old expand code is moved
18773 (*arc_clzsi2): ... here.
18774 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
18775 the CC register. The old expand code is moved to ...
18776 (arc_ctzsi2): ... here.
18778 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18780 * config/arc/arc.opt (mindexed-loads): Use initial value
18781 TARGET_INDEXED_LOADS_DEFAULT.
18782 (mauto-modify-reg): Use initial value
18783 TARGET_AUTO_MODIFY_REG_DEFAULT.
18784 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18785 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18786 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
18787 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
18789 2017-07-17 Martin Liska <mliska@suse.cz>
18792 * opts.c (finish_options): Do not allow -fgnu-tm
18793 w/ -fsanitize={kernel-,}address. Say sorry.
18795 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18798 * tree-loop-distribution.c (classify_partition): Only assert on
18799 numer of iterations.
18800 (merge_dep_scc_partitions): Delete prameter. Update function call.
18801 (distribute_loop): Remove code handling loop with unknown niters.
18802 (pass_loop_distribution::execute): Skip loop with unknown niters.
18804 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18807 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
18808 function sort_partitions_by_post_order.
18810 2017-07-17 Bin Cheng <bin.cheng@arm.com>
18812 PR tree-optimization/81374
18813 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
18814 the max index of basic blocks, rather than number of basic blocks.
18816 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18818 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
18820 (arc_legitimate_pic_operand_p): Likewise.
18821 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
18823 (arc_needs_pcl_p): Likewise.
18824 (arc_legitimate_pc_offset_p): Likewise.
18825 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
18826 function is also used in constrains.md.
18827 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
18828 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
18829 PLUS. Only return true/false in known cases, otherwise assert.
18830 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
18831 is already called in arc_legitimate_constant_p.
18832 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
18834 (LEGITIMATE_PIC_OPERAND_P): Use
18835 arc_raw_symbolic_reference_mentioned_p function.
18836 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
18841 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
18842 Andrew Burgess <andrew.burgess@embecosm.com>
18844 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
18845 (arc_return_address_register): New function.
18846 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
18847 (arc_handle_fndecl_attribute): Add naked attribute.
18848 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18849 (TARGET_WARN_FUNC_RETURN): Likewise.
18850 (arc_allocate_stack_slots_for_args): New function.
18851 (arc_warn_func_return): Likewise.
18852 (machine_function): Change type fn_type.
18853 (arc_compute_function_type): Consider new naked function type,
18854 change function return type.
18855 (arc_must_save_register): Adapt to handle new
18856 arc_compute_function_type's return type.
18857 (arc_expand_prologue): Likewise.
18858 (arc_expand_epilogue): Likewise.
18859 (arc_return_address_regs): Delete.
18860 (arc_return_address_register): New function.
18861 (arc_epilogue_uses): Use above function.
18862 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
18863 (arc_function_type): Change encoding, add naked type.
18864 (ARC_INTERRUPT_P): Change to handle the new encoding.
18865 (ARC_FAST_INTERRUPT_P): Likewise.
18866 (ARC_NORMAL_P): Define.
18867 (ARC_NAKED_P): Likewise.
18868 (arc_compute_function_type): Delete prototype.
18869 * config/arc/arc.md (in_ret_delay_slot): Use
18870 arc_return_address_register function.
18871 (simple_return): Likewise.
18872 (p_return_i): Likewise.
18874 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18876 PR tree-optimization/81428
18877 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
18878 can't be built for those types.
18880 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18882 Remove stuff dead since r239246.
18884 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18885 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18886 (avr_inform_devices): Remove dead stuff.
18888 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18890 * config/arm/arm_neon.h: Fix softp typo.
18892 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18894 PR tree-optimization/81365
18895 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18896 aggregate moves onto bb predecessor edges, make sure there are no
18897 loads that could alias the lhs in between the start of bb and the
18900 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18903 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18904 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18905 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18907 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18909 PR tree-optimization/81396
18910 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18911 (init_symbolic_number): Initialize it to 1.
18912 (perform_symbolic_merge): Add n_ops from both operands into the new
18914 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18915 without base_addr as useless if they need more than one operation.
18916 (bswap_replace): Handle !bswap case for NULL base_addr.
18918 2017-07-17 Tom de Vries <tom@codesourcery.com>
18921 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18924 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18926 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18927 conditional builtin define __FIX_LEON3FT_B2BST.
18929 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18931 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18932 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18935 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18937 PR rtl-optimization/81424
18938 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18939 to remove potential trapping from operands if -fnon-call-exceptions.
18941 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18943 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18944 profile_proability for scalling.
18945 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18947 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18949 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18951 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18953 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18954 fixpoint arithmetics.
18956 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18958 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18959 fixpoint arithmetics.
18961 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18963 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18964 fixpoint arithmetics.
18966 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18968 * profile-count.h (profile_probability::from_reg_br_prob_note,
18969 profile_probability::to_reg_br_prob_note): New functions.
18970 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18971 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18972 * predict.c (probability_reliable_p): Update.
18973 (edge_probability_reliable_p): Update.
18974 (br_prob_note_reliable_p): Update.
18975 (invert_br_probabilities): Update.
18976 (add_reg_br_prob_note): New function.
18977 (combine_predictions_for_insn): Update.
18978 * asan.c (asan_clear_shadow): Update.
18979 * cfgbuild.c (compute_outgoing_frequencies): Update.
18980 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18981 (update_br_prob_note): Update.
18982 (rtl_verify_edges): Update.
18983 (purge_dead_edges): Update.
18984 (fixup_reorder_chain): Update.
18985 * emit-rtl.c (try_split): Update.
18986 * ifcvt.c (cond_exec_process_insns): Update.
18987 (cond_exec_process_if_block): Update.
18988 (dead_or_predicable): Update.
18989 * internal-fn.c (expand_addsub_overflow): Update.
18990 (expand_neg_overflow): Update.
18991 (expand_mul_overflow): Update.
18992 * loop-doloop.c (doloop_modify): Update.
18993 * loop-unroll.c (compare_and_jump_seq): Update.
18994 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18995 * predict.h: Update.
18996 * reorg.c (mostly_true_jump): Update.
18998 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18999 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19000 * config/arc/arc.c: (emit_unlikely_jump): Update.
19001 * config/arm/arm.c: (emit_unlikely_jump): Update.
19002 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19003 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19004 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19005 (ix86_print_operand): Update.
19006 (ix86_split_fp_branch): Update.
19007 (predict_jump): Update.
19008 * config/ia64/ia64.c (ia64_print_operand): Update.
19009 * config/mmix/mmix.c (mmix_print_operand): Update.
19010 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19011 (rs6000_expand_split_stack_prologue): Update.
19012 * config/rs6000/rs6000.c: Update.
19013 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19014 (s390_expand_vec_movstr): Update.
19015 (s390_expand_cs_tdsi): Update.
19016 (s390_expand_split_stack_prologue): Update.
19017 * config/sh/sh.c (sh_print_operand): Update.
19018 (expand_cbranchsi4): Update.
19019 (expand_cbranchdi4): Update.
19020 * config/sparc/sparc.c (output_v9branch): Update.
19021 * config/spu/spu.c (get_branch_target): Update.
19022 (ea_load_store_inline): Update.
19023 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19024 * config/tilepro/tilepro.c: Update.
19026 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19028 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19029 (gimplify_save_expr): Likewise.
19031 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19033 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19035 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19037 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19039 * timevar.def (TV_IPA_FNSUMMARY): Define.
19041 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19043 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19044 to back store errata sensitive sequence from being generated.
19045 (sqrtdf2_fix): Likewise.
19047 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19049 * tree-ssa-threadupdate.c (compute_path_counts,
19050 update_joiner_offpath_counts): Use profile_probability.
19052 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19055 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19057 * config/arm/arm-c.c (arm_cpu_builtins): Define
19058 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19060 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19062 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19063 array entries to represent __ieee128 versions of the
19064 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19065 scalar_extract_sig, and scalar_insert_exp built-in functions.
19066 (altivec_resolve_overloaded_builtin): Add special case handling
19067 for the __builtin_scalar_insert_exp function, as represented by
19068 the P9V_BUILTIN_VEC_VSIEDP constant.
19069 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19070 exponent support for __ieee128 argument.
19071 (VSESQP): Add scalar extract signature support for __ieee128
19073 (VSTDCNQP): Add scalar test negative support for __ieee128
19075 (VSIEQP): Add scalar insert exponent support for __int128 argument
19076 with __ieee128 result.
19077 (VSIEQPF): Add scalar insert exponent support for __ieee128
19078 argument with __ieee128 result.
19079 (VSTDCQP): Add scalar test data class support for __ieee128
19081 (VSTDCNQP): Add overload support for scalar test negative with
19082 __ieee128 argument.
19083 (VSTDCQP): Add overload support for scalar test data class
19084 __ieee128 argument.
19085 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19086 UNSPEC_VSX_SXSIGDP.
19087 (UNSPEC_VSX_SIEXPQP): New constant.
19088 (xsxexpqp): New insn for VSX scalar extract exponent quad
19090 (xsxsigqp): New insn for VSX scalar extract significand quad
19092 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19093 precision with floating point argument.
19094 (xststdcqp): New expand for VSX scalar test data class quad
19096 (xststdcnegqp): New expand for VSX scalar test negative quad
19098 (xststdcqp): New insn to match expansions for VSX scalar test data
19099 class quad precision and VSX scalar test negative quad precision.
19100 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19101 special case operand checking to enforce that second operand of
19102 VSX scalar test data class with quad precision argument is a 7-bit
19104 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19105 prototypes and descriptions of __ieee128 versions of
19106 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19107 scalar_test_data_class, and scalar_test_neg built-in functions.
19109 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19111 PR tree-optimization/81162
19112 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19113 replace a negate with an add.
19115 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19117 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19119 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19121 * config/arm/arm-c.c (arm_cpu_builtins): Define
19122 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19124 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19126 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19127 (armv8-r): Set ARM Cortex-R52 as default CPU.
19128 * config/arm/arm-tables.opt: Regenerate.
19129 * config/arm/arm-tune.md: Regenerate.
19130 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19132 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19133 extension for -mcpu=cortex-r52.
19135 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19137 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19138 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19139 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19140 (fp-armv8): Define it as FP_ARMv8 only.
19141 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19142 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19144 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19145 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19146 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19147 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19148 than TARGET_FPU_ARMV8.
19149 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19150 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19151 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19153 * config/arm/neon.md (neon_vrint): Likewise.
19154 (neon_vcvt): Likewise.
19155 (neon_<fmaxmin_op><mode>): Likewise.
19156 (<fmaxmin><mode>3): Likewise.
19157 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19158 * config/arm/predicates.md (arm_cond_move_operator): Check against
19159 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19161 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19163 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19164 to top of function.
19166 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19168 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19169 loop in comment with memset.
19171 2017-07-14 Martin Liska <mliska@suse.cz>
19173 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19174 * dwarf2out.c (is_java): Remove the function.
19175 (output_pubname): Remove usage of the function.
19176 (lower_bound_default): Remove usage of DW_LANG_Java.
19177 (gen_compile_unit_die): Likewise.
19178 * gcc.c: Remove compiler defaults for .java and .zip files.
19179 * gimple-expr.c (remove_suffix): Change as there's no longer
19180 extension than 4-letter one.
19181 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19182 (gimplify_save_expr): Likewise.
19183 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19184 as it's possible even for other languages than Java.
19185 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19186 * lto-opts.c (lto_write_options): Remove reference to Java.
19187 * opts.c (strip_off_ending): Update file extension handling.
19188 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19189 * tree-eh.c (lower_resx): Likewise.
19190 * tree.c (free_lang_data_in_type): Remove dead code.
19191 (find_decls_types_r): Likewise.
19192 (build_common_builtin_nodes): Remove Java from a comment.
19193 (verify_type): Remove dead code.
19194 * varasm.c (assemble_external): Remove Java from a comment.
19196 2017-07-14 Martin Liska <mliska@suse.cz>
19198 * opts.c (finish_options): Add quotes.
19199 (common_handle_option): Likewise.
19201 2017-07-14 Martin Liska <mliska@suse.cz>
19203 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19204 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19205 Remove N_SO_PASCAL.
19206 * dwarf2out.c (lower_bound_default): Do not handle
19208 (gen_compile_unit_die): Likewise.
19209 * gcc.c: Remove default extension binding for GNU Pascal.
19210 * stmt.c: Remove Pascal language from a comment.
19211 * xcoffout.c: Likewise.
19213 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19216 * diagnostic-show-locus.c (fixit_cmp): New function.
19217 (layout::layout): Sort m_fixit_hints.
19218 (column_range::column_range): Assert that the values are valid.
19219 (struct char_span): New struct.
19220 (correction::overwrite): New method.
19221 (struct source_line): New struct.
19222 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19223 calls in terms of classes source_line and char_span, and
19224 correction::overwrite.
19225 (selftest::test_overlapped_fixit_printing_2): New function.
19226 (selftest::diagnostic_show_locus_c_tests): Call it.
19228 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19230 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19231 early if there is no lhs.
19233 2017-07-13 Martin Liska <mliska@suse.cz>
19235 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19236 (gen_reference_type_die): Likewise.
19237 * stor-layout.c: Remove Pascal-related comment.
19239 2017-07-13 Martin Liska <mliska@suse.cz>
19241 * opts.c (finish_options): Add quotes to error messages.
19242 (parse_sanitizer_options): Likewise.
19244 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19246 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19248 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19250 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19252 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19254 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19256 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19258 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19261 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19262 provides the hardware capability bits, define the macro
19263 __BUILTIN_CPU_SUPPORTS__.
19264 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19265 if GLIBC does not provide the hardware capability bits. Add a
19266 gcc_unreachable call if the built-in cpu function is neither
19267 __builtin_cpu_is nor __builtin_cpu_supports.
19268 (rs6000_get_function_versions_dispatcher): Change the warning
19269 that an old GLIBC is used which does not export the capability
19270 bits to be an error.
19271 * doc/extend.texi (target_clones attribute): Document the
19272 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19273 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19274 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19275 the macros defined by GCC if the newer GLIBC is available.
19277 2017-07-12 Jeff Law <law@redhat.com>
19279 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19280 remaining includes slightly.
19281 * config/riscv/riscv-builtins.c: Include profile-count.h.
19283 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19286 * config/avr/avr.c (avr_set_current_function): In diagnostic
19287 messages: Quote keywords and (parts of) identifiers.
19288 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19291 2017-07-12 Carl Love <cel@us.ibm.com>
19293 * config/rs6000/rs6000-c.c: Add support for built-in functions
19294 vector bool char vec_revb (vector bool char);
19295 vector bool short vec_revb (vector short char);
19296 vector bool int vec_revb (vector bool int);
19297 vector bool long long vec_revb (vector bool long long);
19298 * doc/extend.texi: Update the built-in documentation file for the
19299 new built-in functions.
19301 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19303 * config/s390/s390.md: Remove movcc splitter.
19305 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19307 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19308 load/store on condition.
19310 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19313 * config/avr/avr.c (avr_encode_section_info)
19314 [progmem && !TREE_READONLY]: Error if progmem object needs
19317 2017-07-11 Michael Collison <michael.collison@arm.com>
19319 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19322 2017-07-11 Carl Love <cel@us.ibm.com>
19324 * config/rs6000/rs6000-c.c: Add support for builtins
19325 vector unsigned int vec_parity_lsbb (vector signed int);
19326 vector unsigned int vec_parity_lsbb (vector unsigned int);
19327 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19328 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19329 vector unsigned long long vec_parity_lsbb (vector signed long long);
19330 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19331 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19332 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19333 * doc/extend.texi: Update the built-in documentation file for the
19334 new built-in functions.
19336 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19338 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19339 (layout::m_primary_loc): New field.
19340 (layout::layout): Initialize new field. Move location filtering
19341 logic from here to...
19342 (layout::maybe_add_location_range): ...this new method. Add
19343 support for filtering to just the lines already specified by other
19345 (layout::will_show_line_p): New method.
19346 (gcc_rich_location::add_location_if_nearby): New method.
19347 (selftest::test_add_location_if_nearby): New test function.
19348 (selftest::diagnostic_show_locus_c_tests): Call it.
19349 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19352 2017-07-11 Tom de Vries <tom@codesourcery.com>
19354 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19355 (bb_first_real_insn): New function.
19356 (nvptx_single): Add extra initialization of broadcasted condition
19359 2017-07-11 Nathan Sidwell <nathan@acm.org>
19361 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19363 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19365 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19366 Binutils doc as TEXI will mess them up.
19367 * doc/invoke.texi (AVR Options): Same here.
19369 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19371 * config/sparc/sparc.opt (mfix-ut700): New option.
19372 (mfix-gr712rc): Likewise.
19373 (sparc_fix_b2bst): New variable.
19374 * doc/invoke.texi (SPARC options): Document them.
19375 (ARM options): Fix warnings.
19376 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19377 instructions to prevent sequences that can trigger the store-store
19378 errata for certain LEON3FT processors.
19379 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19380 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19381 * config/sparc/sparc.md (fix_b2bst): New attribute.
19382 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19384 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19387 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19389 (*rsqrtsf2_sse): Ditto.
19391 (div<mode>3): Macroize insn from divdf3 and divsf3
19392 using MODEF mode iterator.
19394 2017-07-10 Martin Sebor <msebor@redhat.com>
19396 PR tree-optimization/80397
19397 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19398 instead of testing for equality to INTEGER_TYPE.
19400 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19402 * config.gcc: Remove uclibc from arc target spec.
19404 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19406 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19408 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19411 * lto-wrapper.c (remove_option): New function.
19412 (merge_and_complain): Merge PIC/PIE options more realistically.
19414 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19416 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19420 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19421 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19422 * config.in: Regenerate.
19423 * configure: Regenerate.
19424 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19425 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19426 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19427 (TARGET_GASISR_PROLOGUES): ...target mask.
19428 * common/config/avr/avr-common.c
19429 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19430 Set -mgas-isr-prologues.
19431 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19432 INSERT_PASS_BEFORE for it.
19433 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19434 * config/avr/avr.c (avr_option_override)
19435 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19436 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19437 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19438 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19439 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19441 (make_avr_pass_pre_proep): New function.
19442 (emit_push_sfr) <treg>: Add argument to function and use it
19443 instead of TMP_REG.
19444 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19445 and set machine->gasisr.yes.
19446 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19447 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19448 __gcc_isr.n_pushed to .L__stack_usage.
19449 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19450 (avr_asm_final_postscan_insn): ...this new static function.
19451 * config/avr/avr.h (machine_function)
19452 <is_no_gccisr, use_L__stack_usage>: New fields.
19453 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19454 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19455 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19456 (gasisr, *gasisr): New expander and insn.
19457 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19458 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19459 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19461 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19463 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19466 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19468 Move jump-tables out of .text again.
19471 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19472 (ASM_OUTPUT_ADDR_VEC): New function.
19473 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19474 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19475 INSN_ADDRESSes as asm comment.
19476 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19477 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19478 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19479 * config/avr/avr.md (*tablejump): Adjust comment.
19480 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19481 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19483 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19484 (avr_output_addr_vec): New proto.
19485 (avr_log_t) <insn_addresses>: New field.
19487 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19490 * config/i386/i386.c (ix86_function_arg_advance): Set
19491 outgoing_args_on_stack to true if there are outgoing arguments
19493 (ix86_function_arg): Likewise.
19494 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19495 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19496 * config/i386/i386.h (machine_function): Add
19497 outgoing_args_on_stack.
19499 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19501 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19502 supporting pthreds.
19503 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19505 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19507 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19508 (REAL_H): Remove $(MACHMODE_H).
19509 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19511 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19512 $(MACHMODE_H) and double-int.h.
19513 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19515 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19518 2017-07-07 Andrew Pinski <apinski@cavium.com>
19520 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19521 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19523 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19525 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19526 Add warning if GCC was not configured to link against a GLIBC that
19527 exports the hardware capability bits.
19528 (make_resolver_func): Make resolver function private and not a
19529 COMDAT function. Create the name with clone_function_name instead
19530 of make_unique_name.
19533 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19534 correct operand in doing the split.
19536 2017-07-07 Carl Love <cel@us.ibm.com>
19538 * config/rs6000/rs6000-c: Add support for built-in function
19539 vector unsigned short vec_pack_to_short_fp32 (vector float,
19541 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19542 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19543 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19544 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19545 (convert_4f32_8i16): Add define_expand.
19546 * doc/extend.texi: Update the built-in documentation file for the
19547 new built-in function.
19549 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19551 * config/sparc/m8.md: New file.
19552 * config/sparc/sparc.md: Include m8.md.
19554 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19556 * config/sparc/sparc.opt: New option -mvis4b.
19557 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19558 (sparc_option_override): Handle VIS4B.
19559 (enum sparc_builtins): Define
19560 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19561 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19562 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19563 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19564 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19565 (check_constant_argument): New function.
19566 (sparc_vis_init_builtins): Define builtins
19567 __builtin_vis_dictunpack{8,16,32},
19568 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19569 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19570 __builtin_vis_fpcmpde{8,16,32}shl and
19571 __builtin_vis_fpcmpur{8,16,32}shl.
19572 (sparc_expand_builtin): Check that the constant operands to
19573 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19574 constant and in range.
19575 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19577 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19578 (SPARC_IMM5_P): Likewise.
19579 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19580 (enabled): Handle vis4b.
19581 (UNSPEC_DICTUNPACK): New unspec.
19582 (UNSPEC_FPCMPSHL): Likewise.
19583 (UNSPEC_FPUCMPSHL): Likewise.
19584 (UNSPEC_FPCMPDESHL): Likewise.
19585 (UNSPEC_FPCMPURSHL): Likewise.
19586 (cpu_feature): New CPU feature `vis4b'.
19587 (dictunpack{8,16,32}): New insns.
19588 (FPCSMODE): New mode iterator.
19589 (fpcscond): New code iterator.
19590 (fpcsucond): Likewise.
19591 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19592 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19593 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19594 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19595 * config/sparc/constraints.md: Define constraints `q' for unsigned
19596 2-bit integer constants and `t' for unsigned 5-bit integer
19598 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19600 (imm5_operand_dictunpack16): Likewise.
19601 (imm5_operand_dictunpack32): Likewise.
19602 (imm2_operand): Likewise.
19603 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19604 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19605 ditunpack* and fpcmp*shl builtins.
19607 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19609 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19610 * config.in: Add HAVE_AS_SPARC6 define.
19611 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19613 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19615 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19616 (CPP_CPU_SPEC): Handle m8.
19617 (ASM_CPU_SPEC): Likewise.
19618 * config/sparc/sparc-opts.h (enum processor_type): Add
19620 * config/sparc/sparc.c (m8_costs): New struct.
19621 (sparc_option_override): Handle TARGET_CPU_m8.
19622 (sparc32_initialize_trampoline): Likewise.
19623 (sparc64_initialize_trampoline): Likewise.
19624 (sparc_issue_rate): Likewise.
19625 (sparc_register_move_cost): Likewise.
19626 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19627 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19628 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19629 (CPP_CPU_SPEC): Handle M8.
19630 (ASM_CPU_SPEC): Likewise.
19631 (AS_M8_FLAG): Define.
19632 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19633 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19634 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19636 * configure: Regenerate.
19637 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19640 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19642 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19644 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19645 ("*movdi_insn_sp32"): Do not set v3pipe.
19646 ("*movsi_insn"): Likewise.
19647 ("*movdi_insn_sp64"): Likewise.
19648 ("*movsf_insn"): Likewise.
19649 ("*movdf_insn_sp32"): Likewise.
19650 ("*movdf_insn_sp64"): Likewise.
19651 ("*zero_extendsidi2_insn_sp64"): Likewise.
19652 ("*sign_extendsidi2_insn"): Likewise.
19653 ("*mov<VM32:mode>_insn"): Likewise.
19654 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19655 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19656 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19657 ("<vlop:code><VL:mode>3"): Likewise.
19658 ("*not_<vlop:code><VL:mode>3"): Likewise.
19659 ("*nand<VL:mode>_vis"): Likewise.
19660 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19661 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19662 ("one_cmpl<VL:mode>2"): Likewise.
19663 ("faligndata<VM64:mode>_vis"): Likewise.
19664 ("alignaddrsi_vis"): Likewise.
19665 ("alignaddrdi_vis"): Likweise.
19666 ("alignaddrlsi_vis"): Likewise.
19667 ("alignaddrldi_vis"): Likewise.
19668 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19669 ("bmaskdi_vis"): Likewise.
19670 ("bmasksi_vis"): Likewise.
19671 ("bshuffle<VM64:mode>_vis"): Likewise.
19672 ("cmask8<P:mode>_vis"): Likewise.
19673 ("cmask16<P:mode>_vis"): Likewise.
19674 ("cmask32<P:mode>_vis"): Likewise.
19675 ("pdistn<P:mode>_vis"): Likewise.
19676 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19678 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19680 * config/sparc/sparc.md ("subtype"): New insn attribute.
19681 ("*wrgsr_sp64"): Set insn subtype.
19682 ("*rdgsr_sp64"): Likewise.
19683 ("alignaddrsi_vis"): Likewise.
19684 ("alignaddrdi_vis"): Likewise.
19685 ("alignaddrlsi_vis"): Likewise.
19686 ("alignaddrldi_vis"): Likewise.
19687 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19688 ("fexpand_vis"): Likewise.
19689 ("fpmerge_vis"): Likewise.
19690 ("faligndata<VM64:mode>_vis"): Likewise.
19691 ("bshuffle<VM64:mode>_vis"): Likewise.
19692 ("cmask8<P:mode>_vis"): Likewise.
19693 ("cmask16<P:mode>_vis"): Likewise.
19694 ("cmask32<P:mode>_vis"): Likewise.
19695 ("fchksm16_vis"): Likewise.
19696 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
19697 ("fmean16_vis"): Likewise.
19698 ("fp<plusminus_insn>64_vis"): Likewise.
19699 ("<plusminus_insn>v8qi3"): Likewise.
19700 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19701 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
19702 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
19703 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
19704 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
19705 ("*movqi_insn"): Likewise.
19706 ("*movhi_insn"): Likewise.
19707 ("*movsi_insn"): Likewise.
19708 ("movsi_pic_gotdata_op"): Likewise.
19709 ("*movdi_insn_sp32"): Likewise.
19710 ("*movdi_insn_sp64"): Likewise.
19711 ("movdi_pic_gotdata_op"): Likewise.
19712 ("*movsf_insn"): Likewise.
19713 ("*movdf_insn_sp32"): Likewise.
19714 ("*movdf_insn_sp64"): Likewise.
19715 ("*zero_extendhisi2_insn"): Likewise.
19716 ("*zero_extendqihi2_insn"): Likewise.
19717 ("*zero_extendqisi2_insn"): Likewise.
19718 ("*zero_extendqidi2_insn"): Likewise.
19719 ("*zero_extendhidi2_insn"): Likewise.
19720 ("*zero_extendsidi2_insn_sp64"): Likewise.
19721 ("ldfsr"): Likewise.
19722 ("prefetch_64"): Likewise.
19723 ("prefetch_32"): Likewise.
19724 ("tie_ld32"): Likewise.
19725 ("tie_ld64"): Likewise.
19726 ("*tldo_ldub_sp32"): Likewise.
19727 ("*tldo_ldub1_sp32"): Likewise.
19728 ("*tldo_ldub2_sp32"): Likewise.
19729 ("*tldo_ldub_sp64"): Likewise.
19730 ("*tldo_ldub1_sp64"): Likewise.
19731 ("*tldo_ldub2_sp64"): Likewise.
19732 ("*tldo_ldub3_sp64"): Likewise.
19733 ("*tldo_lduh_sp32"): Likewise.
19734 ("*tldo_lduh1_sp32"): Likewise.
19735 ("*tldo_lduh_sp64"): Likewise.
19736 ("*tldo_lduh1_sp64"): Likewise.
19737 ("*tldo_lduh2_sp64"): Likewise.
19738 ("*tldo_lduw_sp32"): Likewise.
19739 ("*tldo_lduw_sp64"): Likewise.
19740 ("*tldo_lduw1_sp64"): Likewise.
19741 ("*tldo_ldx_sp64"): Likewise.
19742 ("*mov<VM32:mode>_insn"): Likewise.
19743 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19744 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19746 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19748 * config/sparc/sparc.md ("type"): New insn type viscmp.
19749 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
19751 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
19752 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
19753 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19754 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
19756 ("n7_vis_logical_11cycle"): Likewise.
19757 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
19758 * config/sparc/niagara2.md ("niag3_vis": Likewise.
19759 * config/sparc/niagara.md ("niag_vis"): Likewise.
19760 * config/sparc/ultra3.md ("us3_fga"): Likewise.
19761 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
19763 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19765 * config/sparc/sparc.md: New instruction type `bmask'.
19766 (bmaskdi_vis): Use the `bmask' type.
19767 (bmasksi_vis): Likewise.
19768 * config/sparc/ultra3.md (us3_array): Likewise.
19769 * config/sparc/niagara7.md (n7_array): Likewise.
19770 * config/sparc/niagara4.md (n4_array): Likewise.
19771 * config/sparc/niagara2.md (niag2_vis): Likewise.
19772 (niag3_vis): Likewise.
19773 * config/sparc/niagara.md (niag_vis): Likewise.
19775 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19777 * ipa-comdats.c: Remove optimize check from gate.
19778 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
19779 for functions not optimized.
19780 (ipa_fn_summary_read): Skip optimize check.
19781 (ipa_fn_summary_write): Likewise.
19782 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
19784 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
19786 (can_inline_edge_p): Check flag_pcc_struct_return for match.
19787 (check_callers): Give up on caller which is not optimized.
19788 (inline_small_functions): Likewise.
19789 (ipa_inline): Do not give up when not optimizing.
19790 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
19791 away unoptimizes cdtors.
19792 (whole_program_function_and_variable_visibility): Do
19793 ipa_discover_readonly_nonaddressable_vars in LTO mode.
19794 * ipa.c (process_references): Do not check optimize.
19795 (symbol_table::remove_unreachable_nodes): Update optimize check.
19796 (set_writeonly_bit): Update optimize check.
19797 (pass_ipa_cdtor_merge::gate): Do not check optimize.
19798 (pass_ipa_single_use::gate): Remove.
19800 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19802 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
19803 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
19804 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
19805 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
19806 permute_load, permute_store, adjust_extract, adjust_splat,
19807 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
19808 replace_swap_with_copy, dump_swap_insn_table,
19809 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
19810 recombine_lvx_pattern, recombine_stvx_pattern,
19811 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
19812 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
19813 to file rs6000-p8swap.c.
19814 * config/rs6000/rs6000-p8swap.c: New file.
19815 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
19816 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
19817 and rs6000*-*-* targets.
19819 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19821 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
19823 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19825 * lto-wrapper.c (merge_and_complain): Do not merge
19826 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
19827 fsigned_zeros, ftrapping_math, fwrapv.
19828 (append_compiler_options): Do not track these options.
19829 (append_linker_options): Likewie
19831 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19833 * cgraphunit.c (cgraph_node::finalize_function): When
19834 !flag_toplevel_reorde set no_reorder flag.
19835 (varpool_node::finalize_decl): Likewise.
19836 (symbol_table::compile): Drop no toplevel reorder path.
19838 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19840 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
19841 edges; zero probability is not better than uninitialized.
19843 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19845 * asan.h (asan_sanitize_allocas_p): Declare.
19846 * asan.c (asan_sanitize_allocas_p): New function.
19847 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
19848 (handle_builtin_alloca): Likewise.
19849 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
19850 if !asan_sanitize_allocas_p.
19851 * params.def (asan-instrument-allocas): Add new option.
19852 * params.h (ASAN_PROTECT_ALLOCAS): Define.
19853 * opts.c (common_handle_option): Disable allocas sanitization for
19856 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
19858 * asan.c: Include gimple-fold.h.
19859 (get_last_alloca_addr): New function.
19860 (handle_builtin_stackrestore): Likewise.
19861 (handle_builtin_alloca): Likewise.
19862 (asan_emit_allocas_unpoison): Likewise.
19863 (get_mem_refs_of_builtin_call): Add new parameter, remove const
19864 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
19865 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
19866 (instrument_builtin_call): Pass gimple iterator to
19867 get_mem_refs_of_builtin_call.
19868 (last_alloca_addr): New global.
19869 * asan.h (asan_emit_allocas_unpoison): Declare.
19870 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
19871 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
19872 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
19873 if function calls alloca.
19874 * gimple-fold.c (replace_call_with_value): Remove static keyword.
19875 * gimple-fold.h (replace_call_with_value): Declare.
19876 * internal-fn.c: Include asan.h.
19877 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
19878 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
19880 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19882 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19883 (C_SELFTEST_FLAGS): New.
19884 (CPP_SELFTEST_FLAGS): New.
19885 (SELFTEST_DEPS): New, from deps of s-selftest.
19886 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19887 (CPP_SELFTEST_DEPS): New.
19888 (selftest): Add dependency on s-selftest-c++.
19889 (s-selftest): Rename to...
19890 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19891 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19892 than SELFTEST_FLAGS.
19893 (selftest-gdb): Rename to...
19894 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19896 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19897 (selftest-valgrind): Rename to...
19898 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19900 (selftest-valgrind): Reintroduce as an alias for
19901 selftest-c-valgrind.
19902 (s-selftest-c++): New.
19903 (selftest-c++-gdb): New.
19904 (selftest-c++-valgrind): New.
19906 2017-07-06 Olivier Hainque <hainque@adacore.com>
19908 * gcc.c (process_command): When deciding if undefined variables
19909 should be ignored when processing specs, accept "gcc -v" as well.
19911 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19913 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19914 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19916 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19918 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19919 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19920 * config/arm/arm-tables.opt: Regenerate.
19921 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19923 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19925 2017-07-06 Carl Love <cel@us.ibm.com>
19927 * ChangeLog: Clean up from mid air collision
19929 2017-07-06 Carl Love <cel@us.ibm.com>
19931 * config/rs6000/rs6000-c.c: Add support for built-in functions
19932 vector signed int vec_subc (vector signed int, vector signed int);
19933 vector signed __int128 vec_subc (vector signed __int128,
19934 vector signed __int128);
19935 vector unsigned __int128 vec_subc (vector unsigned __int128,
19936 vector unsigned __int128);
19937 vector signed int vec_sube (vector signed int, vector signed int,
19938 vector signed int);
19939 vector unsigned int vec_sube (vector unsigned int,
19940 vector unsigned int,
19941 vector unsigned int);
19942 vector signed __int128 vec_sube (vector signed __int128,
19943 vector signed __int128,
19944 vector signed__int128);
19945 vector unsigned __int128 vec_sube (vector unsigned __int128,
19946 vector unsigned __int128,
19947 vector unsigned __int128);
19948 vector signed int vec_subec (vector signed int, vector signed int,
19949 vector signed int);
19950 vector unsigned int vec_subec (vector unsigned int,
19951 vector unsigned int,
19952 vector unsigned int);
19953 vector signed __int128 vec_subec (vector signed __int128,
19954 vector signed __int128,
19955 vector signed__int128);
19956 vector unsigned __int128 vec_subec (vector unsigned __int128,
19957 vector unsigned __int128,
19958 vector unsigned __int128);
19959 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19960 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19961 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19962 BU_ALTIVEC_OVERLOAD_X definitions.
19963 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19964 * doc/extend.texi: Update the built-in documentation file for the new
19965 built-in functions.
19967 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19970 * diagnostic-show-locus.c (layout::layout): Use start and finish
19971 spelling location for the start and finish of each range.
19972 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19973 Add unused aspect param.
19974 * input.c (expand_location_1): Add "aspect" param, and use it
19975 to access the correct part of the location.
19976 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19978 (expand_location_to_spelling_point): Likewise.
19979 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19980 param, and pass it to expand_location_1.
19982 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19984 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19985 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19986 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19987 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19988 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19989 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19990 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19991 _mm_maskz_getmant_ss): New intrinsics.
19992 (__builtin_ia32_getexpss128_mask): Changed to ...
19993 __builtin_ia32_getexpss128_round ... this.
19994 (__builtin_ia32_getexpsd128_mask): Changed to ...
19995 __builtin_ia32_getexpsd128_round ... this.
19996 * config/i386/i386-builtin-types.def
19997 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19998 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19999 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20000 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20001 __builtin_ia32_getmantss_mask_round): New builtins.
20002 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20003 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20004 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20005 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20006 * config/i386/sse.md
20007 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20008 avx512f_sgetexp<mode><mask_scalar_name>
20009 <round_saeonly_scalar_name> ... this.
20010 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20011 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20012 vgetexp<ssescalarmodesuffix>
20013 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20014 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20015 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20016 avx512f_vgetmant<mode><mask_scalar_name>
20017 <round_saeonly_scalar_name> ... this.
20018 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20019 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20020 vgetmant<ssescalarmodesuffix>
20021 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20022 %0<mask_scalar_operand4>, %1, %2
20023 <round_saeonly_scalar_mask_op4>, %3} ... this.
20024 * config/i386/subst.md (mask_scalar_operand4,
20025 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20026 round_saeonly_scalar_nimm_predicate): New subst attributes.
20028 2017-07-06 Julia Koval <julia.koval@intel.com>
20030 * config/i386/i386.c (ix86_erase_embedded_rounding):
20031 Remove code for old rounding pattern.
20033 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20035 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20037 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20039 * doc/sourcebuild.texi (Test Directives, Variants of
20040 dg-require-support): Add documentation for dg-require-stack-check.
20042 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20044 * config/i386/subst.md (mask_scalar, round_scalar,
20045 round_saeonly_scalar): New meta-templates.
20046 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20047 round_scalar_mask_operand3, round_scalar_mask_op3,
20048 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20049 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20050 round_saeonly_scalar_constraint,
20051 round_saeonly_scalar_prefix): New subst attribute.
20052 * config/i386/sse.md
20053 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20054 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20055 <round_scalar_name> ... this.
20056 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20057 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20058 <round_scalar_name> ... this.
20059 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20060 <sse>_vm<code><mode>3<mask_scalar_name>
20061 <round_saeonly_scalar_name> ... this.
20062 (v<plusminus_mnemonic><ssescalarmodesuffix>
20063 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20064 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20065 v<plusminus_mnemonic><ssescalarmodesuffix>
20066 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20067 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20068 (v<multdiv_mnemonic><ssescalarmodesuffix>
20069 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20070 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20071 v<multdiv_mnemonic><ssescalarmodesuffix>
20072 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20073 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20074 (v<maxmin_float><ssescalarmodesuffix>
20075 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20076 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20077 v<maxmin_float><ssescalarmodesuffix>
20078 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20079 %0<mask_scalar_operand3>, %1, %<iptr>2
20080 <round_saeonly_scalar_mask_op3>} ... this.
20082 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20084 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20085 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20087 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20088 Alan Hayward <alan.hayward@arm.com>
20089 David Sherwood <david.sherwood@arm.com>
20091 * combine.c (simplify_if_then_else): Remove "enum" before
20093 * compare-elim.c (can_eliminate_compare): Likewise.
20094 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20096 (aarch64_lookup_simd_builtin_type): Likewise.
20097 (aarch64_simd_builtin_type): Likewise.
20098 (aarch64_init_simd_builtin_types): Likewise.
20099 (aarch64_simd_expand_args): Likewise.
20100 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20102 (aarch64_reverse_mask): Likewise.
20103 (aarch64_simd_emit_reg_reg_move): Likewise.
20104 (aarch64_gen_adjusted_ldpstp): Likewise.
20105 (aarch64_ccmp_mode_to_code): Likewise.
20106 (aarch64_operands_ok_for_ldpstp): Likewise.
20107 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20108 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20110 (aarch64_min_divisions_for_recip_mul): Likewise.
20111 (aarch64_reassociation_width): Likewise.
20112 (aarch64_get_condition_code_1): Likewise.
20113 (aarch64_simd_emit_reg_reg_move): Likewise.
20114 (aarch64_simd_attr_length_rglist): Likewise.
20115 (aarch64_reverse_mask): Likewise.
20116 (aarch64_operands_ok_for_ldpstp): Likewise.
20117 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20118 (aarch64_gen_adjusted_ldpstp): Likewise.
20119 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20121 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20122 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20123 (arm_lookup_simd_builtin_type): Likewise.
20124 (arm_simd_builtin_type): Likewise.
20125 (arm_init_simd_builtin_types): Likewise.
20126 (arm_expand_builtin_args): Likewise.
20127 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20128 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20129 (ft32_setup_incoming_varargs): Likewise.
20130 (ft32_function_arg): Likewise.
20131 (ft32_function_arg_advance): Likewise.
20132 (ft32_pass_by_reference): Likewise.
20133 (ft32_arg_partial_bytes): Likewise.
20134 (ft32_valid_pointer_mode): Likewise.
20135 (ft32_addr_space_pointer_mode): Likewise.
20136 (ft32_addr_space_legitimate_address_p): Likewise.
20137 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20139 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20140 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20141 (iamcu_alignment): Likewise.
20142 (canonicalize_vector_int_perm): Likewise.
20143 (ix86_noce_conversion_profitable_p): Likewise.
20144 (ix86_mpx_bound_mode): Likewise.
20145 (ix86_operands_ok_for_move_multiple): Likewise.
20146 * config/microblaze/microblaze-protos.h
20147 (microblaze_expand_conditional_branch_reg): Likewise.
20148 * config/microblaze/microblaze.c
20149 (microblaze_expand_conditional_branch_reg): Likewise.
20150 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20152 (rs6000_reassociation_width): Likewise.
20153 (rs6000_invalid_binary_op): Likewise.
20154 (fusion_p9_p): Likewise.
20155 (emit_fusion_p9_load): Likewise.
20156 (emit_fusion_p9_store): Likewise.
20157 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20159 (riscv_hard_regno_mode_ok_p): Likewise.
20160 (riscv_address_insns): Likewise.
20161 (riscv_split_symbol): Likewise.
20162 (riscv_legitimize_move): Likewise.
20163 (riscv_function_value): Likewise.
20164 (riscv_hard_regno_nregs): Likewise.
20165 (riscv_expand_builtin): Likewise.
20166 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20167 (riscv_build_integer): Likewise.
20168 (riscv_split_integer): Likewise.
20169 (riscv_legitimate_constant_p): Likewise.
20170 (riscv_cannot_force_const_mem): Likewise.
20171 (riscv_regno_mode_ok_for_base_p): Likewise.
20172 (riscv_valid_base_register_p): Likewise.
20173 (riscv_valid_offset_p): Likewise.
20174 (riscv_valid_lo_sum_p): Likewise.
20175 (riscv_classify_address): Likewise.
20176 (riscv_legitimate_address_p): Likewise.
20177 (riscv_address_insns): Likewise.
20178 (riscv_load_store_insns): Likewise.
20179 (riscv_force_binary): Likewise.
20180 (riscv_split_symbol): Likewise.
20181 (riscv_force_address): Likewise.
20182 (riscv_legitimize_address): Likewise.
20183 (riscv_move_integer): Likewise.
20184 (riscv_legitimize_const_move): Likewise.
20185 (riscv_legitimize_move): Likewise.
20186 (riscv_address_cost): Likewise.
20187 (riscv_subword): Likewise.
20188 (riscv_output_move): Likewise.
20189 (riscv_canonicalize_int_order_test): Likewise.
20190 (riscv_emit_int_order_test): Likewise.
20191 (riscv_function_arg_boundary): Likewise.
20192 (riscv_pass_mode_in_fpr_p): Likewise.
20193 (riscv_pass_fpr_single): Likewise.
20194 (riscv_pass_fpr_pair): Likewise.
20195 (riscv_get_arg_info): Likewise.
20196 (riscv_function_arg): Likewise.
20197 (riscv_function_arg_advance): Likewise.
20198 (riscv_arg_partial_bytes): Likewise.
20199 (riscv_function_value): Likewise.
20200 (riscv_pass_by_reference): Likewise.
20201 (riscv_setup_incoming_varargs): Likewise.
20202 (riscv_print_operand): Likewise.
20203 (riscv_elf_select_rtx_section): Likewise.
20204 (riscv_save_restore_reg): Likewise.
20205 (riscv_for_each_saved_reg): Likewise.
20206 (riscv_register_move_cost): Likewise.
20207 (riscv_hard_regno_mode_ok_p): Likewise.
20208 (riscv_hard_regno_nregs): Likewise.
20209 (riscv_class_max_nregs): Likewise.
20210 (riscv_memory_move_cost): Likewise.
20211 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20212 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20213 (rl78_addr_space_address_mode): Likewise.
20214 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20216 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20217 (rs6000_reassociation_width): Likewise.
20218 (rs6000_invalid_binary_op): Likewise.
20219 (fusion_p9_p): Likewise.
20220 (emit_fusion_p9_load): Likewise.
20221 (emit_fusion_p9_store): Likewise.
20222 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20223 (ok_for_simple_move_operands): Likewise.
20224 (ok_for_simple_move_strict_operands): Likewise.
20225 (ok_for_simple_arith_logic_operands): Likewise.
20226 (visium_legitimize_reload_address): Likewise.
20227 (visium_select_cc_mode): Likewise.
20228 (output_cbranch): Likewise.
20229 (visium_split_double_move): Likewise.
20230 (visium_expand_copysign): Likewise.
20231 (visium_expand_int_cstore): Likewise.
20232 (visium_expand_fp_cstore): Likewise.
20233 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20234 (visium_function_arg): Likewise.
20235 (visium_function_arg_advance): Likewise.
20236 (visium_libcall_value): Likewise.
20237 (visium_setup_incoming_varargs): Likewise.
20238 (visium_legitimate_constant_p): Likewise.
20239 (visium_legitimate_address_p): Likewise.
20240 (visium_legitimize_address): Likewise.
20241 (visium_secondary_reload): Likewise.
20242 (visium_register_move_cost): Likewise.
20243 (visium_memory_move_cost): Likewise.
20244 (prepare_move_operands): Likewise.
20245 (ok_for_simple_move_operands): Likewise.
20246 (ok_for_simple_move_strict_operands): Likewise.
20247 (ok_for_simple_arith_logic_operands): Likewise.
20248 (visium_function_value_1): Likewise.
20249 (rtx_ok_for_offset_p): Likewise.
20250 (visium_legitimize_reload_address): Likewise.
20251 (visium_split_double_move): Likewise.
20252 (visium_expand_copysign): Likewise.
20253 (visium_expand_int_cstore): Likewise.
20254 (visium_expand_fp_cstore): Likewise.
20255 (visium_split_cstore): Likewise.
20256 (visium_select_cc_mode): Likewise.
20257 (visium_split_cbranch): Likewise.
20258 (output_cbranch): Likewise.
20259 (visium_print_operand_address): Likewise.
20260 * expmed.c (flip_storage_order): Likewise.
20261 * expmed.h (emit_cstore): Likewise.
20262 (flip_storage_order): Likewise.
20263 * genrecog.c (validate_pattern): Likewise.
20264 * hsa-gen.c (gen_hsa_addr): Likewise.
20265 * internal-fn.c (expand_arith_overflow): Likewise.
20266 * ira-color.c (allocno_copy_cost_saving): Likewise.
20267 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20268 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20269 (process_invariant_for_inheritance): Likewise.
20270 * lra-eliminations.c (move_plus_up): Likewise.
20271 * omp-low.c (lower_oacc_reductions): Likewise.
20272 * simplify-rtx.c (simplify_subreg): Likewise.
20273 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20274 (TARGET_CHKP_BOUND_MODE): Likewise..
20275 * targhooks.c (default_chkp_bound_mode): Likewise.
20276 (default_setup_incoming_vararg_bounds): Likewise.
20277 * targhooks.h (default_chkp_bound_mode): Likewise.
20278 (default_setup_incoming_vararg_bounds): Likewise.
20279 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20280 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20281 (have_whole_vector_shift): Likewise.
20282 * tree-vect-stmts.c (vectorizable_load): Likewise.
20283 * doc/tm.texi: Regenerate.
20285 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20287 Graceful degrade if Binutils PR21472 is not available.
20290 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20291 .rodata in flash test fails.
20292 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20293 * confgure: Regenerate.
20294 * config.in: Regenerate.
20295 * config/avr/avr.c (avr_asm_named_section)
20296 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20297 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20298 (avr_asm_init_sections): Same.
20300 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20302 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20303 (fma<VH:mode>4_intrinsic): Likewise.
20304 (*fmsub<VCVTF:mode>4): Likewise.
20305 (*fmsub<VH:mode>4_intrinsic): Likewise.
20307 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20310 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20311 Don't depend on "optimize > 0".
20312 (out_movhi_r_mr, out_movqi_mr_r): Same.
20313 (out_movhi_mr_r, out_movqi_r_mr): Same.
20314 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20315 io_address_operand on "optimize > 0".
20317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20319 * tree-loop-distribution.c: Add general explanantion on the pass.
20320 (generate_loops_for_partition): Mark distributed loop.
20321 (pg_add_dependence_edges): New parameter. Handle alias data
20322 dependence specially and record it in the parameter if asked.
20323 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20324 (init_partition_graph_vertices, add_partition_graph_edge): New.
20325 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20326 (free_partition_graph_vdata, build_partition_graph): New.
20327 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20328 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20329 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20330 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20331 (version_for_distribution_p, finalize_partitions): New.
20332 (distribute_loop): Handle alias data dependence specially. Factor
20333 out loop fusion code as functions and call these functions.
20335 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20337 * tree-loop-distribution.c (classify_partition): New parameter and
20338 better handle reduction statement.
20339 (rdg_build_partitions): Revise comment.
20340 (distribute_loop): Compute statements in all partitions and pass it
20341 to classify_partition.
20343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20345 * tree-loop-distribution.c (enum partition_type): New.
20346 (struct partition): New field type.
20347 (partition_merge_into): Add parameter. Update partition type.
20348 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20349 (build_rdg_partition_for_vertex): Compute partition type.
20350 (rdg_build_partitions): Dump partition type.
20351 (distribute_loop): Update calls to partition_merge_into.
20353 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20355 * tree-loop-distribution.c (struct ddr_hasher): New.
20356 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20358 (classify_partition): Call get_data_dependence.
20359 (pg_add_dependence_edges): Ditto.
20360 (distribute_loop): Release data dependence hash table.
20362 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20364 * tree-loop-distribution.c (ref_base_address): Delete.
20365 (similar_memory_accesses): Rename ...
20366 (share_memory_accesses): ... to this. Check if partitions access
20367 the same memory reference.
20368 (distribute_loop): Call share_memory_accesses.
20370 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20372 * tree-loop-distribution.c (struct partition): New field recording
20373 its data reference.
20374 (partition_alloc, partition_free): Init and release data refs.
20375 (partition_merge_into): Merge data refs.
20376 (build_rdg_partition_for_vertex): Collect data refs for partition.
20377 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20379 (distribute_loop): Remve data refs from vertice data of partition
20382 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20384 * tree-loop-distribution.c (params.h): Include header file.
20385 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20386 (datarefs_vec): New global var.
20387 (create_rdg_vertices): Use datarefs_vec directly.
20388 (free_rdg): Don't free data references.
20389 (build_rdg): Update use. Don't free data references.
20390 (distribute_loop): Compute global variable for data references.
20391 Bail out if there are too many data references.
20393 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20395 * tree-loop-distribution.c (loop_nest): New global var.
20396 (build_rdg): Use loop directly, rather than loop nest.
20397 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20399 (distribute_loop): Compute global variable loop nest. Update use.
20401 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20403 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20404 (partition_merge_into): New parameter. Dump reason for fusion.
20405 (distribute_loop): Update use of partition_merge_into.
20407 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20409 * tree-loop-distribution.c (bb_top_order_index): New.
20410 (bb_top_order_index_size, bb_top_order_cmp): New.
20411 (stmts_from_loop): Use topological order.
20412 (pass_loop_distribution::execute): Compute and release topological
20413 order for basic blocks.
20415 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20417 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20420 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20422 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20423 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20424 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20425 * internal-fn.def (LOOP_DIST_ALIAS): New.
20426 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20427 (fold_loop_internal_call): ... this.
20428 (vect_loop_dist_alias_call): New function.
20429 (set_uid_loop_bbs): Call fold_loop_internal_call.
20430 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20433 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20436 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20437 Require dead FLAGS_REG at the beginning of a peephole.
20439 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20442 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20443 arguments in the call to __builtin_ia32_sbb_u32.
20444 (_subborrow_u64): Swap _X and _Y arguments in the call to
20445 __builtin_ia32_sbb_u64.
20447 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20450 * tree-vrp.c (compare_assert_loc): Turn into a function template
20451 with stable template parameter. Only test if a->e is NULL,
20452 !a->e == !b->e has been verified already. Use e == NULL or
20453 e != NULL instead of e or ! e tests. If stable is true, don't use
20454 iterative_hash_expr, on the other side allow a or b or both NULL
20455 and sort the NULLs last.
20456 (process_assert_insertions): Sort using compare_assert_loc<false>
20457 instead of compare_assert_loc, later sort using
20458 compare_assert_loc<true> before calling process_assert_insertions_for
20459 in a loop. Use break instead of continue once seen NULL pointer.
20461 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20463 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20464 Cortex-R7 and Cortex-R8 processors.
20466 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20468 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20469 uninitialized while src is not.
20471 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20473 * common/config/arm/arm-common.c: Adjust include path for
20475 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20476 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20477 (arm-cpu-data.h): Likewise.
20478 (arm-cpu-cdata.h): Likewise.
20479 * config/arm/arm-cpu.h: Delete.
20480 * config/arm/arm-cpu-cdata.h: Delete.
20481 * config/arm/arm-cpu-data.h: Delete.
20483 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20485 * config/arm/arm-cpus.in (cortex-a55): New.
20486 (cortex-a75): Likewise.
20487 (cortex-a75.cortex-a55): Likewise.
20488 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20490 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20491 * config/arm/arm-cpu-cdata.h: Regenerate.
20492 * config/arm/arm-cpu-data.h: Regenerate.
20493 * config/arm/arm-cpu.h: Regenerate.
20494 * config/arm/arm-tables.opt: Regenerate.
20495 * config/arm/arm-tune.md: Regenerate.
20497 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20499 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20501 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20503 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20506 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20508 PR tree-optimization/81292
20509 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20510 full_string_p, also call adjust_related_strinfos if the adjustment
20511 is simple, otherwise invalidate related strinfos.
20513 2017-07-04 Martin Liska <mliska@suse.cz>
20516 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20517 newly created variable as DECL_IGNORED_P.
20519 2017-07-04 Martin Liska <mliska@suse.cz>
20522 * ipa-inline.c (inline_small_functions):
20523 Use xstrdup_for_dump.
20525 2017-07-04 Tom de Vries <tom@codesourcery.com>
20527 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20529 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20532 * config/darwin.c (darwin_function_switched_text_sections):
20535 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20537 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20539 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20541 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20543 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20545 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20546 min_profitable_iters, and th as inclusive lower bounds.
20547 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20548 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20549 for min_profitable_iters and min_profitable_estimate.
20550 (vect_transform_loop): Treat th as an inclusive lower bound.
20551 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20553 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20556 * config/darwin.c (darwin_function_switched_text_sections):
20557 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20558 in two pieces, and suppress the use of buf.
20560 2017-07-03 Nathan Sidwell <nathan@acm.org>
20562 * hash-table.h (hash_table_mod1): Fix indentation.
20564 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20566 PR middle-end/81290
20567 * predict.c (force_edge_cold): Be more careful about propagation
20569 * profile-count.h (profile_probability::guessed,
20570 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20572 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20574 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20576 * doc/invoke.texi (rcpc architecture extension): Document it.
20578 2017-07-03 Richard Biener <rguenther@suse.de>
20580 PR tree-optimization/60510
20581 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20582 the scalar reduction PHI and use it.
20583 (vectorizable_reduction): Properly guard the single_defuse_cycle
20584 path for non-SLP reduction chains where we cannot use it.
20585 Rework reduc_def/index and vector type deduction. Rework
20586 vector operand gathering during reduction op code-gen.
20587 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20588 chains dissolve the chain and leave it to non-SLP reduction
20591 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20593 * tree-data-ref.h (dr_alignment): Declare.
20594 * tree-data-ref.c (dr_alignment): New function.
20595 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20596 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20598 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20600 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20602 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20603 and base_misalignment fields.
20604 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20605 * tree-data-ref.c: Include builtins.h.
20606 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20607 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20608 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20609 * tree-vect-data-refs.c: Include tree-cfg.h.
20610 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20611 fields instead of calculating an alignment here.
20612 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20613 innermost_loop_behavior fields.
20615 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20617 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20619 (DR_STEP_ALIGNMENT): New macro.
20620 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20621 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20622 (create_data_ref): Print it.
20623 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20624 to tell whether the step preserves vector (mis)alignment.
20625 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20626 Move the check for an integer step and generalise to all INTEGER_CST.
20627 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20628 Print the outer step alignment.
20630 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20632 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20633 with offset_alignment.
20634 (DR_ALIGNED_TO): Delete.
20635 (DR_OFFSET_ALIGNMENT): New macro.
20636 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20637 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20638 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20639 (create_data_ref): Likewise.
20640 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20641 (vect_analyze_data_refs): Likewise.
20642 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20643 creating dummy innermost behavior.
20645 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20647 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20648 with a "innermost_loop_behavior *" and refeence tree.
20649 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20650 (create_data_ref): Update call accordingly.
20651 * tree-predcom.c (find_looparound_phi): Likewise.
20653 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20655 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20656 fields with dr_wrt_vec_loop.
20657 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20658 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20659 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20660 (vect_dr_behavior): New function.
20661 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20662 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20663 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20664 track whether the step preserves the misalignment.
20665 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20666 Use vect_dr_behavior.
20667 (vect_setup_realignment): Update call accordingly.
20668 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20669 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20670 call to vect_create_addr_base_for_vector_ref.
20671 (vect_create_cond_for_align_checks): Likewise.
20672 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20673 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20674 (vect_recog_mask_conversion_pattern): Likewise.
20675 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20676 (new_stmt_vec_info): Remove redundant zeroing.
20678 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20680 * common/config/arm/arm-common.c (arm_be8_option): New function.
20681 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20682 (ISA_ARMv6): Add isa_bit_be8.
20683 * config/arm/arm.h (arm_be8_option): Add prototype.
20684 (BE8_SPEC_FUNCTION): New define.
20685 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20686 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20687 (mlittle-endian): Similarly.
20688 (mbe8, mbe32): New options.
20689 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20690 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20692 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20694 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20696 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20698 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
20699 (cleanup_tree_cfg_bb): Use it.
20700 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
20702 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
20704 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20707 * loop-doloop.c (add_test): Update profile.
20709 2017-07-03 Martin Liska <mliska@suse.cz>
20712 * sanopt.c (rewrite_usage_of_param): New function.
20713 (sanitize_rewrite_addressable_params): Likewise.
20714 (pass_sanopt::execute): Call rewrite_usage_of_param.
20716 2017-07-03 Richard Biener <rguenther@suse.de>
20718 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
20719 back to using VIEW_CONVERT_EXPR.
20721 2017-07-03 Martin Liska <mliska@suse.cz>
20724 * doc/extend.texi: Document when a resolver function is
20725 generated for target_clones.
20727 2017-07-03 Martin Liska <mliska@suse.cz>
20729 * asan.c (asan_emit_stack_protection): Unpoison just red zones
20730 and shadow memory of auto variables which are subject of
20731 use-after-scope sanitization.
20732 (asan_expand_mark_ifn): Add do set only when is_poison.
20734 2016-07-03 Richard Biener <rguenther@suse.de>
20736 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
20738 (vect_force_simple_reduction): Record reduction def -> phi mapping.
20739 (vectorizable_reduction): Perform reduction PHI creation when
20740 visiting a reduction PHI and adjust and simplify code generation
20741 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
20742 (vect_transform_loop): Visit reduction PHIs.
20743 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
20744 defs into the SLP tree.
20745 (vect_build_slp_tree): Reduction defs terminate the recursion.
20746 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
20748 (vect_get_vec_defs_for_stmt_copy): Export.
20749 (vect_get_vec_defs): Likewise.
20750 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
20752 (vect_get_vec_defs_for_stmt_copy): Declare.
20753 (vect_get_vec_defs): Likewise.
20755 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20757 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
20758 parameter with a "loop" parameter and use it instead of the
20759 loop containing DR_STMT. Don't check simple_iv when doing
20760 BB analysis. Describe the two analysis modes in the comment.
20762 2017-07-03 Tom de Vries <tom@codesourcery.com>
20764 PR tree-optimization/69468
20765 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
20766 (find_same_succ_bb): Handle ignore_edge_flags.
20768 2017-07-03 Tom de Vries <tom@codesourcery.com>
20770 PR tree-optimization/81192
20771 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
20773 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
20775 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
20777 2017-07-03 Tom de Vries <tom@codesourcery.com>
20779 PR tree-optimization/81192
20780 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
20781 BB_SAME_SUCC (bb) == NULL.
20783 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20785 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
20788 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20790 * dumpfile.c: Include profile-count.h
20791 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
20793 (insert_cond_bb): Update profile.
20794 * tree-cfg.h (insert_cond_bb): Update prototype.
20795 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
20796 * tree-dump.c: Do not include tree-cfg.
20798 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20800 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
20802 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20804 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
20807 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20809 * tree-complex.c (expand_complex_div_wide): update profile.
20811 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20812 Alan Hayward <alan.hayward@arm.com>
20813 David Sherwood <david.sherwood@arm.com>
20815 * Makefile.in (MACHMODE_H): Remove insn-modes.h
20816 (CORETYPES_H): New define.
20817 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
20818 (insn-modes-inline.h, s-modes-inline-h): New rules.
20819 (generated_files): Add insn-modes-inline.h.
20820 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
20821 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
20822 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
20823 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
20824 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
20825 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
20826 (build/gencodes.o, build/genconditions.o): Likewise.
20827 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
20828 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
20829 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
20830 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
20831 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
20832 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
20833 * coretypes.h: Include everything up to real.h for generators.
20834 Include insn-modes.h first. Include wide-int-print.h after
20835 wide-int.h. Include insn-modes-inline.h and then machmode.h.
20836 * machmode.h: Don't include insn-modes.h here.
20837 * function-tests.c: Remove includes of signop.h, machmode.h,
20838 double-int.h and wide-int.h.
20840 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
20842 * optc-save-gen.awk: Likewise.
20843 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
20844 * godump.c: Remove include of wide-int-print.h.
20845 * pretty-print.h: Likewise.
20846 * wide-int-print.cc: Likewise.
20847 * wide-int.cc: Likewise.
20848 * hash-map-tests.c: Remove include of signop.h.
20849 * hash-set-tests.c: Likewise.
20850 * rtl-tests.c: Likewise.
20851 * mkconfig.sh: Remove include of machmode.h.
20852 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
20854 (emit_insn_modes_inline_h): ...this new function. Emit the code
20855 into an insn-modes-inline.h header file, adding appropriate
20856 include guards and end comments.
20857 (emit_insn_modes_c_header): Remove include of machmode.h.
20858 (emit_min_insn_modes_c_header): Include coretypes.h rather than
20860 (main): Handle -i flag and call emit_insn_modes_inline_h when
20863 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20865 * tree-ssa-strlen.c (strinfo): Rename the length field to
20866 nonzero_chars. Add a full_string_p field.
20867 (compare_nonzero_chars, zero_length_string_p): New functions.
20868 (get_addr_stridx): Add an offset_out parameter.
20869 Use compare_nonzero_chars.
20870 (get_stridx): Update accordingly. Use compare_nonzero_chars.
20871 (new_strinfo): Update after above changes to strinfo.
20872 (set_endptr_and_length): Set full_string_p.
20873 (get_string_length): Update after above changes to strinfo.
20874 (unshare_strinfo): Update call to new_strinfo.
20875 (maybe_invalidate): Likewise.
20876 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
20877 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
20878 as a uhwi instead of an shwi. Update after above changes to
20879 strinfo and new_strinfo.
20880 (zero_length_string): Assert that chainsi contains full strings.
20881 Use zero_length_string_p. Update call to new_strinfo.
20882 (adjust_related_strinfos): Update after above changes to strinfo.
20883 Copy full_string_p from origsi.
20884 (adjust_last_stmt): Use zero_length_string_p.
20885 (handle_builtin_strlen): Update after above changes to strinfo and
20886 new_strinfo. Install the lhs as the string length if the previous
20887 entry didn't describe a full string.
20888 (handle_builtin_strchr): Update after above changes to strinfo
20890 (handle_builtin_strcpy): Likewise.
20891 (handle_builtin_strcat): Likewise.
20892 (handle_builtin_malloc): Likewise.
20893 (handle_pointer_plus): Likewise.
20894 (handle_builtin_memcpy): Likewise. Track nonzero characters
20895 that aren't necessarily followed by a nul terminator.
20896 (handle_char_store): Likewise.
20898 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20900 PR tree-optimization/80769
20901 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20902 for malloc and calloc. Document the new invariant that all related
20903 strinfos have delayed lengths or none do.
20904 (verify_related_strinfos): Move earlier in file.
20905 (set_endptr_and_length): New function, split out from...
20906 (get_string_length): ...here. Also set the lengths of related
20908 (zero_length_string): Assert that chainsi has known (rather than
20910 (adjust_related_strinfos): Likewise.
20912 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20914 PR tree-optimization/81136
20915 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20916 assert that two references with the same misalignment have the same
20917 compile-time misalignment if those compile-time misalignments
20920 2017-07-01 Andi Kleen <ak@linux.intel.com>
20922 * print-tree.c (print_node): Print all attributes.
20924 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20926 * cfg.c (scale_bbs_frequencies): New function.
20927 * cfg.h (scale_bbs_frequencies): Declare it.
20928 * cfgloopanal.c (single_likely_exit): Cleanup.
20929 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20931 (scale_loop_profile): Likewise.
20932 (loop_version): Likewise.
20933 (create_empty_loop_on_edge): Update.
20934 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20935 scale_loop_frequencies, scale_loop_profile, loopify,
20936 loop_version): Update prototypes.
20937 * modulo-sched.c (sms_schedule): Update.
20938 * predict.c (unlikely_executed_edge_p): Also check probability.
20939 (probably_never_executed_edge_p): Fix typo.
20940 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20941 * tree-parloops.c (gen_parallel_loop): Update.
20942 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20943 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20944 * tree-ssa-loop-split.c (split_loop): Update.
20945 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20946 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20947 (vect_loop_versioning): Update.
20948 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20950 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20952 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20954 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20956 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20957 to keep profile consistent.
20959 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20961 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20962 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20963 * profile-count.h (max_safe_multiplier): Make unsigned.
20964 (profile_count::guessed_zero): New.
20966 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20968 * bb-reorder.c (fix_up_crossing_landing_pad,
20969 fix_crossing_conditional_branches): Use make_single_succ_edge
20970 to keep profile consistent.
20972 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20974 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20977 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20980 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20981 the right scopes, make sure cond_jump isn't preserved between multiple
20982 iterations. Search for fallthru edge whenever there are 3+ edges and
20983 use find_fallthru_edge for it.
20985 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20987 Patch by Alexander Monakov <amonakov@ispras.ru>
20988 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20989 probabilities consistently.
20991 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20993 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20994 emit_cmp_and_jump_insns.
20996 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20999 * tree-inline.c (expand_call_inline): Combine profile statuses.
21001 2017-06-30 Andrew Pinski <apinski@cavium.com>
21003 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21004 fold_stmt returned true.
21006 2017-06-30 Nathan Sidwell <nathan@acm.org>
21008 * ggc.h (empty_string): Delete.
21009 * cfgexpand.c (expand_asm_stmt): Use plain "".
21010 * optabs.c (expand_asm_memory_barrier): Likewise.
21011 * stringpool.c (empty_string): Delete.
21012 (digit_vector, digit_string): Delete.
21013 (ggc_alloc_string): Use plain "", don't optimize single digit
21014 strings. Use ggc_alloc_atomic.
21016 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21018 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21019 comparison set and one other set, use the cost of the non-comparison
21022 2017-06-30 Nathan Sidwell <nathan@acm.org>
21024 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21027 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21029 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21030 loops. Remove now unneeded calls to gimple_switch_set_label() that
21031 just set removed labels to NULL_TREE.
21033 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21035 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21036 (set_range_info): ...here. Only call set_range_info_raw if domain
21038 (set_nonzero_bits): Call set_range_info_raw.
21039 * tree-ssanames.h (set_range_info_raw): New.
21041 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21044 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21045 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21046 of nonimmediate_operand and <store_mask_constraint> instead of m
21047 for the input operand. For V8FI iterator, always split if input
21048 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21049 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21050 <store_mask_predicate> instead of register_operand and
21051 <store_mask_constraint> instead of v for the input operand. Make
21052 sure both operands aren't MEMs for if not <mask_applied>.
21054 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21056 * lto-wrapper.c (copy_file) Close both file descriptors before
21059 2017-06-30 Martin Liska <mliska@suse.cz>
21062 * multiple_target.c (create_dispatcher_calls): Make ifunc
21063 also for function that don't have calls or are not referenced.
21065 2017-06-30 Richard Biener <rguenther@suse.de>
21067 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21068 analyze the first scalar stmt. Move vector type computation
21069 for the BB case here from ...
21070 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21071 live operation processing in the SLP case properly.
21073 2017-06-30 Richard Biener <rguenther@suse.de>
21075 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21077 2017-06-30 Martin Liska <mliska@suse.cz>
21080 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21081 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21083 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21085 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21088 2017-06-30 Richard Biener <rguenther@suse.de>
21090 PR tree-optimization/81249
21091 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21092 condition reduction result to original scalar type.
21094 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21096 * profile-count.h (enum profile_quality): Fix typos and whitespace
21099 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21101 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21102 type for branch probabilities.
21104 2017-06-29 Julian Brown <julian@codesourcery.com>
21105 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21107 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21108 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21109 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21110 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21112 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21114 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21115 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21116 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21117 CC usage from generic code to here.
21118 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21119 CC usage into the target macros.
21121 2017-06-29 Maya Rashish <coypu@sdf.org>
21123 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21126 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21128 * arm/arm-builtins.c: Include profile-count.h
21129 * except.c (sjlj_emit_function_enter): Use
21130 profile_probability::unlikely.
21132 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21134 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21135 and tocrel_offset be pointer args rather than implicitly using
21137 (legitimate_constant_pool_address_p, rs6000_emit_move,
21138 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21139 tocrel_offset and use in toc_relative_expr_p call.
21140 (print_operand, print_operand_address): Use static tocrel_base_oac
21141 and tocrel_offset_oac.
21142 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21145 2017-06-29 Maya Rashish <coypu@sdf.org>
21147 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21149 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21151 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21152 objects, take into account only the alignment of 'op0' and 'mode1' if
21155 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21157 * ccmp.c (ccmp_tree_comparison_p): New function.
21158 (ccmp_candidate_p): Update to use above function.
21159 (get_compare_parts): New function.
21160 (expand_ccmp_next): Update to use new functions.
21161 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21163 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21164 take mode as argument.
21165 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21166 * expr.c (expand_expr_real_1): Pass mode as argument.
21168 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21170 * combine.c (combine_instructions): Print insns to dump_file, together
21173 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21175 * asan.c (asan_emit_stack_protection): Update.
21176 (create_cond_insert_point): Update.
21177 * auto-profile.c (afdo_propagate_circuit): Update.
21178 * basic-block.h (struct edge_def): Turn probability to
21179 profile_probability.
21180 (EDGE_FREQUENCY): Update.
21181 * bb-reorder.c (find_traces_1_round): Update.
21182 (better_edge_p): Update.
21183 (sanitize_hot_paths): Update.
21184 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21185 (make_single_succ_edge): Update.
21186 (check_bb_profile): Update.
21187 (dump_edge_info): Update.
21188 (update_bb_profile_for_threading): Update.
21189 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21190 probabilitycount to 0.
21191 * cfgbuild.c (compute_outgoing_frequencies): Update.
21192 * cfgcleanup.c (try_forward_edges): Update.
21193 (outgoing_edges_match): Update.
21194 (try_crossjump_to_edge): Update.
21195 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21196 (expand_gimple_tailcall): Update.
21197 (construct_init_block): Use make_single_succ_edge.
21198 (construct_exit_block): Use make_single_succ_edge.
21199 * cfghooks.c (verify_flow_info): Update.
21200 (redirect_edge_succ_nodup): Update.
21201 (split_edge): Update.
21202 (account_profile_record): Update.
21203 * cfgloopanal.c (single_likely_exit): Update.
21204 * cfgloopmanip.c (scale_loop_profile): Update.
21205 (set_zero_probability): Remove.
21206 (duplicate_loop_to_header_edge): Update.
21207 * cfgloopmanip.h (loop_version): Update prototype.
21208 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21209 (force_nonfallthru_and_redirect): Update.
21210 (update_br_prob_note): Update.
21211 (rtl_verify_edges): Update.
21212 (purge_dead_edges): Update.
21213 (rtl_lv_add_condition_to_bb): Update.
21214 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21215 * cgraphunit.c (init_lowered_empty_function): Update.
21216 (cgraph_node::expand_thunk): Update.
21217 * cilk-common.c: Include profile-count.h
21218 * dojump.c (inv): Remove.
21219 (jumpifnot): Update.
21220 (jumpifnot_1): Update.
21221 (do_jump_1): Update.
21223 (do_jump_by_parts_greater_rtx): Update.
21224 (do_compare_rtx_and_jump): Update.
21225 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21226 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21227 * dwarf2cfi.c: Include profile-count.h
21228 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21229 (sjlj_emit_dispatch_table): Likewise.
21230 * explow.c: Include profile-count.h
21231 * expmed.c (emit_store_flag_force): Update.
21232 (do_cmp_and_jump): Update.
21233 * expr.c (compare_by_pieces_d::generate): Update.
21234 (compare_by_pieces_d::finish_mode): Update.
21235 (emit_block_move_via_loop): Update.
21236 (store_expr_with_bounds): Update.
21237 (store_constructor): Update.
21238 (expand_expr_real_2): Update.
21239 (expand_expr_real_1): Update.
21240 * expr.h (try_casesi, try_tablejump): Update prototypes.
21241 * gimple-pretty-print.c (dump_probability): Update.
21242 (dump_profile): New.
21243 (dump_gimple_label): Update.
21244 (dump_gimple_bb_header): Update.
21245 * graph.c (draw_cfg_node_succ_edges): Update.
21246 * hsa-gen.c (convert_switch_statements): Update.
21247 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21248 (find_if_case_1): Update.
21249 (find_if_case_2): Update.
21250 * internal-fn.c (expand_arith_overflow_result_store): Update.
21251 (expand_addsub_overflow): Update.
21252 (expand_neg_overflow): Update.
21253 (expand_mul_overflow): Update.
21254 (expand_vector_ubsan_overflow): Update.
21255 * ipa-cp.c (good_cloning_opportunity_p): Update.
21256 * ipa-split.c (split_function): Use make_single_succ_edge.
21257 * ipa-utils.c (ipa_merge_profiles): Update.
21258 * loop-doloop.c (add_test): Update.
21259 (doloop_modify): Update.
21260 * loop-unroll.c (compare_and_jump_seq): Update.
21261 (unroll_loop_runtime_iterations): Update.
21262 * lra-constraints.c (lra_inheritance): Update.
21263 * lto-streamer-in.c (input_cfg): Update.
21264 * lto-streamer-out.c (output_cfg): Update.
21265 * mcf.c (adjust_cfg_counts): Update.
21266 * modulo-sched.c (sms_schedule): Update.
21267 * omp-expand.c (expand_omp_for_init_counts): Update.
21268 (extract_omp_for_update_vars): Update.
21269 (expand_omp_ordered_sink): Update.
21270 (expand_omp_for_ordered_loops): Update.
21271 (expand_omp_for_generic): Update.
21272 (expand_omp_for_static_nochunk): Update.
21273 (expand_omp_for_static_chunk): Update.
21274 (expand_cilk_for): Update.
21275 (expand_omp_simd): Update.
21276 (expand_omp_taskloop_for_outer): Update.
21277 (expand_omp_taskloop_for_inner): Update.
21278 * omp-simd-clone.c (simd_clone_adjust): Update.
21279 * optabs.c (expand_doubleword_shift): Update.
21280 (expand_abs): Update.
21281 (emit_cmp_and_jump_insn_1): Update.
21282 (expand_compare_and_swap_loop): Update.
21283 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21284 * predict.c (predictable_edge_p): Update.
21285 (edge_probability_reliable_p): Update.
21286 (set_even_probabilities): Update.
21287 (combine_predictions_for_insn): Update.
21288 (combine_predictions_for_bb): Update.
21289 (propagate_freq): Update.
21290 (estimate_bb_frequencies): Update.
21291 (force_edge_cold): Update.
21292 * profile-count.c (profile_count::dump): Add missing space into dump.
21293 (profile_count::debug): Add newline.
21294 (profile_count::differs_from_p): Explicitly convert to unsigned.
21295 (profile_count::stream_in): Update.
21296 (profile_probability::dump): New member function.
21297 (profile_probability::debug): New member function.
21298 (profile_probability::differs_from_p): New member function.
21299 (profile_probability::differs_lot_from_p): New member function.
21300 (profile_probability::stream_in): New member function.
21301 (profile_probability::stream_out): New member function.
21302 * profile-count.h (profile_count_quality): Rename to ...
21303 (profile_quality): ... this one.
21304 (profile_probability): New.
21305 (profile_count): Update.
21306 * profile.c (compute_branch_probabilities): Update.
21307 * recog.c (peep2_attempt): Update.
21308 * sched-ebb.c (schedule_ebbs): Update.
21309 * sched-rgn.c (find_single_block_region): Update.
21310 (compute_dom_prob_ps): Update.
21311 (schedule_region): Update.
21312 * sel-sched-ir.c (compute_succs_info): Update.
21313 * stmt.c (struct case_node): Update.
21314 (do_jump_if_equal): Update.
21315 (get_outgoing_edge_probs): Update.
21316 (conditional_probability): Update.
21317 (emit_case_dispatch_table): Update.
21318 (expand_case): Update.
21319 (expand_sjlj_dispatch_table): Update.
21320 (emit_case_nodes): Update.
21321 * targhooks.c: Update.
21322 * tracer.c (better_p): Update.
21323 (find_best_successor): Update.
21324 * trans-mem.c (expand_transaction): Update.
21325 * tree-call-cdce.c: Update.
21326 * tree-cfg.c (gimple_split_edge): Upate.
21327 (move_sese_region_to_fn): Upate.
21328 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21329 * tree-eh.c (lower_resx): Upate.
21330 (cleanup_empty_eh_move_lp): Upate.
21331 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21332 * tree-inline.c (copy_edges_for_bb): Update.
21333 (copy_cfg_body): Update.
21334 * tree-parloops.c (gen_parallel_loop): Update.
21335 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21336 (gimple_gen_time_profiler): Update.
21337 * tree-ssa-dce.c (remove_dead_stmt): Update.
21338 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21339 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21340 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21341 (unloop_loops): Update.
21342 (try_peel_loop): Update.
21343 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21344 * tree-ssa-loop-split.c (connect_loops): Update.
21345 (split_loop): Update.
21346 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21347 (hoist_guard): Update.
21348 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21349 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21350 (value_replacement): Update.
21351 * tree-ssa-reassoc.c (branch_fixup): Update.
21352 * tree-ssa-tail-merge.c (replace_block_by): Update.
21353 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21354 (create_edge_and_update_destination_phis): Update.
21355 (compute_path_counts): Update.
21356 (recompute_probabilities): Update.
21357 (update_joiner_offpath_counts): Update.
21358 (freqs_to_counts_path): Update.
21359 (duplicate_thread_path): Update.
21360 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21361 (struct switch_conv_info): Update.
21362 (gen_inbound_check): Update.
21363 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21364 (vect_do_peeling): Update.
21365 (vect_loop_versioning): Update.
21366 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21367 (optimize_mask_stores): Update.
21368 * ubsan.c (ubsan_expand_null_ifn): Update.
21369 * value-prof.c (gimple_divmod_fixed_value): Update.
21370 (gimple_divmod_fixed_value_transform): Update.
21371 (gimple_mod_pow2): Update.
21372 (gimple_mod_pow2_value_transform): Update.
21373 (gimple_mod_subtract): Update.
21374 (gimple_mod_subtract_transform): Update.
21375 (gimple_ic): Update.
21376 (gimple_stringop_fixed_value): Update.
21377 (gimple_stringops_transform): Update.
21378 * value-prof.h: Update.
21380 2017-06-29 Carl Love <cel@us.ibm.com>
21382 * config/rs6000/rs6000-c.c: Add support for built-in functions
21383 vector signed int vec_signed (vector float);
21384 vector signed long long vec_signed (vector double);
21385 vector signed int vec_signed2 (vector double, vector double);
21386 vector signed int vec_signede (vector double);
21387 vector signed int vec_signedo (vector double);
21388 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21389 instruction generator.
21390 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21391 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21392 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21394 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21395 vunsignede_v2df): Add define_expands.
21396 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21397 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21398 VEC_UNSIGNEDO): Add definitions.
21399 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21400 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21401 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21402 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21403 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21404 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21405 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21406 vec_unsignede, vec_unsignedo): Add builtin defines.
21407 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21409 * doc/extend.texi: Update the built-in documentation file for the
21410 new built-in functions.
21412 2017-06-29 Richard Biener <rguenther@suse.de>
21414 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21415 reduction chains to LOOP_VINFO_REDUCTIONS.
21416 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21417 SLP reductions after processing reduction chains.
21419 2017-06-29 Nathan Sidwell <nathan@acm.org>
21421 * builtins.c (fold_builtin_FUNCTION): Use
21422 lang_hooks.decl_printable_name.
21424 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21426 PR middle-end/81194
21427 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21428 with only one label.
21429 * stmt.c (expand_case): Assert NCASES is greater than one.
21431 2017-06-29 Richard Biener <rguenther@suse.de>
21433 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21435 (group_case_labels): Likewise.
21436 (find_taken_edge): Push sanity checking on val to workers...
21437 (find_taken_edge_cond_expr): ... here
21438 (find_taken_edge_switch_expr): ... and here, handle cases
21439 with just a default label.
21440 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21441 (group_case_labels): Likewise.
21442 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21443 group_case_labels does anything cleanup the CFG again.
21445 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21447 PR tree-optimization/81196
21448 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21449 exit condition comparing two IVs.
21451 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21453 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21454 profile to the dummy entry at the end of the list of architectures.
21455 * config/arm/arm-cpu-cdata.h: Regenerated.
21457 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21458 Michael Collison <michael.collison@arm.com>
21461 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21463 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21464 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21465 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21466 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21467 of shift when the shift amount is masked with constant equal to
21468 the size of the mode.
21469 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21472 2017-06-29 Martin Liska <mliska@suse.cz>
21474 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21476 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21479 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21482 (edge_growth_cache_entry::edge_growth_cache_entry): New
21484 (reset_edge_growth_cache): Update to use constructor.
21486 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21488 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21489 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21490 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21492 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21494 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21495 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21497 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21499 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21500 (*-linux-uclibc*): Add t-uclibc tmake_file.
21501 * config/t-musl: New.
21502 * config/t-uclibc: New.
21504 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21506 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21508 (gen_comm_data): Emit architectural setting of arch_prof.
21509 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21511 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21512 (armv8-m.base, armv8-m.main): Likewise.
21513 * arm-protos.h (arm_build_target): Add profile field.
21514 (arch_option): Likewise.
21515 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21517 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21518 arm_active_target.profile.
21520 2017-06-28 Richard Biener <rguenther@suse.de>
21522 PR middle-end/81227
21523 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21524 TYPE_OVERFLOW_WRAPS.
21525 * match.pd (negate_expr_p): Likewise.
21526 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21527 fold_build2, not fold_binary.
21529 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21531 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21532 Convert memory address to Pmode.
21533 (aarch64_print_operand): Assert MEM operands are always Pmode.
21535 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21538 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21539 Remove redundant if.
21540 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21541 * config/arm/aarch-common-protos.h
21542 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21543 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21545 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21548 * multiple_target.c (create_dispatcher_calls): Set the default
21549 clone to be static, not public.
21551 2017-06-28 Richard Biener <rguenther@suse.de>
21553 * tree-vect-loop.c (vectorizable_reduction): Move special
21554 cond reduction IV var creation ...
21555 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21556 parameter. Use STMT_VINFO_VECTYPE.
21557 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21560 2017-06-28 Martin Liska <mliska@suse.cz>
21563 * ipa-visibility.c (non_local_p): Handle visibility.
21565 2017-06-28 Martin Liska <mliska@suse.cz>
21568 * common.opt: Add IntegerRange to various options.
21569 * opt-functions.awk (integer_range_info): New function.
21570 * optc-gen.awk: Add integer_range_info to cl_options struct.
21571 * opts-common.c (decode_cmdline_option): Handle
21572 CL_ERR_INT_RANGE_ARG.
21573 (cmdline_handle_error): Likewise.
21574 * opts.c (print_filtered_help): Show valid interval in
21575 when --help is provided.
21576 * opts.h (struct cl_option): Add range_min and range_max fields.
21577 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21579 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21581 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21582 (x * C EQ/NE y * C): New transformation.
21584 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21586 * genmultilib (combination_space): Accept '+' in option names.
21588 2017-06-28 Martin Liska <mliska@suse.cz>
21591 * asan.c (instrument_derefs): Bail out inner references
21592 that are hard register variables.
21594 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21597 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21598 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21600 2017-06-28 Richard Biener <rguenther@suse.de>
21602 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21603 (vect_get_slp_defs): Adjust.
21604 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21606 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21608 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21609 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21610 (vectorizable_reduction): Adjust.
21611 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21613 (vect_get_slp_defs): Likewise.
21614 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21615 (vectorizable_bswap): Adjust.
21616 (vectorizable_call): Likewise.
21617 (vectorizable_conversion): Likewise.
21618 (vectorizable_assignment): Likewise.
21619 (vectorizable_shift): Likewise.
21620 (vectorizable_operation): Likewise.
21621 (vectorizable_store): Likewise.
21622 (vectorizable_condition): Likewise.
21623 (vectorizable_comparison): Likewise.
21625 2017-06-28 Michael Collison <michael.collison@arm.com>
21628 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21630 (arm_gen_movmemqi): Removed unused variable 'i'.
21631 Convert 'for' loop into 'while' loop.
21632 (arm_expand_prologue): Remove last unnecessary set of insn.
21633 (thumb_pop): Remove unused variable 'pushed_words'.
21634 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21636 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21638 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21639 * config/s390/s390-protos.h: Add prototype of
21640 s390_rel_address_ok_p.
21641 * config/s390/s390.c (s390_got_symbol): New function.
21642 (s390_rel_address_ok_p): New function.
21643 (legitimize_pic_address): Use s390_rel_address_ok_p.
21644 (s390_load_got): Use s390_got_symbol.
21645 (s390_option_override): Issue error if
21646 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21647 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21649 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21651 2017-06-27 Andrew Pinski <apinski@cavium.com>
21653 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21654 (X * copysign (1.0, X)): New pattern.
21655 (X * copysign (1.0, -X)): New pattern.
21656 (copysign (-1.0, CST)): New pattern.
21658 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21660 * genmultilib (combination_space): Remove variable.
21661 Validate reuse rules against regular expression for any sequence
21662 of multilib options in any order.
21664 2017-06-27 Michael Collison <michael.collison@arm.com>
21666 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21667 call aarch64_split_simd_combine.
21668 * (aarch64_combine_internal<mode>): Delete pattern.
21669 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21670 Allow register and subreg operands.
21672 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21674 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21675 specific need, just fallback on defaults.
21676 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21678 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21679 Olivier Hainque <hainque@adacore.com>
21681 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21683 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21684 targets. Pick a default if no particular attempt applied.
21685 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21688 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21690 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21691 (x86_64-wrs-vxworks7): Likewise.
21693 2017-06-27 Marek Polacek <polacek@redhat.com>
21696 * ubsan.c (instrument_null): Check get_base_address's result for null.
21698 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21700 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
21702 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
21704 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
21705 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
21706 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
21707 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
21708 New function types.
21709 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
21710 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
21711 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
21712 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
21713 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
21714 BUILT_IN_FEUPDATEENV): New builtins.
21715 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
21716 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
21717 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
21718 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
21720 (builtin_structptr_types): Adjust size.
21721 * tree.c (builtin_structptr_types): Add four entries.
21723 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21724 Olivier Hainque <hainque@adacore.com>
21726 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
21727 (TLS_SYM): New local macro, forcing reference to __tls__ on
21728 link command lines for VxWorks 7 RTPs, triggering initialization
21730 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
21731 OS features TLS support, true for RTPs on VxWorks 7.
21732 * config/vxworks.c (vxworks_override_options): Setup emutls
21735 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21737 * predict.c (test_prediction_value_range): Use -1U instead of -1
21738 to avoid narrowing conversion warning.
21739 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
21740 to avoid narrowing conversion warning.
21741 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
21743 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
21745 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21747 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
21748 64bit configurations.
21749 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
21750 (SIZE_TYPE): Likewise.
21751 * config/vxworks.c (vxworks_emutls_var_fields): Use
21752 long_unsigned_type_node instead of unsigned_type_node as the offset
21753 field type, which is "pointer" mode in emutls.c.
21755 2017-06-27 Jakub Jelinek <jakub@redhat.com>
21758 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
21760 PR middle-end/81207
21761 * gimple-fold.c (replace_call_with_call_and_fold): Handle
21762 gimple_vuse copying separately from gimple_vdef copying.
21764 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21766 * value-prof.c (free_hist): Remove call to memset and the enclosing if
21769 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21770 Olivier Hainque <hainque@adacore.com>
21772 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
21773 for all vxworks7 targets.
21774 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
21775 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
21776 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
21777 variations for VX6/VX7 and 32/64bits later on in ...
21778 (VXWORKS_LIB_SPEC): Leverage new macros.
21779 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
21780 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
21782 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
21784 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
21785 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
21787 2017-06-26 Carl Love <cel@us.ibm.com>
21789 * config/rs6000/rs6000-c.c: Add support for built-in functions
21790 vector bool char vec_reve (vector bool char);
21791 vector signed char vec_reve (vector signed char);
21792 vector unsigned char vec_reve (vector unsigned char);
21793 vector bool int vec_reve (vector bool int);
21794 vector signed int vec_reve (vector signed int);
21795 vector unsigned int vec_reve (vector unsigned int);
21796 vector bool long long vec_reve (vector bool long long);
21797 vector signed long long vec_reve (vector signed long long);
21798 vector unsigned long long vec_reve (vector unsigned long long);
21799 vector bool short vec_reve (vector bool short);
21800 vector signed short vec_reve (vector signed short);
21801 vector double vec_reve (vector double);
21802 vector float vec_reve (vector float);
21803 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
21804 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
21805 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
21806 (altivec_vreve): New pattern.
21807 * config/rs6000/altivec.h (vec_reve): New define.
21808 * doc/extend.texi (vec_rev): Update the built-in documentation file
21809 for the new built-in functions.
21811 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21813 PR tree-optimization/71815
21814 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
21816 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
21818 (slsr_process_phi): Likewise.
21819 (replace_uncond_cands_and_profitable_phis): Don't replace a
21820 multiply candidate with a stride of 1 (copy or cast).
21821 (phi_incr_cost): Call uses_consumed_by_stmt rather than
21823 (lowest_cost_path): Likewise.
21824 (total_savings): Likewise.
21826 2017-06-26 Richard Biener <rguenther@suse.de>
21829 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
21830 Use def_builtin_pure for all gather builtins.
21832 2017-06-26 Richard Biener <rguenther@suse.de>
21834 PR tree-optimization/81203
21835 * tree-tailcall.c (find_tail_calls): Do not move stmts into
21836 non-dominating BBs.
21838 2017-06-26 Marek Polacek <polacek@redhat.com>
21841 * doc/invoke.texi: Document -Wmultistatement-macros.
21843 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
21845 * doc/sourcebuild.texi (ARM-specific attributes): Document new
21846 arm_neon_ok_no_float_abi effective target.
21848 2017-06-26 Richard Biener <rguenther@suse.de>
21850 PR tree-optimization/80928
21851 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
21852 (copy_bbs): Set BB_DUPLICATED flag early.
21853 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
21855 (execute_on_shrinking_pred): Likewise.
21856 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
21857 BB_DUPLICATED blocks.
21858 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
21859 iterate over all PHIs considering removal of *gsi.
21861 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
21863 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
21866 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21868 * config/rs6000/rs6000-string.c: (expand_block_clear,
21869 do_load_for_compare, select_block_compare_mode,
21870 compute_current_alignment, expand_block_compare,
21871 expand_strncmp_align_check, expand_strn_compare,
21872 expand_block_move, rs6000_output_load_multiple)
21873 Move functions related to string/block move/compare
21874 to a separate file.
21875 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
21876 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
21877 for this function which is now used in two files.
21878 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
21879 * config.gcc: Add rs6000-string.o to extra_objs for
21880 targets powerpc*-*-* and rs6000*-*-*.
21882 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21885 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21886 32-bit, since indexed is not valid for DImode.
21887 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21888 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21889 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21890 (define_peephole2 for Altivec d-form store): Likewise.
21893 * multiple_target.c (create_dispatcher_calls): Only create the
21894 dispatcher call if the function is the default clone of a
21895 versioned function.
21897 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21899 PR middle-end/80902
21900 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21901 a call, force the call to not be a tail call.
21903 2017-06-23 Jeff Law <law@redhat.com>
21905 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21908 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21910 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21911 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21912 handling for early expansion of vector shifts (sl,sr,sra,rl).
21913 (builtin_function_type): Add vector shift right instructions
21914 to the unsigned argument list.
21916 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21918 rtl-optimizatoin/79286
21919 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21920 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21921 trap. PIC register plus a const unspec without offset can never trap.
21923 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21925 * tree.h (builtin_structptr_type): New type.
21926 (builtin_structptr_types): Declare new array.
21927 * tree.c (builtin_structptr_types): New array.
21928 (free_lang_data, build_common_tree_nodes): Use it.
21930 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21933 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21936 2017-06-22 Matt Turner <mattst88@gmail.com>
21938 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21939 Lake models to skylake case. Assume skylake for unknown
21940 models with clflushopt.
21942 2017-06-22 Jeff Law <law@redhat.com>
21944 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21945 frame sizes that do not satisfy aarch64_uimm12_shift.
21947 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21949 * profile-count.h (apply_probability,
21950 apply_scale, probability_in): Fix checks for zero.
21952 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21954 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21955 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21957 2016-06-22 Richard Biener <rguenther@suse.de>
21959 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21960 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21961 REDUC_MAX_EXPR support.
21962 (vectorizable_reduction): Likewise.
21963 (vect_create_epilog_for_reduction): Likewise.
21965 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21967 * match.pd (A / (1 << B) -> A >> B): New.
21968 * generic-match-head.c: Include optabs-tree.h.
21969 * gimple-match-head.c: Likewise.
21970 * optabs-tree.h (target_supports_op_p): New.
21971 * optabs-tree.c (target_supports_op_p): New.
21973 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21975 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21976 $gcc_cv_ld --help output.
21977 (gcc_cv_ld_demangle): Likewise.
21978 (gcc_cv_ld_eh_frame_hdr): Likewise.
21979 (gcc_cv_ld_pie): Likewise.
21980 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21981 (gcc_cv_ld_buildid): Likewise.
21982 (gcc_cv_ld_sysroot): Likewise.
21983 (ld_bndplt_support): Likewise.
21984 (ld_pushpopstate_support): Likewise.
21985 * configure: Regenerate.
21986 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21988 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21991 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21992 operands indexes to avoid gap between operands and match_dups.
21994 2017-06-21 Andrew Pinski <apinski@cavium.com>
21996 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21997 Increment Arith_shift and Arith_shift_reg by 1.
21998 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22000 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22001 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22002 (aarch64_strip_extend): Add new argument and test for it.
22003 (aarch64_cheap_mult_shift_p): New function.
22004 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22005 add a cost if it is true.
22006 Update calls to aarch64_strip_extend.
22007 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22009 2017-06-21 Andrew Pinski <apinski@cavium.com>
22011 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22013 (thunderxt88): Likewise.
22014 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22015 (thunderx_prefetch_tune): New variable.
22016 (thunderx2t99_prefetch_tune): Update for the correct values.
22017 (thunderxt88_tunings): New variable.
22018 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22019 generic_prefetch_tune.
22020 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22022 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22024 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22025 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22026 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22027 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22028 (aarch64_atomic_cas<mode>, GPI): Likewise.
22030 2017-06-21 Martin Liska <mliska@suse.cz>
22032 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22033 statements on cold and hot labels.
22034 * predict.c (tree_estimate_probability_bb): Remove the
22035 prediction from this place.
22037 2017-06-21 Martin Liska <mliska@suse.cz>
22039 PR tree-optimization/79489
22040 * gimplify.c (maybe_add_early_return_predict_stmt): New
22042 (gimplify_return_expr): Call the function.
22043 * predict.c (tree_estimate_probability_bb): Remove handling
22045 * predict.def: Update comment about early return predictor.
22046 * gimple-predict.h (is_gimple_predict): New function.
22047 * predict.def: Change default value of early return to 66.
22048 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22050 * passes.def: Put pass_strip_predict_hints to the beginning of
22053 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22055 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22056 FUNCTION_DECL declarations.
22057 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22059 (dwaf2out_decl): Likewise.
22060 * godump.c (go_early_global_decl): Skip call to the real debug hook
22061 for FUNCTION_DECL declarations.
22062 * passes.c (rest_of_decl_compilation): Skip call to the
22063 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22064 -fdump-go-spec is passed.
22066 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22068 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22069 Reorder for compactness.
22070 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22071 (def_builtin_pure, def_builtin_pure2): New functions.
22072 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22074 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22076 * match.pd (nop_convert): New predicate.
22077 ((A +- CST1) +- CST2): Allow some NOP conversions.
22079 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22082 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22083 with ctors/dtors if GOVD_SHARED is set.
22085 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22087 * config/aarch64/aarch64.md (movti_aarch64):
22088 Emit mov rather than orr.
22089 (movtf_aarch64): Likewise.
22090 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22091 Emit mov rather than orr.
22093 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22095 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22096 Swap alternatives, make integer dup more expensive.
22098 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22100 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22101 Return true for non-tls symbols.
22103 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22105 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22106 (cortex-a75): Likewise.
22107 (cortex-a75.cortex-a55): Likewise.
22108 * config/aarch64/aarch64-tune.md: Regenerate.
22109 * doc/invoke.texi (-mtune): Document new values for -mtune.
22111 2017-06-21 Tom de Vries <tom@codesourcery.com>
22113 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22114 stack_size feature.
22115 (Effective-Target Keywords, Other attributes): Suggest using
22116 dg-add-options stack_size feature to get stack limit in stack_size
22117 effective target documentation.
22119 2017-06-21 Julian Brown <julian@codesourcery.com>
22120 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22122 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22123 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22124 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22126 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22127 attribute type list for neon_multiply.
22128 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22129 attribute type list for neon_multiply.
22130 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22131 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22132 attribute type list for neon_multiply.
22133 * config/arm/types.md (crypto_pmull): Add.
22134 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22135 attribute type list.
22137 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22139 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22142 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22144 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22145 to make sure not to dereference a NULL cost_classes_ptr pointer.
22147 2017-06-20 Carl Love <cel@us.ibm.com>
22149 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22150 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22151 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22152 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22153 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22154 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22155 VMULOSW): New enum "unspec" values.
22156 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22157 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22158 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22159 altivec_vmulosw): New patterns.
22160 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22161 VMULOSW): Add definitions.
22163 2017-06-20 Julia Koval <julia.koval@intel.com>
22165 * config/i386/i386.c: Fix rounding expand for new pattern.
22166 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22168 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22170 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22171 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22173 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22175 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22178 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22180 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22181 architecture, then by implementer ID.
22182 * config/aarch64/aarch64-tune.md: Regenerate.
22184 2017-06-20 Richard Biener <rguenther@suse.de>
22186 PR middle-end/81097
22187 * fold-const.c (split_tree): Fold to type before negating.
22189 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22191 * diagnostic-show-locus.c
22192 (selftest::test_fixit_deletion_affecting_newline): New function.
22193 (selftest::diagnostic_show_locus_c_tests): Call it.
22195 2017-06-20 Andreas Schwab <schwab@suse.de>
22198 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22201 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22203 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22204 __ARM_FEATURE_COPROC according to support.
22206 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22208 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22209 Rewritten to avoid overflow for > 32-bit pointers.
22212 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22213 by removing enum keyword.
22214 (ubsan_type_descriptor): Likewise. Formatting fix.
22217 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22218 splitter): Require TARGET_SSE2 in the condition.
22220 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22223 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22224 for doing vector set of SFmode on ISA 3.0.
22225 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22226 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22228 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22229 SFmode value into a V4SF variable that was extracted from another
22230 V4SF variable without converting the element to double precision
22231 and back to single precision vector format.
22232 (vsx_insert_extract_v4sf_p9_2): Likewise.
22234 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22236 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22237 in UWHI to avoid undefined overflow.
22240 * ubsan.h (enum ubsan_encode_value_phase): New.
22241 (ubsan_encode_value): Change second argument to
22242 enum ubsan_encode_value_phase with default value of
22243 UBSAN_ENCODE_VALUE_GENERIC.
22244 * ubsan.c (ubsan_encode_value): Change second argument to
22245 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22246 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22247 create_tmp_var_raw instead of create_tmp_var and use a
22249 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22250 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22251 ubsan_encode_value callers.
22254 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22255 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22256 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22258 2017-06-19 Richard Biener <rguenther@suse.de>
22260 PR middle-end/81118
22261 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22262 estimates if we changed anything.
22264 2017-06-19 Richard Biener <rguenther@suse.de>
22266 PR tree-optimization/80887
22267 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22268 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22269 simplified lookups, then reset mprts_hook.
22270 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22272 (try_to_simplify): Likewise.
22274 2017-06-19 Martin Liska <mliska@suse.cz>
22277 * gimplify.c (gimplify_switch_expr):
22278 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22280 2017-06-19 Martin Liska <mliska@suse.cz>
22282 * doc/install.texi: Document that PGO runs in 4 stages.
22284 2017-06-19 Martin Liska <mliska@suse.cz>
22287 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22288 to dispatcher function name.
22289 * multiple_target.c (replace_function_decl): New function.
22290 (create_dispatcher_calls): Redirect both edges and references.
22292 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22294 * profile-count.c (profile_count::dump): Dump quality.
22295 (profile_count::differs_from_p): Update for unsigned val.
22296 * profile-count.h (profile_count_quality): New enum.
22297 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22299 2017-06-19 Richard Biener <rguenther@suse.de>
22301 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22302 struct function as arg.
22303 (estimate_numbers_of_iterations): Export overload with loop arg.
22304 (free_numbers_of_iterations_estimates_loop): Use an overload of
22305 free_numbers_of_iterations_estimates instead.
22306 * tree-cfg.c (remove_bb): Adjust.
22307 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22308 * tree-parloops.c (gen_parallel_loop): Likewise.
22309 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22311 (tree_unroll_loops_completely): Likewise.
22312 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22313 Use an overload instead and export.
22314 (estimated_loop_iterations): Adjust.
22315 (max_loop_iterations): Likewise.
22316 (likely_max_loop_iterations): Likewise.
22317 (estimate_numbers_of_iterations): Take struct function as arg
22319 (loop_exits_before_overflow): Adjust.
22320 (free_numbers_of_iterations_estimates_loop): Use an overload.
22321 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22322 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22324 2017-06-19 Richard Biener <rguenther@suse.de>
22327 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22328 RANGE_EXPR conservatively.
22330 2017-06-16 Carl Love <cel@us.ibm.com>
22332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22333 definitions for vec_float, vec_float2, vec_floato,
22334 vec_floate built-ins.
22335 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22336 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22338 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22339 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22340 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22341 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22342 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22343 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22344 vec_floato): Add builtin defines.
22345 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22346 Update the built-in documentation file for the new built-in
22349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22351 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22352 (mthumb): Mark as the negative of -marm.
22354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22356 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22358 (ARM Options, -mtune): Document that this accepts the same
22359 extension options as -mcpu.
22360 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22364 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22365 permitted extensions.
22367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22369 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22370 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22371 (armv8-m.main): Add option +nodsp.
22372 * config/arm/arm-cpu-cdata.h: Regenerated.
22374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22376 * config/arm/t-fuchsia: New file.
22377 * config.gcc (arm*-*-fuchsia*): Use it.
22379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22381 * config/arm/t-symbian: Rewrite for new option infrastructure.
22383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22385 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22386 (MULTILIB_REQUIRED): Likewise.
22388 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22390 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22391 (MULTILIB_RESUE): Likewise.
22392 (MULTILIB_MATCHES): Likewise.
22393 (MULTLIB_REQUIRED): Likewise.
22395 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22397 * config/arm/t-rtems: Rewrite for new option framework.
22399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22401 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22402 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22403 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22404 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22405 * config/arm/t-multilib: ... here.
22406 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22407 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22408 armv7-a and armv8*-a when A-profile libraries have not been built.
22409 * config/arm/t-rmprofile: Rewrite.
22411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22413 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22414 with a backslash. Remove the backslash after substituting unescaped
22416 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22420 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22421 the driver to pass through the default setting of -mfloat-abi.
22422 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22424 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22425 (all_feat_combs): New rule.
22426 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22428 * config/arm/t-aprofile: Rewrite.
22430 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22432 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22433 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22434 fpu is not specified by the user/command-line.
22435 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22436 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22437 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22438 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22439 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22440 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22444 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22445 * config/arm/t-arm-elf: Rewritten.
22447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22449 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22450 have some floating-point instructions.
22451 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22452 (TARGET_MAYBE_HARD_FLOAT): New macro.
22453 * config/arm/arm-builtins.c (arm_init_builtins): Use
22454 TARGET_MAYBE_HARD_FLOAT.
22455 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22459 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22460 (configargs.h): Include it.
22461 (arm_print_hint_for_fpu_option): New function.
22462 (arm_parse_fpu_option): New function.
22463 (candidate_extension): New class.
22464 (arm_canon_for_multilib): New function.
22465 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22466 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22467 (ARCH_CANONICAL_SPECS): New macro.
22468 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22472 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22473 are set after handling multilib fragments. Set target_cpu_default2
22476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22478 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22480 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22481 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22483 (check_arch): Likewise.
22484 * config/arm/arm.c (arm_configure_build_target): Handle
22485 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22486 options in the default.
22488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22490 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22491 when an option is an alias of another.
22492 * config/arm/parsecpu.awk (optalias): New parser token.
22493 (gen_comm_data): Mark non-alias options as such. Emit entries
22494 for extension aliases.
22495 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22496 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22497 (armv6kz, armv6zk, armv6t2): Likewise.
22498 (armv7): Make vfpv3-d16 an alias.
22499 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22501 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22502 Sort in canonical order.
22503 (armv8-a): Sort in canonical order.
22504 (armv8.1-a, armv8.2-a): Likewise.
22505 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22507 (cortex-a9): Sort in canonical order.
22508 * config/arm/arm.c (selftests.h): Include it.
22509 (arm_test_cpu_arch_data): New function.
22510 (arm_run_self_tests): New function.
22511 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22512 (targetm): Move declaration to the end of the file.
22513 * arm-cpu-cdata.h: Regenerated.
22515 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22517 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22518 call to target_mode_check describing the type of option passed.
22519 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22520 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22521 arm_parse_cpu_option_name to match parameters against list of
22523 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22524 arm_arch_core_flags data structure.
22525 * config/arm/arm-cpu_cdata.h: Regenerated.
22527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22529 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22531 (arm_print_hint_for_cpu_option): Likewise.
22532 (arm_print_hint_for_arch_option): Likewise.
22533 (arm_parse_cpu_option_name): Likewise.
22534 (arm_parse_arch_option_name): Likewise.
22535 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22536 of entries in the all_fpus list.
22537 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22538 (arm_parse_cpu_option_name): Declare.
22539 (arm_parse_arch_option_name): Declare.
22540 (arm_parse_option_features): Declare.
22541 (arm_intialize_isa): Declare.
22542 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22544 (gen_comm_data): ... here. Make definitions non-static.
22545 * config/arm/arm-cpu-data.h: Regenerated.
22546 * config/arm/arm-cpu-cdata.h: Regenerated.
22548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22550 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22551 (cpu_arch_extension): New structure.
22552 (cpu_arch_option, arch_option, cpu_option): New structures.
22553 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22554 architecture types.
22555 (gen_data): Generate new format data tables.
22556 * config/arm/arm.c (cpu_tune): New structure.
22557 (cpu_option, processors): Delete.
22558 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22559 (arm_print_hint_for_cpu_option): ... this and ...
22560 (arm_print_hint_for_arch_option): ... this.
22561 (arm_parse_arch_cpu_name): Delete. Replace with ...
22562 (arm_parse_cpu_option_name): ... this and ...
22563 (arm_parse_arch_option_name): ... this.
22564 (arm_unrecognized_feature): Change type of target parameter to
22566 (arm_parse_arch_cpu_features): Delete. Replace with ...
22567 (arm_parse_option_features): ... this.
22568 (arm_configure_build_target): Rework to use new configuration data
22570 (arm_print_tune_info): Rework for new configuration data tables.
22571 * config/arm/arm-cpu-data.h: Regenerated.
22572 * config/arm/arm-cpu.h: Regenerated.
22574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22576 * Makefile.in (OBJS): Move sbitmap.o from here ...
22577 (OBJS-libcommon): ... to here.
22579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22581 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22582 (ISA_ALL_CRYPTO): New macro.
22583 (ISA_ALL_SIMD): New macro
22584 (ISA_ALL_FP): New macro.
22585 * config/arm/arm.c (fpu_bitlist): Update initializer.
22586 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22588 (arm9e): Add fpu. Add option for nofp
22589 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22590 (arm926ej-s, arm1026ej-s): Likewise.
22591 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22592 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22593 neon-fp16, neon-vfpv4, nofp and nosimd.
22594 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22595 (cortex-a8): Add fpu. Add option for nofp.
22596 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22597 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22598 (cortex-r4f): Add fpu.
22599 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22600 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22602 (cortex-r8): Likewise.
22603 (cortex-m4): Add fpu. Add option for nofp.
22604 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22605 (cortex-a17.cortex-a7): Likewise.
22606 (cortex-a32): Add fpu. Add options for crypto and nofp.
22607 (cortex-a35, cortex-a53): Likewise.
22608 (cortex-a57): Add fpu. Add option for crypto.
22609 (cortex-a72, cortex-a73): Likewise.
22610 (exynos-m1): Likewise.
22611 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22612 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22613 (cortex-m33): Add fpu. Add option for nofp.
22614 * config/arm/arm-cpu-cdata.h: Regenerated
22615 * config/arm/arm-cpu-data.h: Regenerated.
22617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22619 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22620 (armv5te, armv5tej): Likewise.
22621 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22622 (armv7): Add options fp and vfpv3-d16.
22623 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22624 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22626 (armv7ve): Likewise.
22627 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22628 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22629 (armv8-a): Add nocrypto option.
22630 (armv8.1-a, armv8.2-a): Likewise.
22631 (armv8-m.main): add options fp, fp.dp and nofp.
22633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22635 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22637 (armv8-a+crc): Delete.
22638 (armv8.1-a): Add options simd, crypto and nofp.
22639 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22640 (armv8.2-a+fp16): Delete.
22641 (armv8-m.main): Add option dsp.
22642 (armv8-m.main+dsp): Delete.
22643 (cortex-a8): Add fpu. Add nofp option.
22644 (cortex-a9): Add fpu. Add nofp and nosimd options.
22645 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22646 link to main cpu and architecture data structures.
22647 (gen_comm_data): Only put isa attributes from the main architecture
22649 (option): New statement for architecture and CPU entries.
22650 * arm.c (struct cpu_option): New structure.
22651 (struct processors): Add entry for options.
22652 (arm_unrecognized_feature): New function.
22653 (arm_parse_arch_cpu_name): Ignore any characters after the first
22655 (arm_parse_arch_cpu_feature): New function.
22656 (arm_configure_build_target): Separate out any CPU and architecture
22657 features and parse separately. Don't error out if -mfpu=auto is
22658 used with only an architecture string.
22659 (arm_print_asm_arch_directives): New function.
22660 (arm_file_start): Call it.
22661 * config/arm/arm-cpu-cdata.h: Regenerated.
22662 * config/arm/arm-cpu-data.h: Likewise.
22663 * config/arm/arm-tables.opt: Likewise.
22665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22667 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22668 assembler when it is not -mfpu=auto.
22670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22672 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22673 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22674 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22675 (ASM_CPU_SPEC): Rewrite.
22676 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22677 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22678 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22679 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22680 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22681 copied string is NUL-terminated. Also strip any characters prefixed
22683 (arm_rewrite_selected_arch): New function.
22684 (arm_rewrite_march): New function.
22686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22688 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22689 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22690 (march, mcpu, mtune): Convert to string-based options.
22691 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22692 (arm_parse_arch_cpu_name): New function.
22693 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22694 identify selected architecture or CPU.
22695 (arm_option_save): New function.
22696 (TARGET_OPTION_SAVE): Redefine.
22697 (arm_option_restore): Restore string options.
22698 (arm_option_print): Print string options.
22700 2017-06-16 Martin Sebor <msebor@redhat.com>
22702 PR tree-optimization/80933
22703 PR tree-optimization/80934
22704 * builtins.c (fold_builtin_3): Do not handle bcmp here.
22705 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
22706 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
22707 (gimple_fold_builtin): Call them.
22709 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22711 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
22712 as unlikely; update profile.
22714 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22716 * predict.c (force_edge_cold): Handle declaring edges impossible
22719 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22721 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
22723 (try_unroll_loop_completely): Fix reporting.
22725 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22727 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
22729 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
22732 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
22733 if given a non-constant argument for an intrinsic which requires a
22736 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22738 * profile.c (compare_freqs): New function.
22739 (branch_prob): Sort edge list.
22740 (find_spanning_tree): Assume that the list is priority sorted.
22742 2017-06-16 Richard Biener <rguenther@suse.de>
22744 PR tree-optimization/81090
22745 * passes.def (pass_record_bounds): Remove.
22746 * tree-pass.h (make_pass_record_bounds): Likewise.
22747 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
22748 make_pass_record_bounds): Likewise.
22749 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
22750 not free niter estimates at the beginning but at the end.
22751 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
22753 2017-06-16 Richard Biener <rguenther@suse.de>
22755 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
22756 initializer to workaround ICE in host GCC 4.8.
22758 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
22760 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
22762 (clone_inlined_nodes): Update.
22764 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22766 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
22767 prefetch settings, and enable prefetching by default at -O3.
22769 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22771 * config/aarch64/aarch64.c (aarch64_override_options_internal):
22772 Set flag_prefetch_loop_arrays according to tuning data.
22774 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22776 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
22777 New tune structure.
22778 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
22779 [Unrelated to main purpose of the patch] Place the pointer field last
22780 to enable type checking errors when tune structure are wrongly merged.
22781 * config/aarch64/aarch64.c (generic_prefetch_tune,)
22782 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
22783 (thunderx2t99_prefetch_tune): New tune constants.
22784 (tune_params *_tunings): Update all tunings (no functional change).
22785 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
22786 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
22787 from tunings structures.
22789 2017-06-16 Jakub Jelinek <jakub@redhat.com>
22792 * ubsan.c (instrument_null): Add T argument, use it instead
22793 of computing it based on IS_LHS.
22794 (instrument_object_size): Likewise.
22795 (pass_ubsan::execute): Adjust instrument_null and
22796 instrument_object_size callers to pass gimple_get_lhs or
22797 gimple_assign_rhs1 result to it. Use instrument_null instead of
22798 calling get_base_address and instrument_mem_ref. Handle
22799 aggregate call arguments for object-size sanitization.
22801 2017-06-16 Yury Gribov <tetra2005@gmail.com>
22803 PR tree-optimization/81089
22804 * tree-vrp.c (is_masked_range_test): Validate operands of
22807 2017-06-15 Martin Sebor <msebor@redhat.com>
22810 * dumpfile.c (dump_register): Avoid calling memset to initialize
22811 a class with a default ctor.
22812 * gcc.c (struct compiler): Remove const qualification.
22813 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
22814 * hash-table.h: Ditto.
22815 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
22817 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
22818 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
22820 * params.h (struct param_info): Make struct members non-const.
22821 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
22822 with default initialization.
22823 * vec.h (vec_copy_construct, vec_default_construct): New helper
22825 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
22826 with vec_copy_construct.
22827 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
22828 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
22829 * doc/invoke.texi (-Wclass-memaccess): Document.
22831 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22833 * emit-rtl.h (is_leaf): Update comment about local
22834 register allocator.
22836 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22839 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
22840 for a variable to have a section before checking if the section has a
22842 Set section to.persistent if persistent attribute is set.
22843 Warn if .persistent attribute is used on an automatic variable.
22845 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
22847 PR rtl-optimization/80474
22848 * reorg.c (update_block): Do not ignore instructions in a delay slot.
22850 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
22852 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
22855 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
22857 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
22858 (casesi): Emit bounds checking as RTL.
22859 (casesi_internal_mips16_<mode>): Remove bounds checking.
22861 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
22863 * config/xtensa/xtensa.c (xtensa_option_override): Append
22864 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
22865 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
22866 xtensa_doloop_hooks): Define unconditionally.
22867 (xtensa_reorg_loops): Only call reorg_loops in the presence of
22869 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
22870 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
22871 for it in xtensa_option_override.
22872 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
22873 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
22875 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
22877 * doc/cppopts.texi: Document '-' special value to -MF.
22879 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
22881 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22882 (cortex_a53_fconst): Likewise.
22883 (cortex_a53_fpmul): Likewise.
22884 (cortex_a53_f_load_64): Likewise.
22885 (cortex_a53_f_load_many): Likewise.
22886 (cortex_a53_advsimd_alu): Likewise.
22887 (cortex_a53_advsimd_alu_q): Likewise.
22888 (cortex_a53_advsimd_mul): Likewise.
22889 (cortex_a53_advsimd_mul_q): Likewise.
22890 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22891 Add missing fmul, r2f_cvt and fconst cases.
22893 2017-06-14 Richard Biener <rguenther@suse.de>
22895 PR middle-end/81088
22896 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22898 (fold_binary_loc): When associating do not treat pre-existing
22899 TREE_OVERFLOW on literal constants as a reason to allow
22900 TREE_OVERFLOW on associated literal constants.
22902 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22904 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22905 (MASK_FEATURES): New macro.
22906 * config/sparc/sparc.c (sparc_option_override): Remove the special
22907 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22909 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22911 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22912 a division of 0 if non-call exceptions are enabled.
22914 2017-06-14 Andrew Pinski <apinski@cavium.com>
22915 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22918 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22919 Improve vector initialization code gen for only variable case.
22921 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22923 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22925 2017-06-14 Richard Biener <rguenther@suse.de>
22927 PR tree-optimization/81083
22928 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22931 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22933 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22934 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22935 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22936 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22937 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22938 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22940 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22942 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22943 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22945 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22947 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22949 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22951 * config/rs6000/t-rtems: Don't handle SPE.
22953 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22955 * config/rs6000/t-linux: Don't handle SPE.
22957 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22959 * config/rs6000/eabispe.h: Delete file.
22961 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22963 * config/rs6000/t-spe: Delete file.
22965 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22967 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22968 (rs6000_legitimate_offset_address_p): Return false for anything in
22969 V2SImode or V2SFmode.
22971 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22973 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22974 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22975 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22977 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22978 (rs6000_legitimate_offset_address_p): Ditto.
22979 (rs6000_emit_move): Ditto.
22980 (rs6000_init_builtins): Remove V4HI_type_node.
22982 2017-06-13 Martin Liska <mliska@suse.cz>
22985 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22986 (gate_asan): Likewise.
22987 * asan.h (asan_no_sanitize_address_p): Remove the function.
22988 (sanitize_flags_p): New function.
22989 * builtins.def: Fix coding style.
22990 * common.opt: Use renamed enum value.
22991 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22992 * doc/extend.texi: Document no_sanitize attribute.
22993 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22994 to SANITIZE_UNDEFINED_NONDEFAULT.
22995 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22996 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22997 Use sanitize_flags_p.
22998 * gimplify.c (gimplify_function_tree): Likewise.
22999 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23000 * opts.c (parse_no_sanitize_attribute): New function.
23001 (common_handle_option): Use renamed enum value.
23002 * opts.h (parse_no_sanitize_attribute): Declare.
23003 * tree.c (sanitize_flags_p): New function.
23004 * tree.h: Declared here.
23005 * tsan.c: Use sanitize_flags_p.
23006 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23007 (instrument_mem_ref): Likewise.
23008 (instrument_bool_enum_load): Likewise.
23009 (do_ubsan_in_current_function): Remove the function.
23010 (pass_ubsan::execute): Use sanitize_flags_p.
23011 * ubsan.h: Remove do_ubsan_in_current_function
23012 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23013 (dump_function_to_file): Use it here.
23015 2017-06-13 Martin Jambor <mjambor@suse.cz>
23017 PR tree-optimization/80803
23018 PR tree-optimization/81063
23019 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23020 (propagate_subaccesses_across_link): Enqueue subtree whenever
23021 necessary instead of relying on the caller.
23023 2017-06-13 Martin Jambor <mjambor@suse.cz>
23025 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23026 that have a first_link.
23027 (sort_and_splice_var_accesses): Do not check first_link before
23029 (subtree_mark_written_and_enqueue): Likewise.
23030 (propagate_all_subaccesses): Likewise and do not stop at first
23031 parent with a first_link.
23033 2017-06-13 Martin Jambor <mjambor@suse.cz>
23035 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23038 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23040 * match.pd: New pattern.
23042 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23044 * tree-vrp.c (is_masked_range_test): New function.
23045 (register_edge_assert_for): Determine ranges for
23048 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23050 PR tree-optimization/67328
23051 * fold-const.c (maskable_range_p): New function.
23052 (build_range_check): Generate bittests if possible.
23054 2017-06-13 Martin Liska <mliska@suse.cz>
23056 * gimple-pretty-print.c (dump_probability): Add new argument.
23057 (dump_edge_probability): Dump both probability and count.
23058 (dump_gimple_label): Likewise.
23059 (dump_gimple_bb_header): Likewise.
23061 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23064 * config/avr/avr-devices.c: Fix indentation.
23065 * config/avr/gen-avr-mmcu-specs.c: Dito.
23067 2017-06-13 Richard Biener <rguenther@suse.de>
23069 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23070 instead get vector type from stmt_info.
23071 (vectorizable_reduction): Adjust. Remove dead code.
23073 2017-06-13 Richard Biener <rguenther@suse.de>
23075 PR middle-end/81065
23076 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23077 case of C * (x * C2 + C3).
23078 (fold_addr_of_array_ref_difference): Properly fold index difference.
23080 2017-06-12 David S. Miller <davem@davemloft.net>
23083 * config/sparc/sparc.md (return expander): Emit frame blockage if
23084 function uses alloca.
23086 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23088 * combine.c (make_field_assignment): Check len rather than the mode
23089 precision when calling force_to_mode.
23091 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23093 Support multilibs and devices that see flash in RAM address range.
23096 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23097 (avr_mcu_t) <flash_pm_offset>: New field.
23098 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23099 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23100 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23101 (AVR_TINY_PM_OFFSET): Remove macro.
23102 * config/avr/avr.opt (-mshort-calls): New option.
23103 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23104 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23105 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23106 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23107 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23108 instead of avr_arch->have_jmp_call.
23109 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23110 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23111 avr_arch->flash_pm_offset to define.
23112 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23113 new field flash_pm_offset. Add entry for avrxmega3.
23114 (avr_texinfo): Add entry for avrxmega3.
23115 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23116 attiny212, attiny214,
23117 attiny412, attiny414, attiny416, attiny417,
23118 attiny814, attiny816, attiny817,
23119 attiny1614, attiny1616, attiny1617,
23120 attiny3214, attiny3216, attiny3217.
23121 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23122 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23123 (avr_print_operand_address) [AVR_TINY]: Same.
23124 (avr_asm_init_sections) <readonly_data_section>: Only patch
23125 callback if avr_arch->flash_pm_offset = 0.
23126 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23127 for rodata if avr_arch->flash_pm_offset != 0.
23128 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23129 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23130 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23131 (m_options): Append opt_rcall.
23132 (m_dirnames): Append dir_rcall.
23133 * config/avr/t-multilib: Regenerate.
23135 * configure.ac [target=avr]: Check whether avrxmega3 default
23136 linker description file works as needed.
23137 * configure: Regenerate.
23138 * doc/avr-mmcu.texi: Regenerate.
23139 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23140 <__AVR_ARCH__>: Document avrxmega3 and 103.
23141 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23142 <__AVR_SHORT_CALLS__>: Document it.
23143 <__AVR_PM_BASE_ADDRESS__>: Document it.
23144 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23145 (AVR Variable Attributes) <progmem>: Document this is
23146 not needed for avrxmega3.
23147 (AVR Named Address Spaces) <__flash>: Dito.
23149 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23151 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23153 2017-06-12 Doug Rupp <rupp@adacore.com>
23155 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23156 Append vxworks-stdint.h to the tm_file list.
23157 * config/vxworks-stdint.h: New file.
23159 2017-06-12 Martin Liska <mliska@suse.cz>
23161 PR tree-optimization/81041
23162 * tree-profile.c (gimple_gen_ic_func_profiler):
23163 Create an extra BB in profile-generate
23164 (gimple_gen_time_profiler): Likewise.
23166 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23168 PR tree-optimization/81003
23169 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23170 (update_range_test): Use it instead of force_gimple_operand_gsi.
23172 2017-06-12 Richard Biener <rguenther@suse.de>
23174 PR tree-optimization/81053
23175 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23176 with backedge value not defined in loop. Simplify def stmt
23179 2017-06-11 Tom de Vries <tom@codesourcery.com>
23182 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23184 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23185 nvptx_cannot_force_const_mem.
23187 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23189 * opts.c (finish_options): Move test for flag_split_stack after
23190 it has been initialized.
23192 2017-06-11 Jason Merrill <jason@redhat.com>
23194 * tree.h (id_equal): New.
23195 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23196 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23197 instead of strcmp of IDENTIFIER_POINTER.
23199 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23201 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23202 (mark_all_inlined_calls_cdtor): Fix formating.
23203 (inline_transform): Rescale profile before inlining.
23205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23207 * cgraph.h (cgraph_edge::clone): Update prototype.
23208 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23209 (cgraph_node::create_clone): Update.
23210 (cgraph_node::create_version_clone): Update.
23211 * tree-inline.c (copy_bb): Update.
23212 (expand_call_inline): Update.
23214 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23216 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23217 factored out from ...
23218 (rs6000_emit_prologue): ... here.
23220 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23222 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23223 factored out from ...
23224 (rs6000_emit_prologue): ... here.
23226 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23228 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23230 (handle_missing_profiles): Fix computation of tp_first_run.
23231 (counts_to_freqs): Do not touch freqs when count is 0.
23233 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23235 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23238 2017-06-10 Tom de Vries <tom@codesourcery.com>
23240 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23241 attributes): Document signal effective target.
23243 2017-06-10 Tom de Vries <tom@codesourcery.com>
23245 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23246 Document effective target stack_size.
23248 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23250 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23251 to the edit_context if they can be auto-applied.
23253 2017-06-9 Ian Lance Taylor <iant@golang.org>
23255 * opts.c (finish_options): If -fsplit-stack, disable implicit
23256 -forder-blocks-and-partition.
23257 * doc/invoke.texi (Optimize Options): Document that when using
23258 -fsplit-stack -forder-blocks-and-partition is not implicitly
23261 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23263 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23264 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23265 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23266 * builtins.def (abort, trap, unreachable): Declare cold.
23267 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23268 * tree-core.h (ECF_COLD): New.
23269 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23270 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23272 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23274 * predict.c (unlikely_executed_stmt_p): Cleanup.
23276 2017-06-09 Richard Biener <rguenther@suse.de>
23278 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23279 model if the ref is always written to.
23281 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23283 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23285 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23287 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23290 2017-06-09 Tom de Vries <tom@codesourcery.com>
23293 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23294 "target cannot support label values" when encountering LABEL_REF.
23296 2017-06-09 Martin Liska <mliska@suse.cz>
23298 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23299 (gimple_gen_ic_func_profiler): Emit direct comparison
23300 of __gcov_indirect_call_callee with NULL.
23301 (gimple_gen_time_profiler): Change probability from
23302 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23304 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23306 * profile.c (edge_gcov_counts): Turn to pointer.
23307 (compute_branch_probabilities, compute_branch_probabilities): Update.
23308 (branch_prob): Do not clear edge_gcov_count.
23309 * profile.h (edge_gcov_counts): Turn to pointer.
23310 (edge_gcov_count): Update.
23312 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23314 * gimple.h (gimple_check_failed): Mark cold.
23316 2017-06-09 Richard Biener <rguenther@suse.de>
23318 PR tree-optimization/66623
23319 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23320 refactor check_reduction into two parts, properly computing
23321 whether we have to check reduction validity for outer loop
23324 2017-06-09 Richard Biener <rguenther@suse.de>
23326 PR tree-optimization/79483
23327 * graphite-scop-detection.c (order): New global.
23328 (get_order): Compute bb to order mapping that satisfies code
23329 generation constraints.
23330 (cmp_pbbs): New helper.
23331 (build_scops): Start domwalk at entry block, sort generated
23334 2017-06-09 Richard Biener <rguenther@suse.de>
23336 PR middle-end/81007
23337 * ipa-polymorphic-call.c
23338 (ipa_polymorphic_call_context::restrict_to_inner_class):
23339 Skip FIELD_DECLs with error_mark_node type.
23340 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23343 2017-06-09 Martin Liska <mliska@suse.cz>
23345 * predict.c (struct branch_predictor): New struct.
23346 (test_prediction_value_range): New test.
23347 (predict_c_tests): New function.
23348 * selftest-run-tests.c (selftest::run_tests): Run the function.
23349 * selftest.h: Declare new tests.
23351 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23354 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23355 gen_add3_insn did not fail.
23356 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23357 r0, construct that number in a temporary reg and add that reg to r0.
23358 If asked to put the result in r0 as well, fail.
23360 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23362 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23363 for early expansion of vec_eqv.
23365 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23367 PR middle-end/81005
23368 * ubsan.c (instrument_null): Avoid pointless code temporary.
23369 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23371 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23375 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23378 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23381 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23383 * predict.c (unlikely_executed_edge_p): Move ahead.
23384 (probably_never_executed_edge_p): Use it.
23386 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23388 PR middle-end/79988
23389 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23390 gimple_call_builtin_p call.
23392 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23394 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23395 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23396 rtl_check_failed_type2, rtl_check_failed_code1,
23397 rtl_check_failed_code2, rtl_check_failed_code_mode,
23398 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23399 rtvec_check_failed_bounds, rtl_check_failed_flag,
23400 _fatal_insn_not_found, _fatal_insn): Likewise.
23401 * tree.h (tree_contains_struct_check_failed,
23402 tree_check_failed, tree_not_check_failed,
23403 tree_class_check_failed, tree_range_check_failed,
23404 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23405 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23406 tree_operand_check_failed, omp_clause_check_failed,
23407 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23410 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23412 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23413 flag_branch_probabilities.
23414 * ipa-inline.c (edge_badness): Likewise.
23415 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23416 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23417 * predict.c (maybe_hot_frequency_p): Likewise.
23418 (probably_never_executed): Likewise.
23419 * sched-ebb.c (schedule_ebbs): Likewise.
23420 * sched-rgn.c (find_single_block_region): Likewise.
23421 * tracer.c (tail_duplicate): Likewise.
23423 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23425 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23426 longer requires x_flag_profile_use.
23428 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23430 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23431 instead of flag_reorder_blocks_and_partition.
23432 * dbxout.c (dbxout_function_end): Likewise.
23433 * dwarf2out.c (gen_subprogram_die): Likewise.
23434 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23435 * hw-doloop.c (reorg_loops): Likewise.
23436 * varasm.c (assemble_start_function,
23437 assemble_end_function): Likewise.
23438 (decide_function_section): Do not check for
23439 flag_reorder_blocks_and_partition.
23441 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23443 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23445 (chkp_get_hard_register_fake_addr_expr): Ditto.
23446 (chkp_build_addr_expr): Add check for hard reg case.
23447 (chkp_parse_array_and_component_ref): Ditto.
23448 (chkp_find_bounds_1): Ditto.
23449 (chkp_process_stmt): Don't generate bounds store for
23452 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23454 * predict.c (maybe_hot_bb_p): Do not check profile status.
23455 (maybe_hot_edge_p): Likewise.
23456 (probably_never_executed): Check for zero counts even if profile
23458 (unlikely_executed_edge_p): New function.
23459 (unlikely_executed_stmt_p): New function.
23460 (unlikely_executed_bb_p): New function.
23461 (set_even_probabilities): Use unlikely predicates.
23462 (combine_predictions_for_bb): Likewise.
23463 (predict_paths_for_bb): Likewise.
23464 (predict_paths_leading_to_edge): Likewise.
23465 (determine_unlikely_bbs): New function.
23466 (estimate_bb_frequencies): Use it.
23467 (compute_function_frequency): Use zero counts even if profile is
23469 * profile-count.h: Fix typo.
23471 2017-08-08 Julia Koval <julia.koval@intel.com>
23473 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23474 _mm512_mask_cvtsepi16_storeu_epi8,
23475 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23476 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23477 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23478 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23479 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23480 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23481 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23482 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23483 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23484 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23485 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23486 __builtin_ia32_pmovuswb256mem_mask,
23487 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23488 __builtin_ia32_pmovwb512mem_mask): New builtins.
23490 2017-08-08 Julia Koval <julia.koval@intel.com>
23492 PR target/73350,80862
23493 * config/i386/subst.md (round): Fix round pattern.
23494 * config/i386/i386.c (ix86_erase_embedded_rounding):
23495 Fix erasing rounding for the fixed pattern.
23497 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23499 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23501 2017-06-08 Martin Liska <mliska@suse.cz>
23503 PR gcov-profile/80911
23504 * gcov.c (block_info::block_info): New constructor.
23506 2017-06-07 Carl Love <cel@us.ibm.com>
23508 * config/rs6000/rs6000-c: The return type of the following
23509 built-in functions was implemented as int not long long. Fix sign
23510 of return value for the unsigned version of vec_mulo and vec_mule.
23511 vector unsigned long long vec_bperm (vector unsigned long long,
23512 vector unsigned char)
23513 vector signed long long vec_mule (vector signed int,
23515 vector unsigned long long vec_mule (vector unsigned int,
23516 vector unsigned int)
23517 vector signed long long vec_mulo (vector signed int,
23519 vector unsigned long long vec_mulo (vector unsigned int,
23520 vector unsigned int)
23521 * doc/extend.texi: Fix the documentation for the built-in
23524 2017-06-07 Carl Love <cel@us.ibm.com>
23527 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23530 2017-06-07 Carl Love <cel@us.ibm.com>
23532 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23533 support, Generate doublehv for signed int/float for BE case only.
23535 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23537 * doc/invoke.texi (mcx16): Rewrite.
23539 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23541 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23542 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23543 *mov<mode>_softfloat, and an anonymous splitter): Use
23544 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23548 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23550 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23551 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23552 (rs6000_debug_reg_global): Adjust.
23553 (rs6000_init_hard_regno_mode_ok): Adjust.
23554 (rs6000_dbx_register_number): Adjust.
23555 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23556 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23557 Remove SPE_ACC and SPEFSCR.
23558 (REG_ALLOC_ORDER): Ditto.
23559 (FRAME_POINTER_REGNUM): Change to 111.
23560 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23561 (REG_CLASS_NAMES): Ditto.
23562 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23563 (REGISTER_NAMES): Ditto.
23564 (ADDITIONAL_REG_NAMES): Ditto.
23565 (rs6000_reg_names): Ditto.
23566 * config/rs6000/rs6000.md: Renumber some register number
23569 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23571 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23573 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23574 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23576 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23577 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23578 Delete the SPE high registers.
23579 (REG_ALLOC_ORDER): Ditto.
23580 (enum reg_class): Remove SPE_HIGH_REGS.
23581 (REG_CLASS_NAMES): Ditto.
23582 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23583 (REGISTER_NAMES): Ditto.
23584 (rs6000_reg_names): Ditto.
23585 * doc/tm.texi.in: Remove SPE as example.
23586 * doc/tm.texi: Regenerate.
23588 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23590 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23591 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23592 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23593 * config/rs6000/rs6000.md (type): Remove "brinc".
23595 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23597 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23598 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23599 * config/rs6000/linuxspe.h: Delete file.
23600 * config/rs6000/rs6000.md: Don't include spe.md.
23601 * config/rs6000/spe.h: Delete file.
23602 * config/rs6000/spe.md: Delete file.
23603 * config/rs6000/t-rs6000: Remove spe.md.
23605 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23607 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23608 (reg_or_none500mem_operand): Delete.
23609 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23610 instead of reg_or_none500mem_operand.
23612 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23615 handling of SPE flags.
23616 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23620 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23622 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23623 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23624 paired_divv2sf3): Similar.
23625 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23626 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23627 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23629 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23630 Rename the paired_* instruction patterns.
23631 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23633 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23634 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23635 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23636 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23637 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23638 PAIRED_VECTOR_MODE.
23639 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23640 (spe_func_has_64bit_regs_p): Delete.
23641 (spe_expand_predicate_builtin): Delete.
23642 (spe_expand_evsel_builtin): Delete.
23643 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23644 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23645 (invalid_e500_subreg): Delete.
23646 (rs6000_legitimize_address): Always force_reg op2 as well, for
23647 paired single memory accesses.
23648 (rs6000_member_type_forces_blk): Delete.
23649 (rs6000_spe_function_arg): Delete.
23650 (rs6000_expand_unop_builtin): Delete SPE handling.
23651 (rs6000_expand_binop_builtin): Ditto.
23652 (spe_expand_stv_builtin): Delete.
23653 (bdesc_2arg_spe): Delete.
23654 (spe_expand_builtin): Delete.
23655 (spe_expand_predicate_builtin): Delete.
23656 (spe_expand_evsel_builtin): Delete.
23657 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23658 (spe_init_builtins): Delete.
23659 (spe_func_has_64bit_regs_p): Delete.
23660 (savres_routine_name): Delete "info" parameter. Adjust callers.
23661 (rs6000_emit_stack_reset): Ditto.
23662 (rs6000_dwarf_register_span): Delete.
23663 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23664 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23665 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23667 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23669 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23670 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23671 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23672 mulv2sf3, divv2sf3): Delete expanders.
23674 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23676 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23678 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23680 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23681 * config/rs6000/rs6000.c: Ditto.
23683 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23685 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23686 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23687 comparison_operator.
23689 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23691 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23692 * config/rs6000/rs6000.opt: Ditto.
23693 * config/rs6000/t-rtems: Ditto.
23695 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23697 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
23698 TARGET_E500_SINGLE by 0, simplify.
23699 * config/rs6000/rs6000.c: Ditto.
23700 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
23701 (spe_build_register_parallel): Delete.
23702 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
23703 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
23704 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
23705 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
23706 (E500_CONVERT): Delete.
23707 * config/rs6000/spe.md: Remove many patterns and all define_constants.
23709 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23711 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
23712 * config/rs6000/dfp.md: Ditto.
23713 (negdd2, *negdd2_fpr): Merge.
23714 (absdd2, *absdd2_fpr): Merge.
23715 (negtd2, *negtd2_fpr): Merge.
23716 (abstd2, *abstd2_fpr): Merge.
23717 * config/rs6000/e500.h: Delete file.
23718 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
23719 TARGET_FPRS by 1 and simplify.
23720 * config/rs6000/rs6000-c.c: Ditto.
23721 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
23722 TARGET_DF_SPE by 0.
23723 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
23725 * config/rs6000/rs6000.md: Ditto.
23726 (floatdidf2, *floatdidf2_fpr): Merge.
23727 (move_from_CR_gt_bit): Delete.
23728 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
23729 (E500_CR_IOR_COMPARE): Delete.
23730 (All patterns that require !TARGET_FPRS): Delete.
23731 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
23733 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23735 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
23737 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23739 * graphds.c (add_edge): Intitialize edge's attached data.
23740 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
23741 pointer parameter. Call pointed function on each edge during
23742 graph traversing. Skip traversing the edge when the function
23744 (graphds_dfs, graphds_scc): Ditto.
23745 (for_each_edge): New parameter. Pass the new parameter to callback
23747 * graphds.h (skip_edge_callback): New function pointer type.
23748 (graphds_dfs, graphds_scc): New function pointer parameter.
23749 (graphds_edge_callback, for_each_edge): New parameter.
23751 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23753 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
23754 out code checking if runtime alias check is possible to below ...
23755 Call the new function.
23756 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
23757 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
23759 2017-06-07 Marek Polacek <polacek@redhat.com>
23762 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
23763 TYPE_OVERFLOW_WRAPS check.
23765 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23767 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
23768 if versioning is required.
23769 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
23770 peeling with the check for versioning.
23772 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23774 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
23775 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
23776 Set true to new parameter if new ssa variable is defined.
23777 (vect_gen_vector_loop_niters): Refactor. Set range information
23778 for the new vector loop bound variable.
23779 (vect_do_peeling): Ditto.
23781 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23783 * tree-affine.c (ssa.h): Include header file.
23784 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
23785 has wrapping overflow behavior.
23787 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23789 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
23791 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23793 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
23794 (tree_to_aff_combination): ... here.
23796 2017-06-07 Bin Cheng <bin.cheng@arm.com>
23798 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
23799 reg_pressure model function.
23800 (ivopts_global_cost_for_size): Delete.
23801 (determine_set_costs, iv_ca_recount_cost): Call new model function
23802 ivopts_estimate_reg_pressure.
23804 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23806 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
23807 expensive than udiv. Remove floating point cases from mod.
23809 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23811 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
23812 Increase idiv cost.
23814 2017-06-07 Tamar Christina <tamar.christina@arm.com>
23816 * config/aarch64/aarch64.md
23817 (copysignsf3): Fix mask generation.
23819 2017-06-07 Jakub Jelinek <jakub@redhat.com>
23821 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
23823 (class dump_manager): Add register_dumps method.
23824 * dumpfile.c: Include langhooks.h.
23825 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
23826 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
23827 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
23828 (dump_manager::dump_register): Start with 512 entries instead of 32.
23829 (dump_manager::register_dumps): New method.
23830 * toplev.c (general_init): Instead of invoking register_dumps
23831 langhook, invoke register_dumps method on the dump manager.
23832 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
23835 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
23837 * doc/md.texi: Clarify the restrictions on a define_insn condition.
23838 Say that # requires an associated define_split to exist, and that
23839 the define_split must be suitable for use after register allocation.
23841 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23843 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
23844 (compute_outgoing_frequencies): Also initialize zero counts.
23845 (find_many_sub_basic_blocks): Do not produce uninitialized profile
23846 around loops; preserve more of profile when nothing changes.
23848 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
23850 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
23852 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
23853 * config/arm/arm-cpu-cdata.h: Regenerate.
23854 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
23855 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
23856 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
23857 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
23858 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
23860 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
23862 * config/arm/t-rmprofile: Likewise.
23863 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
23865 2017-06-06 David S. Miller <davem@davemloft.net>
23868 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
23869 blockage if function uses alloca.
23871 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23873 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
23874 New "uid" fields to hold pretty-print IDs of group and ref.
23875 Memory references are now identified as <group_id>:<ref_id>
23876 instead of using [random] addresses.
23877 (dump_mem_details): Simplify, no functional change.
23878 (dump_mem_ref): Simplify and make output more concise.
23879 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
23880 (find_or_create_group): Initialize group uid.
23881 (record_ref): Initialize ref uid. Improve debug output.
23882 (prune_group_by_reuse, should_issue_prefetch_p,)
23883 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23884 (mark_nontemporal_store, determine_loop_nest_reuse):
23885 Improve debug output.
23887 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23889 * dbgcnt.def (prefetch): New debug counter.
23890 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23891 (schedule_prefetches): Stop issueing prefetches if debug counter
23894 2017-06-06 Tom de Vries <tom@codesourcery.com>
23896 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23897 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23899 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23901 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23902 Use aarch64_reg_or_zero predicate for operand 4.
23903 (aarch64_compare_and_swap<mode> define_insn_and_split):
23904 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23905 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23907 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23909 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23910 (arm_compute_save_core_reg_mask): This.
23911 (thumb1_compute_save_reg_mask): Rename into ...
23912 (thumb1_compute_save_core_reg_mask): This.
23913 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23914 (arm_compute_frame_layout): Likewise.
23916 2017-06-06 Richard Biener <rguenther@suse.de>
23918 PR tree-optimization/80974
23919 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23920 keep or clear leaders SSA info.
23922 2017-06-06 Tom de Vries <tom@codesourcery.com>
23924 * config/nvptx/nvptx.c (split_mode_p): New function.
23925 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23927 2017-06-06 Tom de Vries <tom@codesourcery.com>
23929 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23931 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23934 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23937 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23939 * shrink-wrap.c (handle_simple_exit): Update profile.
23940 (try_shrink_wrapping): Upate profile.
23942 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23944 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23945 (tree_guess_outgoing_edge_probabilities): New.
23946 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23947 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23949 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23951 * ipa-split.c (split_function): Initialize return bb profile.
23953 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23955 * profile.c (compute_branch_probabilities): Also initialize
23956 EXIT_BLOCK profile.
23958 2017-06-06 Richard Biener <rguenther@suse.de>
23960 PR tree-optimization/80928
23961 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23962 (vect_analyze_loop_operations): Properly guard analysis for
23964 (vect_transform_loop): Likewise.
23965 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23966 (vect_model_induction_cost): Do not cost for pure SLP.
23967 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23968 of induction in inner loop vectorization.
23969 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23970 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23971 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23973 (vect_analyze_slp_cost_1): Cost induction.
23974 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23975 (vect_get_slp_vect_defs): Likewise.
23976 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23977 (vect_transform_stmt): Handle SLP reductions.
23978 * tree-vectorizer.h (vectorizable_induction): Adjust.
23980 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23982 * config/rs6000/rs6000.c (make_resolver_func): Update
23983 init_lowered_empty_function call.
23985 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23987 * doc/invoke.texi: Document the -fprofile-abs-path option.
23988 * common.opt (fprofile-abs-path): New option.
23989 * gcov-io.h (gcov_write_filename): Declare.
23990 * gcov-io.c (gcov_write_filename): New function.
23991 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23992 * profile.c (output_location): Likewise.
23994 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23996 * shring-wrap.c: Revert accidental commit.
23998 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24000 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24002 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24004 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24006 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24007 profile in callgraph edge.
24008 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24009 (apply_scale): Likewise.
24010 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24011 Also scale profile when inlining function with zero profile.
24012 (initialize_cfun): Update exit block profile even when it is zero.
24013 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24014 when profile is read.
24016 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24018 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24019 (CLONE_*): New constants to define the processors we can generate
24020 code for with the target_clone attribute.
24021 (rs6000_clone_map): New array to identify which clone processors
24022 the current program is running on.
24023 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24024 target_clone attribute.
24025 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24026 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24027 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24028 (cpu_expand_builtin): Add support for target_clone attribute.
24029 (rs6000_valid_attribute_p): Allow "default" attribute.
24030 (get_decl_name): New debug function to simplify printing the
24031 current function name in debugging statements.
24032 (rs6000_clone_priority): New functions to support the target_clone
24033 attribute, and be able to generate code to switch between ISA 2.05
24034 through ISA 3.0 (power6 through power9).
24035 (rs6000_compare_version_priority): Likewise.
24036 (rs6000_get_function_versions_dispatcher): Likewise.
24037 (make_resolver_func): Likewise.
24038 (add_condition_to_bb): Likewise.
24039 (dispatch_function_versions): Likewise.
24040 (rs6000_generate_version_dispatcher_body): Likewise.
24041 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24042 (fusion_gpr_load_p): Fix a spacing issue.
24043 * doc/extend.texi (Common Function Attributes): Document that the
24044 PowerPC supports the target_clone attribute.
24046 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24048 * config/arm/arm.h: explain F symbol found in description of ARM
24049 register allocation in its legend.
24051 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24053 * config/mips/frame-header-opt.c: Include profile-count.h.
24054 * config/riscv/riscv.c: Include profile-count.h
24056 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24058 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24060 (sm_set_flag_if_changed): Add bbs field.
24061 (execute_sm_if_changed_flag_set): Pass BBS.
24062 (execute_sm): Update.
24064 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24066 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24069 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24071 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24072 (peephole2): New peephole2 to emit the above.
24073 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24075 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24077 * config/aarch64/aarch64.c (define_peephole2 above
24078 *sub_<shift>_<mode>): New peephole.
24080 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24082 * config/i386/i386.c (make_resolver_func): Update.
24083 * Makefile.in: Add profile-count.h and profile-count.o
24084 * auto-profile.c (afdo_indirect_call): Update to new API.
24085 (afdo_set_bb_count): Update.
24086 (afdo_propagate_edge): Update.
24087 (afdo_propagate_circuit): Update.
24088 (afdo_calculate_branch_prob): Update.
24089 (afdo_annotate_cfg): Update.
24090 * basic-block.h: Include profile-count.h
24091 (struct edge_def): Turn count to profile_count.
24092 (struct basic_block_def): Likewie.
24093 (REG_BR_PROB_BASE): Move to profile-count.h
24094 (RDIV): Move to profile-count.h
24095 * bb-reorder.c (max_entry_count): Turn to profile_count.
24096 (find_traces): Update.
24097 (rotate_loop):Update.
24098 (connect_traces):Update.
24099 (sanitize_hot_paths):Update.
24100 * bt-load.c (migrate_btr_defs): Update.
24101 * cfg.c (RDIV): Remove.
24102 (init_flow): Use alloc_block.
24103 (alloc_block): Uninitialize count.
24104 (unchecked_make_edge): Uninitialize count.
24105 (check_bb_profile): Update.
24106 (dump_edge_info): Update.
24107 (dump_bb_info): Update.
24108 (update_bb_profile_for_threading): Update.
24109 (scale_bbs_frequencies_int): Update.
24110 (scale_bbs_frequencies_gcov_type): Update.
24111 (scale_bbs_frequencies_profile_count): New.
24112 * cfg.h (update_bb_profile_for_threading): Update.
24113 (scale_bbs_frequencies_profile_count): Declare.
24114 * cfgbuild.c (compute_outgoing_frequencies): Update.
24115 (find_many_sub_basic_blocks): Update.
24116 * cfgcleanup.c (try_forward_edges): Update.
24117 (try_crossjump_to_edge): Update.
24118 * cfgexpand.c (expand_gimple_tailcall): Update.
24119 (construct_exit_block): Update.
24120 * cfghooks.c (verify_flow_info): Update.
24121 (dump_bb_for_graph): Update.
24122 (split_edge): Update.
24123 (make_forwarder_block): Update.
24124 (duplicate_block): Update.
24125 (account_profile_record): Update.
24126 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24127 (get_estimated_loop_iterations): Update.
24128 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24129 (single_likely_exit): Update.
24130 * cfgloopmanip.c (scale_loop_profile): Update.
24132 (set_zero_probability): Update.
24133 (lv_adjust_loop_entry_edge): Update.
24134 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24135 (purge_dead_edges): Update.
24136 (rtl_account_profile_record): Update.
24137 * cgraph.c (cgraph_node::create): Uninitialize count.
24138 (symbol_table::create_edge): Uninitialize count.
24139 (cgraph_update_edges_for_call_stmt_node): Update.
24140 (cgraph_edge::dump_edge_flags): Update.
24141 (cgraph_node::dump): Update.
24142 (cgraph_edge::maybe_hot_p): Update.
24143 * cgraph.h: Include profile-count.h
24144 (create_clone), create_edge, create_indirect_edge): Update.
24145 (cgraph_node): Turn count to profile_count.
24146 (cgraph_edge0: Likewise.
24147 (make_speculative, clone): Update.
24148 (create_edge): Update.
24149 (init_lowered_empty_function): Update.
24150 * cgraphclones.c (cgraph_edge::clone): Update.
24151 (duplicate_thunk_for_node): Update.
24152 (cgraph_node::create_clone): Update.
24153 * cgraphunit.c (cgraph_node::analyze): Update.
24154 (cgraph_node::expand_thunk): Update.
24155 * final.c (dump_basic_block_info): Update.
24156 * gimple-streamer-in.c (input_bb): Update.
24157 * gimple-streamer-out.c (output_bb): Update.
24158 * graphite.c (print_global_statistics): Update.
24159 (print_graphite_scop_statistics): Update.
24160 * hsa-brig.c: Include basic-block.h.
24161 * hsa-dump.c: Include basic-block.h.
24162 * hsa-gen.c (T sum_slice): Update.
24163 (convert_switch_statements):Update.
24164 * hsa-regalloc.c: Include basic-block.h.
24165 * ipa-chkp.c (chkp_produce_thunks): Update.
24166 * ipa-cp.c (struct caller_statistics): Update.
24167 (init_caller_stats): Update.
24168 (gather_caller_stats): Update.
24169 (ipcp_cloning_candidate_p): Update.
24170 (good_cloning_opportunity_p): Update.
24171 (get_info_about_necessary_edges): Update.
24172 (dump_profile_updates): Update.
24173 (update_profiling_info): Update.
24174 (update_specialized_profile): Update.
24175 (perhaps_add_new_callers): Update.
24176 (decide_about_value): Update.
24177 (ipa_cp_c_finalize): Update.
24178 * ipa-devirt.c (struct odr_type_warn_count): Update.
24179 (struct decl_warn_count): Update.
24180 (struct final_warning_record): Update.
24181 (possible_polymorphic_call_targets): Update.
24182 (ipa_devirt): Update.
24183 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24184 * ipa-icf.c (sem_function::merge): Update.
24185 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24186 * ipa-inline.c (compute_uninlined_call_time): Update.
24187 (compute_inlined_call_time): Update.
24188 (want_inline_small_function_p): Update.
24189 (want_inline_self_recursive_call_p): Update.
24190 (edge_badness): Update.
24191 (lookup_recursive_calls): Update.
24192 (recursive_inlining): Update.
24193 (inline_small_functions): Update.
24194 (dump_overall_stats): Update.
24195 (dump_inline_stats): Update.
24196 * ipa-profile.c (ipa_profile_generate_summary): Update.
24197 (ipa_propagate_frequency): Update.
24198 (ipa_profile): Update.
24199 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24200 * ipa-utils.c (ipa_merge_profiles): Update.
24201 * loop-doloop.c (doloop_modify): Update.
24202 * loop-unroll.c (report_unroll): Update.
24203 (unroll_loop_runtime_iterations): Update.
24204 * lto-cgraph.c (lto_output_edge): Update.
24205 (lto_output_node): Update.
24206 (input_node): Update.
24207 (input_edge): Update.
24208 (merge_profile_summaries): Update.
24209 * lto-streamer-in.c (input_cfg): Update.
24210 * lto-streamer-out.c (output_cfg): Update.
24211 * mcf.c (create_fixup_graph): Update.
24212 (adjust_cfg_counts): Update.
24213 (sum_edge_counts): Update.
24214 * modulo-sched.c (sms_schedule): Update.
24215 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24216 * predict.c (maybe_hot_count_p): Update.
24217 (probably_never_executed): Update.
24218 (dump_prediction): Update.
24219 (combine_predictions_for_bb): Update.
24220 (propagate_freq): Update.
24221 (handle_missing_profiles): Update.
24222 (counts_to_freqs): Update.
24223 (rebuild_frequencies): Update.
24224 (force_edge_cold): Update.
24225 * predict.h: Include profile-count.h
24226 (maybe_hot_count_p, counts_to_freqs): UPdate.
24227 * print-rtl-function.c: Do not include cfg.h
24228 * print-rtl.c: Include basic-block.h
24229 * profile-count.c: New file.
24230 * profile-count.h: New file.
24231 * profile.c (is_edge_inconsistent): Update.
24232 (correct_negative_edge_counts): Update.
24233 (is_inconsistent): Update.
24234 (set_bb_counts): Update.
24235 (read_profile_edge_counts): Update.
24236 (compute_frequency_overlap): Update.
24237 (compute_branch_probabilities): Update; Initialize and deinitialize
24239 (branch_prob): Update.
24240 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24241 (edge_gcov_count): New.
24242 (bb_gcov_count): New.
24243 * shrink-wrap.c (try_shrink_wrapping): Update.
24244 * tracer.c (better_p): Update.
24245 * trans-mem.c (expand_transaction): Update.
24246 (ipa_tm_insert_irr_call): Update.
24247 (ipa_tm_insert_gettmclone_call): Update.
24248 * tree-call-cdce.c: Update.
24249 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24250 (gimple_duplicate_sese_tail): Update.
24251 (gimple_account_profile_record): Update.
24252 (execute_fixup_cfg): Update.
24253 * tree-inline.c (copy_bb): Update.
24254 (copy_edges_for_bb): Update.
24255 (initialize_cfun): Update.
24256 (freqs_to_counts): Update.
24257 (copy_cfg_body): Update.
24258 (expand_call_inline): Update.
24259 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24260 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24261 (try_unroll_loop_completely): Update.
24262 (try_peel_loop): Update.
24263 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24264 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24265 * tree-ssa-loop-split.c (connect_loops): Update.
24266 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24267 * tree-ssa-reassoc.c (branch_fixup): Update.
24268 * tree-ssa-tail-merge.c (replace_block_by): Update.
24269 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24270 (compute_path_counts): Update.
24271 (update_profile): Update.
24272 (recompute_probabilities): Update.
24273 (update_joiner_offpath_counts): Update.
24274 (estimated_freqs_path): Update.
24275 (freqs_to_counts_path): Update.
24276 (clear_counts_path): Update.
24277 (ssa_fix_duplicate_block_edges): Update.
24278 (duplicate_thread_path): Update.
24279 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24280 (struct switch_conv_info): Update.
24281 * tree-tailcall.c (decrease_profile): Update.
24282 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24283 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24284 * value-prof.c (check_counter): Update.
24285 (gimple_divmod_fixed_value): Update.
24286 (gimple_mod_pow2): Update.
24287 (gimple_mod_subtract): Update.
24288 (gimple_ic_transform): Update.
24289 (gimple_stringop_fixed_value): Update.
24290 * value-prof.h (gimple_ic): Update.
24292 2017-06-02 Carl Love <cel@us.ibm.com>
24294 * config/rs6000/rs6000-c: Add support for built-in functions
24295 vector double vec_doublee (vector signed int);
24296 vector double vec_doublee (vector unsigned int);
24297 vector double vec_doublee (vector float);
24298 vector double vec_doubleh (vector signed int);
24299 vector double vec_doubleh (vector unsigned int);
24300 vector double vec_doubleh (vector float);
24301 vector double vec_doublel (vector signed int);
24302 vector double vec_doublel (vector unsigned int);
24303 vector double vec_doublel (vector float);
24304 vector double vec_doubleo (vector signed int);
24305 vector double vec_doubleo (vector unsigned int);
24306 vector double vec_doubleo (vector float);.
24307 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24308 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24310 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24311 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24312 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24314 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24315 vec_doublel, vec_doubleh.
24316 * doc/extend.texi: Update the built-in documentation file for the
24317 new built-in functions.
24319 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24322 * ipa-inline-analysis.c (free_growth_caches): Set
24323 edge_removal_hook_holder to NULL after removing it.
24325 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24327 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24328 comparision with zero.
24330 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24331 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24332 for early expansion of vec_min and vec_max builtins.
24333 (builtin_function_type): Add min/max unsigned variants to those
24334 identified as having unsigned arguments.
24336 2017-06-02 Olivier Hainque <hainque@adacore.com>
24338 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24340 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24342 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24343 Use VALL_F16 iterator rather than VALL.
24345 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24347 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24348 Emit CBNZ inside loop when doing a strong exchange and comparing
24349 against zero. Generate the CC flags after the loop.
24351 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24353 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24354 (dl_section_ref): New.
24355 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24356 On AIX, append an expression to subtract the size of the
24357 section length to dl_section_ref.
24359 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24362 for early expansion of vector absolute builtins.
24364 2017-06-02 Richard Biener <rguenther@suse.de>
24366 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24367 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24369 2017-06-02 Richard Biener <rguenther@suse.de>
24371 PR tree-optimization/80948
24372 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24373 stmt order as well.
24375 2017-06-02 Richard Biener <rguenther@suse.de>
24377 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24379 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24380 for inductions as relevant.
24382 2017-06-02 Richard Biener <rguenther@suse.de>
24384 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24385 (vectorizable_induction): ... this. Remove dead code.
24387 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24389 * builtins. (expand_builtin_alloca): Remove second parameter and
24390 infer its value from the first parameter instead.
24391 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24393 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24395 PR rtl-optimization/80903
24396 * loop-doloop.c (add_test): Unshare sequence.
24398 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24400 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24402 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24404 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24406 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24407 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24408 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24409 (xlogue_layout::get_stub_rtx): Make static.
24410 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24411 (xlogue_layout::compute_stub_managed_regs): Rename to...
24412 (xlogue_layout::count_stub_managed_regs): ...this.
24413 (xlogue_layout::is_stub_managed_reg): New function.
24414 (xlogue_layout::m_stub_names): Rename to...
24415 (xlogue_layout::s_stub_names): ...this, make static.
24416 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24417 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24418 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24419 xlogue_layout::s_stub_names): Instantiate statics.
24420 (stub_managed_regs): Remove.
24421 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24422 (disable_call_ms2sysv_xlogues): Rename to...
24423 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24424 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24426 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24427 change after reload_completed.
24428 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24430 (ix86_expand_prologue): Likewise.
24431 (ix86_expand_epilogue): Likewise.
24432 (ix86_expand_split_stack_prologue): Likewise.
24433 (ix86_compute_frame_layout): Remove frame parameter ...
24434 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24435 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24437 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24438 (ix86_frame): Move from here ...
24439 * config/i386/i386.h (ix86_frame): ... to here.
24440 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24441 complete ix86_frame data structure instead. Remove some_ld_name.
24443 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24445 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24446 symbols that hold a DECL_VALUE_EXPR.
24448 2017-06-01 Martin Jambor <mjambor@suse.cz>
24450 PR tree-optimization/80898
24451 * tree-sra.c (process_subtree_disqualification): Removed.
24452 (disqualify_candidate): Do not acll
24453 process_subtree_disqualification.
24454 (subtree_mark_written_and_enqueue): New function.
24455 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24456 RHS has been disqualified and re-queue LHS if necessary. Apart
24457 from that, ignore disqualified RHS.
24459 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24461 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24462 address fetch for z10 or later.
24464 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24466 * config/arc/arc.md (tst_movb): Add guard when splitting.
24468 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24470 * config/arc/arc.c (arc_can_eliminate): Test against
24471 arc_frame_pointer_needed.
24473 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24475 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24476 to prevent store reordering.
24477 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24478 (type): Add block type.
24479 (stack_tie): Define special instruction to be used in
24482 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24484 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24485 constraint. It is not valid for the pattern.
24486 (noncommutative_binary_comparison): Likewise.
24488 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24490 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24493 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24495 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24496 be used by the reg-alloc.
24498 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24500 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24501 reg-alloc when having mul64 or mul32x16 instructions.
24502 (mulsidi3): Likewise.
24503 (umulsidi3): Likewise.
24504 (mulsi32x16): New pattern.
24505 (mulsi64): Likewise.
24506 (mulsidi64): Likewise.
24507 (umulsidi64): Likewise.
24508 (MUL32x16_REG): Define.
24509 (mul64_600): Use MUL32x16_REG.
24510 (mac64_600): Likewise.
24511 (umul64_600): Likewise.
24512 (umac64_600): Likewise.
24514 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24516 * config/arc/arc.md (mulsi3_700): Make it commutative.
24518 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24520 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24522 (*sign_extendsidi2_insn): Likewise for movstosw.
24524 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24526 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24527 the type of the input discriminant value. Convert the
24528 discriminant value of signedness vary.
24530 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24532 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24533 Add to -Wall section.
24535 2017-06-01 Richard Biener <rguenther@suse.de>
24537 PR middle-end/66313
24538 * fold-const.c (fold_plusminus_mult_expr): If the factored
24539 factor may be zero use a wrapping type for the inner operation.
24540 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24541 and handle moved defs.
24542 (process_assignment): Properly guard the unary op case. Return a
24543 tri-state indicating that moving the stmt before the call may allow
24544 to continue. Pass through to_move.
24545 (find_tail_calls): Handle moving unrelated defs before
24548 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24551 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24552 splitter result in the canonical way.
24554 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24556 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24557 also for 32bit target. Update insn attributes.
24558 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24560 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24562 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24563 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24564 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24565 (_mm_maskz_min_ss): New intrinsics.
24567 2017-05-31 Martin Liska <mliska@suse.cz>
24569 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24570 Change comment style to one we normally use.
24571 (vectorizable_reduction): Likewise.
24572 (vectorizable_induction): Likewise.
24573 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24574 (vectorizable_call): Likewise.
24575 (vectorizable_simd_clone_call): Likewise.
24576 (vectorizable_conversion): Likewise.
24577 (vectorizable_assignment): Likewise.
24578 (vectorizable_shift): Likewise.
24579 (vectorizable_operation): Likewise.
24580 (vectorizable_store): Likewise.
24581 (vectorizable_load): Likewise.
24582 * tree-vectorizer.h: Likewise.
24584 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24586 * passes.c (emergency_dump_function): New.
24587 * tree-pass.h (emergency_dump_function): Declare.
24588 * plugin.c (plugins_internal_error_function): Remove.
24589 * plugin.h (plugins_internal_error_function): Remove declaration.
24590 * toplev.c (internal_error_function): New static function. Use it...
24591 (general_init): ...here.
24593 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24595 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24596 (arc_rtx_costs): Add costs for new patterns.
24597 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24598 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24600 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24602 * tree-ssa-strlen.c (get_next_strinfo): New function.
24603 (get_stridx_plus_constant): Use it.
24604 (zero_length_string): Likewise.
24605 (adjust_related_strinfos): Likewise.
24606 (adjust_last_stmt): Likewise.
24608 2017-05-31 Richard Biener <rguenther@suse.de>
24611 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24612 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24614 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24616 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24617 loop_vinfo argument and use of dependence distance vectors.
24618 Check instead whether the two references differ only in their
24619 initial value and assume that they have the same alignment if the
24620 difference is a multiple of the vector alignment.
24621 (vect_analyze_data_refs_alignment): Update call accordingly.
24623 2017-05-31 Martin Liska <mliska@suse.cz>
24626 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24628 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24630 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24631 (create_intersect_range_checks): Move from ...
24632 * tree-data-ref.c (create_intersect_range_checks_index)
24633 (create_intersect_range_checks): ... to here.
24634 (create_runtime_alias_checks): New function factored from ...
24635 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24636 here. Call above function.
24637 * tree-data-ref.h (create_runtime_alias_checks): New function.
24639 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24641 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24642 segment length for dr_b and compute it in wide_int.
24644 2017-05-31 Richard Biener <rguenther@suse.de>
24646 PR tree-optimization/80906
24647 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24648 and pass through iv_map.
24649 (copy_bb_and_scalar_dependences): Adjust.
24650 (translate_pending_phi_nodes): Likewise.
24651 (copy_loop_close_phi_args): Handle code-generating IVs instead
24654 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24656 * diagnostic-color.c (color_dict): Add "type-diff".
24657 (parse_gcc_colors): Update comment.
24658 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24659 -fdiagnostics-show-template-tree and -fno-elide-type.
24660 (GCC_COLORS): Add type-diff to example.
24662 (-fdiagnostics-show-template-tree): New.
24663 (-fno-elide-type): New.
24664 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24665 the pp_format_decoder callback. Call any m_format_postprocessor's
24667 (pretty_printer::pretty_printer): Initialize
24668 m_format_postprocessor.
24669 (pretty_printer::~pretty_printer): Delete any
24670 m_format_postprocessor.
24671 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24672 (class format_postprocessor): New class.
24673 (struct pretty_printer::format_decoder): Document the new parameters.
24674 (struct pretty_printer::m_format_postprocessor): New field.
24675 * tree-diagnostic.c (default_tree_printer): Update for new
24676 bool and const char ** params.
24677 * tree-diagnostic.h (default_tree_printer): Likewise.
24679 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24681 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24682 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24683 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24684 rs6000_gen_cell_microcode code.
24685 (rs6000_final_prescan_insn): Delete.
24686 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24687 "warn-cell-microcode" entries.
24688 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24689 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24690 throughout. Change cc_reg_not_micro_cr0_operand to
24691 cc_reg_not_cr0_operand throughout.
24692 (*extendhi<mode>2_noload): Delete.
24693 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24694 (mwarn-cell-microcode): Delete.
24695 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24696 -mgen-cell-microcode and -mwarn-cell-microcode.
24698 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
24701 * config/i386/constraints.md (Yd): New constraint.
24703 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
24704 and (?Yd, r) alternatives. Update insn attributes.
24705 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
24706 and (?*Yd, r) alternatives. Update insn attributes.
24707 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
24709 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
24711 * gimplify.c (gimplify_modify_expr): Don't create a
24712 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
24715 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
24717 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
24719 2017-05-30 Richard Biener <rguenther@suse.de>
24721 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
24722 and reduc_def fields.
24723 (STMT_VINFO_REDUC_TYPE): New define.
24724 (STMT_VINFO_REDUC_DEF): Likewise.
24725 (vect_force_simple_reduction): Adjust prototype.
24726 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
24727 (vect_is_simple_reduction): Remove check_reduction argument.
24728 (vect_force_simple_reduction): Adjust and set
24729 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24730 (vectorizable_reduction): Do not re-do reduction analysis
24731 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
24732 * tree-parloops.c (gather_scalar_reductions): Adjust.
24734 2017-05-30 Richard Biener <rguenther@suse.de>
24736 PR middle-end/80901
24737 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
24740 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
24742 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
24743 Introduce unknown_misalignment parameter and remove vf.
24744 (vect_peeling_hash_get_lowest_cost):
24745 Pass unknown_misalignment parameter.
24746 (vect_enhance_data_refs_alignment):
24747 Fix unsupportable data ref treatment.
24749 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24751 * tree-vect-data-refs.c (vect_get_data_access_cost):
24752 Workaround for SLP handling.
24753 (vect_enhance_data_refs_alignment):
24754 Compute costs for doing no peeling at all, compare to the best
24755 peeling costs so far and avoid peeling if cheaper.
24757 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24759 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
24760 Return peeling info and set costs to zero for unlimited cost
24762 (vect_enhance_data_refs_alignment): Also inspect all datarefs
24763 with unknown misalignment. Compute and costs for unknown
24764 misalignment, compare them to the costs for known misalignment
24765 and choose the cheapest for peeling.
24767 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24769 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
24770 (vect_get_peeling_costs_all_drs): Create function.
24771 (vect_peeling_hash_get_lowest_cost):
24772 Use vect_get_peeling_costs_all_drs.
24773 (vect_peeling_supportable): Create function.
24774 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
24776 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
24778 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
24779 DR_HAS_NEGATIVE_STEP.
24780 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
24781 (vect_enhance_data_refs_alignment): Use.
24782 (vect_duplicate_ssa_name_ptr_info): Use.
24783 * tree-vectorizer.h (dr_misalignment): Use.
24784 (known_alignment_for_access_p): Use.
24786 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24789 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
24791 (has_section_name): New function.
24793 2017-05-30 Martin Liska <mliska@suse.cz>
24796 * auto-profile.c (get_function_decl_from_block): Fix
24799 2017-05-30 Richard Biener <rguenther@suse.de>
24801 PR middle-end/80876
24802 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
24804 2017-05-30 Martin Liska <mliska@suse.cz>
24806 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
24807 * dumpfile.h (struct dump_file_info): Remove ctors.
24809 2017-05-30 Martin Liska <mliska@suse.cz>
24811 * predict.def: Fix GNU coding style.
24813 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24815 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24816 Mark 'to' argument with ATTRIBUTE_UNUSED.
24818 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
24820 * config/xtensa/xtensa.c (xtensa_emit_call): Use
24821 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
24822 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
24825 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
24827 * doc/install.texi (Options specification): Restore entry of
24828 --enable-sjlj-exceptions.
24830 2017-05-27 Michael Eager <eager@eagercon.com>
24833 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
24835 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
24837 * config/microblaze/microblaze.h
24838 (FIXED_REGISTERS): Update in macro.
24839 (CALL_USED_REGISTERS): Update in macro.
24841 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
24843 * doc/install.texi: Add links to macOS binary distributions.
24845 2017-05-27 Jakub Jelinek <jakub@redhat.com>
24849 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24851 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24853 2017-05-26 Martin Liska <mliska@suse.cz>
24855 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
24857 2017-05-26 Martin Liska <mliska@suse.cz>
24859 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
24860 always leading ';; '.
24861 (dump_bb_info): Likewise.
24862 (brief_dump_cfg): Likewise.
24863 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
24864 * dumpfile.c: Remove usage of TDF_VERBOSE.
24865 * dumpfile.h (enum dump_kind): Likewise.
24866 (dump_gimple_bb_header): Do not use TDF_COMMENT.
24867 * print-tree.c (debug_verbose): Remove.
24868 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
24869 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
24870 * tree-diagnostic.c (default_tree_printer): Replace
24871 TDF_DIAGNOSTIC with TDF_SLIM.
24873 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24875 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
24876 in parameter loop, rather than loop_vinfo.
24877 (create_intersect_range_checks): Ditto.
24878 (vect_create_cond_for_alias_checks): Update call to above functions.
24880 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24882 PR tree-optimization/80815
24883 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24884 for merging runtime alias checks. Handle negative DR_STEPs.
24886 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24888 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24890 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24891 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24892 out code pruning runtime alias checks.
24893 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24894 factored out from above.
24895 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24897 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24899 (prune_runtime_alias_test_list): New decalaration.
24901 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24903 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24904 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24905 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24906 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24907 (operator==, comp_dr_with_seg_len_pair): Ditto.
24908 (vect_prune_runtime_alias_test_list): Ditto.
24910 2017-05-26 Martin Liska <mliska@suse.cz>
24913 * params.def: Bound partial-inlining-entry-probability param.
24915 2017-05-26 Marek Polacek <polacek@redhat.com>
24918 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24921 2017-05-26 Richard Biener <rguenther@suse.de>
24923 PR tree-optimization/80842
24924 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24927 2017-05-26 Richard Biener <rguenther@suse.de>
24929 PR tree-optimization/80844
24930 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24932 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24934 * doc/md.texi (Machine Constraints): Update x86 family
24935 machine constraints section to match 'config/i386/constraints.md'.
24937 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24939 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24941 2017-05-25 Nathan Sidwell <nathan@acm.org>
24943 * doc/invoke.texi (--enable-languages): Update documentation.
24945 2017-05-25 Martin Liska <mliska@suse.cz>
24947 * dumpfile.c: Add TDF_FOLDING.
24948 * dumpfile.h (enum dump_kind): Likewise.
24949 * genmatch.c (dt_simplify::gen_1): Use it.
24951 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24953 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24955 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24957 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24958 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24960 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24962 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24963 * match.pd (X == C): Rewrite it here.
24964 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24965 with_certain_nonzero_bits2): New predicates.
24966 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24968 2017-05-24 Nathan Sidwell <nathan@acm.org>
24970 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24973 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24976 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24978 * config/powerpcspe: New port. Files are copied from the rs6000
24979 port, with "rs6000" in filenames replaced by "powerpcspe".
24981 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24983 PR rtl-optimization/80754
24984 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24986 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24988 * config/sparc/sparc.md (length): Return the correct value for -mflat
24989 sibcalls to match output_sibcall.
24991 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24995 * config/rs6000/rs6000.c (struct machine_function): Add new field
24997 (rs6000_get_separate_components): Init that field, use it.
24998 (rs6000_components_for_bb): Use the field.
25000 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25002 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25004 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25006 PR middle-end/80823
25007 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25009 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25012 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25013 address against address_operand predicate.
25014 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25016 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25018 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25019 parameters passed indirectly.
25021 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25023 * config/i386/i386.md (*movdi_internal): Remove SSE4
25024 alternative 18 (?r, *v). Update insn attributes.
25025 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25026 Update insn attributes.
25027 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25028 Update insn attributes.
25029 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25030 alternative 1 (r, v). Remove isa attribute.
25031 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25032 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25033 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25035 2017-05-23 Tom de Vries <tom@codesourcery.com>
25037 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25040 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25042 * cgraphunit.c (symbol_table::process_new_functions): Update.
25043 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25044 (inline_generate_summary): Rename to ...
25045 (ipa_fn_summary_generate): ... this one.
25046 (inline_read_summary): Rename to ...
25047 (ipa_fn_summary_read): ... this one.
25048 (inline_write_summary): Rename to ...
25049 (ipa_fn_summary_write): ... this one.
25050 (inline_free_summary): Rename to ...
25051 (ipa_free_fn_summary): ... this one.
25052 (pass_data_local_fn_summary, pass_local_fn_summary,
25053 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25054 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25055 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25056 make_pass_ipa_fn_summary): New.
25057 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25058 inline_write_summary, inline_free_summary): Remove.
25059 (ipa_free_fn_summary) : New.
25060 * ipa-inline.c (ipa_inline): Update.
25061 (pass_ipa_inline): Do not generate summaries.
25062 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25064 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25065 and add pass_ipa_fn_summary.
25066 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25068 (make_pass_inline_parameters): Remove.
25070 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25072 * omp-low.c (struct omp_context): Remove "default_kind" member.
25075 * omp-offload.c (execute_oacc_device_lower): Remove the
25076 parallelism dimensions function attributes for unparallelized
25077 OpenACC kernels constructs.
25079 2017-05-23 Martin Liska <mliska@suse.cz>
25081 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25083 (cgraph_edge::make_speculative): Likewise.
25084 (cgraph_edge::resolve_speculation): Likewise.
25085 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25086 (cgraph_node::dump): Likewise.
25087 * cgraph.h: Likewise.
25088 * cgraphunit.c (analyze_functions): Likewise.
25089 (symbol_table::compile): Likewise.
25090 * ipa-cp.c (print_all_lattices): Likewise.
25091 (determine_versionability): Likewise.
25092 (initialize_node_lattices): Likewise.
25093 (ipcp_verify_propagated_values): Likewise.
25094 (estimate_local_effects): Likewise.
25095 (update_profiling_info): Likewise.
25096 (create_specialized_node): Likewise.
25097 (perhaps_add_new_callers): Likewise.
25098 (decide_about_value): Likewise.
25099 (decide_whether_version_node): Likewise.
25100 (identify_dead_nodes): Likewise.
25101 (ipcp_store_bits_results): Likewise.
25102 * ipa-devirt.c (dump_targets): Likewise.
25103 (ipa_devirt): Likewise.
25104 * ipa-icf.c (sem_item::dump): Likewise.
25105 (sem_function::equals): Likewise.
25106 (sem_variable::equals): Likewise.
25107 (sem_item_optimizer::read_section): Likewise.
25108 (sem_item_optimizer::execute): Likewise.
25109 (congruence_class::dump): Likewise.
25110 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25111 (dump_inline_summary): Likewise.
25112 (estimate_node_size_and_time): Likewise.
25113 (inline_analyze_function): Likewise.
25114 * ipa-inline-transform.c (inline_call): Likewise.
25115 * ipa-inline.c (report_inline_failed_reason): Likewise.
25116 (want_early_inline_function_p): Likewise.
25117 (edge_badness): Likewise.
25118 (update_edge_key): Likewise.
25119 (inline_small_functions): Likewise.
25120 * ipa-profile.c (ipa_profile): Likewise.
25121 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25122 (ipa_make_edge_direct_to_target): Likewise.
25123 (remove_described_reference): Likewise.
25124 (ipa_impossible_devirt_target): Likewise.
25125 (propagate_controlled_uses): Likewise.
25126 (ipa_print_node_params): Likewise.
25127 (ipcp_transform_function): Likewise.
25128 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25129 (propagate_pure_const): Likewise.
25130 * ipa-reference.c (generate_summary): Likewise.
25131 (read_write_all_from_decl): Likewise.
25132 (propagate): Likewise.
25133 (ipa_reference_read_optimization_summary): Likewise.
25134 * ipa-utils.c (ipa_merge_profiles): Likewise.
25135 * ipa.c (walk_polymorphic_call_targets): Likewise.
25136 (symbol_table::remove_unreachable_nodes): Likewise.
25137 (ipa_single_use): Likewise.
25138 * passes.c (execute_todo): Likewise.
25139 * predict.c (drop_profile): Likewise.
25140 * symtab.c (symtab_node::get_dump_name): New function.
25141 (symtab_node::dump_name): Likewise.
25142 (symtab_node::dump_asm_name): Likewise.
25143 (symtab_node::dump_references): Likewise.
25144 (symtab_node::dump_referring): Likewise.
25145 (symtab_node::dump_base): Likewise.
25146 (symtab_node::debug_symtab): Likewise.
25147 * tree-sra.c (convert_callers_for_node): Likewise.
25148 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25149 * value-prof.c (init_node_map): Likewise.
25151 2017-05-23 Martin Liska <mliska@suse.cz>
25153 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25154 and symtab_node::debug_symtab to symbol_table::debug.
25155 * cgraphunit.c (analyze_functions): Use the renamed function.
25156 (symbol_table::compile): Likewise.
25157 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25158 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25159 * passes.c (execute_todo): Likewise.
25160 * symtab.c (symbol_table::dump): New function.
25161 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25163 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25165 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25166 that nonconst implies exec.
25168 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25170 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25171 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25172 (inline_edge_summary_vec): Turn into ...
25173 (ipa_call_summaries): ... this one.
25174 (redirect_to_unreachable, edge_set_predicate,
25175 evaluate_properties_for_edge, inline_summary_alloc,
25176 reset_ipa_call_summary, reset_inline_summary,
25177 inline_summary_t::duplicate): Update.
25178 (inline_edge_duplication_hook): Turn to ...
25179 (ipa_call_summary_t::duplicate): ... this one.
25180 (inline_edge_removal_hook): Turn to ...
25181 (ipa_call_summary_t::remove): ... this one.
25182 (dump_inline_edge_summary): Turn to ...
25183 (dump_ipa_call_summary): ... this one.
25184 (estimate_function_body_sizes): Update.
25185 (inline_update_callee_summaries): Update.
25186 (remap_edge_change_prob): Update.
25187 (remap_edge_summaries): Update.
25188 (inline_merge_summary): Update.
25189 (do_estimate_edge_time): Update.
25190 (inline_generate_summary): Update.
25191 (inline_read_section): Update.
25192 (inline_read_summary): Update.
25193 (inline_free_summary): Update.
25194 * ipa-inline.c (can_inline_edge_p): Update.
25195 (compute_inlined_call_time): Update.
25196 (want_inline_small_function_p): Update.
25197 (edge_badness): Update.
25198 (early_inliner): Update.
25199 * ipa-inline.h (inline_edge_summary): Turn to ...
25200 (ipa_call_summary): ... this one.
25201 (ipa_call_summary_t): New class.
25202 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25203 (ipa_call_summaries): New.
25204 (inline_edge_summary): Remove.
25205 (estimate_edge_growth): Update.
25206 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25207 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25208 * ipa-split.c (execute_split_functions): Update.
25209 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25211 2017-05-23 Tom de Vries <tom@codesourcery.com>
25213 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25214 attributes): Document rdrand effective target.
25216 2017-05-23 Tom de Vries <tom@codesourcery.com>
25218 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25219 attributes): Sort alphabetically.
25221 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25223 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25225 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25228 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25229 V2DF/V2DI splat into two separate patterns, one that handles
25230 registers, and the other that only handles memory. Drop support
25231 for splatting from a GPR on ISA 2.07 and then splitting the
25232 splat into direct move and splat.
25233 (vsx_splat_<mode>_reg): Likewise.
25234 (vsx_splat_<mode>_mem): Likewise.
25236 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25238 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25240 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25242 PR middle-end/80809
25243 * omp-low.c (finish_taskreg_remap): New function.
25244 (finish_taskreg_scan): If unit size of ctx->record_type
25245 is non-constant, unshare the size expression and replace
25246 decls in it with possible outer var refs.
25248 PR middle-end/80809
25249 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25250 GOVD_SHARED rather than GOVD_PRIVATE with it.
25251 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25252 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25254 PR middle-end/80853
25255 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25256 as last argument to build_outer_var_ref for pointer bases of array
25257 section reductions.
25259 2017-05-19 Martin Sebor <msebor@redhat.com>
25261 * print-tree.c (print_node): Print DECL_READ_P flag.
25263 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25265 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25266 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25267 * cgraph.c: Likewise.
25268 * cgraphunit.c: Likewise.
25269 * gengtype.c: Likewise.
25270 * ipa-cp.c: Likewise.
25271 * ipa-devirt.c: Likewise.
25272 * ipa-icf.c: Likewise.
25273 * ipa-predicate.c: Likewise.
25274 * ipa-profile.c: Likewise.
25275 * ipa-prop.c: Likewise.
25276 * ipa-split.c: Likewise.
25278 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25279 edge_predicate_pool, dump_inline_hints,
25280 inline_summary::account_size_time, redirect_to_unreachable,
25281 edge_set_predicate, set_hint_predicate,
25282 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25283 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25284 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25285 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25286 ipa_call_summary_t::remove, initialize_growth_caches,
25287 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25288 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25289 mark_modified, unmodified_parm_1, unmodified_parm,
25290 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25291 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25292 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25293 will_be_nonconstant_predicate, record_modified_bb_info,
25294 get_minimal_bb, record_modified, param_change_prob,
25295 phi_result_unknown_predicate, predicate_for_phi_result,
25296 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25297 estimate_function_body_sizes, compute_inline_parameters,
25298 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25299 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25300 inline_update_callee_summaries, remap_edge_change_prob,
25301 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25302 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25303 inline_analyze_function, inline_summary_t::insert,
25304 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25305 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25306 inline_free_summary): Move to ipa-fnsummary.h
25307 (predicate_t): Remove.
25308 * ipa-fnsummary.c: New file.
25309 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25310 (enum inline_hints_vals, inline_hints, agg_position_info,
25311 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25312 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25313 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25314 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25315 inline_read_summary, inline_write_summary, inline_free_summary,
25316 inline_analyze_function, initialize_inline_failed,
25317 inline_merge_summary, inline_update_overall_summary,
25318 compute_inline_parameters): Move to ipa-fnsummary.h
25319 * ipa-fnsummary.h: New file.
25320 * ipa-inline-transform.h: Include ipa-inline.h.
25321 * ipa-inline.c: LIkewise.
25323 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25325 * ipa-inline.c (edge_badness): Use inlined_time instead of
25326 inline_summaries->get.
25328 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25330 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25332 2017-05-22 Nathan Sidwell <nathan@acm.org>
25334 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25335 (fdump-lang): Document 'raw' option.
25336 * dumpfile.h (TDI_tu): Delete.
25337 * dumpfile.c (dump_files): Remove translation-unit.
25338 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25340 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25342 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25343 command option from $(AWK) call.
25344 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25345 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25346 [FORMAT]: Remove handling of variable.
25347 * config/avr/t-multilib: Regenerate.
25349 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25351 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25353 (dump_inline_summary): Do not print self_time.
25354 (estimate_function_body_sizes): Do not set self_time.
25355 (compute_inline_parameters): Likewise.
25356 (inline_read_section, inline_write_summary): Do not stream self_time.
25357 * ipa-inline.h (inline_summary): Drop self_time.
25359 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25361 * ipa-inline-analysis.c (account_size_time): Rename to ...
25362 (inline_summary::account_size_time): ... this one.
25363 (reset_ipa_call_summary): Turn to ...
25364 (ipa_call_summary::reset): ... this one.
25365 (reset_inline_summary): Turn to ...
25366 (inline_summary::reset): ... this one.
25367 (inline_summary_t::remove): Update.
25368 (inline_summary_t::duplicate): Update.
25369 (ipa_call_summary_t::remove): Update.
25370 (dump_inline_summary): Update.
25371 (estimate_function_body_sizes): Update.
25372 (compute_inline_parameters): Update.
25373 (estimate_node_size_and_time): Update.
25374 (inline_merge_summary): Update.
25375 (inline_update_overall_summary): Update.
25376 (inline_read_section): Update.
25377 (inline_write_summary): Update.
25378 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25379 add account_size_time and reset member functions.
25380 (ipa_call_summary): Add reset function.
25381 * ipa-predicate.h (predicate::operator &): Constify.
25383 2017-05-22 Richard Biener <rguenther@suse.de>
25385 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25387 2017-05-19 Jason Merrill <jason@redhat.com>
25389 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25391 2017-05-19 Marek Polacek <polacek@redhat.com>
25394 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25395 TYPE_OVERFLOW_WRAPS checks.
25397 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25399 * tree-core.h (enum omp_clause_default_kind): Add
25400 "OMP_CLAUSE_DEFAULT_PRESENT".
25401 * tree-pretty-print.c (dump_omp_clause): Handle it.
25402 * gimplify.c (enum gimplify_omp_var_data): Add
25403 "GOVD_MAP_FORCE_PRESENT".
25404 (gimplify_adjust_omp_clauses_1): Map it to
25405 "GOMP_MAP_FORCE_PRESENT".
25406 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25408 * gimplify.c (oacc_default_clause): Clarify.
25410 2017-05-19 Nathan Sidwell <nathan@acm.org>
25412 LANG_HOOK_REGISTER_DUMPS
25413 * toplev.c (general_init): Call register dump lang hook.
25414 * doc/invoke.texi: Document -fdump-lang option family.
25415 * dumpfile.c (dump_files): Remove class dump here.
25416 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25417 * dumpfile.h (tree_dump_index): Remove TDI_class.
25418 * langhooks-def.h (lhd_register_dumps): Declare.
25419 (LANG_HOOKS_REGISTER_DUMPS): Define.
25420 (LANG_HOOKS_INITIALIZER): Add it.
25421 * langhooks.c (lhd_register_dumps): Define.
25422 * langhooks.h (struct lang_hooks): Add register_dumps.
25424 2017-05-19 Nathan Sidwell <nathan@acm.org>
25426 * context.h (context::set_passes): New.
25427 * context.c (context::context): Do not create pass manager.
25428 * toplev.c (general_init): Create pass manager here.
25430 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25432 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25433 use this splitter if two add or or instructions would also work for
25434 the constant we want to generate.
25436 2017-05-19 Richard Biener <rguenther@suse.de>
25439 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25440 predicate evaluation.
25442 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25444 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25446 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25449 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25451 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25452 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25453 (inline_edge_summary_vec): Turn into ...
25454 (ipa_call_summaries): ... this one.
25455 (redirect_to_unreachable, edge_set_predicate,
25456 evaluate_properties_for_edge, inline_summary_alloc,
25457 reset_ipa_call_summary, reset_inline_summary,
25458 inline_summary_t::duplicate): Update.
25459 (inline_edge_duplication_hook): Turn to ...
25460 (ipa_call_summary_t::duplicate): ... this one.
25461 (inline_edge_removal_hook): Turn to ...
25462 (ipa_call_summary_t::remove): ... this one.
25463 (dump_inline_edge_summary): Turn to ...
25464 (dump_ipa_call_summary): ... this one.
25465 (estimate_function_body_sizes): Update.
25466 (inline_update_callee_summaries): Update.
25467 (remap_edge_change_prob): Update.
25468 (remap_edge_summaries): Update.
25469 (inline_merge_summary): Update.
25470 (do_estimate_edge_time): Update.
25471 (inline_generate_summary): Update.
25472 (inline_read_section): Update.
25473 (inline_read_summary): Update.
25474 (inline_free_summary): Update.
25475 * ipa-inline.c (can_inline_edge_p): Update.
25476 (compute_inlined_call_time): Update.
25477 (want_inline_small_function_p): Update.
25478 (edge_badness): Update.
25479 (early_inliner): Update.
25480 * ipa-inline.h (inline_edge_summary): Turn to ...
25481 (ipa_call_summary): ... this one.
25482 (ipa_call_summary_t): New class.
25483 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25484 (ipa_call_summaries): New.
25485 (inline_edge_summary): Remove.
25486 (estimate_edge_growth): Update.
25487 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25488 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25489 * ipa-split.c (execute_split_functions): Update.
25490 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25492 2017-05-19 Richard Biener <rguenther@suse.de>
25494 PR middle-end/80764
25495 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25497 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25499 * config/rs6000/rs6000.c (struct machine_function): Add field
25500 fpr_is_wrapped_separately.
25501 (rs6000_get_separate_components): Use 64 components. Handle the
25502 new FPR components.
25503 (rs6000_components_for_bb): Handle the FPR components.
25504 (rs6000_emit_prologue_components): Handle the FPR components.
25505 (rs6000_emit_epilogue_components): Handle the FPR components.
25506 (rs6000_set_handled_components): Handle the FPR components.
25507 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25508 that are already separately shrink-wrapped.
25509 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25510 that are already separately shrink-wrapped.
25512 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25515 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25518 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25519 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25520 cases where the register allocator uses a move and an offsettable
25521 memory operation to/from a FPR register on ISA 2.06/2.07.
25522 (define_peephole2 for Altivec d-form store): Likewise.
25524 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25527 * config/i386/mmx.md (*mov<mode>_internal): Enable
25528 alternatives 11, 12, 13 and 14 also for 32bit targets.
25529 Remove alternatives 15, 16, 17 and 18.
25530 * config/i386/sse.md (vec_concatv2di): Change
25531 alternative (!x, *y) to (x, ?!*Yn).
25533 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25535 * dumpfile.h (enum dump_kind): Remove stray comma.
25537 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25539 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25540 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25541 predicate::num_conditions
25542 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25543 (CHANGED): turn into predicate::changed.
25544 (agg_position_info): Move to ipa-predicate.h
25545 (add_condition, predicate::add_clause, predicate::operator &=,
25546 predicate::or_with, predicate::evaluate, predicate::probability,
25547 dump_condition, dump_clause, predicate::dump,
25548 predicate::remap_after_duplication, predicate::remap_after_inlining,
25549 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25550 (evaluate_conditions_for_known_args): Update.
25551 (set_cond_stmt_execution_predicate): Update.
25552 * ipa-inline.h: Include ipa-predicate.h
25553 (condition, inline_param_summary, conditions, agg_position_info,
25554 predicate): Move to ipa-predicate.h
25555 * ipa-predicate.c: New file.
25556 * ipa-predicate.h: New file.
25558 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25560 * final.c (leaf_function_p): Check we are not in a sequence.
25562 2017-05-18 Martin Liska <mliska@suse.cz>
25564 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25565 * dumpfile.c (dump_register): Use new enum dump_kind.
25566 (get_dump_file_name): Likewise.
25567 (dump_enable_all): Likewise.
25568 (dump_switch_p_1): Likewise.
25569 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25570 * dumpfile.h (enum dump_kind): New enum type.
25571 (struct dump_file_info): Create constructor and
25572 format fields and comments.
25573 * passes.c (pass_manager::register_one_dump_file):
25575 * statistics.c (statistics_early_init): Likewise.
25576 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25577 TDF_TREE with TDF_SLIM.
25578 (gather_memory_references_ref): Likewise.
25580 2017-05-18 Martin Liska <mliska@suse.cz>
25582 * vec.h (struct vnull): Use it.
25584 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25586 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25587 (true_predicate, false_predicate, true_predicate_p,
25588 false_predicate_p): Remove.
25589 (single_cond_predicate, not_inlined_predicate): Turn to member function
25591 (add_condition): Update.
25592 (add_clause): Turn to...
25593 (predicate::add_clause): ... this one; update; allow passing NULL
25595 (and_predicates): Turn to ...
25596 (predicate::operator &=): ... this one.
25597 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25598 (or_predicates): Turn to ...
25599 (predicate::or_with): ... this one.
25600 (evaluate_predicate): Turn to ...
25601 (predicate::evaluate): ... this one.
25602 (predicate_probability): Turn to ...
25603 (predicate::probability): ... this one.
25604 (dump_condition): Update.
25605 (dump_predicate): Turn to ...
25606 (predicate::dump): ... this one.
25607 (account_size_time): Update.
25608 (edge_set_predicate): Update.
25609 (set_hint_predicate): UPdate.
25610 (evaluate_conditions_for_known_args): Update.
25611 (evaluate_properties_for_edge): Update.
25612 (remap_predicate_after_duplication): Turn to...
25613 (predicate::remap_after_duplication): ... this one.
25614 (remap_hint_predicate_after_duplication): Update.
25615 (inline_summary_t::duplicate): UPdate.
25616 (dump_inline_edge_summary): Update.
25617 (dump_inline_summary): Update.
25618 (set_cond_stmt_execution_predicate): Update.
25619 (set_switch_stmt_execution_predicate): Update.
25620 (compute_bb_predicates): Update.
25621 (will_be_nonconstant_expr_predicate): Update.
25622 (will_be_nonconstant_predicate): Update.
25623 (phi_result_unknown_predicate): Update.
25624 (predicate_for_phi_result): Update.
25625 (array_index_predicate): Update.
25626 (estimate_function_body_sizes): Update.
25627 (estimate_node_size_and_time): Update.
25628 (estimate_ipcp_clone_size_and_time): Update.
25629 (remap_predicate): Rename to ...
25630 (predicate::remap_after_inlining): ... this one.
25631 (remap_hint_predicate): Update.
25632 (inline_merge_summary): Update.
25633 (inline_update_overall_summary): Update.
25634 (estimate_size_after_inlining): Update.
25635 (read_predicate): Rename to ...
25636 (predicate::stream_in): ... this one.
25637 (read_inline_edge_summary): Update.
25638 (write_predicate): Rename to ...
25639 (predicate::stream_out): ... this one.
25640 (write_inline_edge_summary): Update.
25641 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25642 (clause_t): Turn to uint32_t
25643 (predicate): Turn to class; implement constructor and operators
25645 (size_time_entry): Update.
25646 (inline_summary): Update.
25647 (inline_edge_summary): Update.
25649 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25651 * fold-const.c (fold_binary_loc): Move transformation...
25652 * match.pd (C - X CMP X): ... here.
25654 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25656 * config/sparc/sparc.c (sparc_option_override): Set function
25657 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25658 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25660 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25662 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25664 2017-05-18 Marek Polacek <polacek@redhat.com>
25667 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25668 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25669 gimple_assign_load_p.
25671 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25673 PR middle-end/80692
25674 * real.c (do_compare): Give decimal_do_compare preference over
25675 comparing just the signs.
25677 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25679 * doc/md.texi (Canonicalization of Instructions): Describe the
25680 canonical form of instructions that inherently set a condition
25683 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25685 PR middle-end/80775
25686 * tree-cfg.c: Move deletion of unreachable case statements to after
25687 the merging of consecutive case labels.
25689 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25691 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25692 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25693 restoring of callee-saved registers.
25695 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
25697 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
25698 * config/visium/visium.c (single_set_and_flags): Likewise.
25699 * config/visium/visium.md (Substitutions): Likewise.
25701 2017-05-17 Martin Liska <mliska@suse.cz>
25703 * cfg.c: Introduce dump_flags_t type and
25704 use it instead of int type.
25706 * cfghooks.c: Likewise.
25707 * cfghooks.h (struct cfg_hooks): Likewise.
25708 * cfgrtl.c: Likewise.
25709 * cfgrtl.h: Likewise.
25710 * cgraph.c (cgraph_node::get_body): Likewise.
25711 * coretypes.h: Likewise.
25712 * domwalk.c: Likewise.
25713 * domwalk.h: Likewise.
25714 * dumpfile.c (struct dump_option_value_info): Likewise.
25715 (dump_enable_all): Likewise.
25716 (dump_switch_p_1): Likewise.
25717 (opt_info_switch_p): Likewise.
25718 * dumpfile.h (enum tree_dump_index): Likewise.
25719 (struct dump_file_info): Likewise.
25720 * genemit.c: Likewise.
25721 * generic-match-head.c: Likewise.
25722 * gengtype.c (open_base_files): Likewise.
25723 * gimple-pretty-print.c: Likewise.
25724 * gimple-pretty-print.h: Likewise.
25725 * graph.c (print_graph_cfg): Likewise.
25726 * graphite-scop-detection.c (dot_all_sese): Likewise.
25727 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
25728 * loop-unroll.c (report_unroll): Likewise.
25729 * passes.c (pass_manager::register_one_dump_file): Likewise.
25730 * print-tree.c: Likewise.
25731 * statistics.c: Likewise.
25732 * tree-cfg.c: Likewise.
25733 * tree-cfg.h: Likewise.
25734 * tree-dfa.c: Likewise.
25735 * tree-dfa.h: Likewise.
25736 * tree-dump.c (dump_function): Likewise.
25737 * tree-dump.h (struct dump_info): Likewise.
25738 * tree-pretty-print.c: Likewise.
25739 * tree-pretty-print.h: Likewise.
25740 * tree-ssa-live.c: Likewise.
25741 * tree-ssa-live.h: Likewise.
25742 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
25743 * tree-vect-loop.c: Likewise.
25744 * tree-vect-slp.c: Likewise.
25746 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
25747 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25749 PR tree-optimization/80457
25750 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
25751 of all arguments to a statement as scalar_to_vec operations.
25752 (vectorizable_call): Adjust call to vect_model_simple_cost for
25754 (vectorizable_conversion): Likewise.
25755 (vectorizable_assignment): Likewise.
25756 (vectorizable_shift): Likewise.
25757 (vectorizable_operation): Likewise.
25758 (vectorizable_comparison): Likewise.
25759 (vect_is_simple_cond): Record the def types for operands.
25760 (vectorizable_condition): Likewise, call vect_model_simple_cost.
25761 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
25762 for statement argument count.
25764 2017-05-16 Carl Love <cel@us.ibm.com>
25766 * config/rs6000/rs6000-c: Add support for built-in functions
25767 vector unsigned long long vec_bperm (vector unsigned long long,
25768 vector unsigned char)
25769 vector signed long long vec_mule (vector signed int,
25771 vector unsigned long long vec_mule (vector unsigned int,
25772 vector unsigned int)
25773 vector signed long long vec_mulo (vector signed int,
25775 vector unsigned long long vec_mulo (vector unsigned int,
25776 vector unsigned int)
25777 vector signed char vec_sldw (vector signed char,
25778 vector signed char,
25780 vector unsigned char vec_sldw (vector unsigned char,
25781 vector unsigned char,
25783 vector signed short vec_sldw (vector signed short,
25784 vector signed short,
25786 vector unsigned short vec_sldw (vector unsigned short,
25787 vector unsigned short,
25789 vector signed int vec_sldw (vector signed int,
25792 vector unsigned int vec_sldw (vector unsigned int,
25793 vector unsigned int,
25795 vector signed long long vec_sldw (vector signed long long,
25796 vector signed long long,
25798 vector unsigned long long vec_sldw (vector unsigned long long,
25799 vector unsigned long long,
25801 * config/rs6000/rs6000-c: Add support for built-in functions
25802 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
25803 * config/rs6000/altivec.h: Add defintion for vec_sldw.
25804 * doc/extend.texi: Update the built-in documentation for the
25805 new built-in functions.
25807 2017-05-16 Marek Polacek <polacek@redhat.com>
25811 * tree.c (save_expr): Don't fold the expression.
25813 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
25815 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
25816 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
25817 and (?*y,m). Update insn attributes.
25819 2017-05-16 Martin Liska <mliska@suse.cz>
25821 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
25822 flags argument of print_gimple_stmt, print_gimple_expr,
25823 print_generic_stmt and print_generic_expr.
25824 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
25825 * coretypes.h: Likewise.
25826 * except.c (dump_eh_tree): Likewise.
25827 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
25828 * gimple-pretty-print.h: Likewise.
25829 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
25830 (backprop::push_to_worklist): Likewise.
25831 (backprop::pop_from_worklist): Likewise.
25832 (backprop::process_use): Likewise.
25833 (backprop::intersect_uses): Likewise.
25834 (note_replacement): Likewise.
25835 * gimple-ssa-store-merging.c
25836 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
25837 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
25838 (pass_store_merging::execute): Likewise.
25839 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
25840 (ssa_base_cand_dump_callback): Likewise.
25841 (dump_incr_vec): Likewise.
25842 (replace_refs): Likewise.
25843 (replace_mult_candidate): Likewise.
25844 (create_add_on_incoming_edge): Likewise.
25845 (create_phi_basis): Likewise.
25846 (insert_initializers): Likewise.
25847 (all_phi_incrs_profitable): Likewise.
25848 (introduce_cast_before_cand): Likewise.
25849 (replace_one_candidate): Likewise.
25850 * gimplify.c (gimplify_expr): Likewise.
25851 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
25852 (set_rename): Likewise.
25853 (rename_uses): Likewise.
25854 (copy_loop_phi_nodes): Likewise.
25855 (add_close_phis_to_merge_points): Likewise.
25856 (copy_loop_close_phi_args): Likewise.
25857 (copy_cond_phi_args): Likewise.
25858 (graphite_copy_stmts_from_block): Likewise.
25859 (translate_pending_phi_nodes): Likewise.
25860 * graphite-poly.c (print_pdr): Likewise.
25861 (dump_gbb_cases): Likewise.
25862 (dump_gbb_conditions): Likewise.
25863 (print_scop_params): Likewise.
25864 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
25865 (build_cross_bb_scalars_use): Likewise.
25866 (gather_bbs::before_dom_children): Likewise.
25867 * hsa-dump.c (dump_hsa_immed): Likewise.
25868 * ipa-cp.c (print_ipcp_constant_value): Likewise.
25869 (get_replacement_map): Likewise.
25870 * ipa-inline-analysis.c (dump_condition): Likewise.
25871 (estimate_function_body_sizes): Likewise.
25872 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
25873 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
25874 * ipa-prop.c (ipa_dump_param): Likewise.
25875 (ipa_print_node_jump_functions_for_edge): Likewise.
25876 (ipa_modify_call_arguments): Likewise.
25877 (ipa_modify_expr): Likewise.
25878 (ipa_dump_param_adjustments): Likewise.
25879 (ipa_dump_agg_replacement_values): Likewise.
25880 (ipcp_modif_dom_walker::before_dom_children): Likewise.
25881 * ipa-pure-const.c (check_stmt): Likewise.
25882 (pass_nothrow::execute): Likewise.
25883 * ipa-split.c (execute_split_functions): Likewise.
25884 * omp-offload.c (dump_oacc_loop_part): Likewise.
25885 (dump_oacc_loop): Likewise.
25886 * trans-mem.c (tm_log_emit): Likewise.
25887 (tm_memopt_accumulate_memops): Likewise.
25888 (dump_tm_memopt_set): Likewise.
25889 (dump_tm_memopt_transform): Likewise.
25890 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25891 (print_loop): Likewise.
25892 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25893 (chkp_gather_checks_info): Likewise.
25894 (chkp_get_check_result): Likewise.
25895 (chkp_remove_check_if_pass): Likewise.
25896 (chkp_use_outer_bounds_if_possible): Likewise.
25897 (chkp_reduce_bounds_lifetime): Likewise.
25898 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25899 (chkp_mark_completed_bounds): Likewise.
25900 (chkp_register_incomplete_bounds): Likewise.
25901 (chkp_mark_invalid_bounds): Likewise.
25902 (chkp_maybe_copy_and_register_bounds): Likewise.
25903 (chkp_build_returned_bound): Likewise.
25904 (chkp_get_bound_for_parm): Likewise.
25905 (chkp_build_bndldx): Likewise.
25906 (chkp_get_bounds_by_definition): Likewise.
25907 (chkp_generate_extern_var_bounds): Likewise.
25908 (chkp_get_bounds_for_decl_addr): Likewise.
25909 * tree-chrec.c (chrec_apply): Likewise.
25910 * tree-data-ref.c (dump_data_reference): Likewise.
25911 (dump_subscript): Likewise.
25912 (dump_data_dependence_relation): Likewise.
25913 (analyze_overlapping_iterations): Likewise.
25914 * tree-inline.c (expand_call_inline): Likewise.
25915 (tree_function_versioning): Likewise.
25916 * tree-into-ssa.c (dump_defs_stack): Likewise.
25917 (dump_currdefs): Likewise.
25918 (dump_names_replaced_by): Likewise.
25919 (dump_update_ssa): Likewise.
25920 (update_ssa): Likewise.
25921 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25922 * tree-parloops.c (build_new_reduction): Likewise.
25923 (try_create_reduction_list): Likewise.
25924 (ref_conflicts_with_region): Likewise.
25925 (oacc_entry_exit_ok_1): Likewise.
25926 (oacc_entry_exit_single_gang): Likewise.
25927 * tree-pretty-print.h: Likewise.
25928 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25929 (get_scalar_evolution): Likewise.
25930 (add_to_evolution): Likewise.
25931 (get_loop_exit_condition): Likewise.
25932 (analyze_evolution_in_loop): Likewise.
25933 (analyze_initial_condition): Likewise.
25934 (analyze_scalar_evolution): Likewise.
25935 (instantiate_scev): Likewise.
25936 (number_of_latch_executions): Likewise.
25937 (gather_chrec_stats): Likewise.
25938 (final_value_replacement_loop): Likewise.
25939 (scev_const_prop): Likewise.
25940 * tree-sra.c (dump_access): Likewise.
25941 (disqualify_candidate): Likewise.
25942 (create_access): Likewise.
25943 (reject): Likewise.
25944 (maybe_add_sra_candidate): Likewise.
25945 (create_access_replacement): Likewise.
25946 (analyze_access_subtree): Likewise.
25947 (analyze_all_variable_accesses): Likewise.
25948 (sra_modify_assign): Likewise.
25949 (initialize_constant_pool_replacements): Likewise.
25950 (find_param_candidates): Likewise.
25951 (decide_one_param_reduction): Likewise.
25952 (replace_removed_params_ssa_names): Likewise.
25953 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25954 * tree-ssa-copy.c (dump_copy_of): Likewise.
25955 (copy_prop_visit_cond_stmt): Likewise.
25956 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25957 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25958 (record_equivalences_from_stmt): Likewise.
25959 * tree-ssa-dse.c (compute_trims): Likewise.
25960 (delete_dead_call): Likewise.
25961 (delete_dead_assignment): Likewise.
25962 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25963 (forward_propagate_into_cond): Likewise.
25964 (pass_forwprop::execute): Likewise.
25965 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25966 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25968 (move_computations_worker): Likewise.
25969 (execute_sm): Likewise.
25970 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25971 (remove_exits_and_undefined_stmts): Likewise.
25972 (remove_redundant_iv_tests): Likewise.
25973 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25974 (adjust_iv_update_pos): Likewise.
25975 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25976 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25977 (value_replacement): Likewise.
25978 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25979 * tree-ssa-pre.c (print_pre_expr): Likewise.
25980 (get_representative_for): Likewise.
25981 (create_expression_by_pieces): Likewise.
25982 (insert_into_preds_of_block): Likewise.
25983 (eliminate_insert): Likewise.
25984 (eliminate_dom_walker::before_dom_children): Likewise.
25985 (eliminate): Likewise.
25986 (remove_dead_inserted_code): Likewise.
25987 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25988 * tree-ssa-reassoc.c (get_rank): Likewise.
25989 (eliminate_duplicate_pair): Likewise.
25990 (eliminate_plus_minus_pair): Likewise.
25991 (eliminate_not_pairs): Likewise.
25992 (undistribute_ops_list): Likewise.
25993 (eliminate_redundant_comparison): Likewise.
25994 (update_range_test): Likewise.
25995 (optimize_range_tests_var_bound): Likewise.
25996 (optimize_vec_cond_expr): Likewise.
25997 (rewrite_expr_tree): Likewise.
25998 (rewrite_expr_tree_parallel): Likewise.
25999 (linearize_expr): Likewise.
26000 (break_up_subtract): Likewise.
26001 (linearize_expr_tree): Likewise.
26002 (attempt_builtin_powi): Likewise.
26003 (attempt_builtin_copysign): Likewise.
26004 (transform_stmt_to_copy): Likewise.
26005 (transform_stmt_to_multiply): Likewise.
26006 (dump_ops_vector): Likewise.
26007 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26008 (print_scc): Likewise.
26009 (set_ssa_val_to): Likewise.
26010 (visit_reference_op_store): Likewise.
26011 (visit_use): Likewise.
26012 (sccvn_dom_walker::before_dom_children): Likewise.
26013 (run_scc_vn): Likewise.
26014 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26016 (expr_hash_elt::print): Likewise.
26017 (const_and_copies::pop_to_marker): Likewise.
26018 (const_and_copies::record_const_or_copy_raw): Likewise.
26019 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26020 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26021 (dump_predicates): Likewise.
26022 (find_uninit_use): Likewise.
26023 (warn_uninitialized_phi): Likewise.
26024 (pass_late_warn_uninitialized::execute): Likewise.
26025 * tree-ssa.c (verify_vssa): Likewise.
26026 (verify_ssa): Likewise.
26027 (maybe_optimize_var): Likewise.
26028 * tree-vrp.c (dump_value_range): Likewise.
26029 (dump_all_value_ranges): Likewise.
26030 (dump_asserts_for): Likewise.
26031 (register_edge_assert_for_2): Likewise.
26032 (vrp_visit_cond_stmt): Likewise.
26033 (vrp_visit_switch_stmt): Likewise.
26034 (vrp_visit_stmt): Likewise.
26035 (vrp_visit_phi_node): Likewise.
26036 (simplify_cond_using_ranges_1): Likewise.
26037 (fold_predicate_in): Likewise.
26038 (evrp_dom_walker::before_dom_children): Likewise.
26039 (evrp_dom_walker::push_value_range): Likewise.
26040 (evrp_dom_walker::pop_value_range): Likewise.
26041 (execute_early_vrp): Likewise.
26043 2017-05-16 Richard Biener <rguenther@suse.de>
26045 * dwarf2out.c (loc_list_from_tree_1): Do not create
26046 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26048 2017-05-16 Richard Biener <rguenther@suse.de>
26050 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26052 (note_variable_value_in_expr): If we resolved the decl ref
26053 do not push to the stack.
26055 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26057 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26058 operations in fast-math mode.
26059 (vaddq_f16): Likewise.
26060 (vmul_f16): Likewise.
26061 (vmulq_f16): Likewise.
26062 (vsub_f16): Likewise.
26063 (vsubq_f16): Likewise.
26064 * config/arm/neon.md (add<mode>3): New.
26066 (fma:<VH:mode>3): New. Also remove outdated comment.
26069 2017-05-16 Martin Liska <mliska@suse.cz>
26073 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26074 (odr_types_equivalent_p): Likewise.
26076 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26078 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26080 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26083 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26084 non-interunit SSE move alternatives with '?'.
26085 (zero-extendsidi peephole2): New peephole to skip intermediate
26086 general register in SSE zero-extend sequence.
26088 2017-05-15 Jeff Law <law@redhat.com>
26090 * reorg.c (relax_delay_slots): Create a new variable to hold
26091 the temporary target rather than clobbering TARGET_LABEL.
26093 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26094 missing argument to extract_bit_field call.
26095 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26097 2017-05-15 Martin Liska <mliska@suse.cz>
26100 * gcc.c (process_command): Do not allow empty argument of -o option.
26102 2017-05-15 Renlin Li <renlin.li@arm.com>
26104 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26105 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26106 * config/aarch64/constraints.md (Usf): Add long call check.
26107 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26108 (call_value): Likewise.
26109 (sibcall): Likewise.
26110 (sibcall_value): Likewise.
26112 (call_value_insn): New.
26113 (sibcall_insn): Update rtx pattern.
26114 (sibcall_value_insn): Likewise.
26115 (call_internal): Remove.
26116 (call_value_internal): Likewise.
26117 (sibcall_internal): Likewise.
26118 (sibcall_value_internal): Likewise.
26119 (call_reg): Likewise.
26120 (call_symbol): Likewise.
26121 (call_value_reg): Likewise.
26122 (call_value_symbol): Likewise.
26124 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26127 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26129 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26131 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26132 compatible with CCGOCmode and with CCZmode.
26134 2017-05-14 Martin Sebor <msebor@redhat.com>
26136 PR middle-end/77671
26137 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26138 (gimple_fold_builtin_snprintf): Same.
26139 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26140 (gimple_fold_builtin_snprintf): Same.
26141 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26142 of character types.
26143 (is_call_safe): New function.
26144 (try_substitute_return_value): Call it.
26145 (try_simplify_call): New function.
26146 (pass_sprintf_length::handle_gimple_call): Call it.
26148 2017-05-14 Martin Sebor <msebor@redhat.com>
26150 PR middle-end/80669
26151 * builtins.c (expand_builtin_stpncpy): Simplify.
26153 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26155 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26156 * config/i386/i386.h
26157 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26158 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26159 (struct machine_function): Add new members call_ms2sysv,
26160 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26161 (struct machine_frame_state): New fields sp_realigned and
26162 sp_realigned_offset.
26163 * config/i386/i386.c
26164 (enum xlogue_stub): New enum.
26165 (enum xlogue_stub_sets): New enum.
26166 (class xlogue_layout): New class.
26167 (struct ix86_frame): New fields stack_realign_allocate_offset,
26168 stack_realign_offset and outlined_save_offset. Modify comments to
26169 detail stack layout when using out-of-line stubs.
26170 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26171 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26172 -mcall-ms2sysv-xlogues.
26173 (stub_managed_regs): New static variable.
26174 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26175 registers managed by out-of-line stub.
26176 (disable_call_ms2sysv_xlogues): New function.
26177 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26178 m->call_ms2sysv when appropriate and compute frame layout for
26180 (sp_valid_at, fp_valid_at): New inline functions.
26181 (choose_basereg): New function.
26182 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26184 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26185 Use align parameter of choose_baseaddr to generated aligned SSE movs
26187 (pro_epilogue_adjust_stack): Modify to track
26188 machine_frame_state::sp_realigned.
26189 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26190 (ix86_nsaved_sseregs): Likewise.
26191 (ix86_emit_save_regs): Likewise.
26192 (ix86_emit_save_regs_using_mov): Likewise.
26193 (ix86_emit_save_sse_regs_using_mov): Likewise.
26194 (get_scratch_register_on_entry): Likewise.
26195 (gen_frame_set): New function.
26196 (gen_frame_load): Likewise.
26197 (gen_frame_store): Likewise.
26198 (emit_outlined_ms2sysv_save): Likewise.
26199 (emit_outlined_ms2sysv_restore): Likewise.
26200 (ix86_expand_prologue): Modify stack re-alignment code and call
26201 emit_outlined_ms2sysv_save when appropriate.
26202 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26203 parameter rtx_insn *insn, which allows the function to be used to only
26204 generate the notes.
26205 (ix86_expand_epilogue): Modify validity checks of frame and stack
26206 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26207 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26208 * config/i386/predicates.md
26209 (save_multiple): New predicate.
26210 (restore_multiple): Likewise.
26211 * config/i386/sse.md
26212 (save_multiple<mode>): New pattern.
26213 (save_multiple_realign<mode>): Likewise.
26214 (restore_multiple<mode>): Likewise.
26215 (restore_multiple_and_return<mode>): Likewise.
26216 (restore_multiple_leave_return<mode>): Likewise.
26217 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26219 2017-05-14 Julia Koval <julia.koval@intel.com>
26221 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26222 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26223 (__builtin_ia32_xsetbv): New builtins.
26224 * config/i386/i386.c (ix86_expand_special_args_builtin):
26226 (ix86_expand_builtin): Special expand for new intrinsics.
26227 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26228 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26229 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26231 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26233 * cfganal.c (inverted_post_order_compute): Change argument type
26235 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26236 * df-core.c (rest_of_handle_df_initialize): Adjust.
26237 (rest_of_handle_df_finish): Likewise.
26238 (df_analyze_1): Likewise.
26239 (df_analyze): Likewise.
26240 (loop_inverted_post_order_compute): Change argument to be a vec *.
26241 (df_analyze_loop): Adjust.
26242 (df_get_n_blocks): Likewise.
26243 (df_get_postorder): Likewise.
26244 * df.h (struct df_d): Change field to be a vec.
26245 * lcm.c (compute_laterin): Adjust.
26246 (compute_available): Likewise.
26247 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26248 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26249 * tree-ssa-pre.c (compute_antic): Likewise.
26251 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26253 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26254 (depth_first_search::depth_first_search): Change structure init
26255 function to this constructor.
26256 (depth_first_search::add_bb): Rename function to this member.
26257 (depth_first_search::execute): Likewise.
26258 (flow_dfs_compute_reverse_finish): Adjust.
26260 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26262 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26263 (longest_simple_path): Likewise.
26264 * shrink-wrap.c (spread_components): Likewise.
26265 (disqualify_problematic_components): Likewise.
26266 (emit_common_heads_for_components): Likewise.
26267 (emit_common_tails_for_components): Likewise.
26268 (insert_prologue_epilogue_for_components): Likewise.
26270 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26272 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26275 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26277 * df-core.c (df_set_blocks): Start using auto_bitmap.
26278 (df_compact_blocks): Likewise.
26279 * df-problems.c (df_rd_confluence_n): Likewise.
26280 * df-scan.c (df_insn_rescan_all): Likewise.
26281 (df_process_deferred_rescans): Likewise.
26282 (df_update_entry_block_defs): Likewise.
26283 (df_update_exit_block_uses): Likewise.
26284 (df_entry_block_bitmap_verify): Likewise.
26285 (df_exit_block_bitmap_verify): Likewise.
26286 (df_scan_verify): Likewise.
26287 * lra-constraints.c (lra_constraints): Likewise.
26288 (undo_optional_reloads): Likewise.
26289 (lra_undo_inheritance): Likewise.
26290 * lra-remat.c (calculate_gen_cands): Likewise.
26291 (do_remat): Likewise.
26292 * lra-spills.c (assign_spill_hard_regs): Likewise.
26293 (spill_pseudos): Likewise.
26294 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26295 (bitmap_set_subtract_values): Likewise.
26297 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26299 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26300 management with auto_bitmap.
26301 (fix_inter_tick): Likewise.
26302 (fix_recovery_deps): Likewise.
26303 * ira.c (add_store_equivs): Likewise.
26304 (find_moveable_pseudos): Likewise.
26305 (split_live_ranges_for_shrink_wrap): Likewise.
26306 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26307 (rtx_reuse_manager::seen_def_p): Likewise.
26308 (rtx_reuse_manager::set_seen_def): Likewise.
26309 * print-rtl.h (class rtx_reuse_manager): Likewise.
26311 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26313 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26315 (migrate_btr_def): Likewise.
26316 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26317 * df-core.c (loop_post_order_compute): Likewise.
26318 (loop_inverted_post_order_compute): Likewise.
26319 * hsa-common.h: Likewise.
26320 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26321 * init-regs.c (initialize_uninitialized_regs): Likewise.
26322 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26323 (inline_small_functions): Likewise.
26324 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26325 * ira.c (combine_and_move_insns): Likewise.
26326 (build_insn_chain): Likewise.
26327 * loop-invariant.c (find_invariants): Likewise.
26328 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26329 * predict.c (tree_predict_by_opcode): Likewise.
26330 (predict_paths_leading_to): Likewise.
26331 (predict_paths_leading_to_edge): Likewise.
26332 (estimate_loops_at_level): Likewise.
26333 (estimate_loops): Likewise.
26334 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26335 (spread_components): Likewise.
26336 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26337 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26338 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26339 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26340 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26341 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26342 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26343 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26344 (mark_threaded_blocks): Likewise.
26345 (thread_through_all_blocks): Likewise.
26346 * tree-ssa.c (verify_ssa): Likewise.
26347 (execute_update_addresses_taken): Likewise.
26348 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26350 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26352 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26354 (post_order_compute): Likewise.
26355 (inverted_post_order_compute): Likewise.
26356 (pre_and_rev_post_order_compute_fn): Likewise.
26358 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26360 * genrecog.c (int_set::int_set): Explicitly construct our
26361 auto_vec base class.
26362 * vec.h (auto_vec::auto_vec): New constructor.
26364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26366 * bitmap.h (class auto_bitmap): New constructor taking
26367 bitmap_obstack * argument.
26369 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26371 * bitmap.h (class auto_bitmap): Change type of m_bits to
26372 bitmap_head, and adjust ctor / dtor and member operators.
26374 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26376 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26377 when returned register mode doesn't match original mode.
26379 2017-05-12 Jeff Law <law@redhat.com>
26380 Jakub Jelinek <jakub@redhat.com>
26382 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26383 we look for cc setter after the compare-elim changes.
26384 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26385 within the vector to match what compare-elim now expects.
26386 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26387 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26389 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26390 after the compare-elim changes.
26391 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26392 the vector to match what compare-elim now expects.
26393 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26394 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26395 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26396 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26397 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26399 * config/visium/visium.c (single_set_and_flags): Fix where
26400 we look for cc setter after the compare-elim changes.
26401 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26402 with the vector to match what compare-elim now expects.
26403 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26404 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26405 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26406 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26407 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26408 (neg<mode>2_insn_set_overflow): Likewise.
26410 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26412 PR middle-end/79794
26413 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26414 maybe_expand_insn call, set ops[0].target. If still set after call,
26415 set alt_rtl. Add extra arg to recursive calls.
26416 (extract_bit_field): Add alt_rtl argument. Pass to
26418 * expmed.h (extract_bit_field): Fix prototype.
26419 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26420 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26421 to extract_bit_field_calls.
26422 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26423 Pass alt_rtl to extract_bit_field calls.
26424 * calls.c (store_unaligned_arguments_into_psuedos)
26425 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26426 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26428 * optabs.h (struct expand_operand): Add target bitfield.
26430 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26432 * compare-elim.c (try_eliminate_compare): Canonicalize
26433 operation with embedded compare to
26434 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26435 (set (reg) (operation)].
26437 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26439 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26442 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26443 cost of adding a carry flag for ADC instruction.
26444 [case MINUS]: Ignore the cost of subtracting a carry flag
26445 for SBB instruction.
26447 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26449 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26451 * config/rs6000/bmiintrin.h: New file.
26452 * config/rs6000/bmi2intrin.h: New file.
26453 * config/rs6000/x86intrin.h: New file.
26455 2017-05-12 Jeff Law <law@redhat.com>
26457 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26460 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26462 PR middle-end/80707
26463 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26465 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26468 early expansion of vector divide builtins.
26469 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26470 builtins identified as having unsigned arguments.
26472 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26474 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26475 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26476 expansion of vector logical operations (and, andc, or, xor,
26479 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26481 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26482 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26484 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26486 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26487 early GIMPLE expansion of vector multiplies.
26489 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26491 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26492 TARGET_HAVE_MOVT conditional.
26493 (movt splitter): Likewise.
26495 2017-05-12 Richard Biener <rguenther@suse.de>
26497 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26498 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26499 Fold all stmts not inplace.
26501 2017-05-12 Richard Biener <rguenther@suse.de>
26503 PR tree-optimization/80713
26504 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26505 inserted_exprs bit for not removed stmts.
26507 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26509 PR middle-end/69921
26510 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26511 parallelized" attribute for parallelized OpenACC kernels.
26512 * omp-offload.c (execute_oacc_device_lower): Use it.
26514 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26515 Set "oacc kernels" attribute.
26516 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26517 parameter. Adjust all users.
26518 (oacc_fn_attrib_kernels_p): Remove function.
26519 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26520 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26521 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26522 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26523 assert "oacc kernels" attribute is set.
26525 2017-05-11 Carl Love <cel@us.ibm.com>
26527 * config/rs6000/rs6000-c: Add support for built-in functions
26528 vector unsigned char vec_popcnt (vector signed char)
26529 vector unsigned char vec_popcnt (vector unsigned char)
26530 vector unsigned short vec_popcnt (vector signed short)
26531 vector unsigned short vec_popcnt (vector unsigned short)
26532 vector unsigned int vec_popcnt (vector signed int)
26533 vector unsigned int vec_popcnt (vector unsigned int)
26534 vector unsigned long long vec_popcnt (vector signed long long)
26535 vector unsigned long long vec_popcnt (vector unsigned long long)
26536 vector signed long long vec_slo (vector signed long long,
26537 vector signed char)
26538 vector signed long long vec_slo (vector signed long long,
26539 vector unsigned char)
26540 vector unsigned long long vec_slo (vector unsigned long long,
26541 vector signed char)
26542 vector unsigned long long vec_slo (vector unsigned long long,
26543 vector unsigned char)
26544 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26545 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26546 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26547 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26548 * doc/extend.texi: Update the built-in documentation file for the
26549 new built-in functions.
26551 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26553 * attribs.h (sorted_attr_string): Move machine independent
26554 functions for target clone support from the i386 port to common
26555 code. Rename ix86_function_versions to common_function_versions.
26556 Rename make_name to make_unique_name.
26557 (common_function_versions): Likewise.
26558 (make_unique_name): Likewise.
26559 (make_dispatcher_decl): Likewise.
26560 (is_function_default_version): Likewise.
26561 * attribs.c (attr_strcmp): Likewise.
26562 (sorted_attr_string): Likewise.
26563 (common_function_versions): Likewise.
26564 (make_unique_name): Likewise.
26565 (make_dispatcher_decl): Likewise.
26566 (is_function_default_version): Likewise.
26567 * config/i386/i386.c (attr_strcmp): Likewise.
26568 (sorted_attr_string): Likewise.
26569 (ix86_function_versions): Likewise.
26570 (make_name): Likewise.
26571 (make_dispatcher_decl): Likewise.
26572 (is_function_default_version): Likewise.
26573 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26575 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26578 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26579 Account for direct move costs for vec_construct of integer
26582 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26585 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26586 (UNSPEC_STX_ATOMIC): Ditto.
26587 (loaddi_via_sse): New insn.
26588 (storedi_via_sse): Ditto.
26589 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26590 Update corresponding peephole2 patterns.
26591 (atomic_storedi_fpu): Ditto.
26593 2017-05-11 Julia Koval <julia.koval@intel.com>
26595 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26596 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26598 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26599 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26600 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26602 2017-05-11 Nathan Sidwell <nathan@acm.org>
26604 * graphite-poly.c: Include dumpfile.h.
26606 * dumpfle.h (dump_function): Declare here ...
26607 * tree-dump.h (dump_function): ... not here.
26608 * dumpfile.c: #include tree-cfg.h.
26609 (dump_function): Move here from ...
26610 * tree-dump.c (dump_function): ... here.
26611 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26612 * graphite-poly.c: Don't include tree-dump.h.
26613 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26614 * print-tree.c: Likewise.
26615 * stor-layout.c: Likewise.
26616 * tree-nested.c: Likewise.
26618 * dumpfile.c (dump_start): Use TDF_FLAGS.
26619 (dump_enable_all): Fix TDF_KIND check thinko.
26621 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26623 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26624 array entries to represent two legal parameterizations of the
26625 overloaded __builtin_cmpb function, as represented by the
26626 P6_OV_BUILTIN_CMPB constant.
26627 (altivec_resolve_overloaded_builtin): Add special case handling
26628 for the __builtin_cmpb function, as represented by the
26629 P6_OV_BUILTIN_CMPB constant.
26630 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26631 (BU_P6_64BIT_2): New macro.
26632 (BU_P6_OVERLOAD_2): New macro
26633 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26634 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26635 (CMPB): Add overload support to represent both 32-bit and 64-bit
26636 compare-bytes function.
26637 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26638 support for TARGET_CMPB.
26639 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26640 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26641 documentation of the __builtin_cmpb overloaded built-in function.
26643 2017-05-11 Richard Biener <rguenther@suse.de>
26645 PR tree-optimization/80705
26646 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26647 bases are not vectorizable.
26649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26651 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26652 when counting register pressure.
26654 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26656 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26657 (iv_ca_more_deps): Renamed to ...
26658 (iv_ca_compare_deps): ... this.
26659 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26661 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26663 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26665 (determine_group_iv_costs): ... here.
26666 (find_inv_vars_cb): Record inv var if it's not recorded before.
26668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26670 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26671 (get_shiftadd_cost): Ditto.
26673 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26675 * tree-ssa-address.c: Include header file.
26676 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26678 (add_to_parts): Refactor.
26679 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26680 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26683 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26685 PR tree-optimization/53090
26686 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26688 (extract_cond_operands): Detect condition with IV on both sides
26689 and return COMP_IV_EXPR_2.
26690 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26691 (rewrite_use_compare): Simplify by removing call to function
26692 extract_cond_operands.
26694 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26696 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
26697 (extract_cond_operands): Detect condition comparing against non-
26698 invariant bound and return appropriate enum value.
26699 (find_interesting_uses_cond): Update use of extract_cond_operands.
26700 Handle its return value accordingly.
26701 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
26703 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26705 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
26706 nonlinear iv_use computation in loop invariant sensitive way.
26708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26710 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
26711 (find_iv_candidates): Call relate_compare_use_with_all_cands.
26713 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26715 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
26716 (dump_cand): Support iv_cand.inv_exprs.
26717 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
26719 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
26722 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26724 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
26726 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
26727 as local function. Include necessary header files.
26728 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
26730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26732 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
26734 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26736 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
26737 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
26738 RSHIFT_EXPR and BIT_NOT_EXPR.
26740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26742 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
26743 (adjust_setup_cost): New parameter supporting round up adjustment.
26744 (struct address_cost_data): Delete.
26745 (force_expr_to_var_cost): Don't bound cost with spill_cost.
26746 (split_address_cost, ptr_difference_cost): Delete.
26747 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
26748 (struct ainc_cost_data): New struct.
26749 (get_address_cost_ainc): New function.
26750 (get_address_cost, get_computation_cost): Reimplement.
26751 (determine_group_iv_cost_address): Record inv_expr for all uses of
26753 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
26754 (iv_ca_has_deps): Reimplemented to ...
26755 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
26757 (iv_ca_extend): Call iv_ca_more_deps.
26759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26761 * tree-ssa-address.c (struct mem_address): Move to header file.
26762 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
26763 * tree-ssa-address.h (struct mem_address): Move from C file.
26764 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
26766 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26768 * tree-affine.h (aff_combination_type): New interface.
26769 (aff_combination_zero_p): Remove static.
26770 (aff_combination_const_p): New interface.
26771 (aff_combination_singleton_var_p): New interfaces.
26773 2017-05-11 Richard Biener <rguenther@suse.de>
26775 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26776 Skip unreachable blocks and destinations.
26777 (eliminate): Move stmt removal and fixup ...
26778 (fini_eliminate): ... here. Skip inserted exprs.
26779 (pass_pre::execute): Move fini_pre after fini_eliminate.
26780 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
26781 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
26782 PRE to get rid of dead code that has invalid SSA form and
26783 split critical edges again.
26785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26787 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26789 2017-05-11 Richard Biener <rguenther@suse.de>
26791 * passes.c (execute_function_todo): Verify loops if they are
26792 said to be up-to-date.
26793 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
26794 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
26796 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
26799 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
26800 handle calling assemble_external ourself.
26803 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
26804 modes with zero size. Enhance comment.
26806 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26808 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
26809 built-ins for vec_xl and vec_xst with short and char pointer
26812 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26814 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
26815 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
26816 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
26817 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
26818 (_mm_maskz_min_round_ss): New intrinsics.
26819 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
26820 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
26821 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
26822 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
26823 (__builtin_ia32_minss_mask_round): New builtins.
26824 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26825 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26826 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
26828 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
26829 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
26831 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
26834 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
26836 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
26837 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
26838 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
26839 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
26840 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
26841 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
26842 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
26843 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26844 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26845 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
26846 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
26847 (__builtin_ia32_mulss_mask_round): New builtins.
26848 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26849 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26850 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
26852 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
26853 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26855 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26858 2017-05-10 Julia Koval <julia.koval@intel.com>
26860 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
26861 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
26862 (_mm256_setr_m128i): New intrinsics.
26864 2017-05-10 Julia Koval <julia.koval@intel.com>
26866 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
26867 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
26868 (_mm_maskz_rcp14_ss): New intrinsics.
26869 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
26870 (__builtin_ia32_rcp14ss_mask): New builtins.
26871 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
26873 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
26875 PR tree-optimization/51513
26876 * tree-cfg.c (gimple_seq_unreachable_p): New function.
26877 (assert_unreachable_fallthru_edge_p): Use it.
26878 (group_case_labels_stmt): Likewise.
26879 * tree-cfg.h: Prototype it.
26880 * stmt.c: Include cfghooks.h and tree-cfg.h.
26881 (emit_case_dispatch_table) <gap_label>: New local variable.
26882 Use it to fill dispatch table gaps.
26883 Test for default_label before updating probabilities.
26884 (expand_case) <default_label>: Remove unneeded initialization.
26885 Test for unreachable default case statement and remove its edge.
26886 Set default_label accordingly.
26887 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26889 2017-05-10 Carl Love <cel@us.ibm.com>
26891 * config/rs6000/rs6000-c: Add support for built-in functions
26892 vector signed char vec_neg (vector signed char)
26893 vector signed short int vec_neg (vector short int)
26894 vector signed int vec_neg (vector signed int)
26895 vector signed long long vec_neg (vector signed long long)
26896 vector float vec_neg (vector float)
26897 vector double vec_neg (vector double)
26898 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26900 * config/rs6000/altivec.h: Add define for vec_neg
26901 * doc/extend.texi: Update the built-in documentation for the
26902 new built-in functions.
26904 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26906 PR tree-optimization/77644
26907 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26909 2017-05-10 Nathan Sidwell <nathan@acm.org>
26911 * dumpfile.h (TDI_lang_all): New.
26912 (TDF_KIND): New. Renumber others
26913 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26915 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26917 (get_dump_file_name): Adjust suffix generation.
26918 (dump_enable_all): Use TDF_KIND.
26919 * doc/invoke.texi (-fdump-lang-all): Document.
26921 * dumpfile.h: Tabify.
26923 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26926 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26927 Move member access before delete.
26929 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26931 * tree-inline.c (expand_call_inline): Split block at stmt
26934 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26937 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26938 are now unused after splitting mov{sf,sd}_hardfloat.
26939 (f32_lr2): Likewise.
26940 (f32_lm): Likewise.
26941 (f32_lm2): Likewise.
26942 (f32_li): Likewise.
26943 (f32_li2): Likewise.
26944 (f32_lv): Likewise.
26945 (f32_sr): Likewise.
26946 (f32_sr2): Likewise.
26947 (f32_sm): Likewise.
26948 (f32_sm2): Likewise.
26949 (f32_si): Likewise.
26950 (f32_si2): Likewise.
26951 (f32_sv): Likewise.
26952 (f32_dm): Likewise.
26953 (f32_vsx): Likewise.
26954 (f32_av): Likewise.
26955 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26956 For movsf, order stores so the VSX stores occur before the GPR
26957 store which encourages the register allocator to use a traditional
26958 FPR instead of a GPR. For movsd, order the stores so that the GPR
26959 store comes before the VSX stores to allow the power6 to work.
26960 This is due to the power6 not having a 32-bit integer store
26961 instruction from a FPR.
26962 (movsf_hardfloat): Likewise.
26963 (movsd_hardfloat): Likewise.
26965 2017-05-09 Martin Sebor <msebor@redhat.com>
26967 PR translation/80280
26968 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26971 PR translation/80280
26972 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26973 data member added in r247778.
26974 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26976 2017-05-09 Nathan Sidwell <nathan@acm.org>
26978 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26980 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26983 2017-05-09 Marek Polacek <polacek@redhat.com>
26985 * doc/invoke.texi: Fix typo.
26987 2017-05-09 Richard Biener <rguenther@suse.de>
26989 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26990 (vrp_val_is_min): Likewise.
26991 (set_value_range_to_value): Likewise.
26992 (set_value_range_to_nonnegative): Likewise.
26993 (gimple_assign_nonzero_p): Likewise.
26994 (gimple_stmt_nonzero_p): Likewise.
26995 (vrp_int_const_binop): Likewise. Remove unreachable case.
26996 (adjust_range_with_scev): Adjust comments.
26997 (compare_range_with_value): Likewise.
26998 (extract_range_from_phi_node): Likewise.
26999 (test_for_singularity): Likewise.
27001 2017-05-09 Richard Biener <rguenther@suse.de>
27003 * tree-vrp.c (get_single_symbol): Add assert that we don't
27004 get overflowed constants as invariant part.
27005 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27006 checks. Use wi::cmp instead of recursing for integer constants.
27007 (compare_values): Just ignore whether we assumed undefined
27008 overflow instead of failing the compare.
27009 (extract_range_for_var_from_comparison_expr): Add comment before the
27010 TREE_NO_WARNING sets.
27011 (test_for_singularity): Likewise.
27012 (extract_range_from_comparison): Do not disable optimization
27013 when we assumed undefined overflow.
27014 (extract_range_basic): Remove init of unused var.
27016 2017-05-09 Richard Biener <rguenther@suse.de>
27018 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27019 (extract_range_from_multiplicative_op_1): Adjust.
27020 (extract_range_from_binary_expr_1): Use int_const_binop.
27022 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27025 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27026 rs6000_store_data_bypass_p in seven define_bypass directives and
27027 in several comments.
27028 * config/rs6000/rs6000-protos.h: Add prototype for
27029 rs6000_store_data_bypass_p function.
27030 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27031 function implements slightly different (rs6000-specific) semantics
27032 than store_data_bypass_p, returning false rather than aborting
27033 with assertion error when arguments do not satisfy the
27034 requirements of store data bypass.
27035 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27036 rs6000_store_data_bypass_p.
27038 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27040 * config/xtensa/xtensa-protos.h
27041 (xtensa_initial_elimination_offset): New declaration.
27042 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27043 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27044 macro definition, add case for FRAME_POINTER_REGNUM when
27045 FRAME_GROWS_DOWNWARD.
27046 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27047 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27048 xtensa_initial_elimination_offset.
27050 2017-05-08 Nathan Sidwell <nathan@acm.org>
27052 * doc/invoke.texi: Alphabetize -fdump options.
27054 2017-05-08 Martin Sebor <msebor@redhat.com>
27056 PR translation/80280
27057 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27059 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27061 * target.def (compute_frame_layout): New optional target hook.
27062 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27063 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27064 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27066 * reload1.c (verify_initial_elim_offsets): Likewise.
27067 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27068 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27069 (arm_get_frame_offsets): Split up into this ...
27070 (arm_compute_frame_layout): ... and this function.
27072 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27074 * config/aarch64/constraints.md (Usa): New constraint.
27075 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27077 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27079 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27080 with_multilib_list after it has been checked.
27082 2017-05-08 Richard Biener <rguenther@suse.de>
27084 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27085 (bitmap_set_subtract_values): Likewise.
27087 2017-05-08 Richard Biener <rguenther@suse.de>
27089 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27090 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27092 (gimple_stmt_nonzero_warnv_p): Rename to ...
27093 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27095 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27096 (extract_range_basic): Adjust, do not disable propagation on
27097 strict overflow sensitive simplification.
27098 (vrp_visit_cond_stmt): Likewise.
27100 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27102 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27103 body size unconditionally.
27105 2017-05-07 Jeff Law <law@redhat.com>
27108 2017-05-06 Jeff Law <law@redhat.com>
27109 PR tree-optimization/78496
27110 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27113 PR tree-optimization/78496
27114 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27115 (simplify_stmt_using_ranges): Call it.
27116 (vrp_dom_walker::before_dom_children): Extract equivalences
27117 from an ASSERT_EXPR with an equality comparison against a
27120 2017-05-06 Jeff Law <law@redhat.com>
27122 PR tree-optimization/78496
27123 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27126 PR tree-optimization/78496
27127 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27128 (simplify_stmt_using_ranges): Call it.
27129 (vrp_dom_walker::before_dom_children): Extract equivalences
27130 from an ASSERT_EXPR with an equality comparison against a
27133 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27135 * lra-constraints.c (lra_copy_reg_equiv): New function.
27136 (split_reg): Use it to copy equivalence information from the
27137 original register to the spill register.
27139 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27141 PR rtl-optimization/75964
27142 * simplify-rtx.c (simplify_const_relational_operation): Remove
27143 invalid handling of comparisons of integer ABS.
27145 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27147 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27148 initialize to zero.
27149 (init_regs): Remove declaration.
27150 (function_arg_advance_32): Initialize error_p as boolean variable.
27152 2017-05-05 Nathan Sidwell <nathan@acm.org>
27154 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27155 lines. Use for (;;).
27157 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27159 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27160 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27161 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27162 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27163 VF=2 that require versioning.
27165 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27167 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27170 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27172 * diagnostic.h (diagnostic_override_option_index): Convert from
27173 macro to inline function.
27175 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27177 * diagnostic.c (last_module_changed_p): New function.
27178 (set_last_module): New function.
27179 (diagnostic_report_current_module): Convert macro usage to
27180 the above functions.
27181 * diagnostic.h (diagnostic_context::last_module): Strengthen
27182 from const line_map * to const line_map_ordinary *.
27183 (diagnostic_last_module_changed): Delete macro.
27184 (diagnostic_set_last_module): Delete macro.
27186 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27188 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27189 with diagnostic_report_diagnostic.
27190 (diagnostic_n_impl_richloc): Likewise.
27191 * diagnostic.h (report_diagnostic): Delete macro.
27192 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27193 with diagnostic_report_diagnostic.
27194 * substring-locations.c (format_warning_va): Likewise.
27196 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27198 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27199 save/restor of format_spec. Move option-printing code to...
27200 (print_option_information): ...this new function, and
27201 reimplement by simply printing to the pretty_printer,
27202 rather than appending to the format string.
27204 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27206 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27207 handling logic into...
27208 (update_effective_level_from_pragmas): ...this new function.
27210 2017-05-04 Andrew Waterman <andrew@sifive.com>
27212 * config/riscv/riscv.opt (mstrict-align): New option.
27213 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27214 (SLOW_UNALIGNED_ACCESS): Define.
27215 (riscv_slow_unaligned_access): Declare.
27216 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27218 (riscv_slow_unaligned_access): New variable.
27219 (rocket_tune_info): Set slow_unaligned_access to true.
27220 (optimize_size_tune_info): Set slow_unaligned_access to false.
27221 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27222 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27223 (riscv_option_override): Set riscv_slow_unaligned_access.
27224 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27226 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27228 * config/riscv/riscv.md: Unify indentation.
27230 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27235 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27237 (extendsi<mode>2): Add support for doing sign extension via
27238 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27239 don't have ISA 3.0 instructions.
27240 (extendsi<mode>2 splitter): Likewise.
27241 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27242 generate the normal insns since SImode can now go in vector
27243 registers. Disallow the special UNSPECs needed for previous
27244 machines to hide SImode being used. Add new insns
27245 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27246 (fix_trunc<mode>si2_stfiwx): Likewise.
27247 (fix_trunc<mode>si2_internal): Likewise.
27248 (fixuns_trunc<mode>si2): Likewise.
27249 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27250 (fctiw<u>z_<mode>_smallint): Likewise.
27251 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27252 of floating point to 32-bit integer from doing a direct move to
27253 the GPR registers to do a store.
27254 (fctiwz_<mode>): Break long line.
27256 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27258 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27259 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27260 (addr_list, addr_offset_valid_p): New.
27261 (split_address_groups): Check offset validity with above function.
27262 (gt-tree-ssa-loop-ivopts.h): Include header file.
27264 2017-05-05 Nathan Sidwell <nathan@acm.org>
27266 * config.gcc (arm*-*-*): Add missing 'fi'.
27268 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27270 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27271 included in -fopt-info does not matter.
27272 * doc/optinfo.texi (-fopt-info): Fix description of default
27273 behavour. Explicitly say order of options included in -fopt-info
27276 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27278 * config.gcc: Allow combinations of aprofile and rmprofile values for
27279 --with-multilib-list.
27280 * config/arm/t-multilib: New file.
27281 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27282 variables. Remove setting of ISA and floating-point ABI in
27283 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27284 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27285 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27286 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27288 * config/arm/t-rmprofile: Likewise except for the matches changes.
27289 * doc/install.texi (--with-multilib-list): Document the combination of
27290 aprofile and rmprofile values and warn about pitfalls in doing that.
27292 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27294 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27295 (movdi_aarch64): Likewise.
27297 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27299 PR tree-optimization/80632
27300 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27302 (build_arrays): Initialize it for virtual phis.
27303 (fix_phi_nodes): Use it for virtual phis.
27305 PR tree-optimization/80558
27306 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27307 [x, y] op z into [x op, y op z] for op & or | if conditions
27310 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27311 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27314 * config/arm/arm.md (use_literal_pool): Remove.
27315 (64-bit immediate split): No longer takes cost into consideration
27316 if arm_disable_literal_pool is enabled.
27317 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27318 used when arm_disable_literal_pool is enabled.
27319 (arm_max_const_double_inline_cost): Remove use of
27320 arm_disable_literal_pool.
27321 (push_minipool_fix): Add assert.
27322 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27323 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27324 (no_literal_pool_sf_immediate): New.
27326 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27328 PR tree-optimization/80613
27329 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27330 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27332 2017-05-05 Richard Biener <rguenther@suse.de>
27334 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27336 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27338 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27339 of this flag from insn conditions due to removal from r247495.
27341 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27343 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27345 (arm_early_store_addr_dep_ptr): Likewise.
27346 * config/arm/aarch-common-protos.h
27347 (arm_early_load_addr_dep_ptr): Add prototype.
27348 (arm_early_store_addr_dep_ptr): Likewise.
27349 * config/arm/cortex-a53.md: Add new bypasses.
27351 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27353 * tree.c (next_type_uid): Change type to unsigned.
27354 (type_hash_canon): Decrement back next_type_uid if
27355 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27356 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27359 2017-05-04 Martin Sebor <msebor@redhat.com>
27361 * builtins.c: Fix a trivial typo in a comment.
27363 PR middle-end/79234
27364 * builtins.c (check_sizes): Adjust to handle reading past the end.
27365 Avoid printing excessive upper bound of ranges. Use %E to print
27366 tree nodes instead of converting them to %wu.
27367 (expand_builtin_memchr): New function.
27368 (compute_dest_size): Rename...
27369 (compute_objsize): ...to this.
27370 (expand_builtin_memcpy): Adjust.
27371 (expand_builtin_mempcpy): Adjust.
27372 (expand_builtin_strcat): Adjust.
27373 (expand_builtin_strcpy): Adjust.
27374 (check_strncat_sizes): Adjust.
27375 (expand_builtin_strncat): Adjust.
27376 (expand_builtin_strncpy): Adjust and simplify.
27377 (expand_builtin_memset): Adjust.
27378 (expand_builtin_bzero): Adjust.
27379 (expand_builtin_memcmp): Adjust.
27380 (expand_builtin): Handle memcmp.
27381 (maybe_emit_chk_warning): Check strncat just once.
27383 2017-05-04 Martin Sebor <msebor@redhat.com>
27385 PR preprocessor/79214
27386 PR middle-end/79222
27387 PR middle-end/79223
27388 * builtins.c (check_sizes): Add inlining context and issue
27389 warnings even when -Wno-system-headers is set.
27390 (check_strncat_sizes): Same.
27391 (expand_builtin_strncat): Same.
27392 (expand_builtin_memmove): New function.
27393 (expand_builtin_stpncpy): Same.
27394 (expand_builtin): Handle memmove and stpncpy.
27396 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27398 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27399 which is not used any more.
27401 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27403 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27405 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27407 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27408 (cortexa53_tunings): Likewise.
27409 (cortexa57_tunings): Likewise.
27410 (cortexa72_tunings): Likewise.
27411 (cortexa73_tunings): Likewise.
27413 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27415 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27416 Set loop alignment to 8.
27418 2017-05-04 Martin Sebor <msebor@redhat.com>
27420 PR translation/80280
27421 * builtins.c (expand_builtin_object_size): Add missing quoting to
27422 %D and like directives.
27423 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27424 (hsa_type_for_tree_type): Same.
27425 (verify_function_arguments): Same.
27426 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27427 * varasm.c (get_section): Same.
27430 2017-05-04 Martin Sebor <msebor@redhat.com>
27432 PR translation/80280
27433 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27435 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27437 * config/aarch64/aarch64.c (generic_addrcost_table):
27438 Change HI/TI mode setting.
27440 2017-05-04 Martin Jambor <mjambor@suse.cz>
27442 PR tree-optimization/80622
27443 * tree-sra.c (comes_initialized_p): New function.
27444 (build_accesses_from_assign): Only set write lazily when
27445 comes_initialized_p is false.
27446 (analyze_access_subtree): Use comes_initialized_p.
27447 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27448 instead of testing for PARM_DECL.
27450 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27452 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27453 constraint on operand 0 to allow more general addressing modes.
27454 Adjust output template.
27455 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27457 * config/aarch64/aarch64-protos.h
27458 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27459 * config/aarch64/constraints.md (Dp): New address constraint.
27460 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27463 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27465 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27466 update use of estimate_ipcp_clone_size_and_time.
27467 (estimate_local_effects): Update use of
27468 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27469 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27470 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27471 Return nonspecialized time.
27473 2017-05-04 Richard Biener <rguenther@suse.de>
27475 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27476 for the last VUSE which def dominates the PHI. Directly call
27478 (get_continuation_for_phi_1): Remove.
27480 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27482 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27483 to explain the use of truncating division. Cap the number of
27484 iterations to the maximum given by nb_iterations_upper_bound,
27487 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27489 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27490 * configure: Regenerate.
27491 * config.in: Regenerate.
27492 * config/i386/driver-mingw32.c: new file.
27493 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27494 * config.host: Link driver-mingw32.o on MinGW host.
27495 * doc/install.texi: Document new --enable-mingw-wildcard configure
27498 2017-05-04 Marek Polacek <polacek@redhat.com>
27500 PR tree-optimization/80612
27501 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27503 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27504 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27506 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27507 (movt splitter): Likewise.
27508 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27509 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27510 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27511 block for Thumb-1 with MOVT.
27512 (thumb2_legitimate_address_p): Move code block ...
27513 (can_avoid_literal_pool_for_label_p): ... into this new function.
27514 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27516 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27517 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27518 "M-profile targets with the MOVT instruction".
27520 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27522 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27523 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27524 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27526 2017-05-04 Martin Liska <mliska@suse.cz>
27528 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27529 variable cond_code.
27531 2017-05-04 Richard Biener <rguenther@suse.de>
27533 * tree.c (array_at_struct_end_p): Handle arrays at struct
27534 end with flexarrays more conservatively. Refactor and treat
27535 arrays of arrays or aggregates more strict. Fix
27536 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27537 * tree.c (array_at_struct_end_p): Adjust prototype.
27538 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27539 * gimple-fold.c (get_range_strlen): Likewise.
27540 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27542 2017-05-04 Richard Biener <rguenther@suse.de>
27544 PR tree-optimization/31130
27545 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27547 (supports_overflow_infinity): Likewise.
27548 (is_negative_overflow_infinity): Likewise.
27549 (is_positive_overflow_infinity): Likewise.
27550 (is_overflow_infinity): Likewise.
27551 (stmt_overflow_infinity): Likewise.
27552 (overflow_infinity_range_p): Likewise.
27553 (usable_range_p): Remove as always returning true.
27554 (make_overflow_infinity): Remove.
27555 (negative_overflow_infinity): Likewise.
27556 (positive_overflow_infinity): Likewise.
27557 (avoid_overflow_infinity): Likewise.
27558 (set_value_range): Adjust accordingly.
27559 (set_value_range_to_nonnegative): Likewise, remove now unused
27560 overflow_infinity arg.
27561 (vrp_operand_equal_p): Adjust.
27562 (update_value_range): Likewise.
27563 (range_int_cst_singleton_p): Likewise.
27564 (operand_less_p): Likewise.
27565 (compare_values_warnv): Likewise.
27566 (extract_range_for_var_from_comparison_expr): Likewise.
27567 (vrp_int_const_binop): Likewise.
27568 (zero_nonzero_bits_from_vr): Likewise.
27569 (extract_range_from_multiplicative_op_1): Likewise.
27570 (extract_range_from_binary_expr_1): Likewise.
27571 (extract_range_from_unary_expr): Likewise.
27572 (extract_range_from_comparison): Likewise.
27573 (extract_range_basic): Likewise.
27574 (adjust_range_with_scev): Likewise.
27575 (compare_ranges): Likewise.
27576 (compare_range_with_value): Likewise.
27577 (dump_value_range): Likewise.
27578 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27580 (simplify_cond_using_ranges): Adjust.
27582 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27584 * brig-builtins.def: Added a builtin for class_f64.
27585 * builtin-types.def: Added a builtin type needed by class_f64.
27587 2017-05-03 Jason Merrill <jason@redhat.com>
27589 * timevar.def: Add TV_CONSTEXPR.
27591 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27593 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27595 2017-05-03 Martin Jambor <mjambor@suse.cz>
27597 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27598 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27599 * ipa-cp.c (ipcp_propagate_stage): Do not call
27600 ipa_update_after_lto_read.
27601 * ipa-inline.c (ipa_inline): Likewise.
27603 2017-05-03 Martin Jambor <mjambor@suse.cz>
27605 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27606 tag. Added a default constructor and a destructor.
27607 (ipa_edge_args_sum_t): New class;
27608 (ipa_edge_args_sum): Declare.
27609 (ipa_edge_args_vector): Remove declaration.
27610 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27611 (ipa_free_edge_args_substructures): Remove declaration.
27612 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27613 (ipa_edge_args_info_available_for_edge_p): Likewise.
27614 * ipa-prop.c (ipa_edge_args_vector): Removed.
27615 (edge_removal_hook_holder): Likewise.
27616 (edge_duplication_hook_holder): Likewise.
27617 (ipa_edge_args_sum): New variable.
27618 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27619 ipa_edge_args_vector.
27620 (ipa_free_edge_args_substructures): Likewise.
27621 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27622 ipa_edge_args_vector.
27623 (ipa_edge_removal_hook): Turned into method
27624 ipa_edge_args_sum_t::remove.
27625 (ipa_edge_duplication_hook): Turned into method
27626 ipa_edge_args_sum_t::duplicate.
27627 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27628 registering edge hooks.
27629 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27630 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27631 ipa_edge_args_sum instead of ipa_edge_args_vector.
27632 * ipa-profile.c (ipa_profile): Likewise.
27634 2017-05-03 Martin Jambor <mjambor@suse.cz>
27636 * symbol-summary.h (function_summary): New method exists.
27637 (function_summary::symtab_removal): Deallocate through release.
27638 (call_summary): New class.
27639 (gt_ggc_mx): New overload.
27640 (gt_pch_nx): Likewise.
27641 (gt_pch_nx): Likewise.
27643 2017-05-03 Jeff Law <law@redhat.com>
27645 PR tree-optimization/78496
27646 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27647 from simplify_cond_using_ranges. Split off code to walk
27648 backwards through casts into ...
27649 (simplify_cond_using_ranges_2): New function.
27650 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27651 (execute_vrp): After identifying jump threads, call
27652 simplify_cond_using_ranges_2.
27654 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27657 * ipa-inline.h (inline_summary): Add ctor.
27658 (create_ggc): Do not use ggc_cleared_alloc.
27660 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27661 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27663 * gcc.c (handle_braces): Support escaping in switch matching
27665 * doc/invoke.texi (Spec Files): Document it.
27666 Remove superfluous @code markup in items.
27668 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27670 * diagnostic-show-locus.c (struct column_range): New struct.
27671 (get_affected_columns): New function.
27672 (get_printed_columns): New function.
27673 (struct correction): New struct.
27674 (correction::ensure_capacity): New function.
27675 (correction::ensure_terminated): New function.
27676 (struct line_corrections): New struct.
27677 (line_corrections::~line_corrections): New dtor.
27678 (line_corrections::add_hint): New function.
27679 (layout::print_trailing_fixits): Reimplement in terms of the new
27681 (selftest::test_overlapped_fixit_printing): New function.
27682 (selftest::diagnostic_show_locus_c_tests): Call it.
27684 2017-05-03 Nathan Sidwell <nathan@acm.org>
27686 Canonicalize canonical type hashing
27687 * tree.h (type_hash_canon_hash): Declare.
27688 * tree.c (type_hash_list, attribute_hash_list): Move into
27689 type_hash_canon_hash.
27690 (build_type_attribute_qual_variant): Break out hash code calc into
27691 type_hash_canon_hash.
27692 (type_hash_canon_hash): New. Generic type hash computation.
27693 (build_range_type_1, build_array_type_1, build_function_type,
27694 build_method_type_directly, build_offset_type, build_complex_type,
27695 make_vector_type): Call it.
27697 2017-05-03 Richard Biener <rguenther@suse.de>
27699 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
27700 When all DRs have unknown misaligned do not always peel
27701 when there is a store but apply the same costing model as if
27702 there were only loads.
27704 2017-05-03 Richard Biener <rguenther@suse.de>
27707 PR tree-optimization/80492
27708 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27709 compare_base_decls returning dont-know properly.
27711 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
27713 * config/arm/iterators.md (CCSI): New mode iterator.
27714 (arch): New mode attribute.
27715 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
27716 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
27717 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
27718 code iterator for success result mode.
27719 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
27720 the corresponding new insn generators.
27722 2017-05-03 Bin Cheng <bin.cheng@arm.com>
27725 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27726 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27728 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
27730 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
27731 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
27732 (DDR_A): Wrap DDR argument in brackets.
27733 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
27734 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
27735 (DDR_REVERSED_P): Likewise.
27737 2017-05-03 Jakub Jelinek <jakub@redhat.com>
27739 PR tree-optimization/79472
27740 * tree-switch-conversion.c (struct switch_conv_info): Add
27741 contiguous_range and default_case_nonstandard fields.
27742 (collect_switch_conv_info): Compute contiguous_range and
27743 default_case_nonstandard fields, don't clear final_bb if
27744 contiguous_range and only the default case doesn't have the required
27746 (check_all_empty_except_final): Set default_case_nonstandard instead
27747 of failing if contiguous_range and the default case doesn't have empty
27749 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
27750 and only the default case doesn't have the required constants. Skip
27752 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
27753 if default_case_nonstandard.
27754 (build_constructors): Build constant 1 just once. Assert that default
27755 values aren't inserted in between cases if contiguous_range. Skip
27757 (build_arrays): Skip virtual phis.
27758 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
27759 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
27760 Handle virtual phis.
27761 (gen_inbound_check): Handle default_case_nonstandard case.
27762 (process_switch): Adjust check_final_bb caller. Call
27763 gather_default_values with the first non-default case instead of
27764 default case if default_case_nonstandard.
27766 2017-05-02 Nathan Sidwell <nathan@acm.org>
27768 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
27769 check. Fix formatting.
27771 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
27773 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
27774 errors when comparing specialized and unspecialized times.
27776 2017-05-02 David Malcolm <dmalcolm@redhat.com>
27778 * diagnostic-show-locus.c
27779 (layout::should_print_annotation_line_p): Make private.
27780 (layout::print_annotation_line): Make private.
27781 (layout::annotation_line_showed_range_p): Make private.
27782 (layout::show_ruler): Make private.
27783 (layout::print_source_line): Make private. Pass in line and
27784 line_width, rather than calling location_get_source_line. Drop
27786 (layout::print_leading_fixits): New method.
27787 (layout::print_any_fixits): Rename to...
27788 (layout::print_trailing_fixits): ...this, and make private.
27789 Don't print newline fixits.
27790 (diagnostic_show_locus): Move logic for printing one row into...
27791 (layout::print_line): ...this new function. Move the
27792 location_get_source_line call and error-handling from
27793 print_source_line to here. Call print_leading_fixits, and rename
27794 print_any_fixits to print_trailing_fixits.
27795 (selftest::test_fixit_insert_containing_newline): Update now that
27796 newlines are partially supported.
27797 (selftest::test_fixit_insert_containing_newline_2): New test.
27798 (selftest::test_fixit_replace_containing_newline): Update comments.
27799 (selftest::diagnostic_show_locus_c_tests): Call the new test.
27800 * edit-context.c (class added_line): New class.
27801 (class edited_line): Describe newline handling in comment.
27802 (edited_line::actually_edited_p): New method.
27803 (edited_line::print_content): Delete redundant decl.
27804 (edited_line::m_predecessors): New field.
27805 (edited_file::print_content): Call edited_line::print_content.
27806 (edited_file::print_diff): Update to support newlines.
27807 (edited_file::print_diff_hunk): Likewise.
27808 (edited_file::print_run_of_changed_lines): New function.
27809 (edited_file::print_diff_line): Convert to...
27810 (print_diff_line): ...this.
27811 (edited_file::get_effective_line_count): New function.
27812 (edited_line::edited_line): Initialize new field m_predecessors.
27813 (edited_line::~edited_line): Clean up m_predecessors.
27814 (edited_line::apply_fixit): Handle newlines.
27815 (edited_line::get_effective_line_count): New function.
27816 (edited_line::print_content): New function.
27817 (edited_line::print_diff_lines): New function.
27818 (selftest::test_applying_fixits_insert_containing_newline): New
27820 (selftest::test_applying_fixits_replace_containing_newline): New
27822 (selftest::insert_line): New function.
27823 (selftest::test_applying_fixits_multiple_lines): Add example of
27825 (selftest::edit_context_c_tests): Call the new tests.
27827 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27829 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
27830 parameter cand. Update dump information.
27831 (get_computation_cost): Update uses.
27833 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27835 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
27836 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
27837 (get_computation_at, rewrite_use_address): Update use of
27838 get_computation_aff.
27840 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27842 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
27843 (get_computation): Delete.
27844 (get_computation_cost): Implement like get_computation_cost_at.
27845 Use get_computation_at.
27846 (get_computation_cost_at): Delete.
27847 (rewrite_use_nonlinear_expr): Use get_computation_at.
27848 (rewrite_use_compare, remove_unused_ivs): Ditto.
27850 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27852 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
27854 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27856 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
27857 (ivopts_global_cost_for_size): Rename parameter and update uses.
27858 (iv_ca_recount_cost): Update uses.
27859 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
27860 candidates seperately in n_invs and n_cands.
27861 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
27863 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27865 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27866 (find_inv_vars_cb): New.
27867 (find_depends): Renamed to ...
27868 (find_inv_vars): ... this.
27869 (add_candidate_1, force_var_cost): Call find_inv_vars.
27870 (split_address_cost, determine_group_iv_cost_cond): Ditto.
27872 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27874 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
27875 inv_vars. Add inv_exprs.
27876 (struct iv_cand): Rename depends_on to inv_vars.
27877 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
27878 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
27879 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
27880 (dump_cand): Dump inv_vars.
27881 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27882 (record_invariant, find_depends, add_candidate_1): Ditto.
27883 (set_group_iv_cost, force_var_cost): Ditto.
27884 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27885 (get_computation_cost_at, get_computation_cost): Ditto.
27886 (determine_group_iv_cost_generic): Ditto.
27887 (determine_group_iv_cost_address): Ditto.
27888 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27889 (determine_group_iv_costs): Ditto.
27890 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27891 (iv_ca_set_remove_invariants): Renamed to ...
27892 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27893 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27894 (iv_ca_set_add_invariants): Renamed to ...
27895 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27896 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27897 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27898 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27899 (create_new_ivs): Remove useless dump.
27901 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27903 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27905 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27906 (iv_ca_set_no_cp, create_new_iv): Ditto.
27908 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27910 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27912 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27914 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27915 function tree_check2.
27917 2017-05-02 Martin Liska <mliska@suse.cz>
27919 * doc/gcov.texi: Add missing preposition.
27920 * gcov.c (function_info::function_info): Properly fill up
27921 all member variables.
27923 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27925 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27927 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27929 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27931 2017-05-02 Martin Liska <mliska@suse.cz>
27934 * lto-streamer-in.c (lto_read_tree_1): Remove
27935 LTO_STREAMER_DEBUG.
27936 * lto-streamer.c (struct tree_hash_entry): Likewise.
27937 (struct tree_entry_hasher): Likewise.
27938 (tree_entry_hasher::hash): Likewise.
27939 (tree_entry_hasher::equal): Likewise.
27940 (lto_streamer_init): Likewise.
27941 (lto_orig_address_map): Likewise.
27942 (lto_orig_address_get): Likewise.
27943 (lto_orig_address_remove): Likewise.
27944 * lto-streamer.h: Likewise.
27945 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27946 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27948 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27950 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27951 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27952 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27953 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27954 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27955 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27956 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27957 (mm_maskz_sub_ss): New intrinsics.
27958 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27959 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27960 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27961 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27962 (__builtin_ia32_subss_mask_round): New builtins.
27963 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27964 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27965 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27967 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27968 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27970 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27973 2017-05-02 Martin Jambor <mjambor@suse.cz>
27975 PR tree-optimization/78687
27976 * tree-sra.c (access): New field parent.
27977 (process_subtree_disqualification): New function.
27978 (disqualify_candidate): Call it.
27979 (build_accesses_from_assign): Reset write flag if creating an
27981 (build_access_subtree): Fill in parent field and also prpagate
27982 down grp_write flag.
27983 (create_artificial_child_access): New parameter set_grp_write, set
27984 grp_write to its value.
27985 (propagate_subaccesses_across_link): Also propagate grp_write flag
27987 (propagate_all_subaccesses): Push the closest parent back to work
27988 queue if add_access_to_work_queue returned true.
27990 2017-05-02 Richard Biener <rguenther@suse.de>
27992 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27993 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27994 -fstrict-overflow documentation.
27995 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27996 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27997 flag_strict_overflow.
27998 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27999 * lto-opts.c (lto_write_options): Do not stream it.
28000 * lto-wrapper.c (merge_and_complain): Do not handle it.
28001 * opts.c (default_options_table): Do not set -fstrict-overflow.
28002 (finish_options): Likewise do not clear it when sanitizing.
28003 * simplify-rtx.c (simplify_const_relational_operation): Do not
28004 test flag_strict_overflow.
28006 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28008 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28009 using enabled attribute.
28010 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28011 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28012 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28013 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28014 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28015 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28016 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28017 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28018 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28019 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28021 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28023 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28025 2017-05-02 Richard Biener <rguenther@suse.de>
28027 PR tree-optimization/80591
28029 2017-04-10 Richard Biener <rguenther@suse.de>
28031 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28034 2017-05-02 Richard Biener <rguenther@suse.de>
28036 PR tree-optimization/80549
28037 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28038 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28039 headers if they do not have a preheader.
28041 2017-05-02 Martin Liska <mliska@suse.cz>
28044 * common.opt: Fix typo.
28045 * doc/invoke.texi: Likewise.
28047 2017-05-01 Jan Beulich <jbeulich@suse.com>
28049 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28050 swapping, add (x,x,m,x,n) alternative.
28052 2017-05-01 Nathan Sidwell <nathan@acm.org>
28054 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28055 unnecessary unadjusted_alignment check.
28057 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28060 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28062 * gimplify.c (gimplify_cilk_detach): New function.
28063 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28064 * tree-core.h: Document EXPR_CILK_SPAWN.
28065 * tree.h (EXPR_CILK_SPAWN): Define.
28067 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28069 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28070 to use new fixit_hint representation, using the "replace" logic.
28071 (get_line_span_for_fixit_hint): Likewise.
28072 (layout::print_any_fixits): Likewise.
28073 (selftest::test_one_liner_many_fixits): Rename to...
28074 (selftest::test_one_liner_many_fixits_1): ...this, and update
28075 comment and expected output to reflect that the multiple fix-it
28076 hints are now consolidated into one insertion.
28077 (selftest::test_one_liner_many_fixits_2): New test.
28078 (selftest::test_diagnostic_show_locus_one_liner): Update for
28080 (selftest::test_fixit_consolidation): Update for fix-it API
28082 * diagnostic.c (print_parseable_fixits): Likewise.
28083 * edit-context.c (edited_line::m_line_events): Convert from
28084 auto_vec <line_event *> to auto_vec <line_event>.
28085 (class line_event): Convert from abstract base class to a concrete
28086 class, taking over the role of replace_event.
28087 (class insert_event): Delete.
28088 (class replace_event): Rename to class line_event. Convert to
28090 (edit_context::add_fixits): Reimplement.
28091 (edit_context::apply_insert): Delete.
28092 (edit_context::apply_replace): Rename to...
28093 (edit_context::apply_fixit): ...this. Convert to half-open range.
28094 (edited_file::apply_insert): Delete.
28095 (edited_file::apply_replace): Rename to...
28096 (edited_file::apply_fixit): ...this.
28097 (edited_line::~edited_line): Drop deletion of events.
28098 (edited_line::apply_insert): Delete.
28099 (edited_line::apply_replace): Rename to...
28100 (edited_line::apply_fixit): ...this. Convert to half-open range.
28101 Update for change to type of m_line_events.
28102 * edit-context.h (edit_context::apply_insert): Delete.
28103 (edit_context::apply_replace): Rename to...
28104 (edit_context::apply_fixit): ...this.
28106 2017-05-01 Martin Sebor <msebor@redhat.com>
28108 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28111 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28114 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28115 __get_cpuid_max returns 0.
28116 (__get_cpuid_count): Ditto.
28118 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28120 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28121 replacement expression is another instance of one of its arguments.
28123 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28126 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28127 check for stack push/pop autoinc.
28128 * config/i386/i386.c (ix86_agi_dependent): Return false
28129 if the only reason why modified_in_p returned true is that
28130 addr is SP based and set_insn is a push or pop.
28132 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28134 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28137 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28140 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28141 (account_size_time): Use two predicates - exec_pred and
28143 (evaluate_conditions_for_known_args): Compute both clause and
28145 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28146 (inline_summary_t::duplicate): Update.
28147 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28149 (compute_inline_parameters): Likewise.
28150 (estimate_edge_size_and_time): Update caluclation of time.
28151 (estimate_node_size_and_time): Compute both time and nonspecialized
28153 (estimate_ipcp_clone_size_and_time): Update.
28154 (inline_merge_summary): Update.
28155 (do_estimate_edge_time): Update.
28156 (do_estimate_edge_size): Update.
28157 (do_estimate_edge_hints): Update.
28158 (inline_read_section, inline_write_summary): Stream both new predicates.
28159 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28161 (compute_inlined_call_time): Cleanup.
28162 (big_speedup_p): Update.
28163 (edge_badness): Update.
28164 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28165 (size_time_entry): Replace predicate by exec_predicate and
28166 nonconst_predicate.
28167 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28168 (estimate_edge_time): Return also nonspec_time.
28169 (reset_edge_growth_cache): Update.
28171 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28173 PR rtl-optimization/80491
28174 * ifcvt.c (noce_process_if_block): When looking for x setter
28175 with missing else_bb, don't check only the insn right before
28176 cond_earliest, but look for the last insn that x is modified in
28177 within the same bb.
28179 PR rtl-optimization/80491
28180 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28182 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28184 PR tree-optimization/80487
28185 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28187 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28189 PR tree-optimization/79697
28190 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28191 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28192 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28194 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28195 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28197 2017-04-28 Martin Sebor <msebor@redhat.com>
28199 PR tree-optimization/80523
28200 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28201 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28203 (maybe_warn, format_directive, parse_directive): Use new functions.
28204 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28206 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28208 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28210 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28212 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28213 target_header_dir): Set correctly.
28214 * configure: Regenerated.
28215 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28216 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28217 instead of SYSTEM_HEADER_DIR.
28219 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28221 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28222 (estimate_local_effects): Likewise.
28223 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28224 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28225 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28226 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28227 do_estimate_edge_time, estimate_edge_time): Likewise.
28228 * ipa-inline-analysis.c (estimate_node_size_and_time,
28229 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28230 (estimate_time_after_inlining): Remove.
28232 2017-04-28 Martin Liska <mliska@suse.cz>
28234 * doc/gcov.texi: Enhance documentation of gcov.
28236 2017-04-28 Martin Liska <mliska@suse.cz>
28238 * doc/gcov.texi: Sort options in alphabetic order.
28239 * doc/gcov-dump.texi: Likewise.
28240 * doc/gcov-tool.texi: Likewise.
28241 * gcov.c (print_usage): Likewise.
28242 * gcov-dump.c (print_usage): Likewise.
28243 * gcov-tool.c (print_merge_usage_message): Likewise.
28244 (print_rewrite_usage_message): Likewise.
28245 (print_overlap_usage_message): Likewise.
28247 2017-04-28 Martin Liska <mliska@suse.cz>
28249 PR gcov-profile/53915
28250 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28252 2017-04-28 Martin Liska <mliska@suse.cz>
28254 PR gcov-profile/79891
28255 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28256 is marked by compiler as living on a line.
28257 (get_cycles_count): Remove usage of the union.
28258 (output_intermediate_file): Likewise.
28259 (find_source): Fix GNU coding style.
28260 (accumulate_line_counts): Remove old non-all block mode.
28261 (output_lines): Remove usage of the union.
28262 * profile.c (output_location): Include all BBs, even if
28263 belonging to a same line (and file) as a previous BB.
28265 2017-04-28 Martin Liska <mliska@suse.cz>
28267 * gcov.c (process_args): Handle new argument 'w'.
28268 (read_graph_file): Assign ID to BBs.
28269 (output_branch_count): Display BB # if verbose flag is set.
28270 (output_lines): Likewise for arcs.
28271 (print_usage): Add '--verbose' option help.
28272 * doc/gcov.texi: Document --verbose (-w) option.
28274 2017-04-28 Martin Liska <mliska@suse.cz>
28276 * gcov.c (struct block_location_info): New struct.
28277 (process_file): Fill up the new structure.
28278 (read_graph_file): Replace usage of encoding by the newly added
28280 (add_line_counts): Likewise.
28281 (accumulate_line_counts): Remove usage of the union.
28282 (function_info::function_info): New function.
28283 (function_info::~function_info): Likewise.
28284 (process_file): Call delete instead of release_function.
28285 (release_function): Release the function.
28286 (release_structures): Call delete instead of release_function.
28287 (solve_flow_graph): Replace usage of num_blocks.
28288 (find_exception_blocks): Likewise.
28289 (output_lines): Fix GNU coding style.
28291 2017-04-28 Martin Liska <mliska@suse.cz>
28294 * coverage.c (coverage_remove_note_file): New function.
28295 * coverage.h: Declare the function.
28296 * toplev.c (finalize): Clean if an error has been seen.
28298 2017-04-28 Martin Liska <mliska@suse.cz>
28300 PR gcov-profile/80031
28301 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28302 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28303 * gcov.c (read_graph_file): Read just number of blocks.
28304 * profile.c (branch_prob): Do not stream 0 flags per a basic
28307 2017-04-28 Martin Liska <mliska@suse.cz>
28309 * gcov-dump.c (tag_*): Add new argument to declarations.
28310 (dump_gcov_file): Likewise.
28311 (tag_blocks): Add and use new argument depth.
28312 (tag_arcs): Likewise.
28313 (tag_lines): Likewise.
28314 (tag_counters): Likewise.
28315 (tag_summary): Likewise.
28316 (dump_working_sets): Use depth to do a proper indentation.
28318 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28321 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28322 * symtab.c (symtab_node::debug_symtab): Move definition here.
28324 2017-04-28 Richard Biener <rguenther@suse.de>
28326 * lto-streamer.h (LTO_major_version): Bump to 7.
28328 2017-04-28 Richard Biener <rguenther@suse.de>
28330 * tree-vrp.c (assert_info): New struct.
28331 (add_assert_info): New helper.
28332 (register_edge_assert_for_2): Refactor to add asserts to a vector
28334 (register_edge_assert_for_1): Likewise.
28335 (register_edge_assert_for): Likewise.
28336 (finish_register_edge_assert_for): New helper actually registering
28337 asserts where live on edge.
28338 (find_conditional_asserts): Adjust.
28339 (find_switch_asserts): Likewise.
28340 (evrp_dom_walker::try_find_new_range): Generalize.
28341 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28343 2017-04-27 Marek Polacek <polacek@redhat.com>
28346 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28347 arg10 and arg11 to itype.
28349 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28351 * doc/extend.texi (Object Size Checking): Improve grammar.
28353 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28356 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28357 that the logic for permitting reciprocal estimates matches that
28360 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28363 * tree.c (type_cache_hasher::equal): Only compare
28364 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28365 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28366 non-aggregate element types.
28367 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28368 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28370 2017-04-27 Richard Biener <rguenther@suse.de>
28372 PR middle-end/80533
28373 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28374 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28375 keeping a reference to a trailing array.
28377 2017-04-27 Richard Biener <rguenther@suse.de>
28379 PR middle-end/80539
28380 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28381 being in loop-closed SSA form conservatively.
28382 (chrec_fold_multiply_poly_poly): Likewise.
28384 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28386 PR middle-end/79665
28387 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28388 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28390 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28393 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28394 (aarch64_function_arg_alignment): Return unsigned int again, but still
28395 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28396 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28397 Don't emit -Wpsabi note.
28398 (aarch64_function_arg_boundary): Likewise.
28399 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28402 2017-04-26 Nathan Sidwell <nathan@acm.org>
28404 * tree.h (crc32_unsigned_n): Declare.
28405 (crc32_unsigned, crc32_unsigned): Make inline.
28406 * tree.c (crc32_unsigned_bits): Replace with ...
28407 (crc32_unsigned_n): ... this.
28408 (crc32_unsigned, crc32_byte): Remove.
28409 (crc32_string): Remove unnecessary braces.
28411 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28413 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28414 * ipa-inline-analysis.c (MAX_TIME): Remove.
28415 (account_size_time): Use sreal for time.
28416 (dump_inline_summary): Update.
28417 (estimate_function_body_sizes): Update.
28418 (estimate_edge_size_and_time): Update.
28419 (estimate_calls_size_and_time): Update.
28420 (estimate_node_size_and_time): Update.
28421 (inline_merge_summary): Update.
28422 (inline_update_overall_summary): Update.
28423 (estimate_time_after_inlining): Update.
28424 (inline_read_section): Update.
28425 (inline_write_summary): Update.
28426 * ipa-inline.c (compute_uninlined_call_time): Update.
28427 (compute_inlined_call_time): Update.
28428 (recursive_inlining): Update.
28429 (inline_small_functions): Update.
28430 (dump_overall_stats): Update.
28431 * ipa-inline.h: Include sreal.h.
28432 (size_time_entry): Turn time to sreal.
28433 (inline_summary): Turn self_time nad time to sreal.
28435 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28437 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28439 (sreal::stream_out, sreal::stream_in): New.
28440 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28442 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28444 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28447 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28450 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28452 (dimode_scalar_chain::compute_convert_gain): Ditto.
28453 (dimode_scalar_chain::make_vector_copies): Ditto.
28454 (dimode_scalar_chain::convert_reg): Ditto.
28455 (dimode_scalar_chain::convert_insn): Ditto.
28456 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28457 (VI248_AVX512BW_1): New mode iterator.
28458 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28459 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28462 2017-04-25 Martin Sebor <msebor@redhat.com>
28464 PR tree-optimization/80497
28465 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28466 constants are representable in HOST_WIDE_INT.
28467 (parse_directive): Ditto.
28469 2017-04-25 Martin Sebor <msebor@redhat.com>
28472 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28473 (new_zero_array): Adjust signature.
28474 (dom_info::dom_init): Used unsigned rather that size_t.
28475 (dom_info::dom_info): Same.
28477 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28478 Jakub Jelinek <jakub@redhat.com>
28481 * config/arm/arm.c: Include gimple.h.
28482 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28483 returns negative, increment ncrn only if it returned positive.
28484 (arm_needs_doubleword_align): Return int instead of bool,
28485 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28486 members, but if there is any such non-FIELD_DECL
28487 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28488 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28489 returns negative, increment nregs only if it returned positive.
28490 (arm_setup_incoming_varargs): Likewise.
28491 (arm_function_arg_boundary): Emit -Wpsabi note if
28492 arm_needs_doubleword_align returns negative, return
28493 DOUBLEWORD_ALIGNMENT only if it returned positive.
28495 2017-04-25 Marek Polacek <polacek@redhat.com>
28498 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28499 first argument to type.
28501 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28504 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28505 type checks to test for compatibility instead of equality.
28507 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28508 Jakub Jelinek <jakub@redhat.com>
28511 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28513 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28514 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28515 the alignment computation, but return their maximum in warn_alignment.
28516 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28517 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28519 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28520 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28523 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28525 * config/arc/simdext.md (dmpyh): Fix typo.
28527 2017-04-25 Richard Biener <rguenther@suse.de>
28529 PR tree-optimization/80492
28530 * alias.c (compare_base_decls): Handle registers with asm
28531 specification conservatively.
28532 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28533 compare_base_decls returning dont-know properly.
28535 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28537 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28538 (legitimate_offset_address_p): New function.
28539 (arc_legitimate_address_p): Use above function.
28541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28543 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28545 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28547 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28548 ACCH registers whenever they are available.
28550 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28552 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28553 double regs fix when not used.
28555 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28557 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28559 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28560 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28562 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28564 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28565 when emitting short ADD instructions.
28567 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28569 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28571 (cmpsi_cc_c_insn): Likewise.
28572 (cbranchsi4_scratch): Compute proper instruction length using
28573 compact_hreg_operand.
28574 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28576 2017-04-25 Richard Biener <rguenther@suse.de>
28578 PR middle-end/80509
28579 * passes.c (pass_manager::pass_manager): Initialize
28580 m_name_to_pass_map.
28582 2017-04-25 Richard Biener <rguenther@suse.de>
28584 PR tree-optimization/79201
28585 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28587 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28590 * config/s390/vector.md: Split MEM->GPR vector moves for
28591 non-s_operand addresses.
28593 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28596 * config/s390/predicates.md (reload_const_wide_int_operand): New
28598 * config/s390/s390.md ("movti"): Remove d/P alternative.
28599 ("movti_bigconst"): New pattern definition.
28601 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28604 * s390-protos.h (s390_expand_cs_hqi): Removed.
28605 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28606 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28607 modes as well as CCZ1mode and CCZmode.
28608 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28609 signature of s390_emit_compare_and_swap.
28610 (s390_expand_cs_hqi): Likewise, make static.
28611 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28612 compare-and-swap, in some cases.
28613 (s390_expand_cs): Wrapper function.
28614 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28616 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28617 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28618 patterns for small and large integers. Forbid symref memory operands.
28619 Move expander to s390.c. Require cc register.
28620 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28621 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28622 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28623 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28624 symref memory operands. Remove CC mode and call s390_match_ccmode
28626 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28628 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28630 * config/s390/s390.md (define_peephole2): New peephole to help
28631 combining the load-and-test pattern with volatile memory.
28633 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28635 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28636 with CCZmode for TARGET_Z196.
28638 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28640 PR rtl-optimization/80501
28641 * combine.c (make_compound_operation_int): Set subreg_code to SET
28642 even for AND with mask of the sign bit of mode.
28644 PR rtl-optimization/80500
28645 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28646 sum's initial value.
28648 2017-04-25 Julian Brown <julian@codesourcery.com>
28649 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28651 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28653 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28655 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28657 2017-04-25 Julian Brown <julian@codesourcery.com>
28658 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28660 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28661 (thunderx2t99_sha): New Reservation.
28663 2017-04-25 Julian Brown <julian@codesourcery.com>
28664 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28666 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28667 type for 1-element load.
28669 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28671 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28673 2017-04-24 Martin Jambor <mjambor@suse.cz>
28675 PR tree-optimization/80293
28676 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28677 char arrays not totally scalarizable if it is false.
28678 (analyze_all_variable_accesses): Pass correct value in the new
28679 parameter. Add a statistics counter.
28681 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28683 PR middle-end/79931
28684 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28686 2017-04-24 Richard Biener <rguenther@suse.de>
28688 PR tree-optimization/80494
28689 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28690 out for complex types.
28692 2017-04-24 Richard Biener <rguenther@suse.de>
28694 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28695 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28696 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
28697 (DFS): Adjust and never fail.
28698 (sccvn_dom_walker::fail): Remove.
28699 (sccvn_dom_walker::before_dom_children): Adjust.
28700 (run_scc_vn): Likewise and never fail.
28701 * tree-ssa-pre.c (pass_pre::execute): Adjust.
28702 (pass_fre::execute): Likewise.
28704 2017-04-24 Richard Biener <rguenther@suse.de>
28706 PR tree-optimization/79725
28707 * tree-ssa-sink.c (statement_sink_location): Return whether
28708 failure reason was zero uses. Move that check later.
28709 (sink_code_in_bb): Deal with zero uses by removing the stmt
28712 2017-04-24 Richard Biener <rguenther@suse.de>
28715 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
28716 pointer-based references.
28718 2017-04-24 Richard Biener <rguenther@suse.de>
28721 * pass_manager.h (pass_manager::operator new): Remove.
28722 (pass_manager::operator delete): Likewise.
28723 * passes.c (pass_manager::operator new): Remove.
28724 (pass_manager::operator delete): Likewise.
28725 (pass_manager::pass_manager): Zero individual pass members.
28727 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
28730 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
28731 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
28732 Check "XEXP (src, 1)" operand here.
28733 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
28734 Check "XEXP (src, 1)" operand here.
28735 (dimode_scalar_chain::make_vector_copies): Detect count register
28736 of a shift instruction. Zero extend count register from QImode
28737 to DImode to satisfy vector shift pattern count operand predicate.
28738 Substitute vector shift count operand with a DImode copy.
28739 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
28742 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
28744 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
28745 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28746 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
28747 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
28748 (UNSPEC_NOREX_MEM): Remove definition.
28750 2017-04-21 Richard Biener <rguenther@suse.de>
28752 PR tree-optimization/79547
28753 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
28754 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
28755 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
28756 without any constraints.
28758 2017-04-21 Richard Biener <rguenther@suse.de>
28760 PR tree-optimization/78847
28761 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
28763 2017-04-21 Richard Biener <rguenther@suse.de>
28765 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
28766 (build_distinct_type_copy): Likewise.
28767 (build_variant_type_copy): Likewise.
28768 * tree.c (build_qualified_type): Pass down mem-stat info.
28769 (build_distinct_type_copy): Likewise.
28770 (build_variant_type_copy): Likewise.
28772 2017-04-21 Richard Biener <rguenther@suse.de>
28774 PR tree-optimization/80237
28775 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
28777 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
28778 for a simplified result.
28780 2016-04-21 Richard Biener <rguenther@suse.de>
28782 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
28783 sth as strict as a simple_iv but a chrec without symbols and an
28784 operand defined in the loop we are peeling (and not some subloop).
28785 (propagate_constants_for_unrolling): Propagate all constants.
28787 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28790 * config/i386/i386.c (print_reg): Remove assert for disalowed
28791 regno values, call output_operand_lossage instead.
28793 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
28796 * config/i386/constraints.md (Yc): New register constraint.
28797 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
28798 Use Yc constraint for alternative 2 of operand 0. Remove
28799 preferred_for_speed attribute.
28801 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28803 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
28804 lastprivate clauses in SIMT case.
28806 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
28808 * doc/invoke.texi (-Wextra-semi): Document new warning option.
28810 2017-04-20 Richard Biener <rguenther@suse.de>
28812 PR tree-optimization/57796
28813 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
28814 as N scalar stores.
28815 (vect_model_load_cost): Cost gathers as N scalar loads.
28817 2017-04-20 Richard Biener <rguenther@suse.de>
28819 * ggc-page.c (ggc_allocated_p): Rename to ...
28820 (safe_lookup_page_table_entry): ... this and return the lookup
28822 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
28824 2017-04-20 Richard Biener <rguenther@suse.de>
28826 PR tree-optimization/80453
28827 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
28828 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
28829 from the conditions.
28830 (vn_phi_eq): Pass them down.
28831 (vn_phi_lookup): Record them.
28832 (vn_phi_insert): Likewise.
28834 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
28836 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
28837 uninitialized variable warning to avoid buffer overrun.
28839 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
28842 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
28843 is suppressed for '{ 0 }' in C.
28845 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28847 * BASE-VER: Set to 8.0.0.
28849 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
28851 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
28852 priority .init_array and .fini_array section with SECTION_NOTYPE
28855 2017-04-20 Jakub Jelinek <jakub@redhat.com>
28857 PR middle-end/80423
28858 * tree.h (build_array_type): Add typeless_storage default argument.
28859 * tree.c (type_cache_hasher::equal): Also compare
28860 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
28861 (build_array_type): Add typeless_storage argument, set
28862 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
28864 (build_nonshared_array_type): Adjust build_array_type_1 caller.
28865 (build_array_type): Likewise. Add typeless_storage argument.
28867 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28868 Jakub Jelinek <jakub@redhat.com>
28870 PR tree-optimization/80426
28871 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
28872 operation on symbolic operands, also compute the overflow for the
28873 invariant part when the operation degenerates into a negation.
28875 2017-04-19 Jakub Jelinek <jakub@redhat.com>
28878 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
28879 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28882 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28884 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28887 * config/avr/avr.c (tree.h): Include it.
28888 (cgraph.h): Include it.
28889 (avr_encode_section_info): Don't warn for uninitialized progmem
28890 variable if it's just an alias.
28892 2017-04-19 Richard Biener <rguenther@suse.de>
28895 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28896 when needed by AutoPGO.
28898 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28901 * doc/lto.texi: Remove an extra 'that'.
28903 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28905 PR rtl-optimization/80429
28906 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28907 are only used in debug insns.
28909 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28910 Vladimir Makarov <vmakarov@redhat.com>
28912 * config/sparc/predicates.md (input_operand): Add comment. Return
28913 true for any memory operand when LRA is in progress.
28914 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28916 2017-04-18 Jeff Law <law@redhat.com>
28919 * mips.md ({return,simple_return}_internal): Do not overwrite
28922 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28924 PR tree-optimization/80443
28925 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28926 instead of adding 1, subtract -1 and similarly instead of subtracting
28929 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28931 PR rtl-optimization/80357
28932 * haifa-sched.c (tmp_bitmap): New variable.
28933 (model_recompute): Handle duplicate use records.
28934 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28935 (free_global_sched_pressure_data): Free it.
28937 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28940 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28941 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28942 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28943 instead of SYSTEM_HEADER_DIR.
28945 2017-04-18 Jeff Law <law@redhat.com>
28947 PR middle-end/80422
28948 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28949 predecessors after walking up the insn chain.
28951 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28954 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28955 sizetype type into debug info.
28957 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28960 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28961 unneeded test for TARGET_UPPER_REGS_SF.
28962 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28964 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28967 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28968 instead of gsi_after_labels.
28970 2017-04-18 Jeff Law <law@redhat.com>
28972 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28976 2017-04-13 Jeff Law <law@redhat.com>
28977 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28978 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28980 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28983 * config/avr/avr.c (intl.h): Include it.
28984 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28986 2017-04-18 Martin Liska <mliska@suse.cz>
28988 PR gcov-profile/78783
28989 * gcov-tool.c (gcov_output_files): Validate that destination
28990 file is either removed by the tool or by a user.
28992 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28993 Guy Benyei <guybe@mellanox.com>
28995 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28996 block, and do not negate it, the stored id is already negative.
28998 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29000 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29002 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29005 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29006 masks of options that should be turned off if the VSX vector
29007 options are turned off.
29008 (OTHER_P8_VECTOR_MASKS): Likewise.
29009 (OTHER_VSX_VECTOR_MASKS): Likewise.
29010 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29011 rs6000_disable_incompatible_switches to validate no type switches
29013 (rs6000_incompatible_switch): New function to disallow turning on
29014 other vector options if -mno-vsx, -mno-power8-vector, or
29015 -mno-power9-vector are specified.
29017 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29019 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29021 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29023 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29024 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29025 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29026 (ARG_POINTER_CFA_OFFSET): Likewise.
29028 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29030 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29031 conditions to take advantage of various optimizations.
29033 2017-04-13 Jeff Law <law@redhat.com>
29035 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29036 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29037 (zero_extendsidi2_dext): Likewise.
29039 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29042 * fold-const.c (fold_ternary_loc): Revert
29043 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29046 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29048 PR rtl-optimization/80343
29049 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29050 new scratch pseudo.
29052 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29055 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29056 to ubsan_encode_value.
29058 2017-04-13 Jeff Law <law@redhat.com>
29060 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29061 appearing in DEBUG_INSNs.
29063 2017-04-13 Martin Liska <mliska@suse.cz>
29065 PR gcov-profile/80413
29066 * gcov-io.c (gcov_write_string): Copy to buffer just when
29067 allocated size is greater than zero.
29069 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29072 * dwarf2out.c (decls_for_scope): Ignore declarations of
29073 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29075 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29078 * ipa-visibility.c (non_local_p): Fix typos.
29079 (localize_node): When localizing symbol in same comdat group,
29080 dissolve the group only when we know external symbols are going
29082 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29084 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29086 PR tree-optimization/79390
29087 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29088 order does not result in usable sequence, retry with reversed operand
29094 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29095 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29096 op0 instead of fold_convert_loc (loc, type, arg0).
29098 2017-04-12 Jeff Law <law@redhat.com>
29100 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29101 has a delay slot in the generated code.
29103 * config/cris/cris.md (cris_preferred_reload_class): Return
29104 GENNONACR_REGS rather than GENERAL_REGS.
29106 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29109 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29110 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29111 signedness of the result type.
29113 2017-04-12 Richard Biener <rguenther@suse.de>
29114 Jeff Law <law@redhat.com>
29116 PR tree-optimization/80359
29117 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29118 trim stores to TARGET_MEM_REFs.
29120 2017-04-12 Richard Biener <rguenther@suse.de>
29122 PR tree-optimization/79390
29123 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29124 threading case even more.
29126 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29129 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29130 for quad_address_p for TImode, instead of just not indexed_address.
29132 2017-04-12 Richard Biener <rguenther@suse.de>
29133 Bernd Edlinger <bernd.edlinger@hotmail.de>
29135 PR middle-end/79671
29136 * alias.c (component_uses_parent_alias_set_from): Handle
29137 TYPE_TYPELESS_STORAGE.
29138 (get_alias_set): Likewise.
29139 * tree-core.h (tree_type_common): Add typeless_storage flag.
29140 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29141 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29142 for types containing members with TYPE_TYPELESS_STORAGE.
29143 (place_field): Likewise.
29144 (layout_type): Likewise for ARRAY_TYPE.
29145 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29146 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29147 TYPE_TYPELESS_STORAGE.
29148 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29150 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29153 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29154 first argument to type.
29156 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29160 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29161 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29162 (rs6000_expand_binop_builtin): Likewise.
29163 (rs6000_expand_ternop_builtin): Likewise; also add missing
29164 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29165 vshasigma built-ins.
29166 * doc/extend.texi: Document that vec_xxpermdi's third argument
29167 must be a constant.
29169 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29171 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29172 Use shift_const cost parameter when calculating gain of STV shifts.
29174 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29176 PR rtl-optimization/70478
29177 * lra-constraints.c (process_alt_operands): Check memory for
29178 disfavoring memory insn operand.
29180 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29182 PR middle-end/80100
29183 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29184 left shift in unsigned HOST_WIDE_INT type.
29186 PR rtl-optimization/80385
29187 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29188 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29191 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29192 if they have any depend clauses.
29194 2017-04-11 Martin Liska <mliska@suse.cz>
29197 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29198 * ipa-split.c (split_function): Create a local comdat symbol
29199 if caller is in a comdat group.
29201 2017-04-11 Martin Liska <mliska@suse.cz>
29204 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29207 2017-04-11 Martin Sebor <msebor@redhat.com>
29209 PR middle-end/80364
29210 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29211 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29213 (directive::set_width, directive::set_precision, format_character):
29215 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29218 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29221 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29222 conflict, set target->arch_name instead of target->cpu_name.
29224 2017-04-11 Richard Biener <rguenther@suse.de>
29226 PR tree-optimization/80374
29227 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29228 build_zero_cst, remove fold_convertible_p check again.
29230 2017-04-11 Martin Liska <mliska@suse.cz>
29233 * ubsan.c (instrument_object_size): Do not instrument register
29236 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29239 * config/i386/i386-builtin-types.def
29240 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29241 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29242 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29243 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29244 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29245 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29246 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29247 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29248 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29249 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29250 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29251 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29252 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29253 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29254 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29255 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29256 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29257 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29258 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29259 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29260 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29261 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29262 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29263 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29264 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29265 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29266 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29267 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29268 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29269 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29270 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29271 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29272 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29273 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29274 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29275 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29276 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29277 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29278 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29279 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29280 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29281 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29282 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29283 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29284 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29285 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29287 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29288 flag to second_arg_count, handle 4 argument function type _COUNT
29289 aliases, handle second_arg_count on second argument rather than last.
29291 2017-04-10 Jeff Law <law@redhat.com>
29293 PR tree-optimization/80374
29294 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29295 record anything if we can not convert integer_zero_node to the
29298 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29301 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29302 Enhance special handling given to the TARGET_P9_MINMAX option in
29303 relation to certain other options.
29305 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29307 PR tree-optimization/80153
29308 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29309 remove POINTER_PLUS_EXPR's base part directly, rather than through
29312 2017-04-10 Richard Biener <rguenther@suse.de>
29313 Bin Cheng <bin.cheng@arm.com>
29315 PR tree-optimization/80153
29316 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29317 the first element of pointer type aff_tree. Build result expr in
29319 (add_elt_to_tree): Convert to type unconditionally. Remove other
29320 fold_convert calls.
29321 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29322 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29324 2017-04-10 Richard Biener <rguenther@suse.de>
29326 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29329 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29331 PR rtl-optimization/70478
29332 * lra-constraints.c (curr_small_class_check): New.
29333 (update_and_check_small_class_inputs): New.
29334 (process_alt_operands): Update curr_small_class_check. Disfavor
29335 alternative insn memory operands. Check available regs for small
29338 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29341 * config/mips/mips.opt (-mvirt): Update description.
29342 * doc/invoke.texi (-mvirt): Likewise.
29344 2017-04-10 Richard Biener <rguenther@suse.de>
29346 PR middle-end/80362
29347 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29348 looking for NEGATE_EXPR in -A / -B to A / B folding.
29350 2017-04-10 Martin Liska <mliska@suse.cz>
29352 PR gcov-profile/80224
29353 * gcov.c (print_usage): Fix usage string.
29354 (get_gcov_intermediate_filename): Remove.
29355 (output_gcov_file): Use both for normal and intermediate format.
29356 (generate_results): Do not initialize special file for
29357 intermediate format.
29359 2017-04-10 Richard Biener <rguenther@suse.de>
29361 PR tree-optimization/80304
29362 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29365 2017-04-10 Nathan Sidwell <nathan@acm.org>
29368 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29369 (rs6000_init_builtins): Use it.
29371 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29373 * config/arm/arm.md (<mrc>): Add mode to SET source.
29374 (<mrrc>): Likewise.
29376 2017-04-10 Richard Biener <rguenther@suse.de>
29378 PR middle-end/80344
29379 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29381 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29384 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29385 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29386 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29387 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29388 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29389 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29390 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29391 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29392 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29393 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29394 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29395 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29396 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29397 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29398 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29399 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29400 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29401 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29402 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29403 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29404 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29405 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29406 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29408 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29410 PR rtl-optimization/70478
29411 * lra-constraints.c: Reverse the last patch.
29413 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29415 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29416 Add comment for WCHAR_T.
29418 2017-04-08 Martin Liska <mliska@suse.cz>
29421 2017-04-07 Martin Liska <mliska@suse.cz>
29424 * ipa-split.c (split_function): Add function part to a same comdat
29427 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29430 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29432 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29434 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29435 (vec_pairing): ...to this.
29436 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29437 (rs6000_sched_init): Adjust for name change.
29438 (struct rs6000_sched_context): Likewise.
29439 (rs6000_init_sched_context): Likewise.
29440 (rs6000_set_sched_context): Likewise.
29442 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29448 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29450 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29451 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29452 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29454 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29456 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29458 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29460 PR rtl-optimization/70703
29461 * ira-color.c (update_conflict_hard_regno_costs): Use
29462 int64_t instead of HOST_WIDE_INT.
29464 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29466 PR rtl-optimization/70478
29467 * lra-constraints.c (process_alt_operands): Disfavor alternative
29468 insn memory operands.
29470 2017-04-07 Jeff Law <law@redhat.com>
29472 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29473 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29475 2017-04-07 Martin Liska <mliska@suse.cz>
29478 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29479 Show error message instead of an ICE.
29481 2017-04-07 Martin Liska <mliska@suse.cz>
29484 * ipa-split.c (split_function): Add function part to a same comdat
29487 2017-04-07 Richard Biener <rguenther@suse.de>
29489 PR middle-end/80341
29490 * tree.c (get_unwidened): Also handle ! for_type case for
29492 * convert.c (do_narrow): Split out from ...
29493 (convert_to_integer_1): ... here. Do not pass final truncation
29494 type to get_unwidened for TRUNC_DIV_EXPR.
29496 2017-04-07 Richard Biener <rguenther@suse.de>
29498 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29500 (aff_combination_const): Adjust.
29501 (aff_combination_scale): Likewise.
29502 (aff_combination_add_elt): Likewise.
29503 (aff_combination_add_cst): Likewise.
29504 (aff_combination_convert): Likewise.
29505 (add_elt_to_tree): Likewise. Remove unused argument.
29506 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29508 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29510 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29512 * config/arm/arm.c (arm_default_short_enums): Use
29513 ARM_DEFAULT_SHORT_ENUMS.
29514 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29516 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29519 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29520 members with redundant out-of-class redeclaration.
29522 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29525 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29526 * config/i386/i386.md (*zero_extendsidi2):
29527 Add (?*x,*x) and (?*v,*v) alternatives.
29529 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29532 * config/i386/i386.c (ix86_expand_builtin)
29533 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29534 mode from insn data. Convert operands to insn operand mode.
29535 Copy operands that don't satisfy insn predicate to a register.
29537 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29539 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29542 2017-04-06 Richard Biener <rguenther@suse.de>
29544 PR tree-optimization/80334
29545 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29546 preserve alignment of accesses.
29548 2017-04-06 Richard Biener <rguenther@suse.de>
29550 PR tree-optimization/80262
29551 * tree-sra.c (build_ref_for_offset): Preserve address-space
29553 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29554 Drop useless address-space information on MEM_REF offsets.
29556 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29558 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29560 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29562 PR rtl-optimization/70703
29563 * ira-color.c (update_conflict_hard_regno_costs): Use
29564 HOST_WIDE_INT instead of long.
29566 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29569 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29570 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29572 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29573 for x86_64 target. Handle -m3dnowa option.
29575 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29577 PR rtl-optimization/70703
29578 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29579 (update_conflict_hard_regno_costs): Use long instead of unsigned
29580 arithmetic for cost calculation.
29582 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29583 Bernd Edlinger <bernd.edlinger@hotmail.de>
29586 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29589 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29592 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29593 ptr_mode with Pmode throughout.
29594 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29595 into probe_stack_range and use DImode.
29597 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29600 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29601 call_eh_return is true.
29603 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29605 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29606 Initialize last_match_fntype_index.
29608 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29611 * tree-nvr.c: Include internal-fn.h.
29612 (pass_return_slot::execute): Ignore internal calls without
29615 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29616 Richard Biener <rguenther@suse.de>
29619 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29620 captures used multiple times, except for the last use.
29621 * generic-match-head.c: Include gimplify.h.
29623 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29625 PR tree-optimization/79390
29626 * target.h (struct noce_if_info): Declare.
29627 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29628 * target.def (noce_conversion_profitable_p): New target hook.
29629 * ifcvt.h (struct noce_if_info): New type, moved from ...
29630 * ifcvt.c (struct noce_if_info): ... here.
29631 (noce_conversion_profitable_p): Renamed to ...
29632 (default_noce_conversion_profitable_p): ... this. No longer
29634 (noce_try_store_flag_constants, noce_try_addcc,
29635 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29636 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29637 instead of noce_conversion_profitable_p.
29638 * config/i386/i386.c: Include ifcvt.h.
29639 (ix86_option_override_internal): Don't override
29640 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29641 (ix86_noce_conversion_profitable_p): New function.
29642 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29643 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29644 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29645 * doc/tm.texi: Regenerated.
29647 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29649 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29652 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29655 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29656 instructions for small multiply cores.
29658 2017-04-04 Jeff Law <law@redhat.com>
29660 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29662 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29663 that are not set by the loop over the elements.
29665 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29668 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29669 int mode, convert_modes it to mode as unsigned, otherwise use
29670 lowpart_subreg to mode rather than SImode.
29671 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29672 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29673 Use DImode instead of SImode for the shift count operand.
29674 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29677 2017-04-04 Richard Biener <rguenther@suse.de>
29679 PR middle-end/80281
29680 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29681 arithmetic done for the negate or the plus. Simplify.
29682 (A - (-B) -> A + B): Likewise.
29683 * fold-const.c (split_tree): Make sure to not negate pointers.
29685 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29687 PR rtl-optimization/60818
29688 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29689 a compare of comparisons with the thing compared if this results
29690 in a different machine mode.
29692 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29694 * alias.c (base_alias_check): Fix typo in comment.
29695 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29696 * cgraphunit.c (symbol_table::compile): Likewise.
29697 * collect2.c (maybe_run_lto_and_relink): Likewise.
29698 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
29699 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
29700 * config/avr/avr.c (avr_map_op_t): Likewise.
29701 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
29702 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
29703 * config/epiphany/epiphany.md (movcc): Likewise.
29704 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
29705 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
29707 * config/mips/mips.c (mips_save_restore_reg): Likewise.
29708 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
29709 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
29710 * config/sh/sh.c (sh_rtx_costs): Likewise.
29711 * fold-const.c (fold_truth_andor): Likewise.
29712 * genautomata.c (collapse_flag): Likewise.
29713 * gengtype.h (struct type::u::s): Likewise.
29714 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
29715 * input.c (FORMAT_AMOUNT): Likewise.
29716 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
29717 (known_aggs_to_agg_replacement_list): Likewise.
29718 * ipa-inline-analysis.c: Likewise.
29719 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
29720 * ipa-polymorphic-call.c
29721 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
29722 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
29723 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
29725 * modulo-sched.c (apply_reg_moves): Likewise.
29726 * omp-expand.c (build_omp_regions_1): Likewise.
29727 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
29728 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
29729 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
29730 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
29731 * value-prof.c: Likewise.
29732 * var-tracking.c (val_reset): Likewise.
29734 2017-04-03 Richard Biener <rguenther@suse.de>
29736 PR tree-optimization/80275
29737 * fold-const.c (split_address_to_core_and_offset): Handle
29740 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
29742 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
29743 descriptors is at least equal to that of functions.
29745 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29747 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
29749 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
29752 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
29753 (mov<IMOD4:mode>): New expander.
29754 (*mov<IMOD4:mode>_internal): New insn and split pattern.
29756 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
29758 PR rtl-optimization/79405
29759 * fwprop.c (propagations_left): New variable.
29760 (forward_propagate_into): Decrement it.
29761 (fwprop_init): Initialize it.
29762 (fw_prop): If the variable has reached zero, stop propagating.
29763 (fwprop_addr): Ditto.
29765 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29768 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
29769 a FUNCTION_DECL, pass it as decl instead of origin to
29772 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
29774 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
29777 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
29780 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
29781 TARGET_VSX_SMALL_INTEGER.
29783 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29785 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29786 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
29788 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29790 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
29791 extraction from odd-numbered MSA register.
29793 2017-03-31 Jakub Jelinek <jakub@redhat.com>
29795 PR middle-end/80173
29796 * expmed.c (store_bit_field_1): Don't attempt to create
29797 a word subreg out of hard registers wider than word if they
29798 have HARD_REGNO_NREGS of 1 for their mode.
29800 PR middle-end/80163
29801 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
29802 conversions to integer types wider than word and pointer.
29805 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
29806 (rtx_equal_for_cselib_p): Pass 0 to it.
29807 * cselib.c (cselib_hasher::equal): Likewise.
29808 (rtx_equal_for_cselib_1): Add depth argument. If depth
29809 is 128, don't look up VALUE locs and punt. Increment
29810 depth in recursive calls when walking VALUE locs.
29812 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
29814 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
29815 (make_gcov_file_name): Use the canonical path name for generating
29817 (read_line): Fix handling of files with ascii null bytes.
29819 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
29821 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
29822 to initialise a vector register instead
29823 of using a const_int.
29825 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29827 PR translation/80189
29828 * gimplify.c (omp_default_clause): Use %qs instead of %s in
29829 diagnostic messages.
29831 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
29834 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
29835 (dfp_diex_<mode>): Update mode of operand 1.
29836 * doc/extend.texi (dxex, dxexq): Document change to return type.
29837 (diex, diexq): Document change to argument type.
29839 2017-03-30 Martin Jambor <mjambor@suse.cz>
29842 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
29843 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
29844 it reflects the signature changes performed at the callee side.
29845 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
29846 to cgraph_build_function_type_skip_args.
29847 (build_function_decl_skip_args): Adjust call to the above function.
29849 2017-03-30 Jakub Jelinek <jakub@redhat.com>
29852 * config/i386/sse.md
29853 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
29854 register as dest whenever it is a MEM not rtx_equal_p to the
29855 corresponding dup operand, and when forcing into reg move the
29856 reg into the memory afterwards.
29857 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
29858 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
29859 for the force_reg mode.
29860 (avx512vl_vextractf128<mode>): Use register as dest either
29861 always when a MEM, or when it is a MEM not rtx_equal_p to the
29862 corresponding dup operand, or even not when it is a CONST_VECTOR
29863 depending on the mode and lo vs. hi.
29864 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
29866 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
29867 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
29868 Likewise. Require that operands[2] is even.
29869 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
29870 Remove extraneous parens. Require that operands[2] is a multiple
29872 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
29873 operands[0] is a MEM if <mask_applied>, the predicates/constraints
29874 disallow memory then.
29876 2017-03-30 Richard Biener <rguenther@suse.de>
29878 PR tree-optimization/77498
29879 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
29880 to non-constants over backedges.
29882 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29884 PR rtl-optimization/80233
29885 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29886 as last_combined_insn. Do not test for BARRIER_P separately.
29888 2017-03-29 Andreas Schwab <schwab@suse.de>
29891 * calls.c (prepare_call_address): Convert funexp to Pmode before
29892 copying to temp reg.
29894 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29896 PR tree-optimization/80158
29897 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29898 Handle possible future case of more than one alternate
29900 (replace_rhs_if_not_dup): Likewise.
29901 (replace_one_candidate): Likewise.
29903 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29905 PR rtl-optimization/80193
29906 * ira.c (ira): Do not check allocation for LRA.
29908 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29910 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29911 (nvptx_output_simt_exit): Declare.
29912 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29913 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29914 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29915 (nvptx_declare_function_name): ...here. Emit declaration of local
29916 memory space buffer for omp_simt_enter insn.
29917 (nvptx_output_unisimt_switch): New.
29918 (nvptx_output_softstack_switch): New.
29919 (nvptx_output_simt_enter): New.
29920 (nvptx_output_simt_exit): New.
29921 * config/nvptx/nvptx.h (struct machine_function): New fields
29922 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29923 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29924 (UNSPECV_SIMT_EXIT): Ditto.
29925 (omp_simt_enter_insn): New insn.
29926 (omp_simt_enter): New expansion.
29927 (omp_simt_exit): New insn.
29928 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29930 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29931 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29932 (expand_GOMP_SIMT_EXIT): New.
29933 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29934 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29935 (GOMP_SIMT_EXIT): Ditto.
29936 * target-insns.def (omp_simt_enter): New insn.
29937 (omp_simt_exit): Ditto.
29938 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29940 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29941 (lower_rec_input_clauses): Likewise.
29942 (lower_lastprivate_clauses): Handle SIMT privatization.
29944 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29945 (ompdevlow_adjust_simt_enter): New.
29946 (find_simtpriv_var_op): New.
29947 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29948 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29950 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29951 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29952 (copy_decl_for_dup_finish): Ditto.
29954 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29956 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29959 * config/i386/i386.c (ix86_option_override_internal): Always
29960 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29962 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29964 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29966 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29968 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29969 mark new edge's irreducible flag accordign to it.
29970 (vect_do_peeling): Check loop preheader edge's irreducible flag
29971 and pass it to function slpeel_add_loop_guard.
29973 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29975 PR tree-optimization/80218
29976 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29977 Update block frequencies and counts.
29979 2017-03-28 Richard Biener <rguenther@suse.de>
29981 PR tree-optimization/78644
29982 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29983 of a simplification result we may not use it at all.
29985 2017-03-28 Richard Biener <rguenther@suse.de>
29988 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29989 without arguments, generate default definition of a SSA name.
29991 2017-03-28 Richard Biener <rguenther@suse.de>
29993 PR middle-end/80222
29994 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29995 TYPE_REF_CAN_ALIAS_ALL references.
29996 * fold-const.c (fold_indirect_ref_1): Likewise.
29998 2017-03-28 Martin Liska <mliska@suse.cz>
30001 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30002 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30004 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30005 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30007 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30008 (EXTRA_SPECS): Define.
30009 (SUBTARGET_EXTRA_SPECS): Likewise.
30010 (SUBTARGET_CPP_SPEC): Likewise.
30011 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30012 SUBTARGET_EXTRA_SPECS.
30013 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30015 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30017 * config/arc/simdext.md (vst64_insn): Update pattern.
30018 (vld32wh_insn): Likewise.
30019 (vld32wl_insn): Likewise.
30020 (vld64_insn): Likewise.
30021 (vld32_insn): Likewise.
30023 2017-03-28 Marek Polacek <polacek@redhat.com>
30026 * fold-const.c (fold_comparison): Use protected_set_expr_location
30027 instead of SET_EXPR_LOCATION.
30029 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30031 * tree.c (add_expr): Avoid name lookup warning.
30033 2017-03-27 Jeff Law <law@redhat.com>
30035 PR tree-optimization/80216
30036 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30037 function name. Limit recursion depth.
30038 (record_temporary_equivalences): Corresponding changes.
30040 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30042 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30045 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30048 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30050 * cfgcleanup.c (reg_note_cfa_p): New array.
30051 (insns_have_identical_cfa_notes): New function.
30052 (old_insns_match_p): Don't cross-jump in between /f
30053 and non-/f instructions. If both i1 and i2 are frame related,
30054 verify all CFA notes, their order and content.
30056 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30059 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30060 HImode and SImode with zero extend to DImode to one insn.
30061 (bswap<mode>2_extenddi): Likewise.
30062 (bswapsi2_extenddi): Likewise.
30063 (bswaphi2_extendsi): Likewise.
30064 (bswaphi2): Combine bswap HImode and SImode into one insn.
30065 Separate memory insns from swapping register.
30066 (bswapsi2): Likewise.
30067 (bswap<mode>2): Likewise.
30068 (bswaphi2_internal): Delete, no longer used.
30069 (bswapsi2_internal): Likewise.
30070 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30071 store, and gpr<-gpr swap insns.
30072 (bswap<mode>2_store): Likewise.
30073 (bswaphi2_reg): Register only splitter, combine with the splitter.
30074 (bswaphi2 splitter): Likewise.
30075 (bswapsi2_reg): Likewise.
30076 (bswapsi2 splitter): Likewise.
30077 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30078 the insns into load, store, and register/register insns.
30079 (bswapdi2_ldbrx): Likewise.
30080 (bswapdi2_load): Likewise.
30081 (bswapdi2_store): Likewise.
30082 (bswapdi2_reg): Likewise.
30084 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30086 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30087 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30089 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30092 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30094 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30095 special handling for target option conflicts between dform
30096 options (-mpower9-dform, -mpower9-dform-vector,
30097 -mpower9-dform-scalar) and -mno-direct-move.
30099 2017-03-27 Richard Biener <rguenther@suse.de>
30101 PR tree-optimization/80181
30102 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30104 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30106 * config/arc/predicates.md (move_double_src_operand): Replace the
30107 call to move_double_src_operand with a call to address_operand.
30109 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30111 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30112 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30113 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30115 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30117 * config/arc/predicates.md (long_immediate_loadstore_operand):
30118 Consider scaled addresses cases.
30120 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30122 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30123 restored when in interrupt.
30124 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30125 doesn't have delay slot.
30127 2017-03-27 Richard Biener <rguenther@suse.de>
30130 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30131 inlined thunk clones.
30133 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30136 * asan.c (instrument_derefs): Copy over last operand from
30137 original COMPONENT_REF to the new COMPONENT_REF with
30138 DECL_BIT_FIELD_REPRESENTATIVE.
30139 * ubsan.c (instrument_object_size): Likewise.
30141 2017-03-27 Richard Biener <rguenther@suse.de>
30143 PR tree-optimization/80170
30144 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30145 sure DR/SCEV didnt fold in constants we do not see when looking
30146 at the reference base alignment.
30148 2017-03-27 Richard Biener <rguenther@suse.de>
30150 PR middle-end/80171
30151 * gimple-fold.c (fold_ctor_reference): Properly guard against
30152 NULL return value from canonicalize_constructor_val.
30154 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30157 * config/i386/i386.c (ix86_expand_builtin)
30158 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30159 flags reg setting and flags reg using instructions.
30160 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30161 clobbering instructions to zero extend op2.
30163 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30165 * doc/install.texi (Configuration) <--with-aix-soname>:
30166 Update link to AIX ld.
30168 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30170 PR rtl-optimization/80160
30171 PR rtl-optimization/80159
30172 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30173 reg_alternate_class into account.
30175 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30178 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30179 to consider in curr_insn_transform.
30181 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30183 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30184 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30185 and emit_mode_inner.
30187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30189 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30190 argument to the overloaded builtin variants. Use the new flag to
30191 deprecate certain builtin variants.
30192 * config/s390/s390-builtin-types.def: Add new builtin types.
30193 * config/s390/s390-builtins.h: Support new flags field for
30194 overloaded builtins.
30195 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30196 (s390_macro_to_expand): Enable vector float data type.
30197 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30198 builtins by incrementing the __VEC__ version number.
30199 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30201 (s390_resolve_overloaded_builtin): Emit error messages depending
30202 on the builtin flags.
30203 * config/s390/s390.c (s390_expand_builtin): Support additional
30204 flags argument. Change error message to match the messages
30205 emitted in s390-c.c.
30206 * config/s390/s390.md: New UNSPEC_* constants.
30207 (op_type): Add new instruction types.
30208 * config/s390/vecintrin.h: Add new builtins and test data class
30210 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30211 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30212 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30213 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30214 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30215 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30217 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30218 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30219 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30220 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30222 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30223 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30224 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30225 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30227 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30228 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30229 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30230 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30231 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30232 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30233 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30235 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30236 ("vec_scatter_element<V_HW_4:mode>_DI")
30237 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30238 ("vec_fpint<mode>", "vflls")
30239 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30240 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30241 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30242 ("*vec_cmphe<mode>_cc"): ... these.
30244 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30245 mode constant instead of magic value.
30247 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30249 * config/s390/s390.c (s390_expand_vec_compare): Support other
30250 vector floating point modes than just V2DF.
30251 (s390_expand_vcond): Likewise.
30252 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30253 (s390_cannot_change_mode_class): Prevent mode changes between TF
30254 and V1TF in vector registers.
30255 * config/s390/s390.md (DF, SF): New mode attributes.
30256 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30257 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30258 SFmode support for VRs.
30259 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30261 (VFT, VF_HW): New mode iterators.
30262 (vw, sdx): New mode attributes.
30263 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30264 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30265 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30266 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30267 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30268 also the new vector floating point modes. Renaming to ...
30270 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30271 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30272 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30273 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30274 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30275 ("vec_unordered<mode>"): ... these.
30277 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30278 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30279 ("*vec_extendv2df"): New insn definitions.
30281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30283 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30284 ("mulditi3_2", "*muldi3_sign"): New patterns.
30285 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30286 rename the pattern definition.
30288 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30290 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30292 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30296 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30297 instruction if possible.
30298 * config/s390/vector.md (vec_halfnumelts): New mode
30300 ("*vec_vllezlf<mode>"): New pattern.
30302 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30304 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30305 ("popcountv4si2", "popcountv2di2"): Rename to ...
30306 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30307 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30309 ("popcount<mode>2_vxe"): New pattern.
30311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30313 * common/config/s390/s390-common.c (processor_flags_table): Add
30315 * config.gcc: Add arch12.
30316 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30317 Default to arch12 for unknown CPU model numbers.
30318 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30319 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30320 PROCESSOR_max sanity check.
30321 * config/s390/s390-opts.h (enum processor_type): Add
30323 * config/s390/s390.c (processor_table): Add arch12.
30324 (s390_expand_builtin): Add check for B_VXE flag.
30325 (s390_issue_rate): Add PROCESSOR_ARCH12.
30326 (s390_get_sched_attrmask): Likewise.
30327 (s390_get_unit_mask): Likewise.
30328 (s390_sched_score): Enable z13 scheduling for arch12.
30329 (s390_sched_reorder): Likewise.
30330 (s390_sched_variable_issue): Likewise.
30331 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30333 (s390_tune_attr): Use z13 scheduling also for arch12.
30334 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30335 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30336 (TARGET_VXE_P): New macros.
30337 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30338 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30339 * config/s390/s390.opt: Add arch12 as processor_type.
30341 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30343 * config/s390/s390.md
30344 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30345 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30346 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30348 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30349 Rename expanders to ...
30351 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30352 ("fixuns_truncdddi2_emu"): ... these.
30354 ("fixuns_trunc<mode>si2_emu"): New expander.
30356 ("*fixuns_truncdfdi2_z13"): Rename to ...
30357 ("*fixuns_truncdfdi2_vx"): ... this.
30359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30361 * config/s390/2964.md: Remove the single element vector compare
30362 instructions which are no longer used.
30363 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30365 (s390_canonicalize_comparison): Remove handling of DFmode
30367 (s390_expand_vec_compare_scalar): Remove function.
30368 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30369 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30371 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30373 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30375 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30377 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30378 will anyway by matched by mov<mode>_64dfp.
30380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30382 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30383 vlef/vstef. Add missing operand to vleif.
30385 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30387 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30388 pair for all vector types with 64 bit elements.
30389 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30390 * config/s390/vector.md (V_HW_64): ... here.
30391 (V_128_NOSINGLE): New mode iterator.
30392 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30393 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30394 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30395 ("*vec_load_pairv2di"): Change to ...
30396 ("*vec_load_pair<mode>"): ... this one.
30398 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30400 * config/s390/constraints.md: Add comments.
30401 (jKK): Reject element sizes > 8 bytes.
30402 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30404 * config/s390/s390.md: Add the s_operand checks formerly in
30405 s390_split_ok_p to various splitters where they are still
30407 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30408 for 128 bit vectors. Plus two splitters.
30410 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30412 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30418 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30419 error if the boundary argument is not constant.
30421 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30423 PR rtl-optimization/80112
30424 * loop-doloop.c (doloop_condition_get): Don't check condition
30425 if cmp isn't SET with IF_THEN_ELSE src.
30427 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30429 PR tree-optimization/80158
30430 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30431 replacing a candidate statement, also replace it for the
30432 candidate's alternate interpretation.
30433 (replace_rhs_if_not_dup): Likewise.
30434 (replace_one_candidate): Likewise.
30436 2017-03-24 Richard Biener <rguenther@suse.de>
30438 PR tree-optimization/80167
30439 * graphite-isl-ast-to-gimple.c
30440 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30442 (translate_isl_ast_to_gimple::get_rename): Likewise.
30444 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30446 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30447 handling of certain combinations of target options, including the
30448 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30449 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30451 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30454 * config/arm/arm.md (*load_multiple): Add reload_completed to
30455 matching condition.
30457 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30458 Richard Biener <rguenth@suse.de>
30460 PR tree-optimization/79908
30461 PR tree-optimization/80136
30462 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30463 been cast away, gimplify_and_add suffices.
30465 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30467 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30469 2017-03-23 Richard Biener <rguenther@suse.de>
30471 PR tree-optimization/80032
30472 * gimplify.c (gimple_push_cleanup): Forced unconditional
30473 cleanups still have to go to the conditional_cleanups
30476 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30478 PR tree-optimization/80072
30479 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30481 (next_operand_entry_id): Change type to unsigned int.
30482 (sort_by_operand_rank): Make sure to return the right return value
30483 even if unsigned fields are bigger than INT_MAX.
30484 (struct oecount): Change cnt and id type to unsigned int.
30485 (oecount_hasher::equal): Formatting fix.
30486 (oecount_cmp): Make sure to return the right return value
30487 even if unsigned fields are bigger than INT_MAX.
30488 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30491 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30492 TREE_READONLY on result if writing it more than once.
30495 * doc/invoke.texi (-fsanitize=thread): Document that with
30496 -fnon-call-exceptions atomics are not able to throw
30500 * tsan.c: Include tree-eh.h.
30501 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30502 maybe_clean_or_replace_eh_stmt where needed.
30503 (instrument_memory_accesses): Add cfg_changed argument.
30504 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30505 if it returned true.
30506 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30508 PR rtl-optimization/63191
30509 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30510 wrapper function, moved the whole old content into ...
30511 (ix86_delegitimize_address_1): ... this. New inline function.
30512 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30513 true as last argument instead of ix86_delegitimize_address.
30515 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30517 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30518 cortexa57_branch_cost.
30520 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30522 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30524 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30527 * doc/md.texi (Constraints): Document wA constraint.
30528 * config/rs6000/constraints.md (wA): New.
30529 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30530 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30531 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30532 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30534 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30537 * gimplify.c (is_oacc_declared): New function.
30538 (oacc_default_clause): Use it to set default flags for acc declared
30539 variables inside parallel regions.
30540 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30541 declared variables.
30542 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30543 declare attribute to any decl as necessary.
30545 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30548 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30549 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30550 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30551 (arm_arch_lpae): This.
30552 * config/arm/arm.c (arm_arch7ve): Rename into ...
30553 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30554 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30557 2017-03-22 Martin Liska <mliska@suse.cz>
30560 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30561 error message instead of an ICE.
30563 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30565 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30567 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30569 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30571 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30574 2017-03-21 Martin Sebor <msebor@redhat.com>
30576 * doc/extend.texi: Use "cannot" instead of "can't."
30577 * doc/hostconfig.texi: Same.
30578 * doc/install.texi: Same.
30579 * doc/invoke.texi: Same.
30580 * doc/loop.texi: Same.
30581 * doc/md.texi: Same.
30582 * doc/objc.texi: Same.
30583 * doc/rtl.texi: Same.
30584 * doc/tm.texi: Same.
30585 * doc/tm.texi.in: Same.
30586 * doc/trouble.texi: Same.
30588 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30591 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30592 (collect_checksum_attributes): Set it.
30593 (die_checksum_ordered): Use it.
30595 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30597 PR tree-optimization/79908
30598 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30599 change: For a VA_ARG whose LHS has been cast away, use
30600 force_gimple_operand to construct the side effects.
30602 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30604 PR translation/80001
30605 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30606 more amenable to translation.
30607 (oacc_loop_auto_partitions): Likewise.
30609 2017-03-21 Marek Polacek <polacek@redhat.com>
30610 Martin Sebor <msebor@redhat.com>
30612 PR tree-optimization/80109
30613 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30614 on INTEGRAL_TYPE_P.
30616 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30617 Segher Boessenkool <segher@kernel.crashing.org>
30620 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30621 check reg_used_between_p between insn and one of succ or succ2
30622 depending on if succ is artificial insn not inserted into insn
30625 2017-03-21 Martin Liska <mliska@suse.cz>
30627 PR gcov-profile/80081
30628 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30629 * doc/gcc.texi: Include gcov-dump stuff.
30630 * doc/gcov-dump.texi: New file.
30632 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30634 PR rtl-optimization/79150
30635 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30636 conditional jump, if the jump is the last insn of the loop.
30638 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30639 Richard Biener <rguenth@suse.de>
30641 PR tree-optimization/79908
30642 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30643 been cast away, use force_gimple_operand to construct the side
30646 2017-03-21 Martin Liska <mliska@suse.cz>
30648 PR libfortran/79956
30649 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30652 2017-03-21 Brad Spengler <spender@grsecurity.net>
30655 * plugin.c (htab_hash_plugin): New function.
30656 (add_new_plugin): Use it and adjust.
30657 (parse_plugin_arg_opt): Adjust.
30658 (init_one_plugin): Likewise.
30660 2017-03-21 Richard Biener <rguenther@suse.de>
30662 PR tree-optimization/80032
30663 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30664 if set force the cleanup to happen unconditionally.
30665 (gimplify_target_expr): Push inserted clobbers with force_uncond
30666 to avoid them being removed by control-dependent DCE.
30668 2017-03-21 Richard Biener <rguenther@suse.de>
30670 PR tree-optimization/80122
30671 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30672 va_arg_pack_len when the inlined call stmt requires pack
30674 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30676 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30679 * tsan.c (instrument_builtin_call): If the memory model argument
30680 is not a constant, assume it is valid.
30683 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30686 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30688 PR rtl-optimization/79910
30689 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30690 if its dest is used by an insn before I2 (other than the combined
30691 insns themselves, which are properly handled already).
30693 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30696 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30698 * combine.c (record_used_regs): New static function.
30699 (try_combine): Handle situations where there is an additional
30700 instruction between I2 and I3 which needs to have a LOG_LINK
30704 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30706 * combine.c (try_combine): Delete redundant i1 test. Call
30707 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30709 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30712 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
30713 alternatives 13/14.
30715 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30717 PR tree-optimization/80054
30718 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
30719 the optimization if a PHI or any of its arguments is not dominated
30720 by the candidate's basis. Use gphi* rather than gimple* as
30722 (replace_profitable_candidates): Clean up a gimple* variable that
30723 should be a gphi* variable.
30725 2017-03-20 Martin Sebor <msebor@redhat.com>
30728 * doc/extend.texi (attribute constructor): Document present limitation.
30730 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
30733 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
30734 __POWER9_VECTOR__ #ifdef control, change template definition to
30735 use Power9-specific built-in function.
30736 (vec_any_eq): Likewise.
30737 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
30738 to control outcomes from this test.
30739 (vector_ae_<mode>p): For VEC_F modes, likewise.
30741 2017-03-20 Ian Lance Taylor <iant@google.com>
30743 * config/i386/i386.c (ix86_function_regparm): Save an extra
30744 register for -fsplit-stack with DECL_STATIC_CHAIN.
30746 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30749 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
30750 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
30752 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
30754 * config/riscv/riscv.c (riscv_print_operand): Use "fence
30756 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
30759 2017-03-20 Marek Polacek <polacek@redhat.com>
30762 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
30764 2017-03-20 Richard Biener <rguenther@suse.de>
30766 PR tree-optimization/80113
30767 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
30768 allocate extra SSA name for PHI def.
30769 (add_close_phis_to_outer_loops): Likewise.
30770 (add_close_phis_to_merge_points): Likewise.
30771 (copy_loop_close_phi_args): Likewise.
30772 (copy_cond_phi_nodes): Likewise.
30774 2017-03-20 Martin Liska <mliska@suse.cz>
30776 PR middle-end/79753
30777 * tree-chkp.c (chkp_build_returned_bound): Do not build
30778 returned bounds for a LHS that's not a BOUNDED_P type.
30780 2017-03-20 Martin Liska <mliska@suse.cz>
30784 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
30785 COMPLEX_CST and VECTOR_CST.
30787 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30790 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
30791 target operand. A new splitter adds the clobber statement in case
30792 the target operand is dead anyway.
30794 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
30796 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
30797 to age-old versions of binutils and glibc.
30799 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
30801 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
30803 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30805 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
30807 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
30809 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
30810 requirement for binutils 2.13.
30812 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
30814 * combine.c (try_combine): Delete redundant i1 test. Call
30815 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
30817 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
30819 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
30820 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
30822 <riscv64-*-elf>: Re-arrange section
30823 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
30824 <riscv32-*-linux>: Likewise.
30825 <riscv64-*-elf>: Likewise
30826 <riscv64-*-linux>: Likewise.
30828 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
30831 * aarch64.opt(verbose-cost-dump): Fix typo.
30833 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
30836 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
30837 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
30839 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
30841 * reload.c (find_reloads): When reloading a nonoffsettable address,
30842 use RELOAD_OTHER for it and its address reloads.
30844 PR rtl-optimization/79910
30845 * combine.c (record_used_regs): New static function.
30846 (try_combine): Handle situations where there is an additional
30847 instruction between I2 and I3 which needs to have a LOG_LINK
30850 2017-03-17 Jeff Law <law@redhat.com>
30852 PR tree-optimization/71437
30853 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
30854 conditional in the hash table first.
30855 (vrp_dom_walker::before_dom_children): Extract condition from
30856 ASSERT_EXPR. Record condition, its inverion and any implied
30857 conditions as well.
30859 2017-03-17 Marek Polacek <polacek@redhat.com>
30860 Markus Trippelsdorf <markus@trippelsdorf.de>
30862 PR tree-optimization/80079
30863 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
30866 2017-03-17 Richard Biener <rguenther@suse.de>
30868 PR middle-end/80075
30869 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
30870 Properly verify the LHS before the RHS possibly claims to be
30872 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
30875 2017-03-17 Martin Jambor <mjambor@suse.cz>
30877 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
30878 (List of -O2 options): Likewise.
30879 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30882 2017-03-17 Tom de Vries <tom@codesourcery.com>
30884 * gcov-dump.c (print_usage): Print bug_report_url.
30886 2017-03-17 Richard Biener <rguenther@suse.de>
30888 PR middle-end/80050
30889 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30890 (parser::peek): Likewise.
30892 2017-03-17 Richard Biener <rguenther@suse.de>
30894 PR tree-optimization/80048
30895 * sese.c (free_sese_info): Properly release rename_map and
30896 copied_bb_map elements.
30898 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30900 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30901 Add linked-list forward and backlinks. Insert on
30902 construction, remove on destruction.
30903 (class pass_store_merging): Add m_stores_head field.
30904 (pass_store_merging::terminate_and_process_all_chains):
30905 Iterate over m_stores_head list.
30906 (pass_store_merging::terminate_all_aliasing_chains):
30908 (pass_store_merging::execute): Check for debug stmts first.
30909 Push new chains onto the m_stores_head stack.
30911 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30914 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30915 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30916 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30918 2017-03-16 Jeff Law <law@redhat.com>
30920 PR tree-optimization/71437
30921 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30922 member function. Implementation moved into after_dom_children
30923 member function and into the threader's thread_outgoing_edges
30925 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30926 some code into new thread_outgoing_edges.
30927 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30928 definition. Simplify marker handling (do it here). Assume we always
30929 have the available expression and the const/copies tables.
30930 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30932 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30933 * tree-vrp.c (equiv_stack): No longer file scoped.
30934 (vrp_dom_walker): New class.
30935 (vrp_dom_walker::before_dom_children): New member function.
30936 (vrp_dom_walker::after_dom_children): Likewise.
30937 (identify_jump_threads): Setup domwalker. Use it rather than
30938 walking edges in a random order by hand. Simplify setup/finalization.
30939 (finalize_jump_threads): Remove.
30940 (vrp_finalize): Do not call identify_jump_threads here.
30941 (execute_vrp): Do it here instead and call thread_through_all_blocks
30944 PR tree-optimization/71437
30945 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30947 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30949 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30950 (dom_opt_dom_walker::thread_across_edge): Remove
30951 handle_dominating_asserts argument. All callers changed.
30952 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30953 changes. Remove calls to lhs_of_dominating_assert. Other
30954 uses of handle_dominating_asserts turn into unconditional code
30955 (simplify_control_stmt_condition_1): Likewise.
30956 (simplify_control_stmt_condition): Likewise.
30957 (thread_through_normal_block, thread_across_edge): Likewise.
30958 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30959 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30960 object if it is not an SSA_NAME.
30961 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30962 before calling into the VRP specific simplifiers.
30963 (identify_jump_threads): Remove handle_dominating_asserts
30966 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30969 * tree-diagnostic.c (default_tree_printer): No longer static.
30970 * tree-diagnostic.h (default_tree_printer): New prototype.
30972 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30974 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30975 Change ins into fmov.
30977 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30979 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30980 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30981 Use h_con constraint for operand 1.
30982 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30983 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30985 2017-03-15 Jeff Law <law@redhat.com>
30987 PR tree-optimization/71437
30988 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30989 (record_temporary_equivalences): Use it.
30991 PR tree-optimization/71437
30992 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30993 tree-ssa-scopedtables.
30994 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30995 (record_conditions, record_cond, vuse_eq): Likewise.
30996 (record_edge_info): Adjust to API tweak of record_conditions.
30997 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30998 (record_temporary_equivalences, optimize_stmt): Likewise.
30999 (eliminate_redundant_computations): Likewise.
31000 (record_equivalences_from_stmt): Likewise.
31001 * tree-ssa-scopedtables.c: Include options.h and params.h.
31002 (vuse_eq): New function, moved from tree-ssa-dom.c
31003 (build_and_record_new_cond): Likewise.
31004 (record_conditions): Likewise. Accept vector of conditions rather
31005 than edge_equivalence structure for first argument.
31006 for the first argument.
31007 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31008 from tree-ssa-dom.c.
31009 (avail_exprs_stack::record_cond): Likewise.
31010 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31011 from tree-ssa-dom.c.
31012 (avail_exprs_stack): Add new member functions lookup_avail_expr
31014 (record_conditions): Declare.
31016 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31019 * lra-constraints.c (process_alt_operands): Increase reject for
31020 reloading an input/output operand.
31022 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31025 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31026 insns to convert from signed/unsigned char/short to IEEE 128-bit
31028 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31030 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31033 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31034 subreg of inner mode for values already in registers.
31036 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31038 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31039 iteration reg is used after the loop.
31041 2017-03-14 Martin Sebor <msebor@redhat.com>
31043 PR tree-optimization/79800
31044 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31045 precision in negative-positive range.
31046 (format_floating): Call non-const overload with adjusted precision.
31048 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31051 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31054 2017-03-14 Martin Sebor <msebor@redhat.com>
31056 PR middle-end/80020
31057 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31058 * builtins.def (aligned_alloc): Use it.
31061 * Makefile.in (GTFILES): Add calls.c.
31062 * calls.c: Include "gt-calls.h".
31064 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31066 PR rtl-optimization/79728
31067 * regs.h (struct target_regs): New field
31068 x_contains_allocatable_regs_of_mode.
31069 (contains_allocatable_regs_of_mode): New macro.
31070 * reginfo.c (init_reg_sets_1): Initialize it, and change
31071 contains_reg_of_mode so it includes global regs as well.
31072 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31073 rather than contains_regs_of_mode.
31075 2017-03-14 Martin Liska <mliska@suse.cz>
31077 * doc/invoke.texi: Document options that can't be combined with
31078 -fcheck-pointer-bounds.
31080 2017-03-14 Martin Liska <mliska@suse.cz>
31082 PR middle-end/79831
31083 * doc/invoke.texi (-Wchkp): Document the option.
31085 2017-03-14 Martin Liska <mliska@suse.cz>
31087 * Makefile.in: Install gcov-dump.
31089 2017-03-14 Martin Liska <mliska@suse.cz>
31091 * multiple_target.c (expand_target_clones): Bail out for
31092 an invalid attribute.
31094 2017-03-14 Richard Biener <rguenther@suse.de>
31096 * alias.c (struct alias_set_entry): Pack properly.
31097 * cfgloop.h (struct loop): Likewise.
31098 * cse.c (struct set): Likewise.
31099 * ipa-utils.c (struct searchc_env): Likewise.
31100 * loop-invariant.c (struct invariant): Likewise.
31101 * lra-remat.c (struct cand): Likewise.
31102 * recog.c (struct change_t): Likewise.
31103 * rtl.h (struct address_info): Likewise.
31104 * symbol-summary.h (function_summary): Likewise.
31105 * tree-loop-distribution.c (struct partition): Likewise.
31106 * tree-object-size.c (struct object_size_info): Likewise.
31107 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31108 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31109 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31110 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31111 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31112 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31113 (struct _stmt_vec_info): Likewise.
31115 2017-03-14 Martin Liska <mliska@suse.cz>
31118 * multiple_target.c (create_dispatcher_calls): Check that
31119 a target can create a function dispatcher.
31121 2017-03-14 Martin Liska <mliska@suse.cz>
31124 * multiple_target.c (expand_target_clones): Drop local.local
31125 flag for default implementation.
31127 2017-03-14 Richard Biener <rguenther@suse.de>
31129 PR tree-optimization/80030
31130 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31132 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31134 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31135 gcc_fallthrough() instead of __attribute__((fallthrough));
31137 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31139 * doc/gcc.texi: Remove "up" link to (DIR).
31140 * doc/gccint.texi: Ditto.
31142 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31144 * doc/install.texi (Specific) <avr>: Remove reference to
31147 2017-03-13 Jeff Law <law@redhat.com>
31149 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31150 attribute rather than comments.
31152 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31153 match_scratch operand is highest.
31155 2017-03-13 Martin Liska <mliska@suse.cz>
31157 PR middle-end/78339
31158 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31159 is a CHKP clone, use original declaration.
31161 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31163 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31164 (arc_conditional_register_usage): Use a different allocation order
31165 when optimizing for size.
31166 * common/config/arc/arc-common.c (arc_option_optimization_table):
31167 Section anchors default on when optimizing for size.
31169 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31171 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31173 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31175 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31176 * config/arc/arc.md (cpu_facility): Add cd variant.
31177 (*movqi_insn): Add code density variant.
31178 (*movhi_insn): Likewise.
31179 (*movqi_insn): Likewise.
31180 (*addsi3_mixed): Likewise.
31181 (subsi3_insn): Likewise.
31183 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31185 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31187 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31189 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31190 expressions with MINUS and UNARY ops.
31192 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31195 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31197 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31198 between vec_select and vector argument.
31199 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31200 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31201 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31202 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31203 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31204 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31206 2017-03-13 Richard Biener <rguenther@suse.de>
31209 * params.def (vect-max-peeling-for-alignment): Fix typo.
31211 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31213 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31214 issue that only occurred with binutils below 2.18.
31216 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31218 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31219 refer to binutils 2.11/2.12 minimum.
31221 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31223 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31224 ftp.kernel.org and simplify binutils requirement.
31226 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31228 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31230 (Optimize Options): Ditto. Also remove redundancy.
31232 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31234 PR translation/79848
31235 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31237 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31238 to G_ to avoid double translation.
31240 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31242 PR translation/79923
31243 * auto-profile.c (get_combined_location): Convert leading
31244 character of diagnostics to lower case and remove trailing period.
31245 (read_profile): Likewise for various diagnostics.
31246 * config/arm/arm.c (arm_option_override): Remove trailing period
31247 from various diagnostics.
31248 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31249 (msp430_expand_delay_cycles): Likewise.
31251 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31254 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31255 full command-line argument, rather than just "str".
31256 (aarch64_validate_march): Likewise.
31257 (aarch64_validate_mtune): Likewise.
31259 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31261 PR rtl-optimization/78911
31262 * lra-assigns.c (must_not_spill_p): New function.
31263 (spill_for): Use it.
31265 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31267 PR tree-optimization/79981
31268 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31269 ATOMIC_COMPARE_EXCHANGE ifn result.
31270 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31271 IFN_ATOMIC_COMPARE_EXCHANGE.
31273 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31276 * opts.c (parse_sanitizer_options): Add missing question mark to
31277 "did you mean" message.
31279 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31281 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31283 (VMULEUH_UNS): Likewise.
31284 (VMULOUB_UNS): Likewise.
31285 (VMULOUH_UNS): Likewise.
31286 * config/rs6000/rs6000.c (builtin_function_type): Remove
31287 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31289 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31292 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31293 x with result of extra_parsing_for_operand_code_0.
31294 (function_reader::extra_parsing_for_operand_code_0): Convert
31295 return type from void to rtx, returning x. When reading
31296 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31297 larger size containing struct block_symbol.
31299 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31302 -mfloat128-hardware without -m64.
31304 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31307 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31308 entries to the case statement that marks unsigned arguments to
31309 overloaded functions.
31311 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31314 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31316 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31319 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31320 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31322 2017-03-10 Martin Liska <mliska@suse.cz>
31326 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31327 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31330 2017-03-10 Olivier Hainque <hainque@adacore.com>
31332 * tree-switch-conversion (array_value_type): Start by resetting
31333 candidate type to it's main variant.
31335 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31337 PR rtl-optimization/79909
31338 * combine.c (try_combine): Use simplify_replace_rtx on individual
31339 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31340 of the whole CALL_INSN_FUNCTION_USAGE.
31342 PR tree-optimization/79972
31343 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31344 get_range_info on SSA_NAMEs. Formatting fixes.
31346 2017-03-10 Richard Biener <rguenther@suse.de>
31347 Jakub Jelinek <jakub@redhat.com>
31349 PR tree-optimization/77975
31350 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31351 edge to be constant.
31352 (get_val_for): For constant x return it. Formatting fix.
31353 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31354 would use the same bases as the current one.
31356 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31358 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31359 instead of vec_select for V1TImode.
31360 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31362 (VSX_LE_128): Add V1TI to this mode iterator.
31363 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31364 (*vsx_le_perm_store_<mode>): Likewise.
31365 (pre-reload splitter for VSX stores): Likewise.
31366 (post-reload splitter for VSX stores): Likewise.
31367 (*vsx_xxpermdi2_le_<mode>): Likewise.
31368 (*vsx_lxvd2x2_le_<mode>): Likewise.
31369 (*vsx_stxvd2x2_le_<mode>): Likewise.
31371 2017-03-09 Michael Eager <eager@eagercon.com>
31373 Correct failures with --enable-checking=yes,rtl.
31375 * config/microblaze/microblaze.c (microblaze_expand_shift):
31376 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31377 test for const0_rtx.
31378 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31379 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31381 2017-03-09 Richard Biener <rguenther@suse.de>
31383 PR tree-optimization/79977
31384 * graphite-scop-detection.c (scop_detection::merge_sese):
31385 Handle the case of extra exits to blocks dominating the entry.
31387 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31389 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31392 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31394 PR rtl-optimization/79949
31395 * lra-constraints.c (process_alt_operands): Check memory when
31396 trying to predict a cycle. Print about the overall increase.
31398 2017-03-09 Richard Biener <rguenther@suse.de>
31400 PR middle-end/79971
31401 * gimple-expr.c (useless_type_conversion_p): Preserve
31402 TYPE_SATURATING for fixed-point types.
31404 2017-03-09 Richard Biener <rguenther@suse.de>
31407 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31408 alignment of BLKmode params.
31410 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31413 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31414 (VALL_NO_V2Q): Likewise.
31416 * config/aarch64/aarch64-simd.md
31417 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31419 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31420 VALL_NO_V2Q mode iterator.
31421 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31423 2017-03-09 Martin Liska <mliska@suse.cz>
31425 PR tree-optimization/79631
31426 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31427 tree_int_cst_sign_bit just for INTEGER constants.
31429 2017-03-09 Martin Liska <mliska@suse.cz>
31433 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31436 2017-03-09 Marek Polacek <polacek@redhat.com>
31439 * tree.c (inchash::add_expr): Handle TREE_VEC.
31441 2017-03-09 Martin Liska <mliska@suse.cz>
31444 (chkp_narrow_size_and_offset): New function.
31445 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31446 (void chkp_parse_bit_field_ref): New function.
31447 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31448 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31450 2017-03-09 Martin Liska <mliska@suse.cz>
31453 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31454 (chkp_find_bounds_1): Remove gcc_unreachable.
31456 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31459 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31460 BUILT_IN_SYNC*, determine the access type from the size suffix and
31461 always build a MEM_REF with that type. Handle forgotten
31462 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31465 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31466 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31467 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31468 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31469 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31470 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31471 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31472 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31473 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31474 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31475 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31476 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31477 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31478 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31479 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31480 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31481 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31482 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31483 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31484 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31485 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31486 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31487 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31488 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31489 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31490 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31491 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31492 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31493 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31494 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31495 definitions outside of __OPTIMIZE__ guarded section.
31498 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31499 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31500 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31501 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31504 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31506 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31507 ("vfenez<mode>"): Add missing constraints.
31509 2017-03-08 Martin Sebor <msebor@redhat.com>
31512 * config/nds32/nds32.c (nds32_option_override):
31513 Fix misspelled diagnostic.
31515 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31518 * gimplify.c (gimplify_omp_for): Replace index var in outer
31519 taskloop statement with an artificial variable and add
31520 OMP_CLAUSE_PRIVATE clause for it.
31522 2017-03-08 Richard Biener <rguenther@suse.de>
31524 PR tree-optimization/79955
31525 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31526 for accesses that are completely outside of the variable.
31528 2017-03-08 Andrew Haley <aph@redhat.com>
31530 PR tree-optimization/79943
31531 * tree-ssa-loop-split.c (compute_new_first_bound): When
31532 calculating the new upper bound, (END-BEG) should be added, not
31535 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31537 * config/avr/avr.md (setmemhi): Make sure match_dup
31538 operand number comes before match_scratch.
31540 2017-03-08 Richard Biener <rguenther@suse.de>
31542 PR tree-optimization/79920
31543 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31544 with ncopies == 1 to ...
31545 (vect_transform_slp_perm_load): ... here. Properly compute
31546 all element loads by iterating VF times over the group. Do
31547 not handle ncopies (computed in a broken way) in
31548 vect_create_mask_and_perm.
31550 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31553 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31554 is a uniform vector, use uniform_vector_p return value instead of
31555 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31557 2017-03-07 Marek Polacek <polacek@redhat.com>
31559 PR middle-end/79809
31560 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31561 (alloca_call_type): Likewise.
31563 2017-03-07 Martin Liska <mliska@suse.cz>
31565 * gcov.c (process_args): Put comment to correct location.
31567 2017-03-07 Martin Liska <mliska@suse.cz>
31569 PR middle-end/68270
31570 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31571 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31572 (chkp_narrow_bounds_for_field): Likewise.
31573 (chkp_parse_array_and_component_ref): Pass one more argument to
31576 2017-03-07 Richard Biener <rguenther@suse.de>
31578 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31581 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31583 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31584 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31586 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31589 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31590 to end of description.
31591 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31593 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31595 PR rtl-optimization/79901
31596 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31598 (*avx512f_<code><mode>3<mask_name>): ... this.
31599 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31600 iterator instead of VI8_AVX2_AVX512BW.
31602 PR rtl-optimization/79901
31603 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31604 min/max expander, expand it using expand_vec_cond_expr.
31607 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31610 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31613 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31614 to void * for PCH reasons.
31615 * dwarf2out.c (output_loc_operands, output_die): Cast
31616 v.val_vec.array to unsigned char *.
31618 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31621 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31624 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31626 PR rtl-optimization/79571
31627 * lra-constraints.c (process_alt_operands): Calculate static
31628 reject and subtract it from overall when only addresses will be
31631 2017-03-06 Julia Koval <julia.koval@intel.com>
31634 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31635 incoming stack boundary to 128 for 64-bit targets.
31637 2017-03-06 Richard Biener <rguenther@suse.de>
31639 PR tree-optimization/79894
31640 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31641 to NULL after folding it.
31643 2017-03-06 Richard Biener <rguenther@suse.de>
31645 PR tree-optimization/79824
31646 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31647 check disabling peeling for gaps.
31649 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31651 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31652 attributes): Document gettimeofday.
31654 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31656 * config/s390/s390.c (s390_option_override_internal): Set
31657 PARAM_MIN_VECT_LOOP_BOUND
31659 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31661 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31662 * config/s390/s390.md: Likewise.
31664 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31667 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31668 (<avx2_avx512>_perm<mode>): Rename to ...
31669 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31671 (<avx512>_perm<mode>_mask): Rename to ...
31672 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31674 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31675 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31676 instead of VI8F_256_512.
31677 (avx512f_perm<mode>): New define_expand.
31678 (avx512f_perm<mode>_mask): Likewise.
31679 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31680 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31682 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31684 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31685 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31687 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31689 2017-03-06 Martin Liska <mliska@suse.cz>
31692 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31693 when having a SSA NAME w/o VAR_DECL assigned to it.
31695 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31697 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
31698 msa_dpsub_<su>_d): Fix MODE for vec_select.
31700 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31702 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
31704 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
31706 2017-03-06 Richard Biener <rguenther@suse.de>
31708 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
31709 * plugin.c (register_plugin_info): Likewise.
31710 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
31712 2017-03-05 Jakub Jelinek <jakub@redhat.com>
31714 * config/i386/sse.md (sse_storehps, sse_storelps,
31715 avx_<castmode><avxsizesuffix>_<castmode>,
31716 avx512f_<castmode><avxsizesuffix>_<castmode>,
31717 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
31718 in condition that at least one operand is not a MEM.
31720 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31722 PR middle-end/79805
31723 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
31724 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
31726 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
31727 gimple_call_nothrow_p flag based on whether original builtin can throw.
31728 If it can, emit following stmts on the fallthrough edge.
31729 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
31730 don't create new bb if inserting just debug stmts on the edge, try to
31731 insert them on the fallthru bb or just reset debug stmts.
31733 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
31736 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
31737 restore recog_data (including the operand rtxes inside it) around
31738 the call to get_insn_template.
31740 2017-03-03 Martin Sebor <msebor@redhat.com>
31742 PR tree-optimization/79699
31743 * context.c (context::~context): Free MPFR caches to avoid
31744 a memory leak on program exit.
31746 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31748 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
31749 Use wide_int::ulow () instead of .elt (0).
31751 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31753 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
31754 (*pushxf): Limit oF constraint to 32bit targets and add oC
31755 constraint for 64bit targets.
31756 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
31757 (*pushdf): Change rmF constraint to rmC.
31759 2017-03-03 Martin Liska <mliska@suse.cz>
31761 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
31762 Remove unused variable.
31764 2017-03-03 Jakub Jelinek <jakub@redhat.com>
31767 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
31768 is a memory operand, increase num_memory.
31769 (ix86_expand_args_builtin): Likewise.
31771 2017-03-03 Jan Hubicka <jh@suse.cz>
31774 * ipa-devirt.c (maybe_record_node): Properly handle
31775 __cxa_pure_virtual visibility.
31777 2017-03-03 Martin Liska <mliska@suse.cz>
31779 PR tree-optimization/79803
31780 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
31782 (pass_loop_prefetch::execute): Disabled optimization if an
31783 assumption about L1 cache size is not met.
31785 2017-03-03 Martin Liska <mliska@suse.cz>
31787 PR rtl-optimization/79574
31788 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
31789 (hash_scan_set): Likewise.
31790 (dump_hash_table): Likewise.
31791 (hoist_code): Likewise.
31793 2017-03-03 Richard Biener <rguenther@suse.de>
31795 * fixed-value.c (fixed_from_string): Restore use of elt (1)
31796 in place of uhigh ().
31797 (fixed_convert_from_real): Likewise.
31799 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
31802 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
31804 2017-03-03 Richard Biener <rguenther@suse.de>
31806 PR middle-end/79818
31807 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
31808 TYPE_OVERFLOW_UNDEFINED check.
31810 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31812 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
31814 (vector_ae_<mode>_p): Likewise.
31815 (vector_nez_<mode>_p): Likewise.
31816 (vector_ne_v2di_p): Likewise.
31817 (vector_ae_v2di_p): Likewise.
31818 (vector_ne_<mode>_p): Likewise.
31819 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
31821 (vsx_tsqrt<mode>2_fe): Likewise.
31823 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
31826 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
31828 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31830 PR rtl-optimization/79780
31831 * cprop.c (one_cprop_pass): When second and further conditional trap
31832 in a single basic block is turned into an unconditional trap, turn it
31833 into a deleted note to avoid RTL verification failures.
31835 2017-03-02 Richard Biener <rguenther@suse.de>
31837 * fold-const.c (const_binop): Use ulow () instead of elt (0).
31839 2017-03-02 Richard Biener <rguenther@suse.de>
31841 PR tree-optimization/79345
31843 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
31844 param and abort the walk, returning -1 if it is hit.
31845 (walk_aliased_vdefs): Take a limit param and pass it on.
31846 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
31847 defaulting to 0 and return a signed int.
31848 * tree-ssa-uninit.c (struct check_defs_data): New struct.
31849 (check_defs): New helper.
31850 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
31851 about uninitialized memory.
31852 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
31853 bogus uninitialized warning.
31854 (fixed_convert_from_real): Likewise.
31856 2017-03-02 Bin Cheng <bin.cheng@arm.com>
31858 PR tree-optimization/66768
31859 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
31860 iv_use if base object can't be determined.
31862 2017-03-02 Jakub Jelinek <jakub@redhat.com>
31864 PR tree-optimization/79345
31865 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
31866 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
31867 (get_pattern_stats): Initialize it.
31868 * genemit.c (gen_expand): Verify match_scratch numbers come after
31869 match_operand/match_dup numbers.
31870 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
31871 match_scratch numbers.
31872 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
31874 * config/s390/s390.md (trunctdsd2): Likewise.
31876 2017-03-02 Richard Biener <rguenther@suse.de>
31878 * wide-int.h (wide_int_storage::operator=): Implement in terms
31881 2017-03-02 Richard Biener <rguenther@suse.de>
31883 PR tree-optimization/79777
31884 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31885 the to insert expression to sth existing.
31887 2017-03-01 Martin Sebor <msebor@redhat.com>
31889 PR middle-end/79692
31890 * gimple-ssa-sprintf.c
31891 (directive::known_width_and_precision): New function.
31892 (format_integer): Use it.
31893 (get_mpfr_format_length): Consider the full range of precision
31894 when computing %g output with the # flag. Set the likely byte
31895 count to 3 rather than 1 when precision is indeterminate.
31896 (format_floating): Correct the lower bound of precision.
31898 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31900 * doc/invoke.texi: Document default code model for 64-bit Linux.
31902 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31905 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31906 udiv rather than div since input pattern is unsigned.
31908 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31910 * config/i386/i386.c (print_reg): Warn for values of
31911 unsupported size in integer register.
31913 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31916 * config/rs6000/predicates.md (current_file_function_operand): Do
31917 not allow self calls to be local if the function is replaceable.
31919 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31922 * config/rs6000/altivec.h (vec_ctz and others): Change the
31923 preprocessor macro that controls conditional compilation from
31924 _ARCH_PWR9 to __POWER9_VECTOR__.
31925 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31926 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31927 control (instead of _ARCH_PWR9 control) so that template
31928 definition uses power9-specific function.
31929 (vec_any_eq): Likewise.
31930 (vec_all_ne): Change macro definition to use a power9-specific
31931 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31932 _ARCH_PWR9 control).
31933 (vec_any_eq) Likewise.
31934 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31935 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31936 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31937 support for xvcmpnedp instruction.
31938 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31939 macro expansion so that Power9 implementation of vec_all_ne does
31940 not use the AltiVec predicate framework.
31941 (VCMPNEH_P): Likewise.
31942 (VCMPNEW_P): Likewise.
31943 (VCMPNED_P): Likewise.
31944 (VCMPNEFP_P): Likewise.
31945 (VCMPNEDP_P): Likewise.
31946 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31947 implementation of vec_any_eq to not use AltiVec predicate
31949 (VCMPAEH_P): Likewise.
31950 (VCMPAEW_P): Likewise.
31951 (VCMPAED_P): Likewise.
31952 (VCMPAEFP_P): Likewise.
31953 (VCMPAEDP_P): Likewise.
31954 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31955 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31956 not use the AltiVec predicate framework.
31957 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31958 of vec_any_eq to not use AltiVec predicate framework.
31959 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31960 support for predefined __POWER9_VECTOR__ macro to indicate that
31961 Power9 instruction selection is enabled.
31962 (altivec_overloaded_builtins): Remove extraneous
31963 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31964 function argument types RS6000_BTI_bool_V16QI and
31965 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31966 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31967 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31968 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31969 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31970 Power9 for implementations of vec_cmpne. Change the signature for
31971 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31972 (representing vec_all_ne) to remove the previously described first
31973 argument of type RS6000_BTI_INTSI, as this was an artifact of
31974 reliance on the AltiVec predicate framework, which is no longer
31975 used in the implementation of these functions. Add
31976 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31977 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31978 since, unlike the AltiVec predicate framework implementation, we
31979 do not share function descriptors between vec_alle and vec_anyeq.
31980 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31981 set of modes that receive special treatment even when
31982 TARGET_P9_VECTOR is true. The special treatment emits code that
31983 does not depend on Power9 instructions.
31984 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31985 define_expand to not rely on AltiVec predicate framework.
31986 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31988 (vector_ne_v2di_p): Change this define_expand to not rely on
31989 AltiVec predicate framework.
31990 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31992 (vector_ne_<mode>_p): Change this define_expand to not rely on
31993 AltiVec predicate framework.
31994 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31996 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31997 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31998 define_insn pattern.
31999 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32000 define_insn pattern because the xvcmpne<VSs>. instruction is not
32002 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32003 instruction is not supported.
32005 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32007 * config/nvptx/nvptx.c: Include intl.h.
32009 2017-03-01 Martin Jambor <mjambor@suse.cz>
32012 * ipa-prop.h (ipa_bits): Removed field known.
32013 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32014 to pointers. Adjusted their comments to warn about their sharing.
32015 (ipcp_transformation_summary): Change bits to a vector of pointers.
32016 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32017 (ipa_get_ipa_bits_for_value): Declare.
32018 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32019 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32020 (ipa_bits_hash_table): Likewise.
32021 (ipa_vr_ggc_hash_traits): Likewise.
32022 (ipa_vr_hash_table): Likewise.
32023 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32024 being pointers and vr_known being removed.
32025 (ipa_set_jf_unknown): Likewise.
32026 (ipa_get_ipa_bits_for_value): New function.
32027 (ipa_set_jfunc_bits): Likewise.
32028 (ipa_get_value_range): New overloaded functions.
32029 (ipa_set_jfunc_vr): Likewise.
32030 (ipa_compute_jump_functions_for_edge): Use the above functions to
32031 construct bits and vr parts of jump functions.
32032 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32033 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32035 (ipcp_grow_transformations_if_necessary): Also allocate
32036 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32038 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32039 them. Fix too long lines.
32040 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32041 vr_known being removed.
32042 (ipa_read_jump_function): Use new setter functions to construct bits
32043 and vr parts of jump functions or set them to NULL.
32044 (write_ipcp_transformation_info): Adjust for bits being pointers.
32045 (read_ipcp_transformation_info): Likewise.
32046 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32048 Include gt-ipa-prop.h.
32049 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32051 (ipcp_store_bits_results): Likewise.
32052 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32053 Do not write to existing jump functions but use a temporary instead.
32055 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32058 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32059 attempt to use its first operand as BIT_FIELD_REF base.
32061 2017-03-01 Richard Biener <rguenther@suse.de>
32063 PR middle-end/79721
32064 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32065 interpolating formula in wrapping arithmetic.
32066 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32068 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32070 PR tree-optimization/79734
32071 * tree-vect-generic.c (expand_vector_condition): Optimize
32072 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32073 Handle VEC_COND_EXPR where comparison has different inner width from
32074 type's inner width.
32076 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32078 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32079 markup, and similar issues. Remove @opindex entries for things
32080 that aren't options. Add missing -mmpy-option entries.
32082 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32084 PR tree-optimization/79737
32085 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32086 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32087 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32088 instead of byte_size. Formatting fix.
32089 (shift_bytes_in_array_right): Formatting fix.
32091 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32094 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32095 condition on optimize for the leaf function test.
32097 2017-02-28 Martin Liska <mliska@suse.cz>
32100 * read-rtl-function.c (function_reader::handle_unknown_directive):
32101 Bail out when one uses -flto.
32103 2017-02-28 Martin Liska <mliska@suse.cz>
32105 * common.opt: Replace space with tabular for options of <number>
32107 * config/i386/i386.opt: Show <number> value for
32108 -mlarge-data-threshold.
32109 * opts.c (print_filtered_help): Do not display number in hexadecimal
32112 2017-02-28 Martin Liska <mliska@suse.cz>
32114 * common.opt: Fix --help=option -Q for options which are of
32117 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32119 * config/i386/i386.c (print_reg): Error out for values
32120 of 8-bit size in invalid integer register.
32122 2017-02-28 Martin Sebor <msebor@redhat.com>
32124 PR tree-optimization/79691
32125 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32127 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32130 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32131 gcc_unreachable with output_operand_lossage.
32133 2017-02-28 Richard Biener <rguenther@suse.de>
32135 PR tree-optimization/79740
32136 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32138 (visit_nary_op): Insert the nary into the hashtable if we
32139 pattern-matched sth.
32140 * tree-ssa-pre.c (eliminate_insert): Robustify.
32142 2017-02-28 Richard Biener <rguenther@suse.de>
32144 PR middle-end/79731
32145 * fold-const.c (decode_field_reference): Reject out-of-bound
32148 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32150 * config/i386/i386.c: Include intl.h.
32151 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32152 instead of just cond ? "..." : "...".
32153 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32154 * coverage.c (read_counts_file): Likewise.
32155 * omp-offload.c: Include intl.h.
32156 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32157 of just cond ? "..." : "...".
32158 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32159 of just cond ? "..." : "...".
32161 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32164 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32166 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32167 'tune for' CPU name.
32168 * config/arm/arm-cpu-data.h: Regenerated.
32170 2017-02-28 Richard Biener <rguenther@suse.de>
32172 PR tree-optimization/79732
32173 * tree-inline.c (expand_call_inline): Do not shadow var.
32175 2017-02-28 Richard Biener <rguenther@suse.de>
32177 PR tree-optimization/79723
32178 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32179 address-space properly.
32181 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32183 * doc/optinfo.texi (Optimization groups): Fix option used for
32185 * doc/invoke.texi (-fopt-info): Document "omp".
32186 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32187 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32188 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32189 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32190 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32192 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32194 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32195 OPTGROUP_OMP with "omp".
32197 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32200 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32201 for arithmetic shift of unsigned V2DI.
32203 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32205 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32206 arc/linux.h headers.
32207 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32208 (LINK_SPEC): Likewise.
32209 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32210 (EXTRA_SPECS): Likewise.
32211 (STARTFILE_SPEC): Likewise.
32212 (ENDFILE_SPEC): Likewise.
32213 (LIB_SPEC): Likewise.
32214 (TARGET_SDATA_DEFAULT): Likewise.
32215 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32216 (MULTILIB_DEFAULTS): Likewise.
32217 (DWARF2_UNWIND_INFO): Likewise.
32218 * config/arc/big.h: New file.
32219 * config/arc/elf.h: Likewise.
32220 * config/arc/linux.h: Likewise.
32221 * config/arc/t-uClibc: Remove.
32223 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32225 PR tree-optimization/77536
32226 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32227 (tree_transform_and_unroll_loop): Use above function to compute the
32228 estimated niter of unrolled loop and use it when scaling profile.
32229 Also use count info rather than frequency if it's non-zero.
32230 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32231 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32232 (vect_transform_loop): Call above function.
32234 2017-02-27 Richard Biener <rguenther@suse.de>
32236 PR tree-optimization/45397
32237 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32238 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32239 (visit_nary_op): Add pattern matching for CSEing sign-changed
32240 or truncated operations with wider ones.
32242 2017-02-27 Richard Biener <rguenther@suse.de>
32244 PR tree-optimization/79690
32245 * tree-vect-stmts.c (vectorizable_store): Use vector type
32246 built from the DR with address-space.
32248 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32250 * doc/invoke.texi (Optimize Options): Refine the description
32251 of asan-use-after-return.
32253 2017-02-25 Alan Modra <amodra@gmail.com>
32255 PR rtl-optimization/79584
32256 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32257 base, not ad->base_term, the reg within base. Remove assertion
32258 that ad->base == ad->base_term. Replace gen_int_mode using
32259 bogus mode with const0_rtx.
32261 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32263 PR middle-end/79396
32264 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32265 FMA_EXPR like tcc_binary or tcc_unary.
32267 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32270 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32272 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32273 (output_loc_operands): Handle DW_OP_call_ref and
32274 DW_OP_GNU_variable_value.
32275 (struct variable_value_struct): New type.
32276 (struct variable_value_hasher): Likewise.
32277 (variable_value_hash): New variable.
32278 (string_types): Remove.
32279 (copy_loc_descr): New function.
32280 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32281 (prepend_loc_descr_to_each): New function.
32282 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32283 instead of add_loc_descr_to_each if the first argument is single
32284 location list and the second has multiple.
32285 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32286 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32287 when looking for variable value which doesn't have other location info.
32288 (loc_list_from_tree): Formatting fix.
32289 (gen_array_type_die): Simplify DW_AT_string_length handling.
32290 (adjust_string_types): Remove.
32291 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32292 string_types. Call resolve_variable_values.
32293 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32294 (resolve_addr_in_expr): Likewise. Add A argument.
32295 (copy_deref_exprloc): Remove deref argument. Adjust for the
32296 original expression being DW_OP_GNU_variable_value with optionally
32297 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32298 optionally after it.
32299 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32300 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32301 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32302 (variable_value_hasher::hash, variable_value_hasher::equal): New
32304 (resolve_variable_value_in_expr, resolve_variable_value,
32305 resolve_variable_values, note_variable_value_in_expr,
32306 note_variable_value): New functions.
32307 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32310 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32313 * opts.h (handle_generated_option): Add GENERATED_P argument.
32314 * opts-common.c (handle_option): Adjust function comment.
32315 (handle_generated_option): Add GENERATED_P argument, pass it to
32317 (control_warning_option): Pass false to handle_generated_option
32319 * opts.c (maybe_default_option): Pass true to handle_generated_option
32321 * optc-gen.awk: Likewise.
32323 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32325 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32326 a REG, look at the REG it is a SUBREG of.
32327 (splitter for cmpeqsi_t): Ditto.
32329 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32331 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32332 the special USEs with the pattern of the insn, not the insn itself.
32334 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32337 * doc/invoke.texi: Document -mload-store-pairs.
32339 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32340 Sandra Loosemore <sandra@codesourcery.com>
32342 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32343 argument isn't a CONST_INT.
32344 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32345 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32346 (nios2_validate_compare): Bypass alternate compare logic if *op2
32347 is not a CONST_INT.
32348 (ldstwm_operation_p): Return false if first_base is not a REG or
32349 if first_offset is not a CONST_INT.
32351 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32353 * config/cris/cris.md: Use correct operand in a define_peephole2.
32355 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32357 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32359 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32361 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32362 this_insn if it is an INSN or JUMP_INSN.
32363 (force_offsettable): Look at base, not at addr.
32364 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32365 on things that aren't necessarily CONST_INTs.
32367 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32369 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32370 -mfpmath=sse is the default also for x86-32 targets with SSE2
32371 instruction set when @option{-ffast-math} is enabled
32373 2017-02-24 Jeff Law <law@redhat.com>
32375 PR rtl-optimizatoin/79286
32376 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32379 2017-02-24 Richard Biener <rguenther@suse.de>
32381 PR tree-optimization/79389
32382 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32385 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32387 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32388 function comment to reflect reality.
32389 (loop_exits_before_overflow): Fix typo in function description.
32391 2017-02-24 Richard Biener <rguenther@suse.de>
32393 PR tree-optimization/79389
32394 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32395 properly that a threading opportunity exists. Detect conditional
32396 copy/constant propagation opportunities.
32398 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32400 * config/visium/visium.md (type): Add trap.
32401 (b): New mode attribute.
32402 (*btst): Rename into...
32403 (*btst<mode>): ...this and adjust.
32404 (*cbranchsi4_btst_insn): Rename into...
32405 (*cbranch<mode>4_btst_insn): ...this and adjust.
32406 (trap): New define_insn.
32408 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32410 PR tree-optimization/79389
32411 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32412 (noce_reversed_cond_code): New function.
32413 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32414 reversed_comparison_code. Formatting fix.
32415 (noce_try_store_flag): Test rev_cond != NULL in addition to
32416 reversed_comparison_code.
32417 (noce_try_store_flag_constants): Likewise.
32418 (noce_try_store_flag_mask): Likewise.
32419 (noce_try_addcc): Use rev_cond if non-NULL instead of
32420 reversed_comparison_code.
32421 (noce_try_cmove_arith): Likewise. Formatting fixes.
32422 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32423 (noce_find_if_block): Initialize rev_cond.
32424 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32425 instead of false as last argument never attempt to reverse it
32428 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32430 PR tree-optimization/79663
32431 * tree-predcom.c (combine_chains): Process refs in reverse order
32432 only for ZERO length chains, and add explaining comment.
32434 2017-02-23 Jeff Law <law@redhat.com>
32436 PR tree-optimization/79578
32437 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32438 in call to operand_equal_p.
32440 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32443 * config/i386/cpuid.h: Fix another undefined behavior.
32445 2017-02-23 Richard Biener <rguenther@suse.de>
32447 PR tree-optimization/79683
32448 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32449 vector types for data-refs.
32451 2017-02-23 Martin Liska <mliska@suse.cz>
32453 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32455 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32457 PR middle-end/79665
32458 * internal-fn.c (get_range_pos_neg): Moved to ...
32459 * tree.c (get_range_pos_neg): ... here. No longer static.
32460 * tree.h (get_range_pos_neg): New prototype.
32461 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32462 are known to be in between 0 and signed maximum inclusive, try to
32463 expand both unsigned and signed divmod and use the cheaper one from
32466 2017-02-22 Jeff Law <law@redhat.com>
32468 PR tree-optimization/79578
32469 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32470 to compare base operands.
32472 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32475 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32476 gpc_reg_operand instead of fpr_reg_operand.
32478 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32480 * config/mips/mips.c (mips_return_in_memory): Force FP
32481 vector types to be returned in memory for o32 ABI.
32483 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32485 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32486 instead of DW_TAG_member for static data member declarations and don't
32487 set no_linkage_name for static inline data members.
32488 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32491 2017-02-22 Martin Liska <mliska@suse.cz>
32493 * doc/invoke.texi: Replace inequality signs with square brackets
32496 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32499 * lra-constraints.c (simplify_operand_subreg): Handle
32500 WORD_REGISTER_OPERATIONS targets.
32502 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32505 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32506 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32507 elimination by swapping fld*.
32509 2017-02-22 Richard Biener <rguenther@suse.de>
32511 PR tree-optimization/79673
32512 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32513 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32514 irrelevant address-space qualifiers and avoiding a
32515 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32517 2017-02-22 Richard Biener <rguenther@suse.de>
32519 PR tree-optimization/79666
32520 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32521 to not symbolically negate if that may introduce undefined
32524 2017-02-22 Martin Liska <mliska@suse.cz>
32527 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32528 * data-streamer-out.c (streamer_write_gcov_count_stream):
32530 * value-prof.c (stream_out_histogram_value): Make assert more
32531 precise based on type of counter.
32533 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32536 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32537 Use nonimmediate_operand instead of memory_operand for operand 1.
32538 (float-extend standard_x87sse_constant_load splitter): Ditto.
32540 2017-02-21 Jeff Law <law@redhat.com>
32542 PR tree-optimization/79621
32543 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32544 blocks with edges to themselves.
32546 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32549 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32550 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32551 Use gimple_call_builtin_p.
32554 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32555 on temporarily removed DEBUG_INSNs.
32557 PR tree-optimization/79649
32558 * tree-loop-distribution.c (classify_partition): Give up on
32559 non-generic address space loads/stores.
32561 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32563 * doc/loop.texi (Loop manipulation): Remove nonexistent
32564 tree_ssa_loop_version from the documentation.
32565 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32567 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32570 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32571 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32572 * config/rs6000/rs6000.c: Include except.h.
32573 (rs6000_expand_split_stack_prologue): Call
32574 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32576 2017-02-21 Martin Jambor <mjambor@suse.cz>
32579 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32580 have been analyzed.
32582 2017-02-21 Martin Jambor <mjambor@suse.cz>
32584 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32585 for backward compatibility only.
32586 * doc/invoke.texi (Option Summary): Remove all references to
32587 -fipa-cp-alignment.
32589 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32593 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32595 * lra-constraints.c (curr_insn_transform): Handle
32596 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32598 2017-02-21 Martin Liska <mliska@suse.cz>
32600 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32602 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32604 PR translation/79638
32605 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32607 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32610 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32611 (arm_function_ok_for_sibcall): Return false for an indirect call by
32612 descriptor if all the argument registers are used.
32613 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32614 alignment of the function.
32616 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32618 PR tree-optimization/61441
32619 * simplify-rtx.c (simplify_const_unary_operation): For
32620 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32621 the sNaN unmodified.
32623 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32625 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32626 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32627 instead of SYSTEM_HEADER_DIR.
32629 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32630 Martin Liška <mliska@suse.cz>
32632 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32633 Fix typos and grammar, use active voice, and clarify.
32635 2017-02-20 Marek Polacek <polacek@redhat.com>
32637 PR middle-end/79537
32638 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32641 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32643 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32646 * config/i386/i386.c (ix86_expand_builtin): Handle
32647 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32648 ix86_builtins_isa[fcode].isa as a requirement of those
32649 flags and any other flag in the bitmask.
32650 (ix86_init_mmx_sse_builtins): Use 0 instead of
32651 ~OPTION_MASK_ISA_64BIT as mask.
32652 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32653 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32654 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32655 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32657 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32660 * lra-constraints.c (split_reg): Check requested split mode
32661 is supported by the register.
32663 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32665 * lra-constraints.c (simplify_operand_subreg): Remove early
32668 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32671 * lra-constraints.c (curr_insn_transform): Tighten condition
32672 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32674 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32677 * lra-constraints.c (curr_insn_transform): Handle
32678 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32680 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32683 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32685 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32687 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32690 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32693 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32695 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
32698 2017-02-18 Jakub Jelinek <jakub@redhat.com>
32700 * final.c (last_columnnum, override_columnnum): New variables.
32701 (final_start_function): Set last_columnnum, pass it to begin_prologue
32702 hook and pass 0 to dwarf2out_begin_prologue.
32703 (final_scan_insn): Update override_columnnum. Pass last_columnnum
32704 to source_line debug hook.
32705 (notice_source_line): Compute last_columnnum and for debug_column_info
32706 return true on column changes.
32707 * debug.h (struct gcc_debug_hooks): Add column argument to
32708 source_line and begin_prologue hooks.
32709 (debug_nothing_int_charstar_int_bool): Remove prototype.
32710 (debug_nothing_int_int_charstar,
32711 debug_nothing_int_int_charstar_int_bool): New prototypes.
32712 (dwarf2out_begin_prologue): Add column argument.
32713 * debug.c (do_nothing_debug_hooks): Adjust source_line and
32714 begin_prologue hooks.
32715 (debug_nothing_int_charstar_int_bool): Remove.
32716 (debug_nothing_int_int_charstar,
32717 debug_nothing_int_int_charstar_int_bool): New functions.
32718 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
32719 through to dwarf2out_source_line.
32720 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
32721 (dwarf2out_source_line): Add column argument, emit it if requested.
32722 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
32724 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32725 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
32726 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
32727 through to dwarf2out_begin_prologue.
32728 (vmsdbgout_source_line): Add column argument, pass it through to
32729 dwarf2out_source_line.
32730 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
32731 dbxout_source_line caller.
32732 (dbxout_source_line): Add column argument.
32734 * common.opt (gno-column-info, gcolumn-info): New options.
32735 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
32736 (check_die): Also test for multiple DW_AT_decl_column attributes.
32737 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
32738 DW_AT_decl_column if requested.
32739 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
32741 (gen_variable_die): Likewise.
32742 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
32743 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
32746 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
32747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
32748 (ix86_handle_option): Handle OPT_m3dnowa.
32749 * doc/invoke.texi (-m3dnowa): Document.
32750 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
32751 -m3dnowa instead of -m3dnow -march=athlon.
32754 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
32755 instead of gcc_assert for K, r and R code checks. Formatting fixes.
32757 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32760 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
32761 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
32762 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
32763 generator for vsx_xxpermdi_<mode>_be.
32764 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
32765 force big-endian semantics.
32766 (vsx_xxpermdi_<mode>_be): New define_expand with same
32767 implementation as previous version of vsx_xxpermdi_<mode>.
32769 2017-02-17 Jakub Jelinek <jakub@redhat.com>
32771 PR tree-optimization/79327
32772 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
32773 variable, its initialization and use.
32775 2017-02-17 Julia Koval <julia.koval@intel.com>
32777 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
32778 (OPTION_MASK_ISA_PKU_UNSET): New.
32779 (ix86_handle_option): Handle -mrdpid.
32780 * config/i386/cpuid.h (bit_RDPID): New.
32781 * config/i386/driver-i386.c (host_detect_local_cpu):
32782 Detect RDPID feature.
32783 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
32784 * config/i386/i386-c.c (ix86_target_macros_internal):
32786 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
32787 (ix86_valid_target_attribute_inner_p): Add "rdpid".
32788 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
32789 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
32790 * config/i386/i386.md (define_insn "rdpid"): New.
32791 * config/i386/i386.opt Add -mrdpid.
32792 * config/i386/immintrin.h (_rdpid_u32): New.
32794 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
32796 PR rtl-optimization/79541
32797 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
32798 instead of transforming it into USE.
32800 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
32802 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
32803 If HONOR_SNANS (SFmode) force the input to a register.
32804 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
32805 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
32806 an frsp or similar insn.
32808 2017-02-17 Martin Liska <mliska@suse.cz>
32810 PR rtl-optimization/79577
32811 * params.def (selsched-max-sched-times): Increase minimum to 1.
32813 2017-02-17 Martin Liska <mliska@suse.cz>
32815 PR rtl-optimization/79574
32816 * gcse.c (want_to_gcse_p): Prevent integer overflow.
32818 2017-02-17 Martin Liska <mliska@suse.cz>
32820 PR tree-optimization/79529
32821 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
32822 ssa_defined_default_def_p to handle cases which are implicitly
32824 * tree-ssa.c (ssa_defined_default_def_p): New function.
32825 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
32826 which are implicitly defined.
32827 * tree-ssa.h (ssa_defined_default_def_p): Declare.
32829 2017-02-17 Richard Biener <rguenther@suse.de>
32831 PR middle-end/79576
32832 * params.def (max-ssa-name-query-depth): Limit to 10.
32834 2017-02-17 Richard Biener <rguenther@suse.de>
32836 PR tree-optimization/79552
32837 * tree-ssa-structalias.c (visit_loadstore): Properly verify
32840 2017-02-17 Richard Biener <rguenther@suse.de>
32843 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
32845 2017-02-17 Marek Polacek <polacek@redhat.com>
32847 PR middle-end/79536
32848 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
32849 (fold_negate_expr): New wrapper.
32851 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
32853 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
32854 Correct terminology and de-emphasize pre-standard behavior.
32856 2017-02-16 Alan Modra <amodra@gmail.com>
32858 PR rtl-optimization/79286
32859 * ira.c (def_dominates_uses): New function.
32860 (update_equiv_regs): Don't create an equivalence for insns that
32861 may trap where the register def does not dominate the use.
32863 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
32865 PR rtl-optimization/78127
32866 * lra.c (lra): Call lra_eliminate before finish the loop after
32869 2017-02-16 Richard Biener <rguenther@suse.de>
32871 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
32873 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
32874 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
32875 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
32876 (isl_val_int_from_wi): New function.
32877 (extract_affine_gmp): Rename to ...
32878 (extract_affine_wi): ... this, take a widest_int.
32879 (extract_affine_int): Just wrap extract_affine_wi.
32880 (add_param_constraints): Use isl_val_int_from_wi.
32881 (add_loop_constraints): Likewise, and extract_affine_wi.
32883 2017-02-15 Jeff Law <law@redhat.com>
32885 PR middle-end/79521
32886 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32887 ira_init_register_move_cost_if_necessary.
32889 2017-02-15 Martin Sebor <msebor@redhat.com>
32891 PR middle-end/32003
32892 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32893 removed in a prior commit.
32895 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32897 PR tree-optimization/79347
32898 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32899 counters during peeling.
32901 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32903 * Makefile.in (site.exp): Remove "set ISLVER".
32905 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32908 * real.c (real_from_integer): Call real_convert even for decimal.
32910 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32913 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32915 2017-02-14 Andrew Pinski <apinski@cavium.com>
32917 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32918 cores and change the partno/implementer to be correct.
32919 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32920 the 'B" as the implementer.
32921 * config/aarch64/aarch64-tune.md: Regenerate.
32923 2017-02-14 Carl Love <cel@us.ibm.com>
32925 * config/rs6000/rs6000.c: Add case statement entry to make the
32926 xvcvuxdsp built-in argument unsigned.
32927 * config/rs6000/vsx.md: Fix the source and return operand types so they
32928 match the instruction definitions from the ISA document. Fix typo
32929 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32932 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32935 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32936 member early_clobber_alts.
32937 * lra-lives.c (reg_early_clobber_p): New.
32938 (process_bb_lives): Use it.
32939 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32940 (debug_operand_data): Initialize early_clobber_alts.
32941 (setup_operand_alternative): Set up early_clobber_alts.
32942 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32943 alternatives to new_insn_reg.
32944 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32946 (lra_update_insn_regno_info): Pass the new arg.
32948 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32950 PR middle-end/79505
32951 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32952 (new_oacc_loop_raw): Don't clear already cleared fields.
32955 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32956 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32957 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32959 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32962 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32964 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32967 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32968 the extra instruction to the right place to store 128-bit constant
32971 2017-02-14 Martin Sebor <msebor@redhat.com>
32973 PR middle-end/79448
32974 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32975 warning for strings of unknown length.
32977 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32979 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32981 2017-02-14 Jeff Law <law@redhat.com>
32984 * ira-costs.c (scan_one_insn): Initialize register move costs
32985 for pseudos seen in USE/CLOBBER insns.
32987 PR tree-optimization/79095
32988 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32989 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32990 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32991 if the operands are known to be not equal, then the resulting range
32993 (intersect_ranges): If the new range is ~[0,0] and the old range is
32994 wide, then prefer ~[0,0].
32995 * tree-vrp.c (overflow_comparison_p_1): New function.
32996 (overflow_comparison_p): New function.
32997 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32998 if NAME is used in an overflow test.
32999 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33000 overflow check that can be expressed as an equality test, then adjust
33001 ops to be that equality test.
33003 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33005 * config/s390/s390-builtin-types.def: Remove flags argument.
33006 * config/s390/s390.c (s390_init_builtins): Likewise.
33008 2017-02-14 Martin Liska <mliska@suse.cz>
33010 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33011 vector. Fix trailing white spaces.
33013 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33015 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33018 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33020 PR rtl-optimization/68664
33021 * config/arm/arm.c (arm_sched_can_speculate_insn):
33022 New function. Declare prototype.
33023 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33025 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33027 PR rtl-optimization/68664
33028 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33030 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33032 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33034 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33035 max skip bytes for function, loop and jump.
33037 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33039 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33040 ABS_EXPR for gimple dump.
33042 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33045 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33047 PR tree-optimization/79408
33048 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33049 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33050 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33051 also if rhs1 is INTEGER_CST.
33053 2017-02-14 Richard Biener <rguenther@suse.de>
33055 PR middle-end/79432
33056 * tree-into-ssa.c (insert_phi_nodes): When the function can
33057 have abnormal edges rewrite SSA names with broken use-def
33058 dominance out of SSA and register them for PHI insertion.
33060 2017-02-13 Martin Sebor <msebor@redhat.com>
33062 PR middle-end/79496
33063 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33064 clearing info.nowrite flag when snprintf size argument is a range.
33066 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33068 * cprop.c (cprop_jump): Add missing space in string literal.
33069 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33070 (get_constraint_for_component_ref): Likewise.
33071 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33072 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33073 * lra-constraints.c (process_alt_operands): Likewise.
33074 * ipa-inline.c (inline_small_functions): Likewise.
33075 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33076 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33077 * trans-mem.c (diagnose_tm_1_op): Likewise.
33078 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33079 (grid_parallel_clauses_gridifiable): Likewise.
33081 * config/nvptx/mkoffload.c (process): Add space in between
33084 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33085 "MOD4_SSE_REGS" and "ALL_REGS".
33087 * spellcheck.c (test_data): Add , in between "foo" and "food".
33089 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33092 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33093 boundary crossing check and subsequent code generation agree.
33095 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33097 * config/aarch64/aarch64.c (has_memory_op): Delete.
33098 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33101 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33103 PR rtl-optimization/79388
33104 PR rtl-optimization/79450
33105 * combine.c (distribute_notes): When removing TEM_INSN for which
33106 corresponding dest has last value recorded, invalidate that last
33109 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33111 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33112 of explicit '@'. Add missing assembly comment marker on branch costs
33115 2017-02-13 Nathan Sidwell <nathan@acm.org>
33117 * gengtype-lex.l (<in_struct>): Add '/'.
33119 2017-02-13 Martin Liska <mliska@suse.cz>
33122 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33124 2017-02-13 Richard Biener <rguenther@suse.de>
33126 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33128 * configure: Re-generate.
33129 * config.in: Likewise.
33130 * graphite-dependences.c: Simplify as if
33131 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33132 * graphite-isl-ast-to-gimple.c: Likewise.
33133 * graphite-optimize-isl.c: Likewise.
33134 * graphite-poly.c: Likewise.
33135 * graphite-sese-to-poly.c: Likewise.
33136 * graphite.h: Likewise.
33137 * toplev.c: Include isl/version.h and use isl_version () for
33138 printing the ISL version.
33139 * doc/install.texi: Update ISL requirement.
33141 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33143 * doc/standards.texi (Standards): Update reference to
33146 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33148 * doc/extend.texi (Named Address Spaces): sourceware.org now
33150 * doc/install.texi (Binaries): Ditto.
33153 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33155 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33156 standard terminology "stringize"/"stringizing" throughout.
33157 * doc/cppinternals.texi: Likewise.
33159 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33161 * doc/extend.texi: Fix some spelling mistakes and typos.
33162 * doc/invoke.texi: Likewise.
33164 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33167 * params.def (inline-min-speedup) Change from 10 to 8.
33169 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33171 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33174 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33177 * ipa-inline-analysis.c (get_minimal_bb): New function.
33178 (record_modified): Use it.
33179 (remap_edge_change_prob): Handle also ancestor functions.
33181 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33183 * doc/contrib.texi (Contributors): Remove broken link into
33184 the Mauve CVS repository.
33186 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33188 PR middle-end/79454
33189 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33190 result computation whenever lhs doesn't have vector mode, not
33191 just when it has BLKmode.
33193 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33195 * doc/makefile.texi (profiledbootstrap): Refer to the
33196 installation instructions only in textual form.
33198 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33201 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33203 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33205 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33206 (Specific): Update mingw-w64 reference.
33208 (Specific): Remove broken link to Renesas RX processor.
33210 2017-02-10 Richard Biener <rguenther@suse.de>
33212 * toplev.c (process_options): Do not mention obsolete graphite
33213 options when printing sorry message about missing graphite support.
33214 Mention -floop-nest-optimize.
33216 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33218 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33219 (vtst_p16): Likewise.
33220 (vtstq_p8): Likewise.
33221 (vtstq_p16): Likewise.
33223 (vtstq_p64): Likewise.
33224 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33225 (vset_lane_p64): New.
33226 (vsetq_lane_p64): New.
33228 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33230 PR tree-optimization/79411
33231 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33232 stmt operands are SSA_NAMEs used in abnormal phis.
33233 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33236 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33239 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33242 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33244 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33246 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33248 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33252 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33253 not arbitrary TREE_CONSTANT.
33256 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33257 "omp declare target link" attribute unless is_global_var.
33258 * omp-offload.c (find_link_var_op): Likewise.
33260 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33261 Chung-Lin Tang <cltang@codesourcery.com>
33263 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33265 (gimplify_adjust_omp_clauses): Don't delete TILE.
33266 (gimplify_omp_for): Deal with TILE.
33267 * internal-fn.c (expand_GOACC_TILE): New function.
33268 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33270 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33271 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33273 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33274 avoid DIV for outermost collapse var.
33275 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33276 Remove out of date comments, fix whitespace.
33277 * omp-general.c (omp_extract_for_data): Deal with tiling.
33278 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33279 adjust OLF_DIM_BASE value.
33280 (struct omp_for_data): Add tiling field.
33281 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33282 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33283 for auto loops. Remove default auto determining, moved to
33284 oacc_loop_fixed_partitions.
33285 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33286 stmts, add e_mask field.
33287 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33288 (oacc_thread_numbers): Use oacc_dim_call.
33289 (oacc_xform_tile): New.
33290 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33291 (finish_oacc_loop): Adjust for ifns vector.
33292 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33293 add case for GOACC_TILE fns.
33294 (oacc_loop_xform_loop): Delete.
33295 (oacc_loop_process): Iterate over call list directly, and add
33296 handling for GOACC_TILE fns.
33297 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33299 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33300 vector partitioning to outer loops. Assign 2 partitions to loops
33301 when available. Add TILE handling.
33302 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33303 (execite_oacc_device_lower): Process GOACC_TILE fns,
33304 ignore unknown specs.
33305 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33306 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33307 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33309 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33311 * configure.ac (ACX_BUGURL): Update.
33312 * configure: Regenerate.
33314 2017-02-09 Richard Biener <rguenther@suse.de>
33316 PR tree-optimization/69823
33317 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33318 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33320 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33322 * config/arc/arc-c.def: Add __NPS400__ definition.
33323 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33324 (TARGET_NPS400): Define.
33326 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33328 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33330 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33331 pointer, arch_info.
33332 (arc_cpu_types): Fill the arch_info field with a pointer into the
33333 arc_arch_types table.
33334 (arc_selected_cpu): Declare.
33335 * config/arc/arc.c (arc_selected_cpu): Make global.
33336 (arc_selected_arch): Delete.
33337 (arc_base_cpu): Delete.
33338 (arc_override_options): Remove references to deleted variables,
33339 update access to arch information.
33340 (ARC_OPT): Update access to arch information.
33341 (ARC_OPTX): Likewise.
33342 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33343 (TARGET_ARC600): Update access to arch information.
33344 (TARGET_ARC601): Likewise.
33345 (TARGET_ARC700): Likewise.
33346 (TARGET_EM): Likewise.
33347 (TARGET_HS): Likewise.
33348 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33351 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33354 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33355 condition/operands for integer GE/LE/GEU/LEU operations.
33357 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33359 PR translation/79397
33360 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33363 2017-02-08 Martin Jambor <mjambor@suse.cz>
33366 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33367 whether allocation happened.
33368 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33369 nothing was allocated.
33371 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33373 PR tree-optimization/79408
33374 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33375 constant, but SSA_NAME with a known integer range, use the minimum
33376 of that range instead of op1 to determine if modulo can be replaced
33377 with its first operand.
33379 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33381 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33383 2017-02-08 Richard Biener <rguenther@suse.de>
33385 PR tree-optimization/71824
33386 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33387 Check all loops contained in the merged region.
33389 2017-02-07 Andrew Pinski <apinski@cavium.com>
33391 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33393 2017-02-07 Andrew Pinski <apinski@cavium.com>
33395 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33396 (thunderxt88): Likewise.
33397 (thunderxt81): Disable LSE and change v8.1 to v8.
33398 (thunderxt83): Likewise.
33400 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33401 Richard Biener <rguenther@suse.de>
33403 PR middle-end/79399
33404 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33405 type from int to size_t.
33406 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33408 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33410 PR rtl-optimization/79386
33411 * cprop.c (bypass_conditional_jumps): Initialize
33412 bypass_last_basic_block already before splitting bbs after
33413 unconditional traps...
33414 (bypass_conditional_jumps): ... rather than here.
33417 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33418 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33419 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33420 fix -masm=intel patterns.
33422 2017-02-07 Richard Biener <rguenther@suse.de>
33424 PR tree-optimization/79256
33425 PR middle-end/79278
33426 * builtins.c (get_object_alignment_2): Use min_align_of_type
33427 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33428 and ADJUST_FIELD_ALIGN.
33430 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33432 * doc/tm.texi: Regenerate.
33433 * stor-layout.c (layout_decl): Adjust.
33434 (update_alignment_for_field): Likewise.
33435 (place_field): Likewise.
33436 (min_align_of_type): Likewise.
33437 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33438 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33439 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33440 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33441 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33442 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33443 * config/i386/i386.c (x86_field_alignment): Likewise.
33444 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33445 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33446 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33447 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33448 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33449 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33453 2017-01-30 Richard Biener <rguenther@suse.de>
33455 PR tree-optimization/79256
33456 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33457 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33460 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33462 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33463 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33464 builtins to SImode and emit a zero-extend, if necessary.
33466 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33468 * docs/invoke.texi (RISC-V Options): Alphabetize.
33470 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33472 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33475 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33477 * config/riscv/riscv.c: New file.
33478 * common/config/riscv/riscv-common.c: Likewise.
33479 * config.gcc: Likewise.
33480 * config/riscv/constraints.md: Likewise.
33481 * config/riscv/elf.h: Likewise.
33482 * config/riscv/generic.md: Likewise.
33483 * config/riscv/linux.h: Likewise.
33484 * config/riscv/multilib-generator: Likewise.
33485 * config/riscv/peephole.md: Likewise.
33486 * config/riscv/pic.md: Likewise.
33487 * config/riscv/predicates.md: Likewise.
33488 * config/riscv/riscv-builtins.c: Likewise.
33489 * config/riscv/riscv-c.c: Likewise.
33490 * config/riscv/riscv-ftypes.def: Likewise.
33491 * config/riscv/riscv-modes.def: Likewise.
33492 * config/riscv/riscv-opts.h: Likewise.
33493 * config/riscv/riscv-protos.h: Likewise.
33494 * config/riscv/riscv.h: Likewise.
33495 * config/riscv/riscv.md: Likewise.
33496 * config/riscv/riscv.opt: Likewise.
33497 * config/riscv/sync.md: Likewise.
33498 * config/riscv/t-elf-multilib: Likewise.
33499 * config/riscv/t-linux: Likewise.
33500 * config/riscv/t-linux-multilib: Likewise.
33501 * config/riscv/t-riscv: Likewise.
33502 * configure.ac: Likewise.
33503 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33504 Waterman as RISC-V maintainers.
33505 * doc/install.texi: Add RISC-V entries.
33506 * doc/invoke.texi: Add RISC-V options section.
33507 * doc/md.texi: Add RISC-V constraints section.
33508 * configure: Regenerated.
33510 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33513 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33514 false values to be constant vectors with all 0 or all 1 bits set.
33515 (vcondu<mode><mode>): Likewise.
33516 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33518 (fpmask_comparison_operator): Update comment.
33519 (vecint_comparison_operator): New predicate.
33520 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33521 vector conditionals when the true and false values are constant
33522 vectors with all 0 bits or all 1 bits set.
33524 2017-02-06 Martin Sebor <msebor@redhat.com>
33526 PR tree-optimization/79376
33527 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33529 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33531 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33532 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33533 to simplify split condition.
33535 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33537 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33538 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33541 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33543 PR rtl-optimization/68664
33544 * target.def (can_speculate_insn): New hook.
33545 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33546 * doc/tm.texi: Regenerate.
33547 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33548 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33549 (rs6000_sched_can_speculate_insn): New function.
33551 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33553 PR tree-optimization/79284
33554 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33555 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33556 vectorizable_mask_load_store, vectorizable_operation,
33557 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33558 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33559 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33560 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33561 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33562 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33563 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33564 is_gimple_assign (stmt). Replace another such test with
33565 is_gimple_assign (stmt).
33567 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33570 * config/avr/avr.c (rtl-iter.h): Include it.
33571 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33572 (avr_legitimate_combined_insn): ...and implementation.
33574 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33576 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33577 * config/s390/s390.c (s390_const_operand_ok)
33578 (s390_canonicalize_comparison, s390_extract_part)
33579 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33580 (s390_contiguous_bitmask_p, s390_rtx_costs)
33581 (legitimize_pic_address): Likewise.
33582 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33583 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33584 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33585 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33586 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33588 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33590 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33591 REGNO($0) == REGNO($1).
33593 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33595 * config/s390/linux.h(SIZE_TYPE): Add comment.
33597 2017-02-06 Julian Brown <julian@codesourcery.com>
33598 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33599 Virendra Pathak <virendra.pathak@broadcom.com>
33601 * config/aarch64/aarch64-cores.def: Change the scheduler
33603 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33604 * config/aarch64/thunderx2t99.md: New file.
33606 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33608 * doc/standards.texi (Go Language): Update link to language
33611 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33613 * tree-eh.c (lower_resx): Sanitize profile.
33614 (cleanup_empty_eh_move_lp): Likewise.
33616 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33619 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33621 * cfgloopmanip.h (loop_version): Update prototype.
33622 * modulo-sched.c (sms_schedule): Update call of loop_version.
33623 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33624 * tree-parloops.c (gen_parallel_loop): Likewise.
33625 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33626 * tree-ssa-loop-split.c (split_loop): Likewise.
33627 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33628 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33630 2017-02-05 Martin Liska <mliska@suse.cz>
33633 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33635 (print_operand_address): Initialize a struct to zero.
33637 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33639 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33640 garbage collector only in textual form.
33642 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33644 * doc/extend.texi (x86 specific memory model extensions for
33645 transactional memory): Simplify a phrase.
33647 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33650 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33651 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33652 (atomic_storedi_1): Likewise.
33654 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33656 PR tree-optimization/79338
33657 * tree-parloops.c (gather_scalar_reductions): Don't call
33658 vect_analyze_loop_form for loop->inner before destroying loop's
33661 2017-02-03 Martin Sebor <msebor@redhat.com>
33663 PR tree-optimization/79327
33664 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33665 when precision has resulted in leading zeros.
33666 (format_integer): Adjust the likely counter to assume an unknown
33667 argument that may be zero is non-zero.
33669 2017-02-03 Jason Merrill <jason@redhat.com>
33672 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33673 avoid copying non-taken branch.
33675 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33677 PR tree-optimization/79340
33678 * tree-vect-loop.c (vectorizable_reduction): Release
33679 vec_defs elements after safe_splicing them into other vectors.
33682 PR tree-optimization/79327
33683 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33684 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33686 (format_integer): Use wide_int_to_tree instead of build_int_cst
33687 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33688 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33689 of shortest and longest sequence.
33691 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33693 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33694 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33696 2017-02-03 Walter Lee <walt@tilera.com>
33699 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
33700 after initial stackframe link reg save.
33701 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
33703 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33706 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
33707 wu for stxssp alternative.
33709 2017-02-03 Martin Sebor <msebor@redhat.com>
33711 PR tree-optimization/79352
33712 * gimple-fold.c (get_range_strlen): Add argument.
33713 (get_range_strlen): Change return type to bool.
33714 (get_maxval_strlen): Pass in a dummy argument.
33715 * gimple-fold.h (get_range_strlen): Change return type to bool.
33716 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
33717 * tree.h (array_at_struct_end_p): Add argument.
33718 * tree.c (array_at_struct_end_p): Handle it.
33720 2017-02-03 Martin Liska <mliska@suse.cz>
33723 * multiple_target.c (create_dispatcher_calls): Redirect edge
33724 from a caller of a dispatcher.
33725 (expand_target_clones): Make the clones local.
33726 (ipa_target_clone): Do both target clones and resolvers.
33727 (ipa_dispatcher_calls): Remove the pass.
33728 (pass_dispatcher_calls::gate): Likewise.
33729 (make_pass_dispatcher_calls): Likewise.
33730 * passes.def (pass_target_clone): Put as very first IPA early
33733 2017-02-03 Martin Liska <mliska@suse.cz>
33735 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
33736 in case of a function with ifunc attribute.
33738 2017-02-03 Martin Liska <mliska@suse.cz>
33740 * cgraph.c (cgraph_node::dump): Dump function version info.
33741 * symtab.c (symtab_node::dump_base): Add missing new line.
33743 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33745 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
33746 (ifcombine_ifandif): Use it.
33748 2017-02-03 Martin Liska <mliska@suse.cz>
33750 * doc/invoke.texi: Document default value for
33751 use-after-scope-direct-emission-threshold.
33753 2017-02-03 Martin Liska <mliska@suse.cz>
33755 PR tree-optimization/79339
33756 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
33757 (format_floating): Likewise.
33759 2017-02-03 Martin Liska <mliska@suse.cz>
33762 * ipa-prop.c (ipa_node_params_t::insert): Remove current
33764 (ipa_node_params_t::remove): Likewise.
33765 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
33766 initialization from removed ipa_node_params_t::insert.
33767 (ipa_node_params::~ipa_node_params): Move from removed
33768 ipa_node_params_t::release.
33769 * symbol-summary.h (symbol_summary::m_released): New member.
33770 Do not release a summary twice. Do not allow to call finalizer
33771 for types of a summary that live in GGC memory.
33773 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33775 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
33778 2017-02-02 Martin Sebor <msebor@redhat.com>
33780 PR middle-end/79275
33781 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
33782 (format_string): Tighten up the range of output for non-constant
33783 strings and correct the expected range for wide non-constant strings.
33785 2017-02-02 Martin Sebor <msebor@redhat.com>
33787 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
33789 PR middle-end/32003
33790 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
33792 (-fdump-tree-@var): Add to index and document how to come up
33793 with pass-specific option and dump file names.
33794 (-fdump-passes): Clarify where to look for output.
33796 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
33798 PR middle-end/77445
33799 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
33800 statistics of the analyzed path; allow threading for speed when
33801 any of BBs along the path are optimized for speed.
33803 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
33805 PR middle-end/78468
33806 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
33807 settings of the virtual registers.
33810 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
33812 * explow.c (get_dynamic_stack_size): Take known alignment of stack
33813 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
33816 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33818 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
33819 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
33821 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33823 * config/s390/s390.md: Add missing comments with the expanded
33825 * config/s390/vector.md: Likewise.
33826 * config/s390/vx-builtins.md: Likewise.
33828 2017-02-02 Jakub Jelinek <jakub@redhat.com>
33831 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
33832 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
33833 conditions on a single line.
33835 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33837 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33838 __S390_VX__ to __VX__.
33840 2017-02-01 Andrew Pinski <apinski@cavium.com>
33842 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
33843 stmt_info to record_stmt_cost.
33844 (vect_get_known_peeling_cost): Pass stmt_info if known to
33846 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
33847 cpu_vector_cost field into
33848 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
33849 field into vec_int_stmt_cost and vec_fp_stmt_cost.
33850 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
33851 splitting of scalar_stmt_cost and vec_stmt_cost.
33852 (thunderx_vector_cost): Likewise.
33853 (cortexa57_vector_cost): LIkewise.
33854 (exynosm1_vector_cost): Likewise.
33855 (xgene1_vector_cost): Likewise.
33856 (thunderx2t99_vector_cost): Improve after the splitting of the two
33858 (aarch64_builtin_vectorization_cost): Update for the splitting of
33859 scalar_stmt_cost and vec_stmt_cost.
33861 2017-02-01 Torvald Riegel <triegel@redhat.com>
33862 Richard Henderson <rth@redhat.com>
33864 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
33865 conditional on existance of a fast atomic load.
33866 * optabs-query.c (can_atomic_load_p): New function.
33867 * optabs-query.h (can_atomic_load_p): Declare it.
33868 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
33869 no fast atomic load is available for the particular size of access.
33870 (expand_atomic_compare_and_swap): Likewise.
33871 (expand_atomic_load): Likewise.
33872 (expand_atomic_store): Likewise.
33873 (expand_atomic_fetch_op): Likewise.
33874 * testsuite/lib/target-supports.exp
33875 (check_effective_target_sync_int_128): Remove x86 because it provides
33876 no fast atomic load.
33877 (check_effective_target_sync_int_128_runtime): Likewise.
33879 2017-02-01 Richard Biener <rguenther@suse.de>
33881 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33882 (graphite_transform_loops): Provide opt-info for optimized nests.
33883 * tree-parloop.c (parallelize_loops): Provide opt-info for
33884 parallelized loops.
33886 2017-02-01 Richard Biener <rguenther@suse.de>
33888 PR middle-end/79315
33889 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33890 was not set before.
33892 2017-02-01 Richard Biener <rguenther@suse.de>
33894 PR tree-optimization/71824
33895 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33896 Verify the loops are valid in the merged SESE region.
33897 (scop_detection::can_represent_loop_1): Check analyzing the
33898 evolution of the number of iterations in the region succeeds.
33900 2017-01-31 Ian Lance Taylor <iant@golang.org>
33902 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33903 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33905 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33907 PR preprocessor/79210
33908 * input.c (get_substring_ranges_for_loc): Replace line_width
33909 assertion with error-handling.
33911 2017-01-31 Richard Biener <rguenther@suse.de>
33913 PR tree-optimization/77318
33914 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33915 (create_pw_aff_from_tree): Take loop parameter.
33916 (add_condition_to_pbb): Pass loop of the condition to
33917 create_pw_aff_from_tree.
33919 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33921 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33922 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33924 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33928 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33930 (convert_int_to_float128): Likewise.
33931 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33932 (convert_int_to_float128): Likewise.
33933 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33934 (UNSPEC_IEEE128_CONVERT): Likewise.
33935 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33936 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33937 Use local variables for IBM extended format.
33938 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33939 (fix_trunc<mode>si2_fprs): Likewise.
33940 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33941 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33942 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33943 to know that we can now have integers of all sizes in vector
33945 (fix<uns>_<mode>di2_hw): Likewise.
33946 (float<uns>_<mode>si2_hw): Likewise.
33947 (fix_<mode>si2_hw): Likewise.
33948 (fixuns_<mode>si2_hw): Likewise.
33949 (float<uns>_<mode>di2_hw): Likewise.
33950 (float_<mode>di2_hw): Likewise.
33951 (float_<mode>si2_hw): Likewise.
33952 (floatuns_<mode>di2_hw): Likewise.
33953 (floatuns_<mode>si2_hw): Likewise.
33954 (xscvqp<su>wz_<mode>): Delete, no longer used.
33955 (xscvqp<su>dz_<mode>): Likewise.
33956 (xscv<su>dqp_<mode>): Likewise.
33957 (ieee128_mfvsrd_64bit): Likewise.
33958 (ieee128_mfvsrd_32bit): Likewise.
33959 (ieee128_mfvsrwz): Likewise.
33960 (ieee128_mtvsrw): Likewise.
33961 (ieee128_mtvsrd_64bit): Likewise.
33962 (ieee128_mtvsrd_32bit): Likewise.
33964 2017-01-31 Martin Liska <mliska@suse.cz>
33967 * ipa-prop.c (ipa_free_all_node_params): Call release method
33968 instead of ~sumbol_summary to not to trigger double times
33971 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33973 PR tree-optimization/71691
33974 * bitmap.h (class auto_bitmap): New.
33975 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33976 is_maybe_undefined instead of ssa_undefined_value_p.
33978 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33980 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33981 __S390_ARCH_LEVEL__ to __ARCH__.
33983 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33985 PR tree-optimization/79267
33986 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33987 if should_remove_lhs_p is true.
33989 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33992 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33993 (add_alignment_attribute): New.
33994 (base_type_die): Add alignment attribute.
33995 (subrange_type_die): Likewise.
33996 (modified_type_die): Likewise.
33997 (gen_array_type_die): Likewise.
33998 (gen_descr_array_type_die: Likewise.
33999 (gen_enumeration_type_die): Likewise.
34000 (gen_subprogram_die): Likewise.
34001 (gen_variable_die): Likewise.
34002 (gen_field_die): Likewise.
34003 (gen_ptr_to_mbr_type_die): Likewise.
34004 (gen_struct_or_union_type_die): Likewise.
34005 (gen_subroutine_type_die): Likewise.
34006 (gen_typedef_die): Likewise.
34007 (base_type_cmp): Compare alignment attribute.
34009 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34012 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34013 (setb_unsigned) New pattern for setb with CCUNS.
34014 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34015 subfc./subfe sequence to avoid overflow problems. Generate a
34016 shorter sequence with cmpld/setb for power9.
34017 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34018 for generating subfc. instruction.
34019 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34020 now uses this instruction.
34022 2017-01-30 Ian Lance Taylor <iant@google.com>
34025 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34026 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34028 2017-01-30 Martin Sebor <msebor@redhat.com>
34030 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34031 Move constant to the right of a relational operator.
34032 (get_mpfr_format_length, format_character, format_string): Ditto.
34033 (should_warn_p, maybe_warn): Same.
34035 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34037 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34040 * asan.c (get_translation_unit_decl): Remove function.
34041 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34043 2017-01-30 Martin Liska <mliska@suse.cz>
34045 PR gcov-profile/79259
34046 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34047 -fprofile-generate.
34049 2017-01-30 Martin Liska <mliska@suse.cz>
34052 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34053 Initialize variables with NULL value.
34055 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34058 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34060 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34062 2017-01-30 Richard Biener <rguenther@suse.de>
34064 PR tree-optimization/79276
34065 * tree-vrp.c (process_assert_insertions): Properly adjust common
34066 when removing a duplicate.
34068 2017-01-30 Richard Biener <rguenther@suse.de>
34070 PR tree-optimization/79256
34071 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34072 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34074 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34076 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34079 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34080 ("*r<noxa>sbg_<mode>_sll_bitmask")
34081 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34082 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34083 Use contiguous_bitmask_nowrap_operand.
34085 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34088 * config/rs6000/altivec.h (vec_xl): Revise #define.
34089 (vec_xst): Likewise.
34091 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34093 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34095 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34097 PR rtl-optimization/79194
34098 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34099 traps before call to bypass_conditional_jumps.
34101 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34103 PR tree-optimization/71374
34104 * lra-constraints.c (check_conflict_input_operands): New.
34105 (match_reload): Use it.
34107 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34110 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34111 account to calculate conflict_set.
34113 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34115 PR rtl-optimization/78559
34116 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34117 other_insn in combine.
34119 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34121 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34122 uint16_type_node for BT_UINT16.
34124 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34126 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34127 "RTL Tests" to menu.
34128 (GIMPLE Tests): New node.
34129 (RTL Tests): New node.
34131 2017-01-27 Richard Biener <rguenther@suse.de>
34133 PR tree-optimization/79245
34134 * tree-loop-distribution.c (distribute_loop): Apply cost
34135 modeling also to detected patterns.
34137 2017-01-27 Richard Biener <rguenther@suse.de>
34139 PR tree-optimization/71433
34140 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34141 (compare_assert_loc): New function.
34142 (process_assert_insertions): Sort and optimize assert locations
34143 to remove duplicates and push down identical assertions on
34144 edges to their destination block.
34146 2017-01-27 Richard Biener <rguenther@suse.de>
34148 PR tree-optimization/79244
34149 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34150 out SSA names even if abnormal.
34152 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34154 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34155 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34156 instead of MPFR_RNDN.
34158 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34161 * arm.c (arm_option_override): Don't call build_target_option_node
34162 until after doing all option overrides.
34163 (arm_valid_target_attribute_tree): Likewise.
34165 2017-01-27 Martin Liska <mliska@suse.cz>
34167 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34168 for {cd}tors and C++ {cd}tors.
34170 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34172 * config/s390/s390.md ("*setmem_long_and")
34173 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34175 2017-01-26 Martin Sebor <msebor@redhat.com>
34177 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34180 2017-01-26 Martin Sebor <msebor@redhat.com>
34182 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34183 HAVE_DFmode before using XFmode or DFmode.
34184 (parse_directive): Avoid using the z length modifier to avoid
34185 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34187 PR middle-end/78703
34188 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34189 to accept adjustment as an array.
34190 (get_int_range): New function.
34191 (struct directive): Make width and prec arrays.
34192 (directive::set_width, directive::set_precision): Call get_int_range.
34193 (format_integer, format_floating): Handle width and precision ranges.
34194 (format_string, parse_directive): Same.
34196 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34199 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34200 comdat_type_p set, just clone them, but keep the children in the
34204 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34205 which have direct callers with -fvar-tracking-assignments enabled
34207 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34208 inside of type units.
34210 2017-01-26 Martin Sebor <msebor@redhat.com>
34212 PR middle-end/78703
34213 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34215 (struct format_result): Replace number_chars, number_chars_min,
34216 and number_chars_max with a single member of struct result_range.
34218 (format_result::operator+=): Adjust.
34219 (struct fmtresult): Remove bounded. Handle likely and unlikely
34221 (fmtresult::adjust_for_width_or_precision): New function.
34222 (fmtresult:type_max_digits): New function.
34223 (bytes_remaining): Handle likely and unlikely counters.
34224 (min_bytes_remaining): Remove.
34225 (format_percent): Simplify.
34226 (format_integer, format_floating): Set likely and unlikely counters.
34227 (get_string_length, format_character, format_string): Same.
34228 (format_plain, should_warn_p): New function.
34229 (maybe_warn): Call should_warn_p. Update diagnostic messages
34230 and handle those for all directives, including plain strings.
34231 (format_directive): Handle likely and unlikely counters.
34232 Remove unnecessary quoting from diagnostics. Add an informational
34234 (add_bytes): Remove.
34235 (pass_sprintf_length::compute_format_length): Simplify.
34236 (try_substitute_return_value): Handle likely and unlikely counters.
34238 2017-01-26 Carl Love <cel@us.ibm.com>
34240 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34241 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34243 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34246 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34247 endianess for subregs into account.
34248 * lra-constraints.c (lra_constraints): Do risky transformations
34249 always on the first iteration.
34250 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34251 last_call_used_reg_set.
34252 (process_bb_lives): Define and use last_call_used_reg_set.
34253 * lra.c (lra): Always continue after lra_constraints on the first
34256 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34258 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34260 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34262 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34264 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34265 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34266 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34267 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34268 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34269 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34270 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34271 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34272 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34274 2017-01-26 Marek Polacek <polacek@redhat.com>
34277 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34278 for the third operand.
34280 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34282 PR middle-end/79236
34283 * omp-low.c (struct omp_context): Add simt_stmt field.
34284 (scan_omp_for): Return omp_context *.
34285 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34286 context to the _simt_ SIMD stmt.
34287 (lower_omp_for): For combined SIMD with sibling _simt_
34288 SIMD, make sure to use the same decls in _looptemp_
34289 clauses as in the sibling.
34291 2017-01-26 David Sherwood <david.sherwood@arm.com>
34293 PR middle-end/79212
34294 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34297 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34300 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34301 emit fld b; fld a; if possible.
34303 * brig-builtins.def: Update copyright years.
34304 * config/arm/arm_acle_builtins.def: Update copyright years.
34306 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34309 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34310 constraint instead of o for the stxsd instruction.
34312 2017-01-25 Carl Love <cel@us.ibm.com>
34314 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34315 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34317 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34319 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34321 2017-01-25 Richard Biener <rguenther@suse.de>
34323 PR tree-optimization/69264
34324 * target.def (vector_alignment_reachable): Improve documentation.
34325 * doc/tm.texi: Regenerate.
34326 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34328 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34329 earlier changes with respect to TYPE_USER_ALIGN.
34330 (vector_alignment_reachable_p): Likewise. Improve dumping.
34332 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34335 * config/arm/arm.md (xordi3): Force constant operand into a register
34338 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34340 * doc/invoke.texi (-fstore-merging): Correct default optimization
34341 levels at which it is enabled.
34342 (-O): Move -fstore-merging from list to...
34345 2017-01-25 Richard Biener <rguenther@suse.de>
34348 * omp-expand.c: Include debug.h.
34349 (expand_omp_taskreg): Make sure to generate early debug before
34350 outlining anything from a function.
34351 (expand_omp_target): Likewise.
34352 (grid_expand_target_grid_body): Likewise.
34354 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34357 * asan.c (get_translation_unit_decl): New function.
34358 (asan_add_global): Extract modules file name from globals
34359 TRANSLATION_UNIT_DECL name.
34361 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34364 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34365 for long calls with APCS frame and VFP.
34367 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34369 * cfg.c (original_copy_tables_initialized_p): New function.
34370 * cfg.h (original_copy_tables_initialized_p): New decl.
34371 * cfgrtl.c (relink_block_chain): Guard the call to
34372 free_original_copy_tables with a call to
34373 original_copy_tables_initialized_p.
34374 * cgraph.h (symtab_node::native_rtl_p): New decl.
34375 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34376 (symtab_node::needed_p): Don't assert for early assembly output
34377 for __RTL functions.
34378 (cgraph_node::finalize_function): Set "force_output" for __RTL
34380 (cgraph_node::analyze): Bail out early for __RTL functions.
34381 (analyze_functions): Update assertion to support __RTL functions.
34382 (cgraph_node::expand): Bail out early for __RTL functions.
34383 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34385 * function.h (struct function): Update comment for field
34387 * gimple-expr.c: Include "tree-pass.h".
34388 (gimple_has_body_p): Return false for __RTL functions.
34389 * Makefile.in (OBJS): Add run-rtl-passes.o.
34390 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34392 (gcc::pass_manager::get_clean_slate): New accessor.
34393 * passes.c: Include "insn-addr.h".
34394 (should_skip_pass_p): Add logging. Update logic for running
34395 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34396 property-provider override so it is only done for gimple passes.
34398 (skip_pass): New function.
34399 (execute_one_pass): Call skip_pass when skipping passes.
34400 * read-md.c (md_reader::read_char): Support filtering
34401 the input to a subset of line numbers.
34402 (md_reader::md_reader): Initialize fields
34403 m_first_line and m_last_line.
34404 (md_reader::read_file_fragment): New function.
34405 * read-md.h (md_reader::read_file_fragment): New decl.
34406 (md_reader::m_first_line): New field.
34407 (md_reader::m_last_line): New field.
34408 * read-rtl-function.c (function_reader::create_function): Only
34409 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34410 curr_properties. Set DECL_INITIAL to a dummy block.
34411 (read_rtl_function_body_from_file_range): New function.
34412 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34414 * run-rtl-passes.c: New file.
34415 * run-rtl-passes.h: New file.
34417 2017-01-24 Jeff Law <law@redhat.com>
34419 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34422 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34424 PR tree-optimization/79159
34425 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34426 (record_nonwrapping_iv): Improve boundary using above function if no
34427 value range information.
34429 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34430 Martin Jambor <mjambor@suse.cz>
34432 * brig-builtins.def: New file.
34433 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34434 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34435 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34436 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34437 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34438 * builtin-types.def (BT_INT8): New.
34439 (BT_INT16): Likewise.
34440 (BT_UINT8): Likewise.
34441 (BT_UINT16): Likewise.
34442 (BT_FN_ULONG): Likewise.
34443 (BT_FN_UINT_INT): Likewise.
34444 (BT_FN_UINT_ULONG): Likewise.
34445 (BT_FN_UINT_LONG): Likewise.
34446 (BT_FN_UINT_PTR): Likewise.
34447 (BT_FN_ULONG_PTR): Likewise.
34448 (BT_FN_INT8_FLOAT): Likewise.
34449 (BT_FN_INT16_FLOAT): Likewise.
34450 (BT_FN_UINT32_FLOAT): Likewise.
34451 (BT_FN_UINT16_FLOAT): Likewise.
34452 (BT_FN_UINT8_FLOAT): Likewise.
34453 (BT_FN_UINT64_FLOAT): Likewise.
34454 (BT_FN_UINT16_UINT32): Likewise.
34455 (BT_FN_UINT32_UINT16): Likewise.
34456 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34457 (BT_FN_INT_PTR_INT): Likewise.
34458 (BT_FN_UINT_PTR_UINT): Likewise.
34459 (BT_FN_LONG_PTR_LONG): Likewise.
34460 (BT_FN_ULONG_PTR_ULONG): Likewise.
34461 (BT_FN_VOID_UINT64_UINT64): Likewise.
34462 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34463 (BT_FN_INT8_INT8_INT8): Likewise.
34464 (BT_FN_INT16_INT16_INT16): Likewise.
34465 (BT_FN_INT_INT_INT): Likewise.
34466 (BT_FN_UINT_FLOAT_UINT): Likewise.
34467 (BT_FN_FLOAT_UINT_UINT): Likewise.
34468 (BT_FN_ULONG_UINT_UINT): Likewise.
34469 (BT_FN_ULONG_UINT_PTR): Likewise.
34470 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34471 (BT_FN_UINT_UINT_UINT): Likewise.
34472 (BT_FN_VOID_UINT_PTR): Likewise.
34473 (BT_FN_UINT_UINT_PTR: Likewise.
34474 (BT_FN_UINT32_UINT64_PTR): Likewise.
34475 (BT_FN_INT_INT_UINT_UINT): Likewise.
34476 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34477 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34478 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34479 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34480 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34481 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34482 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34483 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34484 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34485 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34486 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34487 * doc/frontends.texi: List BRIG FE.
34488 * doc/install.texi (Testing): Add BRIG tesring requirements.
34489 * doc/invoke.texi (Overall Options): Mention BRIG.
34490 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34492 2017-01-24 Richard Biener <rguenther@suse.de>
34494 PR translation/79208
34495 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34497 2017-01-24 Martin Jambor <mjambor@suse.cz>
34500 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34501 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34502 and known_contexts.
34504 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34506 PR middle-end/79123
34507 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34508 casts from signed to unsigned really don't have a range.
34510 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34512 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34513 GMP_RNDx for compatiblity.
34515 2017-01-24 Martin Liska <mliska@suse.cz>
34518 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34519 that would prevent us to call alloca with -1 as argument.
34521 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34523 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34524 Avoid -Wformat-security warning.
34526 2017-01-23 Andrew Pinski <apinski@cavium.com>
34528 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34531 2017-01-23 Martin Sebor <msebor@redhat.com>
34533 PR middle-end/78703
34534 * gimple-ssa-sprintf.c (warn_level): New global.
34535 (format_integer): Use it here and throughout the rest of the file.
34536 Use the same switch to compute sign as base.
34537 (maybe_warn): New function.
34538 (format_directive): Factor out warnings into maybe_warn.
34539 Add debugging output. Use warn_level.
34540 (add_bytes): Use warn_level.
34541 (pass_sprintf_length::compute_format_length): Add debugging output.
34542 (try_substitute_return_value): Same.
34543 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34545 PR middle-end/78703
34546 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34547 (struct fmtresult, format_integer, format_floating): Adjust.
34548 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34549 (get_string_length, format_string,format_directive): Same.
34550 (pass_sprintf_length::compute_format_length): Same.
34551 (try_substitute_return_value): Simplify slightly.
34553 PR middle-end/78703
34554 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34555 (fmtresult::operator+=): Outlined.
34556 (struct fmtresult): Add ctors.
34557 (struct conversion_spec): Rename...
34558 (struct directive): ...to this. Add and remove data members.
34559 (directive::set_width, directive::set_precision): New functions.
34560 (format_percent): Use fmtresult ctor.
34561 (get_width_and_precision): Remove.
34562 (format_integer): Make naming changes. Avoid computing width and
34564 (format_floating): Same. Adjust indentation.
34565 (format_character, format_none): New functions.
34566 (format_string): Moved character handling to format_character.
34567 (format_directive): Remove arguments, change return type.
34568 (parse_directive): New function.
34569 (pass_sprintf_length::compute_format_length): Move directive
34570 parsing to parse_directive.
34572 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34574 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34575 (assign_assembler_name_if_needed): ... this.
34576 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34577 (assign_assembler_name_if_needed): ... this.
34578 (free_lang_data_in_cgraph): Adjust callers.
34579 * cgraphunit.c (cgraph_node::analyze): Likewise.
34580 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34583 2017-01-23 Richard Biener <rguenther@suse.de>
34585 PR tree-optimization/79088
34586 PR tree-optimization/79188
34587 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34588 resetting loop bounds after last path deletion. Reset loop
34589 bounds of the target loop, make code match the comments.
34590 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34591 Make sure loops need no fixups.
34593 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34595 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34596 exponent support with double type for first argument.
34597 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34598 type returned by __builtin_vec_extract_sig,
34599 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34600 functions from "vector int" to "vector unsigned int" or from
34601 "vector long long int" to "vector unsigned long long int".
34602 Changed type returned by __builtin_vec_extract_exp,
34603 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34604 functions from "vector int" to "vector unsigned int" or from
34605 "vector long long int" to "vector unsigned long long int".
34606 Changed return type of __builtin_vec_test_data_class,
34607 __builtin_vec_test_data_class_sp, and
34608 __builtin_vec_test_data_class_dp from "vector int" to
34609 "vector bool int" or from "vector long long int" to "vector bool
34610 long long int" and changed second argument type from "unsigned
34611 int" to "int". Added new overloaded function forms "vector float
34612 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34613 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34614 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34615 double, vector unsigned long long int)" and "vector double
34616 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34617 long int)". Changed return type of
34618 __builtin_scalar_test_data_class and
34619 __builtin_scalar_test_data_class_sp and
34620 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34621 int" and changed second argument from "unsigned int" to "int".
34622 Changed type returned by __builtin_scalar_test_neg,
34623 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34624 from "int" to "bool int". Added new overloaded function form
34625 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34626 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34627 exponent double-precision with floating point first argument.
34628 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34629 documentation of scalar_test_data_class, scalar_test_neg,
34630 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34631 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34632 vec_test_data_class built-in functions to reflect refinements in
34633 their type signatures.
34635 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34637 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34639 (aarch64_elf_asm_destructor): Likewise.
34641 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34643 PR rtl-optimization/78634
34644 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34645 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34646 * ifcvt.c (noce_try_cmove): Add missing cost check.
34648 PR rtl-optimization/71724
34649 * combine.c (if_then_else_cond): Look for situations where it is
34650 beneficial to undo the work of one of the recursive calls.
34652 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34654 PR tree-optimization/70754
34655 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34656 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34657 combined stmt before it if not NULL.
34658 (combine_chains): Process refs reversely and compute dominance point
34661 2017-01-23 Martin Liska <mliska@suse.cz>
34663 PR tree-optimization/79196
34664 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34665 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34667 (strlen_optimize_stmt): Call the renamed function.
34669 2017-01-23 Michael Matz <matz@suse.de>
34671 PR tree-optimization/78384
34672 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34674 2017-01-23 Richard Biener <rguenther@suse.de>
34676 PR tree-optimization/79186
34677 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34678 both incoming edges before moving an assert.
34680 2017-01-23 Martin Jambor <mjambor@suse.cz>
34682 * ipa-prop.c (load_from_param_1): Removed.
34683 (load_from_unmodified_param): Bits from load_from_param_1 put back
34685 (load_from_param): Removed.
34686 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34687 with stmt. Reverted back to use of load_from_unmodified_param.
34689 2017-01-23 Martin Jambor <mjambor@suse.cz>
34692 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34693 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34694 field a pointer to garbage collected vector, mark lattices and
34695 ipcp_orig_node with GTY((skip)).
34696 (ipa_get_param_count): Adjust to descriptors being a pointer.
34697 (ipa_get_param): Likewise.
34698 (ipa_get_type): Likewise.
34699 (ipa_get_param_move_cost): Likewise.
34700 (ipa_set_param_used): Likewise.
34701 (ipa_get_controlled_uses): Likewise.
34702 (ipa_set_controlled_uses): Likewise.
34703 (ipa_is_param_used): Likewise.
34704 (ipa_node_params_t): Move into garbage collector. New methods insert
34706 (ipa_node_params_sum): Annotate wth GTY(()).
34707 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
34709 (ipa_load_from_parm_agg): Adjust declaration.
34710 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
34711 * ipa-profile.c (ipa_profile): Likewise.
34712 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
34713 (ipa_populate_param_decls): Make descriptors parameter garbage
34715 (ipa_dump_param): Adjust to descriptors being a pointer.
34716 (ipa_alloc_node_params): Likewise.
34717 (ipa_initialize_node_params): Likewise.
34718 (load_from_param_1): Make descriptors parameter garbage collected.
34719 (load_from_unmodified_param): Likewise.
34720 (load_from_param): Likewise.
34721 (ipa_load_from_parm_agg): Likewise.
34722 (ipa_node_params::~ipa_node_params): Removed.
34723 (ipa_free_all_node_params): Remove call to delete operator.
34724 (ipa_node_params_t::insert): New.
34725 (ipa_node_params_t::remove): Likewise.
34726 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
34727 copy known_csts and known_contexts vectors.
34728 (ipa_read_node_info): Adjust to descriptors being a pointer.
34729 (ipcp_modif_dom_walker): Make m_descriptors field garbage
34731 (ipcp_transform_function): Make descriptors variable garbage
34734 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
34736 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
34737 * config/i386/avx512dqintrin.h: Ditto.
34738 * config/i386/avx512fintrin.h: Ditto.
34739 * config/i386/i386.c: Handle new builtins.
34740 * config/i386/i386-builtin.def: Add new builtins.
34741 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
34742 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
34744 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34745 Martin Liska <mliska@suse.cz>
34747 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
34748 * asan.c (asan_expand_poison_ifn): Support stores and use
34749 appropriate ASAN report function.
34750 * internal-fn.c (expand_ASAN_POISON_USE): New function.
34751 * internal-fn.def (ASAN_POISON_USE): Declare.
34752 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
34753 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
34754 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
34755 ASAN_POISON calls w/o LHS.
34756 * tree-ssa.c (execute_update_addresses_taken): Create clobber
34757 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
34758 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
34759 * gimplify.c (asan_poison_variables): Add attribute
34760 use_after_scope_memory to variables that really needs to live
34762 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
34763 having the attribute.
34765 2017-01-23 Martin Liska <mliska@suse.cz>
34767 * asan.c (create_asan_shadow_var): New function.
34768 (asan_expand_poison_ifn): Likewise.
34769 * asan.h (asan_expand_poison_ifn): New declaration.
34770 * internal-fn.c (expand_ASAN_POISON): Likewise.
34771 * internal-fn.def (ASAN_POISON): New builtin.
34772 * sanopt.c (pass_sanopt::execute): Expand
34773 asan_expand_poison_ifn.
34774 * tree-inline.c (copy_decl_for_dup_finish): Make function
34776 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
34777 * tree-ssa.c (is_asan_mark_p): New function.
34778 (execute_update_addresses_taken): Rewrite local variables
34779 (identified just by use-after-scope as addressable) into SSA.
34781 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34783 * doc/install.texi (Specific): opensource.apple.com uses https
34784 now. Remove trailing slash.
34786 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
34788 * README.Portability: Remove note on an Irix compatibility issue.
34790 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
34792 * gcov.c (INCLUDE_ALGORITHM): Define.
34793 (INCLUDE_VECTOR): Define.
34794 No longer include <vector> and <algorithm> directly.
34796 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
34798 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
34800 * doc/invoke.texi (Code Gen Options): Ditto.
34802 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
34805 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
34807 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
34809 rtl-optimization/79125
34810 * cprop.c (local_cprop_pass): Handle cases where we make an
34811 unconditional trap.
34813 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
34817 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
34818 read from, for big endian.
34820 2017-01-20 Jiong Wang <jiong.wang@arm.com>
34822 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
34823 register pauth builtins for LP64 only.
34825 2017-01-20 Marek Polacek <polacek@redhat.com>
34828 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
34831 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34833 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
34835 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
34836 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
34837 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
34839 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34842 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
34843 in big-endian mode when they are not a single duplicated value.
34845 2017-01-20 Richard Biener <rguenther@suse.de>
34847 * BASE-VER: Bump to 7.0.1.
34849 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
34851 * omp-low.c (omplow_simd_context): New struct. Use it...
34852 (lower_rec_simd_input_clauses): ...here and...
34853 (lower_rec_input_clauses): ...here to hold common data. Adjust all
34854 references to idx, lane, max_vf, is_simt.
34856 2017-01-20 Graham Markall <graham.markall@embecosm.com>
34858 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
34861 2017-01-20 Martin Jambor <mjambor@suse.cz>
34863 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
34864 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
34866 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
34867 (GTFILES): Rename hsa.c to hsa-common.c.
34868 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
34869 * hsa-dump.c: Likewise.
34870 * hsa-gen.c: Likewise.
34871 * hsa-regalloc.c: Likewise.
34872 * ipa-hsa.c: Likewise.
34873 * omp-expand.c: Likewise.
34874 * omp-low.c: Likewise.
34875 * toplev.c: Likewise.
34877 2017-01-20 Marek Polacek <polacek@redhat.com>
34880 * doc/invoke.texi: Document -Wduplicated-branches.
34881 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34882 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34883 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34884 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34885 return 0 only when not OEP_LEXICOGRAPHIC.
34886 (fold_build_cleanup_point_expr): Use the expression
34887 location when building CLEANUP_POINT_EXPR.
34888 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34889 * tree.c (add_expr): Handle error_mark_node.
34891 2017-01-20 Martin Liska <mliska@suse.cz>
34894 * tree-profile.c (init_ic_make_global_vars): Do not call
34896 (gimple_init_gcov_profiler): Likewise.
34898 2017-01-20 Martin Liska <mliska@suse.cz>
34901 * cgraph.h (maybe_create_reference): Remove argument and
34903 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34905 * ipa-cp.c (create_specialized_node): Likewise.
34906 * symtab.c (symtab_node::maybe_create_reference): Handle
34907 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34909 2017-01-20 Martin Liska <mliska@suse.cz>
34911 * read-rtl-function.c (function_reader::create_function): Use
34912 build_decl instread of build_decl_stat.
34914 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34916 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34917 * config/i386/avx512dqintrin.h: Ditto.
34918 * config/i386/avx512fintrin.h: Ditto.
34919 * config/i386/i386-builtin-types.def: Add new types.
34920 * config/i386/i386.c: Handle new types.
34921 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34922 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34923 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34924 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34925 (__builtin_ia32_kshiftridi): New.
34926 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34928 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34932 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34933 define to rs6000_init_stack_protect_guard.
34934 (rs6000_init_stack_protect_guard): New function.
34936 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34937 Yunqiang Su <yunqiang.su@imgtec.com>
34939 * config.gcc (supported_defaults): Add madd4.
34940 (with_madd4): Add validation.
34941 (all_defaults): Add madd4.
34942 * config/mips/mips.opt (mmadd4): New option.
34943 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34945 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34947 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34948 (ISA_HAS_FUSED_MADD4): Likewise.
34949 * doc/invoke.texi (-mmadd4): Document the new option.
34950 * doc/install.texi (--with-madd4): Document the new option.
34952 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34954 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34955 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34956 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34957 (aarch64_init_pauth_hint_builtins): New.
34958 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34959 (aarch64_expand_builtin): Expand new builtins.
34961 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34963 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34964 * combine-stack-adj.c (no_unhandled_cfa): Handle
34965 REG_CFA_TOGGLE_RA_MANGLE.
34966 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34967 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34968 info for return address signing.
34969 (aarch64_expand_epilogue): Likewise.
34971 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34973 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34974 * config/aarch64/aarch64-protos.h
34975 (aarch64_return_address_signing_enabled): New declaration.
34976 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34978 (aarch64_expand_prologue): Sign return address before it's pushed onto
34980 (aarch64_expand_epilogue): Authenticate return address fetched from
34982 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34983 (aarch64_attributes): New function attributes for "sign-return-address".
34984 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34985 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34986 ("*do_return"): Generate combined instructions according to key index.
34987 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34988 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34990 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34991 * config/aarch64/aarch64.opt (msign-return-address=): New.
34992 * doc/extend.texi (AArch64 Function Attributes): Documents
34993 "sign-return-address=".
34994 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34996 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34998 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34999 overall option summary.
35001 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35003 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35004 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35005 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35006 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35008 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35010 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35011 -mpower9-minmax by default for -mcpu=power9.
35012 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35013 128-bit floating point.
35015 2017-01-20 Alan Modra <amodra@gmail.com>
35017 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35018 optimizing for size.
35020 2017-01-20 Alan Modra <amodra@gmail.com>
35023 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35024 for strcmp and strncmp from corresponding builtin decl.
35026 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35028 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35029 instead of i386/rtems-64.h.
35030 * config/i386/rtems-64.h: Remove.
35032 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35036 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35038 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35040 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35042 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35043 Change int to HOST_WIDE_INT.
35044 * config/aarch64/aarch64-protos.h
35045 (aarch64_simd_gen_const_vector_dup): Likewise.
35046 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35048 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35050 * langhooks-def.h (lhd_type_for_size): New decl.
35051 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35052 * langhooks.c (lhd_type_for_size): New function, taken from
35055 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35057 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35058 define_bypass for CR latency.
35059 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35060 (power9-alu2): Add define_bypass for CR latency.
35062 (power9-mul): Update insn latency.
35063 (power9-mul-compare): Update insn latency, bypass latency and remove
35066 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35068 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35070 * config/aarch64/aarch64.md
35071 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35072 aarch64_nopcrelative_literal_loads.
35073 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35075 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35077 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35078 TARGET_LOONGSON_3A.
35079 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35081 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35084 * config.gcc (supported_defaults): Add lxc1-sxc1.
35085 (with_lxc1_sxc1): Add validation.
35086 (all_defaults): Add lxc1-sxc1.
35087 * config/mips/mips.opt (mlxc1-sxc1): New option.
35088 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35090 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35091 __mips_no_lxc1_sxc1.
35092 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35093 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35094 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35096 2017-01-19 Richard Biener <rguenther@suse.de>
35098 PR tree-optimization/72488
35099 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35100 sure to restore SSA info.
35101 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35103 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35105 PR rtl-optimization/79121
35106 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35107 of the inner type when shifting an extended value.
35109 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35112 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35113 interposable aliases.
35115 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35118 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35119 Use the evmergelohi instruction.
35120 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35121 (mov_sitf_e500_subreg8_2_be): Likewise.
35122 (mov_sitf_e500_subreg12_2_le): Likewise.
35123 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35124 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35125 (mov_sitf_e500_subreg8_2_le): Likewise.
35126 (mov_sitf_e500_subreg12_2_be): Likewise.
35128 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35130 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35131 attribute from vecsimple to vecperm.
35132 (altivec_vbpermq2): Likewise.
35134 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35137 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35139 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35140 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35141 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35142 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35143 case where N arg is SIZE_MAX.
35144 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35145 (cmpstrsi): Add pattern.
35147 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35149 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35150 __builtin_vec_revb builtins.
35151 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35152 built-in functions to support generation of the ISA 3.0 XXBR<x>
35153 vector byte reverse instructions.
35154 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35155 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35156 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35157 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35158 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35159 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35160 (P9V_BUILTIN_VEC_REVB): Likewise.
35161 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35162 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35163 (p9_xxbrq_v16qi): Likewise.
35164 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35165 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35166 (p9_xxbrh_v8hi): Likewise.
35167 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35168 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35169 vec_revb built-in functions.
35171 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35173 PR rtl-optimization/78952
35174 * config/i386/i386.md (any_extract): New code iterator.
35175 (*insvqi_2): Use any_extract for source operand.
35176 (*insvqi_3): Use any_shiftrt for source operand.
35178 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35180 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35182 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35184 2017-01-18 Matthias Klose <doko@ubuntu.com>
35186 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35188 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35190 * config/rs6000/altivec.h (vec_bperm): Change #define.
35191 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35192 (altivec_vbpermq2): New define_insn.
35193 (altivec_vbpermd): Likewise.
35194 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35195 function interface.
35196 (VBPERMD): Likewise.
35197 (VBPERM): New polymorphic function interface.
35198 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35199 Add entries for P9V_BUILTIN_VEC_VBPERM.
35200 * doc/extend.texi: Add interfaces for vec_bperm.
35202 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35204 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35205 first letter of error messages.
35206 (s390_resolve_overloaded_builtin): Likewise.
35207 * config/s390/s390.c (s390_expand_builtin): Likewise.
35208 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35209 (s390_valid_target_attribute_inner_p): Likewise.
35210 * config/s390/s390.md ("tabort"): Likewise.
35212 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35214 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35215 (ISA_AVOID_DIV_HILO): New macro.
35216 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35217 (ISA_HAS_DDIV): Likewise.
35219 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35221 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35223 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35225 * doc/invoke.texi (fabi-version): Spelling fix.
35227 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35230 * doc/invoke.texi (fabi-version): Mention mangling fix for
35233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35236 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35238 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35241 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35242 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35244 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35246 (stack_protect_test): Ditto.
35247 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35248 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35250 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35251 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35252 -mstack-protector-guard-offset=.
35253 (RS/6000 and PowerPC Options): Ditto.
35255 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35257 * config/i386/i386.h (MASK_CLASS_P): New define.
35258 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35259 there are no registers from different register sets also when
35260 mask registers are used. Update function comment.
35261 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35262 to (*k/*r) and (*k/*km) alternatives.
35264 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35266 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35267 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35268 (EH_RETURN_HANDLER_RTX): New define.
35269 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35270 Force frame pointer in EH return functions.
35271 (aarch64_expand_epilogue): Add barrier for eh_return.
35272 (aarch64_final_eh_return_addr): Remove.
35273 (aarch64_eh_return_handler_rtx): New function.
35274 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35276 (aarch64_eh_return_handler_rtx): New prototype.
35278 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35280 * config/rs6000/altivec.h (vec_rlmi): New #define.
35281 (vec_vrlnm): Likewise.
35282 (vec_rlnm): Likewise.
35283 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35284 (UNSPEC_VRLNM): Likewise.
35285 (VIlong): New mode iterator.
35286 (altivec_vrl<VI_char>mi): New define_insn.
35287 (altivec_vrl<VI_char>nm): Likewise.
35288 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35290 (VRLDNM): Likewise.
35291 (RLNM): New polymorphic function entry.
35292 (VRLWMI): New monomorphic function entry.
35293 (VRLDMI): Likewise.
35294 (RLMI): New polymorphic function entry.
35295 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35296 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35297 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35300 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35303 * dwarf2out.c (field_byte_offset): Restore the
35304 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35305 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35306 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35309 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35312 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35314 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35317 * dwarf2out.c (add_data_member_location_attribute): For constant
35318 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35319 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35320 DW_AT_byte_size attributes.
35322 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35324 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35325 after forcing to constant memory when the code model is medium.
35327 2017-01-17 Julia Koval <julia.koval@intel.com>
35330 * config/i386/avx512fintrin.h
35331 (_mm512_i32gather_ps): Change __addr type to void const*.
35332 (_mm512_mask_i32gather_ps): Ditto.
35333 (_mm512_i32gather_pd): Ditto.
35334 (_mm512_mask_i32gather_pd): Ditto.
35335 (_mm512_i64gather_ps): Ditto.
35336 (_mm512_mask_i64gather_ps): Ditto.
35337 (_mm512_i64gather_pd): Ditto.
35338 (_mm512_mask_i64gather_pd): Ditto.
35339 (_mm512_i32gather_epi32): Ditto.
35340 (_mm512_mask_i32gather_epi32): Ditto.
35341 (_mm512_i32gather_epi64): Ditto.
35342 (_mm512_mask_i32gather_epi64): Ditto.
35343 (_mm512_i64gather_epi32): Ditto.
35344 (_mm512_mask_i64gather_epi32): Ditto.
35345 (_mm512_i64gather_epi64): Ditto.
35346 (_mm512_mask_i64gather_epi64): Ditto.
35347 (_mm512_i32scatter_ps): Change __addr type to void*.
35348 (_mm512_mask_i32scatter_ps): Ditto.
35349 (_mm512_i32scatter_pd): Ditto.
35350 (_mm512_mask_i32scatter_pd): Ditto.
35351 (_mm512_i64scatter_ps): Ditto.
35352 (_mm512_mask_i64scatter_ps): Ditto.
35353 (_mm512_i64scatter_pd): Ditto.
35354 (_mm512_mask_i64scatter_pd): Ditto.
35355 (_mm512_i32scatter_epi32): Ditto.
35356 (_mm512_mask_i32scatter_epi32): Ditto.
35357 (_mm512_i32scatter_epi64): Ditto.
35358 (_mm512_mask_i32scatter_epi64): Ditto.
35359 (_mm512_i64scatter_epi32): Ditto.
35360 (_mm512_mask_i64scatter_epi32): Ditto.
35361 (_mm512_i64scatter_epi64): Ditto.
35362 (_mm512_mask_i64scatter_epi64): Ditto.
35363 * config/i386/avx512pfintrin.h
35364 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35365 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35366 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35367 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35368 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35369 (_mm512_prefetch_i32scatter_ps): Ditto.
35370 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35371 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35372 (_mm512_prefetch_i64scatter_pd): Ditto.
35373 (_mm512_prefetch_i64scatter_ps): Ditto.
35374 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35375 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35376 * config/i386/avx512vlintrin.h
35377 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35378 (_mm_mmask_i32gather_ps): Ditto.
35379 (_mm256_mmask_i32gather_pd): Ditto.
35380 (_mm_mmask_i32gather_pd): Ditto.
35381 (_mm256_mmask_i64gather_ps): Ditto.
35382 (_mm_mmask_i64gather_ps): Ditto.
35383 (_mm256_mmask_i64gather_pd): Ditto.
35384 (_mm_mmask_i64gather_pd): Ditto.
35385 (_mm256_mmask_i32gather_epi32): Ditto.
35386 (_mm_mmask_i32gather_epi32): Ditto.
35387 (_mm256_mmask_i32gather_epi64): Ditto.
35388 (_mm_mmask_i32gather_epi64): Ditto.
35389 (_mm256_mmask_i64gather_epi32): Ditto.
35390 (_mm_mmask_i64gather_epi32): Ditto.
35391 (_mm256_mmask_i64gather_epi64): Ditto.
35392 (_mm_mmask_i64gather_epi64): Ditto.
35393 (_mm256_i32scatter_ps): Change __addr type to void*.
35394 (_mm256_mask_i32scatter_ps): Ditto.
35395 (_mm_i32scatter_ps): Ditto.
35396 (_mm_mask_i32scatter_ps): Ditto.
35397 (_mm256_i32scatter_pd): Ditto.
35398 (_mm256_mask_i32scatter_pd): Ditto.
35399 (_mm_i32scatter_pd): Ditto.
35400 (_mm_mask_i32scatter_pd): Ditto.
35401 (_mm256_i64scatter_ps): Ditto.
35402 (_mm256_mask_i64scatter_ps): Ditto.
35403 (_mm_i64scatter_ps): Ditto.
35404 (_mm_mask_i64scatter_ps): Ditto.
35405 (_mm256_i64scatter_pd): Ditto.
35406 (_mm256_mask_i64scatter_pd): Ditto.
35407 (_mm_i64scatter_pd): Ditto.
35408 (_mm_mask_i64scatter_pd): Ditto.
35409 (_mm256_i32scatter_epi32): Ditto.
35410 (_mm256_mask_i32scatter_epi32): Ditto.
35411 (_mm_i32scatter_epi32): Ditto.
35412 (_mm_mask_i32scatter_epi32): Ditto.
35413 (_mm256_i32scatter_epi64): Ditto.
35414 (_mm256_mask_i32scatter_epi64): Ditto.
35415 (_mm_i32scatter_epi64): Ditto.
35416 (_mm_mask_i32scatter_epi64): Ditto.
35417 (_mm256_i64scatter_epi32): Ditto.
35418 (_mm256_mask_i64scatter_epi32): Ditto.
35419 (_mm_i64scatter_epi32): Ditto.
35420 (_mm_mask_i64scatter_epi32): Ditto.
35421 (_mm256_i64scatter_epi64): Ditto.
35422 (_mm256_mask_i64scatter_epi64): Ditto.
35423 (_mm_i64scatter_epi64): Ditto.
35424 (_mm_mask_i64scatter_epi64): Ditto.
35425 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35426 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35427 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35428 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35429 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35430 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35431 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35432 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35433 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35434 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35435 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35436 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35437 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35438 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35439 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35440 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35441 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35442 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35443 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35444 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35445 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35446 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35447 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35448 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35449 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35450 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35451 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35452 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35453 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35454 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35455 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35456 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35457 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35458 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35459 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35460 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35461 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35462 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35463 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35464 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35465 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35466 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35467 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35468 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35469 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35470 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35471 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35472 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35473 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35474 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35475 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35476 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35477 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35478 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35479 definitions accordingly.
35481 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35482 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35485 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35488 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35491 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35492 hard regs for inner regno.
35494 2017-01-17 Martin Liska <mliska@suse.cz>
35497 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35498 assumption and add comment.
35500 2017-01-17 Nathan Sidwell <nathan@acm.org>
35502 * ipa-visibility.c (localize_node): New function, broken out of ...
35503 (function_and_variable_visibility): ... here. Call it.
35505 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35507 PR middle-end/77445
35508 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35509 correctly set frequency of oudgoing edge.
35510 (duplicate_thread_path): Fix profile updating.
35512 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35515 * configure.ac: Add GCC_BASE_VER.
35516 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35517 version from BASE-VER file.
35518 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35519 (gcc.o): Depend on $(BASEVER).
35520 * common.opt (dumpfullversion): New option.
35521 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35522 * doc/invoke.texi: Document -dumpfullversion.
35523 * doc/install.texi: Document --with-gcc-major-version-only.
35524 * configure: Regenerated.
35526 2017-01-17 Richard Biener <rguenther@suse.de>
35528 PR tree-optimization/71433
35529 * tree-vrp.c (register_new_assert_for): Merge same asserts
35530 on all incoming edges.
35531 (process_assert_insertions_for): Handle insertions at the
35534 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35536 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35537 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35539 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35542 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35545 2017-01-17 Alan Modra <amodra@gmail.com>
35548 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35549 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35550 symbolic stack limit when pic.
35552 2017-01-16 Martin Sebor <msebor@redhat.com>
35554 PR tree-optimization/78608
35555 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35557 2017-01-16 Jeff Law <law@redhat.com>
35560 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35561 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35562 for several include directories that may be relative to sysroot.
35563 * config/i386/x-mingw32 (gplus_includedir): Define.
35564 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35565 (native_system_includedir): Likewise.
35566 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35567 override if TARGET_SYSTEM_ROOT is defined.
35568 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35570 PR tree-optimization/79090
35571 PR tree-optimization/33562
35572 PR tree-optimization/61912
35573 PR tree-optimization/77485
35574 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35575 and computed trims into the dump file.
35577 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35579 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35581 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35584 * gimplify.c (gimplify_init_constructor): If want_value and
35585 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35589 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35590 sequence. Formatting fixes.
35591 (doloop_optimize): Formatting fixes.
35594 * gcc.c (debug_level_greater_than_spec_func): New function.
35595 (static_spec_functions): Add debug-level-gt spec function.
35596 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35598 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35599 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35600 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35601 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35602 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35603 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35605 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35607 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35608 QImode fixups to general and mask registers only.
35610 2017-01-16 Carl Love <cel@us.ibm.com>
35612 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35613 for built-in functions
35614 vector signed char vec_nabs (vector signed char)
35615 vector signed short vec_nabs (vector signed short)
35616 vector signed int vec_nabs (vector signed int)
35617 vector signed long long vec_nabs (vector signed long long)
35618 vector float vec_nabs (vector float)
35619 vector double vec_nabs (vector double)
35620 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35622 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35623 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35624 * doc/extend.texi: Update the documentation file for the new built-in
35627 2017-01-16 Martin Sebor <msebor@redhat.com>
35629 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35632 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35634 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35635 UNSPEC_VSX__XXSPLTD to require special splat handling.
35637 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35640 * system.h: Poison strndup.
35642 2017-01-16 Alan Modra <amodra@gmail.com>
35645 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35648 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35650 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35652 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35654 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35655 call recog here. Assert that INSN_CODE (insn) is non-negative.
35657 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35660 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35662 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35663 in the currently scheduled RTL fragment.
35665 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35667 PR rtl-optimization/78751
35668 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35671 2017-01-14 Jeff Law <law@redhat.com>
35673 PR tree-optimization/79090
35674 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35675 variable length stores.
35676 (compute_trims): Delete dead assignment to *trim_tail.
35677 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35680 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35682 PR rtl-optimization/78626
35683 PR rtl-optimization/78727
35684 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35685 of a block, and split such blocks after everything else is finished.
35687 2017-01-14 Alan Modra <amodra@gmail.com>
35690 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35691 target legitimate_combined_insn.
35692 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35693 (rs6000_legitimate_combined_insn): New function.
35694 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35696 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
35697 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
35698 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
35700 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
35702 * doc/frontends.texi (G++ and GCC): Remove references to Java.
35704 2017-01-13 Jeff Law <law@redhat.com>
35706 PR tree-optimization/33562
35707 PR tree-optimization/61912
35708 PR tree-optimization/77485
35709 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
35711 (delete_dead_assignment): Likewise.
35712 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
35713 statement to delete_dead_call and delete_dead_assignment.
35715 2017-01-13 David Malcolm <dmalcolm@redhat.com>
35718 * substring-locations.c (format_warning_va): Strengthen case 1 so
35719 that both endpoints of the substring must be within the format
35720 range for just the substring to be printed.
35722 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
35724 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
35725 * config/i386/i386.c (ix86_target_string): Add missing options
35726 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
35727 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
35728 flags_other and ix86_target_other to flags2_other. Display unknown
35730 (ix86_valid_target_attribute_inner_p): Add missing options and
35731 reorder options by implied ISAs, as in ix86_target_string.
35733 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35735 * hash-table.h (hash_table::too_empty_p): New function.
35736 (hash_table::expand): Use it.
35737 (hash_table::traverse): Likewise.
35738 (hash_table::empty_slot): Use sizeof (value_type) instead of
35739 sizeof (PTR) to convert bytes to elements. Shrink the table
35740 if the current size is excessive for the current number of
35743 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
35745 * ira-costs.c (record_reg_classes): Break from the inner loop
35746 early once alt_fail is known to be true. Update outer loop
35747 handling accordingly.
35749 2017-01-13 Jeff Law <law@redhat.com>
35751 * tree-ssa-dse.c (decrement_count): New function.
35752 (increment_start_addr, maybe_trim_memstar_call): Likewise.
35753 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
35754 when we know the partially dead statement is a mem* function.
35756 PR tree-optimization/61912
35757 PR tree-optimization/77485
35758 * tree-ssa-dse.c: Include expr.h.
35759 (maybe_trim_constructor_store): New function.
35760 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
35762 PR tree-optimization/33562
35763 PR tree-optimization/61912
35764 PR tree-optimization/77485
35765 * doc/invoke.texi: Document new dse-max-object-size param.
35766 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
35767 * tree-ssa-dse.c: Include params.h.
35768 (dse_store_status): New enum.
35769 (initialize_ao_ref_for_dse): New, partially extracted from
35771 (valid_ao_ref_for_dse, normalize_ref): New.
35772 (setup_live_bytes_from_ref, compute_trims): Likewise.
35773 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
35774 (maybe_trim_partially_dead_store): Likewise.
35775 (maybe_trim_complex_store): Likewise.
35776 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
35777 Track what bytes live from the original store. Return tri-state
35778 for dead, partially dead or live.
35779 (dse_dom_walker): Add constructor, destructor and new private members.
35780 (delete_dead_call, delete_dead_assignment): New extracted from
35782 (dse_optimize_stmt): Make a member of dse_dom_walker.
35783 Use initialize_ao_ref_for_dse.
35785 PR tree-optimization/33562
35786 PR tree-optimization/61912
35787 PR tree-optimization/77485
35788 * sbitmap.h (bitmap_count_bits): Prototype.
35789 (bitmap_clear_range, bitmap_set_range): Likewise.
35790 * sbitmap.c (bitmap_clear_range): New function.
35791 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
35793 2017-01-13 Martin Liska <mliska@suse.cz>
35796 * function.c (set_cfun): Add new argument force.
35797 * function.h (set_cfun): Likewise.
35798 * ipa-inline-transform.c (inline_call): Use the function when
35799 strict alising from is dropped for function we inline to.
35801 2017-01-13 Richard Biener <rguenther@suse.de>
35803 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
35804 for dumping GIMPLE INTEGER_CSTs.
35806 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
35808 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
35809 to 201112L since C++17.
35811 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
35814 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
35815 if -fsanitize=kernel-address is present.
35817 2017-01-13 Richard Biener <rguenther@suse.de>
35819 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
35820 as _Literal ( type ) number in case usual suffixes do not
35821 preserve all information.
35823 2017-01-13 Richard Biener <rguenther@suse.de>
35825 PR tree-optimization/77283
35826 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
35827 and ssa-iterators.h.
35828 (is_feasible_trace): Implement a cost model based on joiner
35831 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
35834 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
35835 char or short to __float128/_Float128 directly.
35837 2017-01-12 Martin Sebor <msebor@redhat.com>
35839 to -Wformat-overflow.
35840 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
35841 (min_bytes_remaining): Same.
35842 (get_string_length): Same.
35843 (format_string): Same.
35844 (format_directive): Same.
35846 (pass_sprintf_length::handle_gimple_call): Same.
35848 2017-01-12 Jakub Jelinek <jakub@redhat.com>
35850 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
35851 info.nowrite calls with no lhs that can't throw. Return bool
35852 whether gsi_remove has been called or not.
35853 (pass_sprintf_length::handle_gimple_call): Return bool whether
35854 try_substitute_return_value called gsi_remove. Formatting fix.
35855 (pass_sprintf_length::execute): Don't use gsi_remove if
35856 handle_gimple_call returned true.
35859 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
35860 be removed due to side-effects, don't remove following barrier nor
35861 turn the successor edge into fallthru edge.
35863 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35866 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
35867 element-reversing loads and stores as not swappable.
35869 2017-01-12 Nathan Sidwell <nathan@acm.org>
35870 Nicolai Stange <nicstange@gmail.com>
35872 * combine.c (try_combine): Don't ignore result of overlap checking
35873 loop. Combine overlap & asm check into single loop.
35875 2017-01-12 Richard Biener <rguenther@suse.de>
35877 * tree-pretty-print.c (dump_generic_node): Provide -gimple
35878 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
35880 2017-01-12 Richard Biener <rguenther@suse.de>
35882 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35883 and TS_TARGET_OPTION directly derive from TS_BASE.
35884 * tree-core.h (tree_optimization_option): Derive from tree_base.
35885 (tree_target_option): Likewise.
35887 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35889 * config/i386/i386.c (memory_address_length): Increase len
35890 only when rip_relative_addr_p returns false.
35892 2017-01-11 Julia Koval <julia.koval@intel.com>
35894 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35895 (OPTION_MASK_ISA_SGX_SET): New.
35896 (ix86_handle_option): Handle OPT_msgx.
35897 * config.gcc: Added sgxintrin.h.
35898 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35899 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35900 * config/i386/i386.c (ix86_target_string): Add -msgx.
35902 (ix86_option_override_internal): Handle new options.
35903 (ix86_valid_target_attribute_inner_p): Add sgx.
35904 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35905 * config/i386/i386.opt: Add msgx.
35906 * config/i386/sgxintrin.h: New file.
35907 * config/i386/x86intrin.h: Add sgxintrin.h.
35909 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35912 * fold-const.c (maybe_nonzero_address): Return 1 for function
35914 (tree_single_nonzero_warnv_p): Don't handle function local objects
35918 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35921 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35924 * opts.c: Include "spellcheck.h"
35925 (struct string_fragment): New struct.
35926 (struct edit_distance_traits<const string_fragment &>): New
35928 (get_closest_sanitizer_option): New function.
35929 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35931 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35933 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35935 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35936 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35937 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35938 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35939 for initial die_offset if dwarf_split_debug_info.
35940 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35941 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35943 (output_skeleton_debug_sections): Formatting fix. Use
35944 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35945 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35947 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35949 * config/arm/cortex-a53.md: Add bypasses for
35950 cortex_a53_r2f_cvt.
35951 (cortex_a53_r2f): Only use for transfers.
35952 (cortex_a53_f2r): Likewise.
35953 (cortex_a53_r2f_cvt): Add reservation for conversions.
35954 (cortex_a53_f2r_cvt): Likewise.
35956 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35958 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35959 to all inlined functions, change static to extern.
35961 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35964 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35966 (arm_assemble_integer): Likewise.
35968 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35970 * config.gcc: Use new awk script to check CPU, FPU and architecture
35971 parameters for --with-... options.
35972 * config/arm/parsecpu.awk: New file
35973 * config/arm/arm-cpus.in: New file.
35974 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35976 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35978 * config/arm/t-arm: Update dependency rules.
35979 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35980 of processing .def files.
35981 * config/arm/genopt.sh: Deleted.
35982 * config/arm/gentune.sh: Deleted.
35983 * config/arm/arm-cores.def: Deleted.
35984 * config/arm/arm-arches.def: Deleted.
35985 * config/arm/arm-fpus.def: Deleted.
35986 * config/arm/arm-tune.md: Regenerated.
35987 * config/arm/arm-tables.opt: Regenerated.
35988 * config/arm/arm-cpu.h: New generated file.
35989 * config/arm/arm-cpu-data.h: New generated file.
35990 * config/arm/arm-cpu-cdata.h: New generated file.
35992 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35995 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35997 (input_varpool_node): Unpack dynamically_initialized bit.
35999 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36001 PR rtl-optimization/79032
36002 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36003 the alignment of the adjusted memory reference against that of MODE,
36004 instead of the alignment of the original memory reference.
36006 2017-01-11 Martin Jambor <mjambor@suse.cz>
36008 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36010 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36011 decorated functions.
36013 2017-01-11 Richard Biener <rguenther@suse.de>
36015 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36016 set range/nonnull info for PHI results. Do not set it on
36017 stmts marked for removal.
36019 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36021 * expr.c (store_field): In the bitfield case, fetch the return value
36022 from the registers before applying a single big-endian adjustment.
36023 Always do a final load for a BLKmode value not larger than a word.
36025 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36028 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36029 that we correctly handle column numbers greater than
36030 LINE_MAP_MAX_COLUMN_NUMBER.
36032 2017-01-10 Martin Sebor <msebor@redhat.com>
36034 PR middle-end/78245
36035 * gimple-ssa-sprintf.c (get_destination_size): Call
36036 {init,fini}object_sizes.
36037 * tree-object-size.c (addr_object_size): Adjust.
36038 (pass_through_call): Adjust.
36039 (pass_object_sizes::execute): Adjust.
36040 * tree-object-size.h (fini_object_sizes): Declare.
36042 2017-01-10 Martin Sebor <msebor@redhat.com>
36044 PR tree-optimization/78775
36045 * builtins.c (get_size_range): Move...
36046 * calls.c: ...to here.
36047 (alloc_max_size): Accept zero argument.
36048 (operand_signed_p): Remove.
36049 (maybe_warn_alloc_args_overflow): Call get_size_range.
36050 * calls.h (get_size_range): Declare.
36052 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36054 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36055 from TI's devices.csv file as of September 2016.
36056 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36058 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36060 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36061 * doc/invoke.texi: Likewise.
36062 * doc/md.texi: Likewise.
36063 * doc/objc.texi: Likewise.
36065 2017-01-10 Joshua Conner <joshconner@google.com>
36067 * config/arm/fuchsia-elf.h: New file.
36068 * config/fuchsia.h: New file.
36069 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36070 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36072 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36074 2016-01-10 Richard Biener <rguenther@suse.de>
36076 PR tree-optimization/79034
36077 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36078 Propagate out degenerate PHIs in the joiner.
36080 2017-01-10 Martin Liska <mliska@suse.cz>
36082 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36083 (sort_congruence_classes_by_decl_uid): Likewise.
36084 (sort_congruence_class_groups_by_decl_uid): Likewise.
36085 (sem_item_optimizer::merge_classes): Sort class, groups in these
36086 classes and members in the groups by DECL_UID of declarations.
36087 This would make merge operations stable.
36089 2017-01-10 Martin Liska <mliska@suse.cz>
36091 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36092 usage of m_classes_vec.
36093 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36094 (sem_item_optimizer::get_group_by_hash): Likewise.
36095 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36096 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36097 (sem_item_optimizer::verify_classes): Likewise.
36098 (sem_item_optimizer::process_cong_reduction): Likewise.
36099 (sem_item_optimizer::dump_cong_classes): Likewise.
36100 (sem_item_optimizer::merge_classes): Likewise.
36101 * ipa-icf.h (congruence_class_hash): Rename from
36102 congruence_class_group_hash. Remove declaration of m_classes_vec.
36104 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36106 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36107 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36108 * config.gcc: Add avx512vpopcntdqintrin.h.
36109 * config/i386/avx512vpopcntdqintrin.h: New.
36110 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36111 * config/i386/i386-builtin-types.def: Add new types.
36112 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36113 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36114 __builtin_ia32_vpopcountq_v8di_mask): New.
36115 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36116 __AVX512VPOPCNTDQ__.
36117 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36118 (PTA_AVX512VPOPCNTDQ): Define.
36119 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36120 TARGET_AVX512VPOPCNTDQ_P): Define.
36121 * config/i386/i386.opt: Add mavx512vpopcntdq.
36122 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36123 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36125 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36127 PR middle-end/77484
36128 * predict.def (PRED_CALL): Set to 67.
36130 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36132 * expr.c (store_field): In the bitfield case, if the value comes from
36133 a function call and is of an aggregate type returned in registers, do
36134 not modify the field mode; extract the value in all cases if the mode
36135 is BLKmode and the size is not larger than a word.
36137 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36140 * config/i386/cpuid.h: Fix undefined behavior.
36142 2017-01-04 Jeff Law <law@redhat.com>
36144 PR tree-optimization/79007
36145 PR tree-optimization/67955
36146 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36147 conservative for pt.null when flag_non_call_exceptions is on.
36149 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36151 PR translation/79019
36152 PR translation/79020
36153 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36154 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36155 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36157 * config/avr/avr.opt (maccumulate-args): Likewise.
36158 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36159 * common.opt (freport-bug): Likewise.
36160 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36161 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36162 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36163 translatable string.
36164 * config/i386/i386.c (function_value_32): Likewise.
36165 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36166 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36168 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36169 * common/config/msp430/msp430-common.c (msp430_handle_option):
36171 * symtab.c (symtab_node::verify_base): Likewise.
36172 * opts.c (set_debug_level): Likewise.
36173 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36174 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36175 missing whitespace to translatable strings.
36176 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36177 * config/sh/superh.h: Likewise.
36178 * config/i386/xopintrin.h: Likewise.
36179 * config/i386/znver1.md: Likewise.
36180 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36181 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36182 * double-int.h (struct double_int): Likewise.
36183 * double-int.c (div_and_round_double): Likewise.
36184 * wide-int.cc: Likewise.
36185 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36186 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36187 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36188 (crossjumps_occurred): ... this.
36189 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36192 PR tree-optimization/78899
36193 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36194 returning bool return struct loop *, NULL for failure and the new
36196 (versionable_outer_loop_p): Don't version outer loop if it has
36197 dont_vectorized bit set.
36198 (tree_if_conversion): When versioning outer loop, ensure
36199 tree_if_conversion is performed also on the inner loop of the
36200 non-vectorizable outer loop copy.
36201 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36202 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36203 prevent vectorization of it.
36204 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36205 the outer loop vectorization of the non-scalar version is attempted
36206 before vectorization of the inner loop in scalar version. If
36207 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36208 vectorization of its inner loop.
36209 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36210 has 2 inner loops, rename also on edges from bb whose single pred
36211 is outer_loop->header. Fix typo in function comment.
36213 2017-01-09 Martin Sebor <msebor@redhat.com>
36216 * asan.c (asan_emit_stack_protection): Increase local buffer size
36217 to avoid snprintf truncation warning.
36219 2017-01-09 Andrew Pinski <apinski@cavium.com>
36221 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36222 to reference thunderx2t99 for the tuning structure
36223 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36225 (thunderx2t99_extra_costs): This.
36226 * config/aarch64/aarch64-tune.md: Regenerate.
36227 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36228 (vulcan_addrcost_table): This.
36229 (vulcan_regmove_cost): Rename to ...
36230 (thunderx2t99_regmove_cost): This.
36231 (vulcan_vector_cost): Rename to ...
36232 (thunderx2t99_vector_cost): this.
36233 (vulcan_branch_cost): Rename to ...
36234 (thunderx2t99_branch_cost): This.
36235 (vulcan_tunings): Rename to ...
36236 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36237 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36239 2017-01-09 Martin Jambor <mjambor@suse.cz>
36243 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36244 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36245 (propagate_vr_accross_jump_function): Use the above function for all
36246 value range computations for pass-through jump functions and type
36247 converasion from explicit value range values.
36248 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36249 parameters from TYPE_ARG_TYPES.
36250 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36251 (ipa_write_node_info): Stream type of the actual argument.
36252 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36254 2017-01-09 Martin Liska <mliska@suse.cz>
36257 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36258 (lookup_compiler): Do not show error message with have_E.
36260 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36262 PR tree-optimization/78938
36263 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36264 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36265 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36266 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36269 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36271 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36274 2017-01-09 Richard Biener <rguenther@suse.de>
36276 PR tree-optimization/78997
36277 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36278 name condition properly.
36280 2017-01-09 Richard Biener <rguenther@suse.de>
36283 * dwarf2out.c (is_cxx): New overload with context.
36284 (is_naming_typedef_decl): Use it.
36286 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36288 * invoke.texi (Option Summary): Correct spacing in option lists
36289 and add line breaks to fix over-long lines.
36291 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36293 PR middle-end/17660
36295 * extend.texi (Common Variable Attributes): Add xref to GCC
36296 Internals manual to explain mode attribute keywords.
36298 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36301 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36302 and Preprocessor Options.
36303 (Options for Linking): Document -pthread here....
36304 (RS/6000 and PowerPC Options): ...not here.
36305 (Solaris 2 Options): ...or here.
36306 * doc/cppopts.texi: Document -pthread.
36308 2017-01-08 Martin Sebor <msebor@redhat.com>
36310 PR middle-end/77708
36311 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36312 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36313 New member functions.
36314 (format_directive): Used them.
36316 (pass_sprintf_length::handle_gimple_call): Same.
36317 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36318 to avoid truncation for any argument.
36319 (extract_affine_mul): Same.
36320 * tree.c (get_file_function_name): Same.
36322 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36324 PR middle-end/77484
36325 * predict.def (PRED_INDIR_CALL): Set to 86.
36327 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36329 PR preprocessor/54124
36330 * doc/cppopts.texi: Reformat -d subtable to list the full name
36331 of the options. Add cross-reference to the docs for the general
36332 compiler -d options.
36333 * doc/invoke.texi (Developer Options): Add cross-reference to the
36334 preprocessor-specific -d option documentation.
36336 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36338 PR preprocessor/13498
36339 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36340 redudant material, and reflect new command-line options.
36341 (System Headers): Likewise.
36343 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36345 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36346 -isystem, and -idirafter. Copy-edit.
36347 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36348 default for -ftrack-macro-expansion. Delete obsolete and
36349 badly-formatted implementation details about -fdebug-cpp output.
36350 * doc/cppwarnopts.texi: Copy-edit.
36352 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36355 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36356 that the transition from a max line width >= 1<<10 to narrower
36357 lines works correctly.
36359 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36361 * doc/options.texi (PerFunction): New.
36362 * opt-functions.awk (switch_flags): Map both Optimization and
36363 PerFunction to CL_OPTIMIZATION.
36364 * opth-gen.awk: Test for PerFunction flag along with
36366 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36367 it only when the latter is present. Skip those that don't in
36368 the hash function generator.
36369 * common.opt (fvar-tracking): Mark as PerFunction instead of
36371 (fvar-tracking-assignments): Likewise.
36372 (fvar-tracking-assignments-toggle): Likewise.
36373 (fvar-tracking-uninit): Likewise.
36375 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36377 PR translation/79018
36378 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36381 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36384 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36385 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36386 TARGET_LONG_JUMP_TABLE_OFFSETS.
36387 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36388 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36389 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36390 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36391 * config/m68k/m68k.md (tablejump expander): Likewise.
36392 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36393 TARGET_LONG_JUMP_TABLE_OFFSETS.
36394 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36395 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36397 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36398 David Holsgrove <david.holsgrove@xilinx.com>
36400 * common/config/microblaze/microblaze-common.c
36401 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36402 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36404 * config/microblaze/microblaze.c (microblaze_must_save_register)
36405 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36407 (microblaze_eh_return): New function.
36408 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36409 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36410 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36411 * config/microblaze/microblaze.md (eh_return): New pattern.
36413 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36415 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36416 GCC_DIAGNOSTIC_STRINGIFY): Define.
36418 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36420 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36422 * config/arm/arm.md (<mcrr>): New.
36424 * config/arm/arm.c (arm_arch5te): New.
36425 (arm_option_override): Set arm_arch5te.
36426 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36428 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36429 (arm_mcrr_qualifiers): ... this. New.
36430 (MRRC_QUALIFIERS): Define to...
36431 (arm_mrrc_qualifiers): ... this. New.
36432 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36434 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36435 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36436 (MRRCI, mrrc, MRRC): New.
36437 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36438 VUNSPEC_MRRC2): New.
36440 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36442 * config/arm/arm.md (<mcr>): New.
36444 * config/arm/arm.c (arm_coproc_builtin_available): Add
36445 support for mcr, mrc, mcr2 and mrc2.
36446 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36447 (arm_mcr_qualifiers): ... this. New.
36448 (MRC_QUALIFIERS): Define to ...
36449 (arm_mrc_qualifiers): ... this. New.
36450 (MCR_QUALIFIERS): Define to ...
36451 (arm_mcr_qualifiers): ... this. New.
36452 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36454 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36455 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36456 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36457 VUNSPEC_MRC2): New.
36459 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36461 * config/arm/arm.md (*ldc): New.
36465 * config/arm/arm.c (arm_coproc_builtin_available): Add
36466 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36467 (arm_coproc_ldc_stc_legitimate_address): New.
36468 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36469 'qualifier_const_pointer'.
36470 (LDC_QUALIFIERS): Define to...
36471 (arm_ldc_qualifiers): ... this. New.
36472 (STC_QUALIFIERS): Define to...
36473 (arm_stc_qualifiers): ... this. New.
36474 * config/arm/arm-protos.h
36475 (arm_coproc_ldc_stc_legitimate_address): New.
36476 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36477 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36478 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36479 stc2, stcl, stc2l): New.
36480 * config/arm/constraints.md (Uz): New.
36481 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36482 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36483 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36484 VUNSPEC_STC2L): New.
36486 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36488 * config/arm/arm.md (<cdp>): New.
36489 * config/arm/arm.c (neon_const_bounds): Rename this ...
36490 (arm_const_bounds): ... this.
36491 (arm_coproc_builtin_available): New.
36492 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36493 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36494 (CDP_QUALIFIERS): Define to...
36495 (arm_cdp_qualifiers): ... this. New.
36497 (arm_expand_builtin_args): Add case for 6 arguments.
36498 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36499 (arm_const_bounds): ... this.
36500 (arm_coproc_builtin_available): New.
36501 * config/arm/arm_acle.h (__arm_cdp): New.
36503 * config/arm/arm_acle_builtins.def (cdp): New.
36505 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36506 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36507 'arm_const_bounds'.
36508 * config/arm/types.md (coproc): New.
36509 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36510 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36511 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36512 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36514 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36516 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36517 (UBINOP_QUALIFIERS): New.
36519 (acle_builtin_data): New. Change comment.
36520 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36521 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36522 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36523 arm_acle_builtins.def.
36524 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36525 (arm_init_acle_builtins): New.
36526 (CRC32_BUILTIN): Remove.
36527 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36528 crc32cb, crc32ch and crc32cw.
36529 (arm_init_crc32_builtins): Remove.
36530 (arm_init_builtins): Use arm_init_acle_builtins rather
36531 than arm_init_crc32_builtins.
36532 (arm_expand_acle_builtin): New.
36533 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36534 * config/arm/arm_acle_builtins.def: New.
36536 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36538 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36539 (arm_builtin_datum): ... this.
36540 (arm_init_neon_builtin): Rename to ...
36541 (arm_init_builtin): ... this. Add a new parameters PREFIX
36542 and USE_SIG_IN_NAME.
36543 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36544 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36545 'arm_builtin_datum'.
36546 (arm_init_vfp_builtins): Likewise.
36547 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36548 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36549 (arm_expand_neon_args): Rename to ...
36550 (arm_expand_builtin_args): ... this. Rename builtin_arg
36551 enum values and differentiate between ARG_BUILTIN_MEMORY
36552 and ARG_BUILTIN_NEON_MEMORY.
36553 (arm_expand_neon_builtin_1): Rename to ...
36554 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36555 values, arm_expand_builtin_args and add bool parameter NEON.
36556 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36557 (arm_expand_vfp_builtin): Likewise.
36558 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36560 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36562 PR middle-end/77484
36563 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36564 * predict.c (tree_estimate_probability_bb): Reverse direction of
36565 polymorphic call predictor.
36567 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36569 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36570 (determine_pass_name_match): ...this new function and...
36571 (should_skip_pass_p): ...this new function.
36573 2017-01-06 Nathan Sidwell <nathan@acm.org>
36575 * ipa-visibility.c (function_and_variable_visibility): Reformat
36576 comments and long lines. Remove extrneous if.
36577 * symtab.c (symtab_node::make_decl_local): Fix code format.
36578 (symtab_node::set_section_for_node): Fix comment typo.
36580 2017-01-06 Martin Liska <mliska@suse.cz>
36583 * lra-constraints.c: Rename invariant to lra_invariant.
36584 * predict.c (set_even_probabilities): Initialize e to NULL.
36586 2017-01-05 Martin Sebor <msebor@redhat.com>
36588 PR tree-optimization/78910
36589 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36590 (format_integer): Correct off-by-one error in the handling
36591 of precision with negative numbers in signed conversions..
36593 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36595 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36597 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36599 PR tree-optimization/71016
36600 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36601 factor_out_conditional_conversion. Formatting fix.
36602 (factor_out_conditional_conversion): Add cond_stmt argument.
36603 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36604 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36607 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36609 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36610 read-rtl-function.o, and selftest-rtl.o.
36611 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36612 (selftest::aarch64_test_loading_full_dump): New function.
36613 (selftest::aarch64_run_selftests): New function.
36614 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36615 selftest::aarch64_run_selftests.
36616 * config/i386/i386.c
36617 (selftest::ix86_test_loading_dump_fragment_1): New function.
36618 (selftest::ix86_test_loading_call_insn): New function.
36619 (selftest::ix86_test_loading_full_dump): New function.
36620 (selftest::ix86_test_loading_unspec): New function.
36621 (selftest::ix86_run_selftests): Call the new functions.
36622 * emit-rtl.c (maybe_set_max_label_num): New function.
36623 * emit-rtl.h (maybe_set_max_label_num): New decl.
36624 * function.c (instantiate_decls): Guard call to
36625 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36626 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36628 * gensupport.c (gen_reader::gen_reader): Pass "false"
36629 for new "compact" param of rtx_reader.
36630 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36631 rather than an empty string for NULL strings.
36632 * read-md.c: Potentially include config.h rather than bconfig.h.
36633 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36634 (have_error): New global, copied from errors.c.
36635 (md_reader::read_name): Rename to...
36636 (md_reader::read_name_1): ...this, adding "out_loc" param,
36637 and converting "missing name or number" to returning false, rather
36639 (md_reader::read_name): Reimplement in terms of read_name_1.
36640 (md_reader::read_name_or_nil): New function.
36641 (md_reader::read_string): Handle "(nil)" by returning NULL.
36642 (md_reader::md_reader): Add new param "compact".
36643 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36644 (md_reader::read_file): New method.
36645 * read-md.h (md_reader::md_reader): Add new param "compact".
36646 (md_reader::read_file): New method.
36647 (md_reader::is_compact): New accessor.
36648 (md_reader::read_name): Convert return type from void to file_location.
36649 (md_reader::read_name_or_nil): New decl.
36650 (md_reader::read_name_1): New decl.
36651 (md_reader::m_compact): New field.
36652 (noop_reader::noop_reader): Pass "false" for new "compact" param
36654 (rtx_reader::rtx_reader): Add new "compact" param.
36655 (rtx_reader::read_rtx_operand): Make virtual and convert return
36656 type from void to rtx.
36657 (rtx_reader::read_until): New decl.
36658 (rtx_reader::handle_any_trailing_information): New virtual function.
36659 (rtx_reader::postprocess): New virtual function.
36660 (rtx_reader::finalize_string): New virtual function.
36661 (rtx_reader::m_in_call_function_usage): New field.
36662 (rtx_reader::m_reuse_rtx_by_id): New field.
36663 * read-rtl-function.c: New file.
36664 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36665 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36666 (selftest::verify_three_block_rtl_cfg): New decl.
36667 * read-rtl-function.h: New file.
36668 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36669 For host, include function.h, memmodel.h, and emit-rtl.h.
36670 (one_time_initialization): New function.
36671 (struct compact_insn_name): New struct.
36672 (compact_insn_names): New array.
36673 (find_code): Handle insn codes in compact dumps.
36674 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36675 (bind_subst_iter_and_attr): Likewise.
36676 (add_condition_to_string): Likewise.
36677 (add_condition_to_rtx): Likewise.
36678 (apply_attribute_uses): Likewise.
36679 (add_current_iterators): Likewise.
36680 (apply_iterators): Likewise.
36681 (initialize_iterators): Guard usage of apply_subst_iterator with
36682 #ifdef GENERATOR_FILE.
36683 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36684 (md_reader::read_mapping): Likewise.
36685 (add_define_attr_for_define_subst): Likewise.
36686 (add_define_subst_attr): Likewise.
36687 (read_subst_mapping): Likewise.
36688 (check_code_iterator): Likewise.
36689 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36691 (one_time_initialization): New function.
36692 (rtx_reader::read_until): New method.
36693 (read_flags): New function.
36694 (parse_reg_note_name): New function.
36695 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36696 Handle reuse_rtx ids.
36697 Wrap iterator lookup within #ifdef GENERATOR_FILE.
36698 Add parsing support for RTL dumps, mirroring the special-cases in
36699 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
36700 values, and calling handle_any_trailing_information.
36701 (rtx_reader::read_rtx_operand): Convert return type from void
36702 to rtx, returning return_rtx. Handle case 'e'. Call
36703 finalize_string on XSTR and XTMPL fields.
36704 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
36705 "(nil)" values were omitted. Call the postprocess vfunc on the
36707 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
36708 class ctor. Initialize m_in_call_function_usage. Call
36709 one_time_initialization.
36710 * rtl-tests.c (selftest::test_uncond_jump): Call
36711 set_new_first_and_last_insn.
36712 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
36713 * selftest-rtl.c: New file.
36714 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
36715 (selftest::get_insn_by_uid): New decl.
36716 * selftest-run-tests.c (selftest::run_tests): Call
36717 read_rtl_function_c_tests.
36718 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
36719 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
36722 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
36724 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
36725 operands in a special way. Assert that pos+len <= mode precision.
36727 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36729 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
36730 3 argument Alias with unlimited for the negative form.
36731 (fno-vect-cost-model): Removed.
36733 2017-01-05 Martin Liska <mliska@suse.cz>
36735 * hsa-gen.c (gen_hsa_divmod): New function.
36736 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
36738 2017-01-05 Martin Liska <mliska@suse.cz>
36741 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
36744 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36746 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
36747 small constant length operands.
36749 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36751 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
36752 between loop iterations.
36754 2017-01-05 Martin Liska <mliska@suse.cz>
36757 * gimplify.c (gimplify_decl_expr): Compare to
36758 asan_poisoned_variables instread of checking flags.
36759 (gimplify_target_expr): Likewise.
36760 (gimplify_expr): Likewise.
36761 (gimplify_function_tree): Conditionally initialize
36762 asan_poisoned_variables.
36764 2017-01-04 Jeff Law <law@redhat.com>
36766 PR tree-optimizatin/78812
36767 * rtl.h (contains_mem_rtx_p): Prototype.
36768 * ifcvt.c (containts_mem_rtx_p): Move from here to...
36769 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
36770 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
36771 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
36772 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
36774 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36776 * input.c (assert_char_at_range): Default-initialize actual_range.
36778 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36780 * df-scan.c (df_ref_create_structure): Make regno unsigned,
36781 to match the caller.
36783 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36785 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
36786 insns after final jump in test to emit dummy move.
36788 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36790 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
36791 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
36793 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
36795 * multiple_target.c (create_dispatcher_calls): Init e_next.
36796 * tree-ssa-loop-split.c (split_loop): Init border.
36797 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
36800 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
36805 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
36806 (altivec_register_operand): Do not return true if the operand
36807 contains a SUBREG mixing SImode and SFmode.
36808 (vsx_register_operand): Likewise.
36809 (vsx_reg_sfsubreg_ok): New predicate.
36810 (vfloat_operand): Do not return true if the operand contains a
36811 SUBREG mixing SImode and SFmode.
36812 (vint_operand): Likewise.
36813 (vlogical_operand): Likewise.
36814 (gpc_reg_operand): Likewise.
36815 (int_reg_operand): Likewise.
36816 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
36817 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
36818 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
36820 (rs6000_emit_move_si_sf_subreg): New helper function.
36821 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
36822 fixup SUBREGs involving SImode and SFmode.
36823 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
36824 numbers for the new peephole2 optimization.
36825 (peephole2 for SFmode unions): New peephole2 to optimize cases in
36826 the GLIBC math library that do AND/IOR/XOR operations on single
36827 precision floating point.
36828 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
36829 target macros to say whether we need to avoid SUBREGs mixing
36831 (TARGET_ALLOW_SF_SUBREG): Likewise.
36832 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
36833 (UNSPEC_SI_FROM_SF): Likewise.
36834 (iorxor): Change spacing.
36835 (and_ior_xor): New iterator for AND, IOR, and XOR.
36836 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
36837 (movdi_from_sf_zero_ext): Likewise.
36838 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
36839 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
36840 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
36841 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
36842 (fms<mode>4): Likewise.
36843 (fnma<mode>4): Likewise.
36844 (fnms<mode>4): Likewise.
36845 (nfma<mode>4): Likewise.
36846 (nfms<mode>4): Likewise.
36848 2017-01-04 Marek Polacek <polacek@redhat.com>
36851 * doc/invoke.texi: Document -Wpointer-compare.
36853 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36855 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
36858 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
36859 descriptions for -gdwarf-5 and emit them as uleb128 instead of
36862 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
36865 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
36866 documentation of the powerpc_popcntb_ok attribute.
36867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36868 code to issue warning messages if a requested CPU configuration is
36869 not supported by the binary (assembler and loader) toolchain.
36870 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
36871 made to define a built-in function that has been disabled.
36872 (paired_init_builtins): Add assertion to prevent ICE if attempt is
36873 made to define a built-in function that has been disabled.
36874 (altivec_init_builtins): Add comment explaining why definition
36875 of the DST built-in functions is not preceded by an assertion
36876 check. Add assertions to prevent ICE if attempts are made to
36877 define an altivec predicate or an abs* built-in function that has
36879 (htm_init_builtins): Add comment explaining why definition of the
36880 htm built-in functions is not preceded by an assertion check.
36882 2017-01-04 Jeff Law <law@redhat.com>
36884 PR tree-optimizatin/67955
36885 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36886 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36887 the points-to solution does not include pt_null. Use DECL_PT_UID
36890 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36892 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36893 Use gen_int_mode instead of gen_lopwart for const_int operands.
36895 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36897 PR tree-optimization/71563
36898 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36899 out of range value - has low bits known to be zero.
36901 2017-01-04 Alan Modra <amodra@gmail.com>
36903 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36904 * configure: Regenerate.
36905 * config.in: Regenerate.
36907 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36910 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36911 a substring of the message, but strcmp with the whole message. Ifdef
36912 ENABLE_NLS, translate the message first using dgettext.
36914 2017-01-03 Jeff Law <law@redhat.com>
36916 PR tree-optimizatin/78856
36917 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36918 (mark_threaded_blocks): Remove code to truncate thread paths that
36919 cross multiple loop headers. Instead invalidate the cached loop
36920 iteration information and handle case of a thread path walking
36921 into an irreducible region.
36923 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36926 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36927 assertions. Add support for doing the signbit if the IEEE 128-bit
36928 floating point value is in a GPR.
36929 * config/rs6000/rs6000.md (Fsignbit): Delete.
36930 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36931 Update the length attribute if the value is in a GPR.
36932 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36933 the sign or zero extension instruction, since the value is always 0/1.
36934 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36937 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36938 extracting SImode to a GPR register so that we can generate a
36939 store, limit the vector to be in a traditional Altivec register
36940 for the vextuwrx instruction.
36942 2017-01-03 Ian Lance Taylor <iant@google.com>
36944 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36946 2017-01-03 Martin Sebor <msebor@redhat.com>
36948 PR tree-optimization/78696
36949 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36950 precision. Use MPFR for %f for greater fidelity. Correct handling
36952 (pass_sprintf_length::compute_format_length): Set width and precision
36953 specified by asrerisk to void_node for vararg functions.
36954 (try_substitute_return_value): Adjust dump output.
36956 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36958 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36960 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36962 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36963 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36964 -mlra/-mno-lra was passed to the compiler.
36966 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36968 PR rtl-optimization/65618
36969 * emit-rtl.c (try_split): Move initialization of "before" and
36970 "after" to just before the call to emit_insn_after_setloc.
36972 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36974 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36976 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36978 * dwarf2out.c (gen_enumeration_type_die): When
36979 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36981 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36983 PR tree-optimization/78965
36984 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36985 Change first argument from const call_info & to call_info &. For %n
36986 set info.nowrite to false.
36988 PR middle-end/78901
36989 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36990 possibly throwing calls.
36992 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36993 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36994 and fns handling, rather than in a separate case SSA_NAME.
36996 2017-01-02 Jeff Law <law@redhat.com>
36998 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36999 fixes for first_period and second_period variables.
37001 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37004 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37005 (*insvqi_1): New insn pattern.
37006 (*insvqi_1_mem_rex64): Ditto.
37007 (*insvqi_2): Ditto.
37008 (*insvqi_3): Rename from *insvqi.
37010 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37012 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37014 * doc/cfg.texi (Edges): Remove reference to Java.
37015 (Maintaining the CFG): Ditto.
37017 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37019 PR middle-end/77674
37020 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37021 transparent aliases.
37023 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37025 PR middle-end/77484
37026 * predict.def (PRED_CALL): Update hitrate.
37027 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37028 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37029 into direct/indirect/polymorphic variants.
37031 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37033 Update copyright years.
37035 * gcc.c (process_command): Update copyright notice dates.
37036 * gcov-dump.c (print_version): Ditto.
37037 * gcov.c (print_version): Ditto.
37038 * gcov-tool.c (print_version): Ditto.
37039 * gengtype.c (create_file): Ditto.
37040 * doc/cpp.texi: Bump @copying's copyright year.
37041 * doc/cppinternals.texi: Ditto.
37042 * doc/gcc.texi: Ditto.
37043 * doc/gccint.texi: Ditto.
37044 * doc/gcov.texi: Ditto.
37045 * doc/install.texi: Ditto.
37046 * doc/invoke.texi: Ditto.
37048 Copyright (C) 2017 Free Software Foundation, Inc.
37050 Copying and distribution of this file, with or without modification,
37051 are permitted in any medium without royalty provided the copyright
37052 notice and this notice are preserved.